|Home | About | Journals | Submit | Contact Us | Français|
We have examined the role of integrins in the development of dendrites in embryonic rat sympathetic neurons in vitro. Previous studies have established that exposure to a basement membrane extract (BME) causes these neurons to form dendrites (Lein and Higgins, 1989). The current experiments demonstrate that the dendrite-promoting activity of BME can be markedly inhibited by antibodies directed against the β1 integrin subunit. Although the specific integrin sub-unit(s) that mediate the effect of BME on dendritic growth were not identified, it was found that neither the 1, or 4 sub-units, nor RGD-binding integrins were involved to any significant extent. These data suggest that, in addition to their well established role in regulating axonal growth, β1 integrins also participate in dendritic growth.
The extracellular matrix (ECM) regulates many aspects of neuronal development, including cell survival and differentiation, migration, axonal pathfinding, synaptogenesis and myelination (rev. by Sanes 1989; LeTourneau et al 1994; Venstrom and Reich-ardt 1993). Evidence from in vitro and in situ studies suggests that interactions between the ECM and developing neurons are mediated primarily by cellular receptors known as integrins (Fernandez-Valle et al 1994; Malek-Hedayat and Rome 1994; Reichardt and Tomaselli 1991). Integrins comprise a large family of heterodimeric glycoproteins, each consisting of an and β subunit that associate noncovalently to form a functional transmembrane receptor (Hynes 1992). There are eight different β subunits, each of which can associate with one or more of at least 15 different subunits. To date, at least three different β integrin subunits and eight different subunits have been identified on neurons or neuronal cell lines (Venstrom and Reichardt 1993; 1995).
Previous studies from this laboratory have demonstrated mat ECM molecules can regulate the morphogenesis of embryonic sympathetic neurons in vitro by selectively modifying axonal or dendritic growth (Lein and Higgins 1989; Lein et al 1991). Thus, when sympathetic neurons are grown in the presence of laminin or collagen IV, they extend only axons; however, when exposed to BME, an extract of basement membrane proteins, these neurons produce a dendritic arbor comparable in size to that formed by their counterparts in situ. Antibody perturb-ation studies have demonstrated that the axon-specific effects of laminin and collagen IV are mediated predominantly by integrins (Lein et al 1991), this is consistent with results from other laboratories which have implicated integrins in the control of axonal growth (Reichardt and Tomaselli 1991). However, mere have been no previous studies examining the role of integrins in dendritic growth. In this study, we demonst-rate that antibodies specific for the β1 integrin subunit significantly inhibit the dendrite promoting activity of BME, suggesting a novel regulatory role for integrins in neuronal morphogenesis.
BME, an urea extract of the Engelbreth-Holm-Swarm (EHS) sarcoma tumor (Kleinman et al 1986), was purchased from Collaborative Research (trade name, Matrigel). RGD and related peptides were-purchased from Bachem, Inc. (Torrance, CA), A goat antiserum (anti-ECMR also known as anti-gp140) which was raised against purified adhesion-related, 140 kD, integral membrane glycoproteins from BHK cells (Knudsen et al 1981), was the generous gift of Dr. K.A. Knudsen (Lankenau Medical Research Center, Philadelphia, PA). The IgG fraction of rabbit antiserum recognizing the rat β1 integrin subunit (Gullberg et al 1989) was kindly provided by Dr. Kristofer Rubin (Uppsala University, Uppsala, Sweden). The monoclonal anti-body (mAb) designated 3A3 which specifically recognizes the 1β1 integrin complex (Turner et al. 1989; Ignatius et al., 1988; Tawil et al 1990) was donated by Dr. D.A. Turner (SUNY Health Science Center at Syracuse, Syracuse, NY). Mouse anti-human VLA4 mAb (clone HP2/1), which blocks the function of the 4 integrin subunit (Yednock et al, 1992) was purchased from AMAC, Inc. (Westbrook, ME).
Suspensions of neurons dissociated from the superior cervical ganglia of Holtzmann (Harlan Sprague-Dawley) rat fetuses (20-21 days) were prepared according to me method of Higgins et al (1991). Neurons were plated at densities ranging from 3 to 8 cells/mm2 onto glass coverslips (Bellco Glass, Inc., Vineland, NJ) precoated with poly-D-lysine (100 μg/ml; Sigma Chemical Co., St. Louis, MO), and maintained in a serum-free medium (Higgins et al. 1991). This medium contains a maximally effective concentration (100 ng/ml) of β-nerve growth factor that supports the long-term survival (≥ 2 months) of sympathetic neurons in culture (Bruckenstein and Higgins 1988). Cytosine-β-D-arabino-furanoside (1 μM) was added to the medium of all cultures on the second and third days; this exposure was usually sufficient to render them virtually free of nonneuronal cells for 30-40 days. To induce dendritic growth, neurons were typically exposed to BME (100 μg/ml) in the culture medium during days 3-8 in vitro.
