Neuron, Vol. 12, 675-690, March, 1994, Copyright y 1994 by Cell Press
TAG-1 Can Mediate Homophilic Binding,
but Neurite Outgrowth on TAG-1 Requires
an L l-like Molecule and 131 Integrins
Dan P. Felsenfeld, Mary A. Hynes,* Karen M. Skoler,
Andrew J. Furley,f and Thomas M. Jessell
Howard Hughes Medical Institute
and Department of Biochemistry
and Molecular Biophysics
Center for Neurobiology and Behavior
New York, New York 10032
Subsets of axons in the embryonic nervous system tran-
siently express the glycoprotein TAG-l, a member of
the subfamily of immunoglobulin (Ig)-Iike proteins that
contain both C2 class Ig and fibronectin type III domains.
TAG-1 is attached to the cell surface by a glycosyl-
phosphatidylinositol linkage and is secreted by neurons.
In vitro studies have shown that substrate-bound TAG-1
promotes neurite outgrowth. We have examined the na-
ture of axonal receptors that mediate the neurite-
outgrowth promoting properties of TAG-1. Although
TAG-1 can mediate homophilic binding, neurite out-
growth on a substrate of TAG-1 does not depend on
the presence of TAG-1 on the axonal surface. Instead,
neurite outgrowth on TAG-1 is inhibited by polyclonal
antibodies directed against L1 and, independently, by
polyclonal and monoclonal antibodies against 131-
containing integrins. These results provide evidence that
TAG-1 can interact with cell surfaces in both a homophi-
lic and heterophilic manner and suggest that neurite
extension on TAG-1 requires the function of both inte-
grins and an Ll-like molecule.
The guidance of axons to their targets in the embry-
onic nervous system is controlled by molecules lo-
cated in the environment of the growing axon (Dodd
and Jessell, 1988; Hynes and Lander, 1992; Goodman
and Shatz, 1993). The recognition of guidance mole-
cules by neurons is thought to be mediated by recep-
tors on the axonal growth cone that, when activated,
lead to changes in the cytoskeleton of the growth
cone and influence the extension and orientation of
axons (Sabry et al., 1991; Payne et al., 1992; Fan et
al., 1993; Lin and Forscher, 1993). In vitro assays that
monitor cell binding or neurite extension have impli-
cated several major classes of surface molecules in
the growth and guidance of axons in vertebrates. Im-
*Present address: Genentech, Inc., 460 Point San Bruno Boule-
vard, South San Francisco, California 94080.
~rPresent address: Laboratory of Developmental Neurobiology,
National Institute of Medical Research, The Ridgeway, Mill Hill,
munoglobulin (Ig)-Iike proteins and members of the
cadherin family can serve both as receptors on the
surface of the growth cone and as substrates for grow-
ing axons (Lagenaur and Lemmon, 1987; Doherty et
al., 1989; Rathjen and Jessell, 1991; Bixby and Zhang,
1990; Takeichi, 1991). In addition, the adhesive and
neurite outgrowth-promoting properties of extracel-
lular matrix glycoproteins such as laminin and fibro-
nectin involve interactions with integrins on the neu-
ronal surface (Hynes, 1992). The restricted expression
of some of these proteins suggests that they function
in the guidance of growth cones as well as in axonal
extension (Dodd et al., 1988; Tanaka et al., 1991; Burns
et al., 1991; Shiga and Oppenheim, 1991).
Insight into the cellular mechanisms of axonal
growth and guidance has emerged from studies of the
developing spinal cord (Landmesser, 1988; Yaginuma
and Oppenheim, 1991; Jessell and Dodd, 1992). Dur-
ing the initial stages of spinal cord development, mo-
tor, sensory relay, and dorsal root ganglion (DRG) neu-
rons differentiate and begin to extend axons to
distinct targets. Each of these neuronal classes tran-
siently express TAG-l, a neuron-specific Ig-like pro-
tein on their axons and growth cones (Yamamoto et
al., 1986; Dodd et al., 1988; Furley et al., 1990; Kara-
gogeos et al., 1991). TAG-1 has been cloned, and its
cDNA predicts a protein that comprises six Ig domains
of the C2 class and four fibronectin type III repeats
(Fu rley et al., 1990). The axonin-1 glycoprotein appears
to be the chick homolog of TAG-1 (Zuellig et al., 1992),
and a human TAG-1/axonin-1 homolog has also been
isolated (Hasler et al., 1993; Tsiotra et al., 1993). The
juxtaposition of Ig-C2 and fibronectin type III domains
places TAG-1 within a subfamily of Ig-like proteins
implicated in neuronal adhesion (Rathjen and Jessell,
1991). Other members of this subfamily include N-CAM
(Cunningham et al., 1987), contactin/F111F3 (Ranscht,
1988; Gennarini et al., 1989; Briimmendorf et al., 1989),
L1 (Moos et al., 1988), Ng-CAM/G4 (Chang et al., 1987;
Burgoon et al., 1991), Nr-CAM (Grumet et al., 1991),
and neurofascin (Volkmer et al., 1992). The second
fibronectin type III repeat of rat and human TAG-l,
but not of chick axonin-1, contains an arginine-
glycine-aspartate (RGD) sequence (Furley et al., 1990;
Zuellig et al., 1992; Tsiotra et al., 1993), in a context
similar to that of the RGD in the cell-binding domain
in fibronectin (Furley et al., 1990; Ruoslahti and
Pierschbacher, 1987; Hynes, 1992).
TAG-1 and axonin-1 exist in two protein isoforms:
a surface form in which cell attachment occurs exclu-
sively by a glycosyl-phosphatidylinositol (GPI) linkage
and a secreted form that appears to derive in part
from a separate precursor protein which lacks the GPI
moeity (Ruegg et al., 1989a, 1989b; Furley et al., 1990;
Karagogeos et al., 1991). Although TAG-1 and axonin-1
are expressed on the cell surface, the proteins are
released from DRG and spinal cord neurons in vitro
1 2 3 4 5 6
X 80 LN
[ ~ "\
~, -- -- Brain TAG-1
.... 293 TAG-1 .~ ~01 ~
40- ",:',,,, "k,
.'< Z 10- :~'-.~.., ~,.~ ~, , . . ~ , ~ " ~ _ ~
0 , , "
, --~ ...... ~'--'-- T - -, -
- ,- TM ,
8 8 o ~ o 8 8
Neurite length in microns
Figure 1. Brain-Derived and Recombinant TAG-1 Promote Neu-
rite Outgrowth from DR(:; Neurons in Vitro
(A) Purification of TAG-1 from different sources. SDS-
polyacrylamide gel electrophoresis of TAG-1 immunoaffinity pu-
rified from E15-E16 rat brains (Brn) after treatment of membranes
with PI-PLC, or separation of the soluble form (Sol), and after
isolation of TAG-1 from transfected human embryonic kidney
293 cells. In each case, a single -135 kd protein is detectable
by silver staining (lanes 1, 3, and 5) and Western blotting (lanes
2, 4, 6) with rabbit anti-TAG-1 antibodies. Molecular size values
are indicated at left: 210 kd, 120 kd, 98 kd, 64 kd, 36 kd, 20 kd,
and 14.3 kd.
(B-E) E14 DRG neurite outgrowth on different nitrocellulose-
immobilized substrates. Neurons were labeled with MAb 3A10,
(Ruegg et al., 1989a; Karagogeos et al., 1991) and in
vivo (Furley et al., 1990). These observations raise the
possibility that TAG-1 and axonin-1 function both on
the surface of developing axons and in their local envi-
ronment. In vitro studies of spinal and sensory neu-
rons have shown that TAG-1 and axonin-1 can pro-
mote neurite outgrowth when presented as an im-
mobilized substrate (Furley et al., 1990; Stoeckli et al.,
1991; Kuhn et al., 1991), suggesting an involvement in
the growth or guidance of developing axons. Sub-
strate-bound TAG-1 may promote neurite outgrowth
through homophilic binding, as has been shown for
other related Ig-like proteins, such as L1 (Lemmon et
al., 1989) and N-CAM (Hoffrnan and Edelman, 1983;
Doherty et al., 1990). Alternatively, TAG-1 may promote
neurite outgrowth through heterophilic interactions
with distinct neuronal receptors. Studies of axonin-1
function in vitro have shown that DRG neurite out-
growth on axonin-1 is inhibited by antibodies against
the Ig-like protein Ng-CAM/G4 (Kuhn et al., 1991).
To define the functions of TAG-1 in more detail, we
have examined the nature of interactions mediated
by TAG-1 when bound as a substrate and when ex-
pressed on the surface of cells. The ability of sub-
strate-bou nd TAG-1 to promote the outgrowth of n eu-
rites from DRG neurons in vitro does not depend on
the presence of TAG-1 on the neuronal surface and
must therefore involve heterophilic interactions. Neu-
rite outgrowth on TAG-1 is blocked independently
by antibodies directed against L1 and the 131 integrin
subunit. Nevertheless, TAG-1 can mediate homophi-
lic binding and promote cell aggregation. These re-
sults provide evidence that TAG-1 can mediate both
homophilic and heterophilic interactions and suggest
that TAG-1 promotes neurite outgrowth by interac-
tions with an Ll-like molecule and with integrins.