Cellular morphology was routinely assessed by the intracellular injection of the fluorescent dye Lucifer yellow (4%). Only neurons whose cell bodies were at least 150 μM from other neuronal somata were injected because density-dependent changes in morphology occur when sympathetic neurons are separated by lesser distances (Bruckenstein and Higgins 1988). Processes whose length was less than the diameter of the cell body were not scored. A minimum of 30 randomly chosen neurons (≥ 10 neurons from three different cultures) were evaluated for each experimental condition. Each experiment was replicated in cultures obtained from at least two different dissect-ions. Data in the text are expressed as the mean ± S.E.M.
Cultures were immunostained with a mAb to the nonphosphorylated forms of the M and H neurofilament subunits (SMI32; Sternberger-Meyer Immunochemicals, Jarre-ttsville, MD). These cytoskeletal proteins have previously been shown to selectively localize to neuronal somata and dendrites (Sternber-ger and Sternberger 1983). Antigens were visualized using indirect immunofluorescence as previously described (Lein and Higgins 1989).
Sympathetic neurons were dissociated from the superior cervical ganglia of prenatal rats plated onto polylysine coated coverslips and maintained in a serum-free medium containing NGF. Nonneuronal cells were eliminated by treatment with an antimitotic agent on days 2 and 3; experimental treatments typic-ally began on day 5 or 6. Morphology was assessed on days 10 to 12 using intracellular dye injections. Dendrites were distinguished from axons using standard light microscopic criteria: dendrites were thicker (up to 5 μM at their base), tapered over their entire length, and ended locally, whereas axons were long and thin (< 1 μM) and of relatively constant diameter.
Sympathetic neurons grown under control conditions typically extend one or occasionally two processes (Table 1). Previous studies have shown that these processes have the cytoskeletal and ultrastructural characteristics of axons (Tropea et al 1988; Lein and Higgins 1989). In contrast, a 5 to 6 day exposure to BME (100 μg/ml) caused these neurons to extend multiple (2 to 4) dendrites in addition to a single axon (Table 1; Figs. 1A and 1B). To ascertain the role of integrins in dendritic growth, sympathetic neurons were exposed to BME in the presence of antibodies or peptides that inhibit integrin function. Anti-ECMR serum has been shown to inhibit integrin function in both nonneuronal and neuronal tissues (Tomaselli et al 1987; Sutherland et al 1988; Albelda et al 1989); and immunoprecipitation studies have demonstrated that this antisera recognizes integrins of the β1 family in rat sympametic neurons and PC12 cells (Tomaselli et al 1987; 1988). Neurons exposed to BME in the presence of anti-ECMR serum (1:500) were generally unipolar, having a single axonal process and no dendrites (Figs. 1C and D). A 5 day exposure to anti-ECMR serum also caused extensive fasciculation of axonal processes (Fig. 1C). Quantitative analyses of neuronal morphology indicated that anti-ECMR serum completely inhibited BME-induced dendritic growth but did not affect the number of axons (Table 1). Nonimmune serum at the same concentration had no significant effects on axonal or dendritic growth.
To ensure that the processes identified as dendrites by light microscopic criteria also exhibited the appropriate cytochemical characteristics, neurons grown under similar culture conditions were immunostained with dendrites of sympathetic neurons in situ (Carden et al 1987) and in vitro (Higgins et al 1991). Neurons exposed to BME in the absence of anti-ECMR serum typically exhibited intense staining of the cell body and dendritic processes with little or no staining of axonal processes (Fig. 2A). A similar pattern of immunoreactivity was observed in BME-treated neurons grown in the presence of nonimmune goat serum (Fig. 2B). In contrast, in cultures treated with both BME and anti-ECMR serum, immunoreactivity was confined to the cell body with few or no reactive processes (Fig. 2C).
To confirm results obtained with anti-ECMR serum, we also examined the effect of a well characterized antiserum raised against the purified rat β1 integrin subunit (Gullberg et al 1989). Sympathetic neurons grown in antibody (SMI 32) to the nonphosphorylated forms of the M and H neurofilament sub-units. These cytoskeletal proteins are localized primarily to the somata andthe presence of the IgG fraction of this rabbit anti-rat β1 serum (475 μg/ml) exhibited a 92% reduction in the percentage of neurons with dendritic growth, and a 94% reduction in me number of dendrites per neuron (Table 2). Nonimmune rabbit IgG at the same concentration had no effect on BME-induced dendritic growth (Table 2).