TAG-1 Promotes DRG Neurite Outgrowth
To identify axonal receptors responsible for neurite
outgrowth on TAG-l, we have used an assaythat moni-
tors neurite extension from embryonic day 14-15
(E14-E15) DRG neurons grown in vitro on substrates
of TAG-1 (Lagenaur and Lemmon, 1987; Furley et al.,
1990). Affi n ity-pu rifled TAG-1 derived from em bryon ic
rat brain and recombinant TAG-1 derived from human
kidney 293 cells or insect cells (Figure 1A) were equally
effective as substrates in promoting DRG neurite out-
growth (Figures 1B, 1C, and 1F and data not shown). In
directed against a filament-associated protein, to visualize neu-
rites. (B) Neurite outgrowth on brain-derived TAG-1. (C) Neurite
outgrowth on recombinant 293 cell-derived TAG-1. (D) Neurite
outgrowth on BSA. (E) Neurite outgrowth on laminin. (F) Quanti-
ration of DRC neurite outgrowth on different protein substrates.
The ordinate shows the percentage of neurites with lengths
greater than the value shown on the abscissa. This plot is repre-
sentative of results from 5 separate experiments.
Bar, 30 I~m (B-E).
TAG-1 Interactions in Neurite Outgrowth
o ~;i )0
Figure 2. TAG-1 Expression on the Surface of Drosophila $2 Cells Induces Cell Aggregation
$2 cells were transfected with a TAG-1 cDNA, and protein expression was induced and assayed by indirect immunofluorescence and
(A and B) Indirect immunofluorescence and phase-contrast micrographs showing TAG-1 expression on the surface of $2 cells detected
with a rabbit anti-TAG-1 antibody. TAG-1 is expressed on the surface of -70% of cells. Cells that express TAG-1 have formed large
aggregates, whereas cells that do not express TAG-1 remain unaggregated.
(C and D) Indirect immunofluorescence and phase-contrast micrographs showing $2 cells transfected with TAG-1 cDNA after treatment
with PI-PLC. TAG-1 is no longer detected on the cell surface, and cells are not aggregated. (E) Confocal image of aggregated $2 cells
showing a concentration of TAG-1 immunofluorescence at regions of cell-cell contact.
Bars, 50 I~m (A-D); 2 pm (E).
control experiments, DRG neurons adhered to bovine
serum albumin (BSA) to an extent similar to that on
TAG-l, but little or no neurite extension was observed
(Figures 1D and 1F), whereas long neurites were ob-
served on laminin (Figures 1E and 1F). These results,
together with additional specificity controls (see Ex-
perimental Procedures), show that substrate-bound
TAG-1 promotes neurite outgrowth.
DRG neurons that extend neurites on TAG-1 ex-
pressed high levels of TAG-1 on their cell bodies, neu-
rites, and growth cones (see Figures 6A-6C). Thus,
TAG-1 expressed on axonal surfaces could function
as a neuronal receptor that mediates the neurite out-
growth-promoting properties of substrate TAG-1
through a homophilic interaction. Alternatively, TAG-1
could promote DRG neurite outgrowth by interacting
with distinct neuronal receptors.
TAG-1 Can Mediate Homophilic Binding
To determine whether TAG-1 can bind homophilically,
we first expressed TAG-1 in the nonadherent Dro-
sophila Schneider type 2 ($2) cell line in which cell
aggregation provides an indicator of binding interac-
tions (Elkins et al., 1990). About 50%-70% of cells trans-
fected with TAG-1 cDNA expressed surface TAG-l, as
detected by indirect immunofluoresence labeling
(Figures 2A, 2B, and 2E). TAG-1 § $2 cells formed large
aggregates of up to -400 cells. Transfected $2 cells
that did not express detectable TAG-1 were not incor-
porated into aggregates (Figures 2A and 2B). In aggre-
gates of TAG-1 § $2 cells, TAG-1 was concentrated at
junctions between cells (Figure 2E), consistent with a
role for TAG-1 in mediating cell adhesion. Phosphati-
dylinositol-specific phospholipase C (PI-PLC) treat-
ment of TAG-1 + $2 cells removed TAG-1 from the cell
surface and completely blocked cell aggregation (Fig-
ures 2C and 2D), whereas the aggregation of cells
transfected with a cDNA encoding a transmembrane
isoform of N-CAM was not affected (data not shown).
These results demonstrate that TAG-1 is expressed on
the surface of $2 cells in a PI-PLC-sensitive, GPI-
anchored form and that surface expression of TAG-1
promotes $2 cell aggregation.
To examine whether the aggregation of TAG-? $2
cells resulted from a homophilic interaction between
TAG-1 molecules, TAG-1 § $2 cells were mixed with
parental $2 cells or with cells transfected with a con-
trol vector. One of the two cell populations was prela-
beled with a nondiffusible fluorescent dye, carboxy-
fluorescein diacetate succinimidyl ester (CFSE), to
distinguish cell types. $2 cell aggregates were com-
posed almost exclusively of TAG-1 § cells, with a low
number of parental or control-transfected $2 cells
(Figures 3A and 3B; Figure 4A). The selective aggrega-
tion of TAG-1 § $2 cells provides strong evidence that
TAG-1 expressed on the surface of $2 cells binds ho-
mophilically to promote aggregation.
We next determined whether surface TAG-1 can
0 ,O ~
TAG-1 Interactions in Neurite Outgrowth
A 1 0 0 -
' Homotypic ' Heterotypic '
I r--] NCAM
T A G - I ]
Homotypic ' Heterotypic
Free Homotypic Heterotypic
Figure 4. Quantitation of $2 Cell Aggregation Assays
Bar graphs derived from an analysis of the contacts made by
untransfected $2 cells and by $2 cells expressing different Ig-like
glycoproteins. Clusters of 3 cells or fewer are defined as free.
A homotypic contact is defined as contact between an individual
cell and a cell with the same surface phenotype. A heterotypic
contact is defined as a contact between an individual cell and a
cell of a different surface phenotype. Under conditions in which
large cell aggregates formed, it was not possible to count all
cell contacts. These bar graphs therefore underrepresent the
number of cells in aggregrates under homotypic aggregation
conditions. Bar graphs are expressed as the percentage of cells
recovered at the end of the assay, which does not take into
account the loss of single cells during centifugation and washes.
Error bars indicate mean + SEM for 2-7 different experiments.
bind to the related neural Ig-like proteins N-CAM and
L1. $2 cells transfected with L1 or N-CAM cDNAs ex-
pressed high levels of protein on their surface and
formed aggregates that almost completely excluded
nonexpressing or parental $2 cells (Figures 3E-3H).
Thus, both L1 and N-CAM can promote cell aggrega-
tion through homophilic binding. TAG-1 + $2 cells
were then mixed with an equal number of L1 + $2 or
N-CAM + $2 cells. The two cell populations were identi-
fied by surface antigen expression or by labeling one
cell population with CFSE. Of cell aggregates that
formed in mixtures of TAG-1 + and N-CAM + $2 cells
or TAG-1 § and L1 + $2 cells, over 85% of aggregated
cells were bound to cells expressing the same Ig-like
protein (Figures 3C and 3D; Figures 4B and 4C). More-
over, the percentage of L1 + $2 cells or N-CAM + $2 cells
that were incorporated into TAG-1 + $2 aggregates was
not greater than that observed with parental $2 cells
(Figure 4), even though large TAG-1 +, L1 +, and N-CAM +
$2 cell aggregates were formed. These results indicate
that under conditions in which the homophilic bind-
ing properties of TAG-l, L1, and N-CAM can readily
be demonstrated, heterophilic interaction between
TAG-1 and two related Ig-like proteins, L1 and N-CAM,
were not detected.
To determine whether homophilic binding activity
is restricted to TAG-1 expressed on cell surfaces, we
assessed the ability of nitrocellulose-bound TAG-1 to
bind homophilically using TAG-l-coated fluorescent
microspheres as probes. TAG-1 microspheres bound
selectively to a substrate of nitrocellulose-immobi-
lized TAG-1 (Figure 5) and did not bind to control
substrates including BSA (Figure 5C). Similarly, micro-
spheres coated with BSA showed no preference for
the TAG-1 su bstrate (Figu re 5D). The binding of TAG-I-
coated microspheres to a TAG-1 substrate was mark-
edly inhibited by addition of solubleTAG-1 (250 p.g/ml)
(Figure 5B), supporting the idea that the homophilic
binding of substrate-bound TAG-1 is not a conse-
quence of the exposure of adhesive sites not present
in the native molecule. These results show that TAG-1
retains its homophilic binding properties when bound
to a substrate and leave open the possibility that a
(A) TAG-1 + $2 cells mixed with parental S2 cells; (B) TAG-1 § S2
cells mixed with N-CAM + $2 cells; (C) TAG-1 + $2 cells mixed with
L1 + $2 cells.
Figure 3. Selective Aggregation of TAG-1 S2 Cells
(A and B) Fluorescence and phase-contrast micrographs of a mixture of TAG-1 $2 cells and untransfected parental $2 cells. The parental
$2 cells have been prelabeled with the fluorescent dye CFSE. Fluorescently labeled cells are largely excluded from aggregates, which
are composed of unlabeled TAG-1 + $2 cells.
(C and D) Selective aggregation of TAG-1 + $2 cells and N-CAM + $2 cells. N-CAM + $2 cells were prelabeled with CFSE and mixed in
equal proportion with TAG-1 + $2 cells. Both cell types form large aggregates that are composed of a single cell type.
(E and F) Fluorescence and phase-contrast micrographs showing N-CAM expression on transfected $2 cells, detected with a rabbit
anti-N-CAM antibody. Cells that do not express N-CAM are not incorporated into aggregates.
(G and H) Fluorescence and phase-contrast micrographs showing L1 expression on transfected $2 cells, detected with a rabbit anti-L1
antiserum. Cells that do not express L1 are not incorporated into aggregates.
Bar, 50 p.m (A-H).
s00 I ............ o .................. ..............................................................................
m ................ ~ .....
B C D
Figure 5. Homophilic Binding of Nitrocellulose-Bound TAG-1
Representative fields of protein-bound covaspheres attached to
different :nitrocellulose-immobilized substrates.
(A) TAG-l-coated covaspheres bound to a substrate of nitrocellu-
(B) Inhibition of the binding of TAG-l-coated covaspheres to
nitrocellulose-bound TAG-1 by soluble TAG-1 (250 p.g/ml).
(C) Binding of TAG-l-coated covaspheres to nitrocellulose-
(D) Binding of BSA-coated covaspheres to nitrocellulose-bound
Quantitation, taken from 6 nonoverlapping fields in 2 different
experiments, is shown in lower panel. Plots show mean + SE.
Unit area = 25,000 p.m 2. Bar, 10 p.m.
homophilic interaction may be involved in neurite
extension on nitrocellulose-bound TAG-1.
Cell Surface TAG-1 Is Not Required for Neurite
Outgrowth on a Substrate of TAG-1
To examine whether substrate-bound TAG-1 pro-
motes neurite outgrowth by interacting with the neu-
ronal cell surface form of TAG-l, we treated DRG neu-
rons with PI-PLC, an enzyme that cleaves GPI anchors
(low and Saltiel, 1988) to eliminate TAG-I (and other
GPl-linked proteins) from the neuronal surface. TAG-I
was removed from the cell surface after 30 min of
PI-P1-C treatment (data not shown), and no surface
TAG-I could be detected by immunofluorescence at
any subsequent time (Figures 6D and 6E). These re-
sults, together with similar biochemical findings (Kar-
agogeos et al., 1991), provide evidence that PI-PIC
removes essentially all TAG-1 from the neuronal cell
surface. In control experiments, N-CAM expression
on DRG neurons, detected by monoclonal antibody
(MAb) 5A5, was not affected by PI-PLC treatment (data
Treatment of DRG neurons with PI-PLC had no de-
tectable effect on the initial adhesion of DRG neurons
(data not shown) and did not alter the median length
of DRG neu rites on substrate-bound TAG-1 oron lami-
nin, when compared with neurons treated with con-
trol buffer (Figures 6F and 6G). Moreover, PI-PLC treat-
ment did not promote the extension of neurites from
DRG neurons that had adhered to a BSA substrate
(data not shown), showing that removal of GPI-
anchored proteins does not unmask a latent sub-
strafe-independent neurite growth potential. These
results indicate that the removal of TAG-I does not
affect the ability of DRG neurites to extend on sub-
strates of either TAG-I or laminin. Some EI4-E15 DRG
neurons that did not express detectable TAG-I, even
in the absence of PI-P1_C treatment (see also Kara-
gogeos et al., 1991), extended neurites on a TAG-1
substrate (data not shown). This result provides fur-
ther evidence that surface TAG-1 expression does not
correlate with the ability to extend neu rites on TAG-1.
Neurite outgrowth on TAG-1 therefore does not re-
quire cell surface TAG-I, suggesting that the neurite
outgrowth-promoting properties of TAG-I depend
on heterophilic interactions.
Anti-L1 Antibodies Inhibit Neurite Outgrowth on
Studies on axonin-1, the chick homolog of TAG-l,
have shown that neurite outgrowth on axonin-1 can
be blocked by antibodies directed
G4 (Kuhn et al., 1991), a molecule that is structurally
related to mouse and rat L1 (Lemmon and Mcl.oon,
1986). The LI glycoprotein is expressed at low levels
on the surface of E14-E15 DRG neurons grown in vitro
for 14-15 hr on TAG-1 or laminin (Karagogeos et al.,
1991, and data not shown). To determine the possible
involvement of L1 in neurite outgrowth on substrate-
bound TAG-I, DRG neurons were grown on a sub-
strate of TAG-I in the presence of polyclonal antibod-
ies directed against LI. DRG neurons still adhered to
TAG-l, but the median length of DRG neurites was
reduced by more than 90% in the presence of anti-L1
antibodies (Figure 7A). Equal concentrations of anti-L1
antibodies did not inhibit the outgrowth of DRG neu-
rites on laminin (Figure 7B), but completely blocked
neurite extension on chick Ng-CAM (Figure 9A).
N-CAM was also expressed on the surface of DRG
neurons grown on TAG-I or laminin (Karagogeos et
al., 1991, and data not shown). Nevertheless, addition
of function-blocking rabbit anti-N-CAM antibodies, at
concentrations at which anti-L1 antibodies were effec-
tive, did not significantly reduce neurite length on
either TAG-1 or laminin (data not shown). These re-
sults suggest that I_1 or an immunologically related
molecule is involved in mediating the outgrowth of
neurites of rat DRG neurons on a substrate of TAG-I.
TAG-1 Interactions in Neurite Outgrowth
~.~ ~ _ TAG-1 + PLC
>1 bo >2'00 >3'o0 >~o ~'oo
Neurite length in microns
- ~ ' ~ ' ~
~'t~ ~ - LN § PLC
>100 >200 >300 >400 >500
Neurile length in microns
Figure 6. PI-PLC Removes TAG-1 from DRG Neurons but Does Not Affect Neurite Outgrowth
(A and B) Phase-contrast and fluorescence images showing TAG-1 expression on the axon and growth cone of an E14 DRG neuron
grown on a substrate of TAG-1. The punctate substrate-associated immunofluorescence indicates the binding of antibody to the TAG-1
substrate. (C) High power immunoiluorescence micrograph showing the expression of TAG-I on the axon, growth cone, and filopodia
of a DRG neuron grown on a laminin substrate in vitro. Similar images are obtained on a TAG-I substrate.
(D and E) Phase-contrast and fluorescence images of DRG neurons grown on a substrate of TAG-I in the presence of 2Ulml PI-PIC.
TAG-I is not detectable on the surface of DRG neurons.
(F) Length of DRG neurites on a substrate of TAG-I. Treatment of DRG neurons with PI-PIC does not significantly alter neu rite lengths.
(G) lengths of DRG neurites on a substrate of laminin. Treatment with PI-PLC does not signficiantly alter neurite length. Neurite length
on a BSA substrate (in the absence of PI-PLC) is also indicated. Plots are representative of at least 10 similar experiments.
Bars, 50 l~m (A, B, D, and E); 10 lJm (C).
This conclusion is consistent with evidence for an
involvement of the Ll-related molecule Ng-CAMIG4
in neurite outgrowth on axonin-1 in chick (Kuhn et
Anti-IS1 Integrin Antibodies Block Neurite Outgrowth
on TAG-1 but Not on the L1-Related Molecule
Integrins have been implicated in the ability of neu-
rons to extend neurites on a variety of substrates, in-
cluding laminin, fibronectin, vitronectin, and throm-
bospondin (Reichardt et al., 1990; Reichardt and
Tomaselli, 1991). In cells of the immune system, mem-
bers of the Ig superfamily can mediate cell interac-
tions by binding to integrins (Diamond et al., 1991).
Since rat and human TAG-I contain an RGD sequence
in a context appropriate for interactions with integrins
(Furley et al., 1990; Hasler et al., 1993; Tsiotra et al.,
1993) it seemed possible that integrins are involved
in neurite outgrowth on TAG-1.
DRG neurons express the l$1 integrin subunit,
which forms heterodimers with a number of different
subunits to mediate the promotion of neurite out-
growth on a variety of substrates (Reichardt and Toma-
selli, 1991). Neurite extension from E14-E15 DRG neu-
rons on substrate-bound TAG-I or laminin was there-
fore examined in the presence of two different poly-
clonal anti-l~l integrin antibodies. Although DRG
O "~ ' ~ ~
Neurite length in microns
.... TAG-1 + anti-L1
i 500 ~g/rn
~ ---- TAG-1 + anh-L1
>1oo ~o >2o0
Neurite length in microns
>~00 >4'0o >5'00 >6'00
~ " t ~ LN
.... LN + anti-L1 30 p.g/ml
i ]~,,t ~..~ LN + anti-L1 500 p.g/ml
>600 >300 >400 >500
Figure 7. DRG Neurite Extension on TAG-1 Is Inhibited by Anti-
(A) Plots of DRG neurite extension on TAG-1 in the presence
and absence of IgG fractions of rabbit anti-L1 antiserum,
(B) Plots of neurite extension on laminin in the presence and
absence of an IgG fractions of rabbit anti-L1 antiserum (30 p.g/
ml and 5001~8/ml). Neurite length on BSA is also indicated. Plots
are representative of 3 different experimentsi
neurons still attached to a TAG-1 substrate, neurite
extension was almost completely abolished in the
presence of Fab fragments of anti-131 integrin anti-
body (antibody Lenny; Buck and Horwitz, 1987) (Fig-
ure 8A). Similar results were observed with a second
anti-J31 integrin antibody (anti-ECMr antibody; Knud-
sen et al., 1981; Tomaselli et al., 1987) (data not shown).
Neurite extension on laminin was also completely in-
hibited by both anti-1]1 integrin antibodies in the ab-
sence of any detectable effect on neuronal attach-
ment (Figure 8B). The inhibition of neurite outgrowth
by anti-~l integrin antibodies on both TAG-1 and lam-
inin was also observed with DRG neurons that had
been stripped of surface TAG-1 with PI-PLC (data not
shown). Thus, antibodies directed against [31 integrins
inhibit neurite outgrowth on a substrate of TAG-l, and
this effect is independent of the presence of TAG-1 on
the cell surface.
To rule out the possibility that the inhibition of neu-
rite outgrowth results from antibodies in the anti-1]1
integrin antisera that cross-react with other surface
molecules expressed by DRG neurons, we examined
the effect of an MAb against 111 integrin on DRG out-
growth on TAG-1 and laminin. Because the only func-
tion-blocking anti-1]1 integrin MAbs available are di-
rected against chick 111 integrin (JG-22; Greve and
Gottlieb, 1982), we grew neurons isolated from E7
chick DRG on substrates of rat TAG-1 and on laminin.
Chick DRG neurons extended neurites on both TAG-1
and laminin, although the median length of chick
DRG neurites on rat TAG-1 was lower than that of
rat DRG neurites (Figure 8C). Nevertheless, anti-1]1
integrin MAbs produced an almost complete inhibi-
tion of neurite outgrowth on both TAG-1 and laminin
(Figure 8D). These results provide evidence that specific
antibodies directed against I]1 integrins block DRG neu-
rite outgrowth on rat TAG-1 and support the results
obtained with polyclonal anti-1]1 integrin antisera.
As one control for specificity of anti-1]1 integrin ef-
fects, we plated DRG neurons on a mixed substrate
that consisted of small islands of rat astrocytes sur-
rounded by laminin. In the presence of anti-1]1 inte-
grin antibodies at concentrations that completely
blocked neurite outgrowth on TAG-l, DRG neurons
with cell bodies located on astrocytes extended neu-
rites over the astrocyte surface, but these neurites
never extended onto the surrounding laminin sub-
strate (data not shown).
To add ress the possibility that integrins are required
for the outgrowth of DRG neu rites on other nitrocellu-
lose-bound neural Ig-like proteins, we monitored rat
DRG neurite outgrowth on a substrate of the chick
Ll-related molecule, Ng-CAM/SD9 (Figure 9A). Previ-
ous studies (Lemmon et al., 1989) have shown that
Ng-CAM/SD9 can promote the outgrowth of neurites
from rodent neurons (Lemmon et al., 1989). Rat DRG
neurons extended long neurites on an Ng-CAM/BD9
substrate (Figure 9B). Neurite extension was inhibited
by anti-L1 antibodies but was not affected by anti-1]1
integrin antibodies at concentrations that produced
strong or complete inhibition of neurite outgrowth
on a laminin substrate (Figures 9B and 9C). Similarly,
rat DRG neurite outgrowth on a substrate of chick
N-cadherin was unaffected by these antibodies (J.
Walter, personal communication). These results show
that the inhibitory effect of anti-1]1 i ntegrin antibodies
occurs on a substrate of TAG-1 but not on a closely
related neuronal Ig-like protein or on a structurally
distinct cell adhesion protein.
Taken together, these findings indicate that the abil-
ity of anti-1]1 integrin antibodies to block neurite out-
growth is selective: inhibition of outgrowth occurs on
TAG-1 but not on Ng-CAM/SD9 or astrocyte sub-
strates. The outgrowth of neurites from DRG neurons
on substrate-bound TAG-1 therefore appears to re-
quire the function of L1 and 1]l-containing integrins.
The present in vitro studies have examined the cellu-
lar interactions and functions of the neuron-specific
Ig-like protein TAG-1. Our results show that TAG-1
TAG-1 Interactions in Neurite Outgrowth
~ 50 -
+ anti-lntegdn 20 i.tg/ml ~
20 >100 >200
Neurite length in microns
>300 >400 >500 >600
Rat DRG / LN
- Rat DRG / LN
+ anti-lntegd;20 pg/m,
- - " "
h E i
Neurite length in microns
>300 ~o >600
i ~ 70-
m . Chick DRG / TAG-1
+ JG22 100 ~zg/ml
---- Rat DRG I TAG-1
Chick DRG / TAG-1
20- 30- 4 0 1 0 , 0 " ~ ~T ~ T "'"~ ~ "-'%"'" "" "-*-'"~" , --
>0 >100 >ZoO
Neurite length in microns
>300 >4o0 >Six) >600
~o >;oo >~
Neurite length in microns
--. Chick DFtG / LN
Chick DRG / LN
~oo ~o0 >5oo
Figure 8. DR(; Neurite Outgrowth on TAG-1 and Laminin Is Blocked by Antibodies to [~1 Integrin
(A) Plots of DRG neurite extension on laminin in the presence and absence of Fab fragments of 131 integrin antibodies (20 Ilg/ml).
Neurite extension on laminin is inhibited. Plots are representative of at least 8 different experiments.
(13) Plots of DRG neurite extension on TAG-1 in the presence and absence of Fab fragments of rabbit anti-I~l integrin. No neurite
extension is observed in the presence of anti-~l integrin antibodies.
(C) Plots of E14 rat and E7 chick DRG neurite extension on rat TAG-1. E7 chick DRG neurons grown in the presence of MAb JG-22
directed against chick 131 integrin do not extend neurites. Note that the extension of neurites from E7 chick DRG neurons is less than
that from E14 rat DRG neurons.
(D) Plots of neurite extension from E7 chick DRG neurons grown on laminin. Neurite extension is greatly reduced in the presence
of MAb IG-22 (100 pg/ml). Plots are representative of 4 different experiments.
can bind in a homophilic manner to promote cell ag-
gregation, whereas heterophilic interactions between
TAG-1 and two related Ig-like proteins, L1 or N-CAM,
were not detected. Homophilic binding, however, is
not required for the extension of DRG neurites on
a substrate of TAG-I. Instead, neurite outgrowth on
TAG-I is blocked by antibodies against L1 and inde-
pendently by antibodies against the I~I integrin sub-
unit. These results suggest that both an L1-1ike mole-
cule and I[$1 integrins are required for neurite out-
growth on a substrate of TAG-1.
TAG-1 Promotes Homophilic Binding, but Neurite
Outgrowth Involves Heterophilic Interactions
TAG-1 expressed on the surface of Drosophila $2 cells
or bound to nitrocellulose mediates homophilic bind-
ing. Homophilic binding has also been reported for
the chick TAG-1 homolog, axonin-1, expressed in a
vertebrate cell line (Rader et al., 1993). In our studies,
the inability of LI+ $2 or N-CAM* $2 cells to bind to
TAG-1 § $2 cells provides evidence for the selectivity
of interactions between different members of the ver-
tebrate neural Ig family under conditions that permit
homophilic binding and are likely to reflect the nor-
mal mode of presentation of these molecules on neu-
ronal cell surfaces. These results do not preclude the
possibility that there are interactions of cell surface
TAG-1 with L1 or N-CAM with binding affinities below
the threshold of detection in the $2 aggregation assay.
Nevertheless, the inability to detect interactions be-
tween TAG-1 and L1 or N-CAM does not simply reflect
a failure of the assay to detect heterophilic interac-
tions, since similar assays have revealed heterophilic
binding between the Drosophila proteins Notch and
A B 100 -
_ ~ S0-
- ~ 40-
-- - NgCAM + anti'lntegrin I
--.-- NgCAM + anti-69A1 I
,-- - ,
>0 >100 >200
Neurite length in microns
~ 50- ~ --- LN+anti-integrin
.--_= 30--,. L,
>5'oo o ~ ,'-~-
Neurite length in microns
>0 >1~00 >300 >400
?9 Figure 9. Absence of Blockade of DRG Neurite Outgrowth on Ng-CAM/SD9 by Anti-I~l Integrin Antibodies
(A) Affinity purification of Ng-CAM/BD9 shows the presence of bands at 200 kd and 135 kd.
(B) Blockade of rat DRG neurite outgrowth on Ng-CAM/GgA1 by anti-L1 antibodies, but not by anti-[~l integrin antibodies.
(C) Anti-~l integrin antibodies markedly inhibit DRG neurite outgrowth on laminin in the same experiment. Results are representative
of 3 separate experiments?9
Delta (Fehon et al., 1990) and between Boss and Sev-
enless (Kramer et al., 1991). Our results therefore, pro-
vide evidence for selectivity in recognition between
closely related vertebrate neural Ig-like proteins ex-
pressed on cell surfaces.
DRG neurons that extend neurites on a TAG-1 sub-
strate express high levels of TAG-1 on their surface.
Despite this, neurite outgrowth on TAG-1 does not
depend on cell surface TAG-1. Treatment of DRG neu-
rons with PI-PLC removed all detectable TAG-1 from
the neuronal surface with no effect on neurite length,
indicating that receptors distinct from TAG-1 mediate
the neurite outgrowth-promoting activity of TAG-1.
The Function of an Ll-like Molecule Is Required for
Neurite Outgrowth on TAG-1
Anti-L1 antibodies inhibit neurite extension on TAG-1.
This finding suggests that L1 or a closely related mole-
cule is a receptor for nitrocellulose-bound TAG-1. This
possibility is consistent with studies carried out on
chick DRG neurons in which anti-Ng-CAM/G4 an-
tibodies inhibited neurite outgrowth on axonin-1
(Kuhn et al., 1991). Moreover, axonin-l-coated beads
can aggregate with G4-coated beads (Kuhn et al.,
1991), indicating that direct interactions between
these two chick Ig-like proteins can occur in vitro un-
der certain conditions.
The ability of anti-L1 antibodies to inhibit neurite
extension on a TAG-1 substrate contrasts with the ab-
sence of a detectable interaction between TAG-1 and
L1 in the $2 aggregation assay. One possible reason
for this is that weak interactions between TAG-1 and
L1 might not be detected in the $2 aggregation assay.
This could be because the binding of TAG-1 to L1
in rat may require coexpression of an integrin (see
below), a situation not found in $2 cells. A second
possibility is that weak interactions which are not suf-
ficient to cause $2 cell aggregation could trigger neu-
rite outgrowth if such interactions were amplified in
the growth cones of neurons. Third, it remains possi-
ble that sequence differences between the chick and
rat proteins may confer different functional proper-
ties to TAG-1 and axonin-1 or to L1 and Ng-CAM. Fi-
nally, our data cou Id also be explained by the fact that
L1 is not the true rat homolog of Ng-CAM (see Grumet
et al., 1991, for a discussion of this issue) and that the
anti-L1 antibodies block the function of an Ll-related
Ig-like molecule on the surface of DRG neurons.
151 Integrin Function Is Required for DRG Neurite
Outgrowth on TAG-1
The present studies also show that DRG neurite out-
growth on TAG-1 is inhibited by anti-151 integrin anti-
bodies. The inhibition of neurite outgrowth in the
presence of anti-151 integrin antibodies could be ex-
plained in several different ways. First, axonal inte-
grins may function as receptors that are bound di-
rectly by TAG-1. Studies of interactions between
immune cells have shown that members of the Ig su-
perfamily, such as the ICAMs, can bind directly with
integrins (Diamond et al., 1991). The extracellular re-
gion of the ICAMs consists exclusively of Ig-like do-
mains, and in ICAM-1 the third Ig domain is involved
in binding to the integrin Mac-1 (Diamond et al., 1991).
In the case of TAG-l, it remains unclear whether its
neurite outgrowth-promoting activity involves the Ig-
like or fibronectin type III domains. The possibility of
an interaction involving fibronectin type III domains
is raised I~y the presence of an RGD sequence within
the second fibronectin type III domain in a context
appropriate for integrin binding (Furley et al., 1990;
Ruoslahti and Pierschbacher, 1987).
TAG-1 Interactions in Neurite Outgrowth
A. Adhesion (fasciculation)
released TAG-1 --
C. Neurite extension: dual receptor requirement
i /,t ,\ ii
iii t ~ iv t 7
"~ I TM
Figure 10. Potential Binding Interactionsand I:unctionsofTAG-1
(A) TAG-1 may promote cell adhesion and axonal fasciculation
by both homophilic and heterophilic interactions. Evidence from
$2 cell aggregation assays indicates that TAG-1 can promote cell
aggregation by homophilic binding. Studies on chick axonin-1
show that the protein can bind to G4 when immobilized to inert
beads. If such an interaction occurs on cell surfaces, adhesion
and fasciculation may also involve heterophilic binding of TAG-1.
(B) The secreted form of TAG-1 may inhibit both homophilic and
heterophilic binding interactions of TAG-l, leading to the loss
of cell adhesion and to defasciculation. The defasciculation of
sensory neurites in the presence of high concentrations of axo-
nin-1 has been demonstrated (Ruegg et al., 1989b).
(C) Blockade of DRG neurite extension by both anti-L1 and anti-
61 integrin antibodies suggests that neurite extension on sub-
strate-bound TAG-1 may require the activation of two distinct
receptor systems. Four possible mechanisms that could contrib-
ute to the dual requirement of Ll-like and integrin receptors are
illustrated. (i) Binding of TAG-1 to integrin and Ll-like molecules
triggers two distinct intracellular signaling pathways, both of
which are required for neurite extension on TAG-1. (ii) A single
TAG-1 molecule may bind to both integrins and an Ll-like mole-
cule. The TAG-l-induced association of these two receptors may
A second possibility is that integrin function is re-
quired for neurite extension in a manner that does
not depend on direct interactions between TAG-1
and integrins. Integrin function could be required
downstream of the initial TAG-l-dependent signal. For
example, the detection of a TAG-1 signal by an inde-
pendent receptor may trigger the secretion of extra-
cellular matrix molecules that may condition the local
environment and interact with axonal integrins. Alter-
natively, anti-I~l integrin antibodies could inhibit neu-
rite outgrowth by activating integrin signaling rather
than by blocking integrin function. Transduction of
an integrin signal could activate intracellular path-
ways that prevent the neuron from responding to the
TAG-1 substrate. Such an effect would have to be a
substrate-specific phenomenon, however, since neu-
rite extension on the related Ig-like protein Ng-CAM/
8D9 or on astrocytes is unaffected by the presence
of anti-I~l integrin antibodies.
The identity of the integrins required for neurite
outgrowth on TAG-1 is not known. The specificity of
the antibodies used indicates that the 131 integrin sub-
unit is involved. Biochemical studies have shown that
DRG neurons express at least four integrin a subunits,
al, a6, a7, and a8 (Reichardt et al., 1990; Bossy et
al., 1991), which represent candidates for heterodimer
formation with 131 integrin.
Homophilic and Heterophilic Interactions of TAG-1
Although neurite outgrowth promoted by TAG-1 in
vitro appears to involve heterophilic interactions, ho-
mophilic binding between TAG-1 molecules on ap-
posing axonal surfaces could contribute to the adhe-
sion and fasciculation of embryonic axons in vivo
(Figure 10A). TAG-1 is expressed in a patchy distribu-
tion on axonal surfaces, and regions of membrane
containing a high density of TAG-1 are frequently
aligned on the surface of apposing axons (Pickford
et al., 1989; Yamamoto et al., 1990), consistent with the
possibility of a homophilic interaction. The signaling
potential of such homophilic interactions, however, is
likely to be limited, since cell surface TAG-1 is present
exclusively as a GPI-anchored molecule and is there-
fore unlikelyto interact directlywith intracel!ular mol-
ecules that trigger changes in axonal growth.
The surface form of TAG-1 could therefore interact
in a heterophilic manner with distinct receptors on
axonal surfaces. Heterophilic binding between TAG-1
and its receptor on an apposing axon could mediate
trigger an intracellular signal that is dependent on such receptor
association. (iii and iv)TAG-1 may bind directly to integrins, lead-
ing indirectly to the activation of an 11 signaling pathway. Alter-
natively, TAG-1 may bind directly to L1 or to an Ll-like molecule,
leading indirectly to activation of an integrin signaling pathway.
The possibility of such a serial pathway of receptor activation
and intracellular signaling is raised by the finding that anti-L1
antibodies alone or anti-131 integrin antibodies alone are effec-
tive in producing a virtually complete inhibition of neurite out-
axonal fasciculation (Figure 10A). Alternatively, TAG-I
could act as a ligand for other receptors that trigger
intracellular changes regulating axonal fasciculation.
In contrast, the secreted form of TAG-1 could inter-
rupt homophilic or heterophilic binding functions
mediated by the surface form of TAG-I (Figure 10B).
Consistent with this, high concentrations of the se-
creted form of axonin-1 have been shown to cause
defasciculation of neurites in vitro (Ruegg et al.,
1989b). Moreover, our present results show directly
that soluble TAG-1 can prevent homophilic binding
of TAG-1. It is possible, therefore, that axonal fascicu-
lation may be controlled by differential regulation of
the surface and secreted forms of these proteins, as
has been observed to occur in vitro (Karagogeos et
.A Dual Receptor Requirement for Neurite Outgrowth
The independent inhibition of DRG neurite extension
by anti-L1 and anti-~l integrin antibodies raises the
possibility that the function of two different receptor
systems is required for DRG neurite extension on
TAG-1. The identity of the axonal receptors that bind
TAG-1 directly is currently not known, but the present
studies and those of Kuhn et al. (1991) suggest an L1-
like molecule and a 131-containing integrin as strong
candidates. Thedual receptor requirement for neurite
outgrowth could be explained in several ways. First,
TAG-1 could bind directly to both integrins and Ll-like
molecules, resulting in the activation of distinct but
convergent intracellular signaling pathways, each of
which is required for neurite growth (Figure 10C, i).
Second, a single TAG-1 molecule could bind both to
L1 or an Ll-like protein and to a 131-containing integrin.
The association of the two surface proteins could in
turn generate a single association-dependent i ntracel-
lular signal. Alternatively, coligation of both L1 and
integrin by TAG-1 could be required for functional
binding if the affinity of TAG-1 for either receptor
alone were insufficient (Figure 10C, ii). Third, TAG-1
could bind directlyto onlyoneof the candidate recep-
tors, with this association resulting in a cis interaction
with the second receptor protein (Figures 10C, iii and
iv). Since both anti-L1 and anti-131 integrin antibodies
alone caused a virtually complete inhibition of neurite
outgrowth, it is possible that such cis interactions op-
erate in a serial manner, with the major component of
intracellular signaling mediated through the surface
receptor that does not directly ligate TAG-1 (Figure
10C, iii and iv).
Although these experiments have examined neurite
outgrowth only on TAG-1 and Ng-CAM/8D9 sub-
strates, it remains possible that neurite outgrowth on
other substrates also involves dual receptor activa-
tion. A combinatorial requirement for neurite out-
growth-promoting molecules would greatly expand
the number of potential guidance cues available in
the developing nervous system. Studies to determine
the involvement of integrins in neurite outgrowth on
other Ig-like proteins should reveal whether the dual
receptor requirement demonstrated with TAG-1 is a
more general feature of axonal outgrowth in vitro and
perhaps also in vivo.
cDNA Cloning and Expression
A rat cDNA encoding TAG-1 was isolated as described (Furley
et al., 1990). Rat cDNAs encoding rat N-CAM (140 kd isoform;
including the VASE exon) and L1 were cloned using oligonucleo-
tides derived from the published rat and mouse sequences
(Small et al., 1987; Moos et al., 1988; Miura et al., 1991), and these
were used to screen rat E13 spinal cord and brain libraries (Furley
et al., 1990; D. Manalo, K. Huang, and A. J. F. unpublished data).
Clones from these screens were used to assemble complete cod-
ing regions for cloning into expression vectors.
For recombinant TAG-I protein productions, the full-length
TAG-I coding region was cloned into the expression vector
pMVT, in which expression is driven by the Moloney retrovirus
long-terminal repeat promoter (see Furley et al., 1990). This con-
struct was transfected into a variant of the human embryonic
kidney cell line, 293, adapted for growth in suspension culture
(a kind gift of C. Gorman, Genentech, Inc.). High expressing
subclones were obtained by fluorescence-activated cell sorting
on the basis of surface protein expression.
For recombinant TAG-I protein production, the full-length
TAG-I coding region was cloned into the expression vector
vitrogen Corp.) using the baculovirus expression system
(O'Reilly et al., 1992). The TAG-I cDNA was truncated by polymer-
ase chain reaction, deleting 142 bases starting at base 3244 (Fu rley
et al., 1990), resulting in a predicted C-terminal peptide sequence
of HIVRNGG. This truncation interrupts the TAG-1 protein 8
amino acids before the predicted GPI addition site, resulting
in a protein that is secreted (D. P. F., K. M. S., A. J. F., and
T. M. J., unpublished data). Recombinant virus was generated
in Sf9 cells by cotransfecting the TAG-I transfer vector and linear-
ized wild-type viral AcMNPV DNA as described (Invitrogen
$2 Cell Aggregation Assays
S2 cells (a gift from Alan Bieber) expressing cell surface cell ad he-
sion molecules were generated essentially as described (Elkins
et al., 1990). TAG-l, L1, or N-CAM cDNAs were inserted into
either pRmHa-3 (cDNA expression driven by a metallothionein
promoter; Bunch et al., 1988) or pCaSpeR (cDNA expression
driven by a Drosophila HSP70 heat shock promoter; Schneuwly
et al., 1987; Rubin and Spradling, 1983). These constructs were
transfected into $2 cells, which were then assayed for transient
protein expression, or established as permanently expressing
cell lines using a-amanitin selection and selection for protein
expression by fluorescence-activated cell sorting.
For assays with transiently transfected cells, cells were washed
and allowed to recover for 24 hr after transfection. Cells con-
taining a metallothionein construct were induced by the addi-
tion of 0.7 mM CuSO4 to the culture medium overnight. Cells
containing a heat shock construct were subject to a 30 min incu-
bation at 37~ followed by a 4 hr recovery at 27~ prior to aggre-
gation. Before the aggregation assay, cells were washed thor-
oughly and resuspended in Hanks' balanced salt solution
(Specialty Media Inc.) with 2 mg/ml BSA (Boehringer). Ceils were
rotated at 100-110 rpm for 45-60 min, a time determined to be
optimal for aggregation. At the end of the aggregation period,
cells were fixed in the assay well by the addition of an equal
volume of 4% paraformaldehyde in 120 mM phosphate buffer
(PB). Cells were then rotated for an additional 10 rain during
For mixing experiments, one of two methods was used to
distinguish the two cell populations. In experiments with stably
transfected cell lines, one population of cells was dye labeled
in advance with CFSE (Molecular Probes). Cells treated with a
1:500 dilution of a 10 mM stock of CFSE for 5 rain were completely
TAG-1 Interactions in Neurite Outgrowth
labeled. In transient transfection experiments, the two different
cell populations were distinguished by antibody labeling.
For antibody staining, $2 cells were treated as described be-
low. Primary antibodies were detected with isotype- and species-
specific antibodies conjugated to either FITC (Boehringer) or
Texas Red (Molecular Probes).
Quantification of $2 Cell Aggregation
Fixed, stained suspensions of $2 cells from aggregation experi-
ments were treated with SlowFade (Molecular Probes), placed
on slides, coverslipped, and sealed to prevent evaporation. $2
cell aggregation was quantified by categorizing cells marked by
the appropriate antibody or dye as either unbound ("free"),
bound homotypically, or bound heterotypically. Serial, nonover-
lapping fields were counted at 250 x magnification. Cell clusters
that contained fewer than 4 cells were not included in aggregate
counts to eliminate a contribution of nonspecific cell clustering
during fixation. Large aggregates (greater than 100 cells) were
counted by treating the aggregate as a two-dimensional object.
This approach results in an underestimate of the number of cells
in larger aggregates. This approach also ignores aggregate size,
which varies with expression levels as well as the percentage of
cells expressing individual proteins.
Covasphere beads (0.5 p.m; Duke Scientific) were coated with
TAG-1 at a concentration of 450 ng of protein per p.I of bead
suspension. Protein binding was carried out for 75 min at 22~
or overnight at 4~ After binding, beads were blocked with 1%
BSA in Dulbecco's phosphate-buffered saline (DPBS) for 75 min
at 22~ and washed with DPBS. The beads were resuspended
in 5 I~1 of PBS per pl of original suspension. This stock was used
at a 1:10 dilution in all experiments.
For covasphere binding to nitrocellulose-bound protein, sub-
strates were prepared as described below. Protein-coated co-
vaspheres, diluted in 0.1% BSA in PBS, were plated onto nitrocel-
lulose-bound protein. Dishes were rocked for 25 rain at 22~
and then allowed to settlewithout agitation for 5 min. For com pe-
titio n experiments, TAG-1 was added to the 0.1% BSA at a concen-
tration of 250 mg/ml. After the binding of beads, dishes were
washed extensively with 1% BSA in PBS and coverslipped in
Beads were counted using a reticule grid. Six random, non-
overlapping fields were counted with each substrate protein to
Purification of TAG-1 and Ng-CAM/8D9
E15-E16 rat brains were dissociated using a Dounce homoge-
nizer in DPBS (Specialty Media Inc.) containing 0.32 M sucrose, 5
mM MgCI2, and 1 mM phenylmethylsulfonyl fluoride. Following
centifugation at 1000 x g to remove large insoluble material,
the membrane suspension was centifuged at 100,000 x g for 1
hr. The supernatant containing the soluble form of TAG-1 was
affinity purified on an anti-TAG-1 MAb (1C12; Dodd et al., 1988).
Membrane-associated TAG-1 was purified as described (Furley
et al., 1990). For production of recombinant TAG-1 protein, super-
natants from TAG-l-transfected 293 cells grown in suspension
were collected, concentrated, and subjected to the same purifi-
Chick Ng-CAM was prepared as described (Lagenaur and Lem-
mon, 1987), using an 8D9 affinity column (antibody obtained
from the Developmental Studies Hybridoma Bank).
Outgrowth experiments were carried out in 4 well dishes (Steri-
lin, Inc.) in which each 10 mm diameter well held a volume of 100
~1. Su bstrates for neu rite outgrowth were prepared as described
(Lagenaur and Lemmon, 1987). Briefly, a4cm 2 piece of nitrocellu-
lose (BA85: Schleicher & Schuell) was dissolved in 6 ml of metha-
nol, and 25 p.I of this solution was pipetted onto a 78 mm 2 area
and air dried. Protein solutions in a volume of 2.0-2.5 gl were
placed onto the nitrocellulose for 5-10 min. Subsequently, the
protein solution was aspirated, and the substrate was blocked
with 1% BSA in PBS. Dishes were incubated in 3% BSA in PBS
for 1-4 h r at 4~ to ensu re the corn plete satu ration of the nitrocel-
lulose. Prior to the plating of DRG neurons, the dishes were
washed three times with PBS and incubated in F12 medium (Spe-
cialty Media Inc.) at 37~ for 15-30 min.
Mixed astrocyte-laminin substrates were made by plating dis-
sociated postnatal day 0 (P0) rat cortical astrocytes at low density
onto a laminin substrate generated as described above.
Astrocytes were cultured in Dulbecco's modified Eagle's medium
containing 10% fetal calf serum. Astrocyte cultures were grown
for 24 hr to permit cell proliferation. Cultures were washed and
plated with a suspension of F14 DRG cells as described below
(Karagogeos et al., 1991).
DRG Neuron Cultures
DRG neurons were isolated from E14-E15 rats and dissociated
with 0.125% trypsin in Ca 2+-, Mg2+-free medium. DRG neurons
were cultured in F12 medium supplemented with N3, vitamins
(GIBCO), 2 mM glutamine, 8 mg/ml glucose, and 0.1% BSA (Kara-
gogeos et al., 1991). For PI-PLC experiments, neurons were pre-
treated with PI-PLC (a gift of Dr. Martin Low) at a concentration
of 2 U/ml for 30 min at 37~ to remove cell surface TAG-1. Control
cells were incubated in parallel. DRG cells were also plated in
the presence of PI-PLC to ensure that cells were exposed to the
enzyme during the course of the experiment. Antibody blockade
experiments were performed by adding antibodies at concentra-
tions described in the text (either Fabs or whole antibody) to
suspensions of neurons prior to plating.
Neurite Outgrowth Assay
For studies of neurite outgrowth from DRG neurons in vitro,
TAG-1 was isolated from embryonic rat brain or from a human
embryonic kidney 293 linethat had been stably transfected with a
rat TAG-1 cDNA. TAG-1 was isolated on a monoclonal anti-TAG-1
affinity column, as described above and, after separation by
SDS-polyacrylamide gel electrophoresis, generated a single 135
kd band, as detected by silver staining and Western blotting
(Figure 4A). The ability of TAG-1 to promote the outgrowth of
DRG neurons was tested by immobilizing affinity-pu rifled TAG-1
(- 50 p.g/ml, 370 nM) on a nitrocellulose substrate (Lagenaur and
Lemmon, 1987). Neu rite outgrowth on TAG-1 was compared with
that of laminin (20 I~g/ml, 50 nM) or BSA (10 mg/ml, 140 p.M)
bound to nitrocellulose. Approximately 60% of added TAG-1 and
laminin bound to the nitrocellulose. To establish that neurite
outgrowth resulted from the interactions of neurons with sub-
strate-bound TAG-1 and not with other minor contaminants that
could have been copurified, we also performed neurite out-
growth experiments with fractions isolated by affinity purifica-
tion on a class-matched (IgG) antibody directed against an irrele-
vant chick antigen and with fractions derived from medium
harvested from untransfected 293 cells recovered from the anti-
TAG-1 affinity column. When immobilized on nitrocellulose, nei-
ther of these fractions promoted DRG neurite outgrowth at lev-
els greater than that obtained with BSA (data not shown).
In addition, the outgrowth of neurites from DRG neurons
treated with PI-PLC to remove surface TAG-1 was inhibited by
addition of rabbit anti-TAG-1 antibodies (data not shown). E14-
E15 DRG neurons were treated before plating with PI-PLC (2 U/
ml) for 30 min at 37~ and then grown in vitro in the continued
presence of PI-PLC for 14-15 hr on TAG-l, BSA, or laminin sub-
Quantification of Neurite Outgrowth
Neurite lengths were quantified using a custom macro written
for the Optimas software package (Bioscan) running on an IBM
PC compatible computer with a PC Vision Plus (Imaging Technol-
ogy, Inc.) frame grabber. Neurite length curves were generated
in Excel (Microsoft) and plotted in DeltaGraph Professional
(Delta Point). Neurite lengths were quantified as the total extent
of the neurite from the cell soma to the growth cone, including
all branches. Distinct processes originating at the cell body were
counted as separate neurites. This method of quantitation was
found to yield results essentially identical to quantitation of total
neurite growth from a given cell. Neurites less than one cell
diameter were counted as zero extension. Neurite lengths were
plotted as described (Chang et al., 1987) with the modification
that zero measurements were included in the quantitation, since
MAb 3A10 staining offered a reliable method of distinguishing
live neurons without processes from dead cells.
Immunocytochemical Localization of Glycoproteins
For immunofluorescence labeling, cultures were removed from
the incubator and rapidly fixed with 4% paraformaldehyde in
0.'12 M PB (pH 7.4) for 15 min. Cultures were then washed two
times with PB before staining. Cultures labeled for detection of
cell surface glycoproteins were preblocked for 5 min with 1%
normal goat serum in PB (NGS-PB). Cultures were incubated in
primary antibodies (diluted in 1% NGS-PB) for 30-60 min at 22~
Cultures were then washed twice in 1% NGS-PB before incuba-
tion in secondary FITC-conjugated isotype-specific antibodies
diluted in 1% NGS-PB for 30-60 min at 22~
washed with 1% NGS-PB followed by PB alone and coverslipped
in SlowFade (Molecular Probes) in glycerol (1:1). Cultures were
visualized on a Zeiss Axioplan microscope by epifluorescence
and phase-contrast optics. Confocal imaging was carried out on
~, Bio-Rad MRC-500 system.
Cultures were processed for staining with the 3A10 antigen
by permeabilizing cells with 0.1% Triton )(-100 in PB for 30 rain
at 22~ Cultures were then washed with PB and preblocked
with 3% NGS-PB for 30 min at 22~
MAb 3A10 for 2 hr at 22~ or overnight at 4~ followed by two
washes in 1% NGS-PB. Cultures were then incubated in second-
ary antibody, washed, and coverslipped as described above.
Cells were incubated in
For immunohistochemistry, TAG-1 was detected by polyclonal
rabbit anti-rat TAG-1 antibodies, by MAb 1C12 (Dodd et al., 1988),
or by MAb 4D7 (Yamamoto et al., 1986). The polysialylated form
of N-CAM was detected by MAb 5A5 (Dodd et al., 1988). N-CAM
was also detected with the polyclonal rabbit antiserum RO25
(gift of Urs Rutishauser). This antibody inhibits N-CAM binding
(Urs Rutishauser, personal communication). L1 was detected by
a mixture of MAb 69A1 (Piggott and Kelly, 1986) and MAb ASCS4
(Sweadner, 1983). For measurement of neurite lengths, DRG cul-
tures were stained with MAb 3A10, which recognizes a filament-
associated protein (Fu rley et al., 1990; Morton, T. M. J., and Dodd,
Inhibition of neurite extension was performed using either
whole antibody dialyzed against DPBS or Fab fragments as de-
scribed below. A number of characterized, function-blocking
anti-integrin antibodies were used. The polyclonal antibody anti-
ECMr (Knudsen et al., 1981; Tomaselli et al. 1987), provided by
Caroline H. Damsky and Louis F. Reichardt, was raised in goat
against proteins derived from BHK cells and binds to the 131
integrin subunit (Tomaselli et al., 1987). The polyclonal antibody
"Lenny" (Buck and Horwitz, 1987), provided by Clayton A. Buck,
was raised in rabbit against material derived from the rat L6A
cell line and binds to the 131 and 133 subunits of the integrin
complex (Albelda et al., 1989). Fab fragments of the rabbit anti-
integrin antibody Lenny were prepared by papain digestion (Har-
low and Lane, 1988). Fc and whole antibody fractions were elimi-
nated from Fab fragments by retention on a protein A column.
Fab fragments were dialyzed against PBS, concentrated, and
stored at 1-5 mg/ml. MAb JG-22 (obtained through the Develop-
mental Studies Hybridoma Bank) was raised against chick mus-
cle (Greve and Gottlieb, 1982) and recognizes the 131 subunit of
the i ntegrin complex (Buck et al., 1986). A polyclonal rabbit serum
raised against rat L1 purified by immunoaffinity with MAb 69A1
(Piggott and Kelly, 1986) was used to test Ll-dependent neurite
We thank C. Buck, C. Damsky, R. Piggott, L. Reichardt, and U.
Rutishauser for providing antibodies and Richard Axel, Jane
Dodd, Andreas Kottmann, Marc Tessier-Lavigne, Susan Morton,
and Jochen Walter for advice, discussions, and comments on
the manuscript. Ira Schieren provided computer assistance and
FACS facilities, Barbara Han, Kim Huang, and Dora Manalo
helped with cDNA cloning and cell culture, and Vicki Leon typed
the manuscript. D. P. F. was supported by NIH training grant
NS 07258. M. A. H. and A. J. F. were Research Associates, and
T. M. J. is an Investigator of the Howard Hughes Medical Insti-
tute. This work was also supported by NIH grant RO1 24880-07.
The first two authors contributed equally to this work.
The costs of publication of this article were defrayed in part
by the payment of page charges. This article must therefore be
hereby marked "advertisement" in accordance with 18 USC Sec-
tion 1734 solely to indicate this fact.
Received August 10, 1993; revised January 19, 1994.
Albelda, S. M., Daise, M., Levine, E. M., and Buck, C. A. (1989).
Identification and characterization of cell-substratu m adhesion
receptors on cultured human endothelial cells. J. Clin. Invest.
Bixby, J. L., and Zhang, R. (1990). Purified N-cadherin is a potent
substrate for the rapid induction of neurite outgrowth. J. Cell
Biol. 110, 1253-1260.
Bossy, B., Bossy-Wetzel, E., and Reichardt, L F. (1991). Character-
ization of the integrin ct8 subunit: a new integrin ~l-associated
subunit, which is prominently expressed on axons and on cells
in contact with basal laminae in chick embryos. EMBO J. 9, 2375-
Br~immendorf, T., Wolff, J. M., Frank, R., and Rathjen, F. G. (1989).
Neural cell recognition molecule Fl1: homologywith fibronectin
type III and immunoglobulin type C domains. Neuron 2, 1351-
Buck, C. A., and Horwitz, A. F, (1987). I ntegrin, a transmem brane
glycoprotein complex mediating cell-substratum adhesion. ].
Cell Sci. 8, 231-250.
Buck, C. A., Shea, E., Duggan, K., and Horwitz, A. F. (1986). Inte-
grin (the CSAT antigen): functionality requires oligomeric integ-
rity. J. Cell Biol. 103, 2421-2428.
Bunch, T. A., Grinblat, Y., and Goldstein, L S. B. (1988). Character-
ization and use of the Drosophila metallothionein promoter in
cultured Drosophila melanogaster cells. Nucl. Acids Res. 16,
Burgoon, M. P., Grumet, M., Mauro, B., Edelman, G. M., and
Cunningham, B. A. (1991). Structure of the chick neuron-glia
cell adhesion molecule, Ng-CAM: origin of the polypeptides and
relation to the Ig superfamily. J. Cell Biol. 110, 1017-1030.
Burns, F. R., von Kannen, S., Guy, L., Raper, J. A., Kamholz, J.,
and Chang, S. (1991). DM-GRASP, a novel immunoglobulin su-
perfamily axonal surface protein that supports neurite exten-
sion. Neuron 7, 209-220.
Chang, S., Rathjen, F. G., and Raper, J. A. (1987). Extension of
neurites on axons is impaired by antibodies against specific neu-
ral cell surface glycoproteins. J. Cell Biol. 104, 355-362.
Cunningham, B. A., Hemperly, J. J., Murray, B. A., Prediger, E. A.,
Brackenbury, R., and Edelman, G. M. (1987). Neural cell adhesion
molecule: structure, irnmunoglobulin-like domains, cell surface
modulation, and alternative RNA splicing. Science 236, 799-806.
Diamond, M. S., Staunton, D. E., Martin, S. D., and Springer,
T. A. (1991). Binding of the integrin Mac-1 (CD11b/CD18) to the
third immunoglobulin-like domain of ICAM-1 (CD54) and its reg-
ulation by glycosylation. Cell 65, 961-971.
Dodd, ]., and Jessell, T. M. (1988). Axon guidance and the pat-
terning of neuronal projections in vertebrates. Science 242, 692-
Dodd, J., Morton, S. B., Karagogeos, D., Yamamoto, M., and Jes-
sell, T. M. (1988). Spatial regulation of axonal glycoprotein expres-
sion on subsets of embryonic spinal neurons. Neuron 1, 105-
TAG-1 interactions in Neurite Outgrowth
Doherty, P., Barton, C. H., Dickson, G., Seaton, P., Rowett,
L H., Moore, S. E., Gower, H. J., and Walsh, F. S. (1989). Neuronal
process outgrowth of human sensory neurons on rnonolayers
of cells transfected with cDNAs for five hu man N-CAM isoforms.
J. Cell Biol. 109, 789-798.
Doherty, P., Cohen, l., and Walsh, F. S. (1990). Neu rite outgrowth
in response to transfected N-CAM changes during development
and is modulated by polysialic acid. Neuron 5, 209-219.
Elkins, T., Hortsch, M., Bieber, A. J., Snow, P. M., and Goodman,
C. S. (1990). Drosophila fasciclin I is a novel homophilic adhesion
molecule that along with fasciclin III can mediate cell sorting.
J. Cell Biol. 110, 1825-1832.
Fan, J., Mansfield, S. G., Redmond, T., Gordon-Weeks, P. R., and
Raper, J. A. (1993). The organization of F-actin and microtubules
in growth cones exposed to a brain-derived collapsing factor. J.
Cell Biol. 121, 867-878.
Fehon, R. G., Kooh, P. ]., Rebay, I., Regan, C. L., Xu, T., Muskav-
itch, M. A., and Artavanis-Tsakonas, S. (1990). Molecular interac-
tions between the protein products of the neu rogenic loci Notch
and Delta, two EGF-homologous genes in Drosophila. Cell 61,
Furley, A. J., Morton, S. B., Manalo, D., Karagogeos, D., Dodd,
l., and Jessell, T. M. (1990). The axonal glycoprotein TAG-I is an
immunoglobulin superfamily member with neurite outgrowth-
promoting activity. Cell 61, 157-170.
Gennarini, G., Cibelli, G., Rougon, G., Mattei, M-G., and Goridis,
C. (1989). The mouse neuronal cell surface glycoprotein F3: a
phosphatidylinositol-anchored member of the immunoglobulin
superfamily related to chick contactin. J. Cell Biol. 109, 775-788.
Goodman, C. S., and Shatz, C. J. (1993). Developmental mecha-
nisms that generate precise patterns of neuronal connectivity.
Cell 721Neuron 10 (Suppl.), 77-98.
Greve, J. M., and Gottlieb, D. (1982). Monoclonal antibodies
which alter the morphology of cultured chick myogenic cells.
J. Cell Biochem. 18, 221-229.
Grumet, M., Mauro, V., Burgoon, M. P., Edelman, G. M., and
Cunningham, B. A. (1991). Structure of a new nervous system
glycoprotein, Nr-CAM, and its relationship to subgroups of neu-
ral cell adhesion molecules. J. Cell Biol. 113, 1399-1412.
Harlow, E., and Lane, D. (1988). Antibodies: A Laboratory Manual
(Cold Spring Harbor, New York: Cold Spring Harbor Labora-
Hasler, 1-. H., Rader, C., Stoeckli, E. T., Zuellig, R. A., and
Sonderegger, P. (1993). cDNA cloning, structural features, and
eucaryotic expression of human TAG-1/axonin-1. Eu r. J. Biochem.
Hoffman, S., and Edelman, G. M. (1983). Kinetics of homophilic
binding by E and A forms of the neural cell adhesion molecule.
Proc. Natl. Acad. Sci. USA 80, 5762-5766.
Hynes, R. O. (1992). Integrins: versatility, modulation, and signal-
ing in cell adhesion. Cell 69, 11-25.
Hynes, R. O., and Lander, A. D. (1992). Contact and adhesive
specificities in the associations, migrations, and targeting of cells
and axons. Cell 68, 303-322.
Jessell, T. M., and Dodd, J. (1992). Floor plate-derived signals and
the control of neural cell pattern in vertebrates. Harvey Lect. 86,
Karagogeos, D., Morton, S. B., Casano, F., Dodd, J., and lessell,
T. M. (1991). Developmental expression of the axonal glycopro-
tein TAG-l: differential regulation by central and peripheral neu-
rons in vitro. Development 112, 51-67.
Knudsen, K. A., Rao, P. E., Damsky, C. H., and Buck, C. A. (1981).
Membrane glycoproteins involved in cell-substratum adhesion.
Proc. Natl. Acad. Sci. USA 78, 6071-6075.
Kramer, H., Cagan, R. L., and Zipursky, S. L. (1991). Interaction
of bride of sevenless membrane-bound ligand and the sevenless
tyrosine-kinase receptor. Nature 352, 207-212.
Kuhn, T. B., Stoeckli, E. T., Condrau, M. A., Rathjen, F. G., and
Sonderegger, P. (1991). Neurite outgrowth on immobilized axo-
nin-1 is mediated by a heterophilic interaction with L1 (G4). ].
Cell Biol. 115, 1113-1126.
Lagenaur, C., and Lemmon, V. (1987). An Ll-like molecule, the
8D9 antigen, is a potent substrate for neurite extension. Proc.
Natl. Acad. Sci. USA 84, 7753-7757.
Landmesser, L. T. (1988). Peripheral guidance cues and the forma-
tion of specific motor projections in the chick. In From Message
to Mind, S. S. Easter, K. F. Barald, and B. M. Carlson, eds. (Sunder-
land, Massachusetts: Sinaur Associates), pp. 121-133.
Lemmon, V., and McLoon, S. C. (1986). The appearance of an
Ll-like molecule in the chick prirnaryvisual pathway. J. Neurosci.
Lemmon, V., Farr, K. L., and Lagenauer, C. (1989). Ll-mediated
axon outgrowth occurs via a homophilic binding mechanism.
Neuron 2, 1597-1603.
Lin, C-H., and Forscher, P. (1993). Cytoskeletal remodeling du ring
growth cone-target interactions. J. Cell Biol. 121, 1369-1383.
Low, M. G., and Saltiel, A. R. (1988). Structural and functional
roles of glycosyl-phosphatidylinositol in membranes. Science
Miura, M., Kobayashi, M., Asou, H., and Uyemura, K. (1991). Mo-
lecular cloning of cDNA encoding the rat neural cell adhesion
molecule L1. Two L1 isoforms in the cytoplasmic region are pro-
duced by differential splicing. FEBS Lett. 289, 91-95.
Moos, M., Tacke, R., Scherer, H., Teplow, D., FriJh, D., and
Schachner, M. (1988). Neural adhesion molecule L1 is a member
of the immunoglobulin superfamily with binding domains simi-
lar to fibronectin. Nature 334, 701-703.
O'Reilly, D. R., Miller, L. K., and Luckow, V. A. (1992). Baculovirus
Expression Vectors: A Laboratory Manual (New York: W. H. Free-
man and Company).
Payne, H. R., Burden, S. M., and Lemmon, V. (1992). Modulation
of growth cone morphology by substrate-bound adhesion mole-
cules. Cell Motil. Cytoskel. 21, 65-73.
Pickford, L. B., Mayer, D. N., Bolin, L. M., and Rouse, R. V. (1989).
Transiently expressed, neural-specific molecule associated with
premigratory granule cells in postnatal mouse cerebellum. J.
Neurocytol. 18, 465-478.
Piggott, R., and Kelly, J. S. (1986). Immunocytochernical and bio-
chemical studies with the monoclonal antibody 69A1: similarities
of the antigen with cell adhesion molecules L1, NILE and Ng-
CAM. Brain Res. 394, 111-122.
Rader, C., Stoeckli, E. T., Ziegler, U., Osterwalder, T., Kunz, B.,
and Sonderegger, P. (1993). Cell-cell adhesion by homophilic
interaction of the neuronal recognition molecule axonin-1. Eur.
J. Biochem. 215, 133-141.
Ranscht, B. (1988). Sequence of contactin, a 130 kd glycoprotein
concentrated in areas of interneu ral contact, defines a new mem-
ber of the immunoglobulin supergene family in the nervous
system. J. Cell Biol. 107, 1561-1573.
Rathjen, F. G., and lessell, T. M. (1991). Glycoproteins that regu-
late the growth and guidance of vertebrate axons: domains and
dynamics of the immunoglobulin/fibronectin type III subfamily.
Sere. Neurosci. 3, 297-307.
Reichardt, L F., and Tomaselli, K. J. (1991). Extracellular matrix
molecules and their receptors: functions in neural development.
Annu. Rev. Neurosci. 14, 531-570.
Reichardt, L F., Bossy, B., Carbonetto, S., De Curtis, I., Emmett,
C., Hall, D. E., Ignatius, M. J., Lefcort, F., Napolitano, E., Neugeb-
auer, K. M., and Tomaselli, K. (1990). Neuronal receptors that
regulate axon growth. Cold Spring Harbor Symp. Quant. Biol.
Rubin, G. M., and Spradling, A. C. (1983). Vectors for P element-
mediated gene transfer in Drosophila. Nucl. Acids Res. 11, 6341.
Ruegg, M. A., Stoeckli, E. T., Kuhn, T. B., Heller, M., Zuellig, R.,
and Sonderegger, P. (1989a). Purification of axonin-1, a protein
that is secreted from axons during neurogenesis. EMBO J. 8, 55-