In an attempt to identify which integrin subunits were involved in the response of sympathetic neurons to BME, monoclonal antibody (mAb) 3A3, which specifically recognizes the rat 1β1 integrin complex (Ignatius et al 1988; Tawil et al 1990), was assessed for its effects on BME-induced dendritic growth. Neurons exposed imultaneously to BME and mAb 3A3 (10 μg/ml) responded with dendritic growth comparable to that of controls exposed to BME in the absence of mAb 3A3 (Table 2). Similar negative results were obtained with mAb to the 4 integrin subunit (not shown).
Many integrins recognize the tripeptide sequence, Arg-Gly-Asp (RGD) which is present in the cell-binding domains of their respective ECM ligand(s) (reviewed by Hynes 1992; Yamada and Kleinman 1992). Since several of the ECM proteins present in BME contain RGD sequences (Brazel et al 1987; 1988; Durkin et al 1988; Sasaki et al 1988; Mann et al 1989), we also tested the ability of soluble RGD peptides to disrupt BME-induced dendrite formation. Neither GRGDS, RGDS, nor a control peptide (RGES) had any inhibitory effects on dendritic growth elicited by BME (Table 3).
Previous studies have shown that integrins mediate the axon-promoting effects of many ECM molecules, including laminin, collagen IV, thrombospondin, and fibronectin (Lein et al 1991; Reichardt and Tomaselli 1991; Letourneau et al 1994). Our data suggest that integrins also regulate dendritic growth. Furthermore, it appears that members of the same integrin subfamily, the β1 integrins, are involved in both axonal and dendritic responses to ECM.
Anti-ECMR serum totally inhibited BME-induced dendritic growth in sympathetic neurons, as determined by bom dye injecti-ons and immunocytochemical analyses. Since previous studies have indicated that anti-ECMR serum recognizes β1 integrins in sympathetic neurons and in neuronal cell lines (Tomaselli et al 1987, 1988), these data were suggestive of a role for β1 integrins in the regulation of dendritic growth. However, this interpretation was subject to the caveat that anti-ECMR serum has been shown to cross react with ß3 integrins in certain nonneuronal cells (Albelda et al 1989). To address this issue, antibody perturbation experiments were performed using an antisera raised specifically against rat ß1 integrin. This anti-ß1 serum also markedly inhibited the dendrite-promoting activity of BME, providing strong support for the conclusion that ß1 integrin(s) play a significant role in dendritic growth.
In addition to the β1 integrin subunit, β3 and β8 integrin subunits have also been identified in neuronal cells (Venstrom and Reichardt 1993; 1995). However, since almost complete inhibition was observed with the anti-β1 serum, it suggests that β1 integrin(s) are the predominant receptor type mediating the dendrite-promoting effects of BME. This observation is consistent with previous inhibition studies of neuronal migration, axonal outgrowth and myelination in which β1 integrins have been found to be the dominant functional receptors mediating neuronal interactions with the ECM (Fernandez-Valle et al 1994; Letourneau et al 1994; Malek-Hedayat and Rome 1994; Reichardt and Tomaselli 1991). Thus, axonal and dendritic growth are similar in that diese processes both utilize integrins of the β1 subfamily. Since axons and dendrites grow at different rates, exhibit very different branching patterns, and follow different trajectories, it seems reasonable to hypothesize that these two types of processes interact with the ECM via different molecular mechanisms. The observation that the β1 integrin subunit is common to both axonal and dendritic outgrowth suggests mat such molecular differences may occur at the level of the integrin subunit.
Our data suggest mat neither the 1, 4, nor any of the subunits mat recognize the RGD binding site when complexed to β1 play a significant role in dendritic growth. Antibody 3A3, which specifically recognizes the 1β1 integrin complex (Tawil et al 1990), had no inhibitory effects on BME-induced dendritic growth when tested at concentrations previously shown to block neuritic outgrowth in sympathetic neurons plated on collagen IV (Lein et al 1991). Similarly, a mAb (HP2/1) known to block the binding of rat 4 integrins to both fibronectin and VCAM (Yednock et al 1992) did not inhibit the dendrite-promoting activity of BME. Soluble RGDS or GRGDS peptides used at concentrations that inhibit RGD-mediated binding to ECM molecules in intact cellular systems (Pierschbacher and Ruoslahti 1984; Yamada and Kennedy 1984; Herbst et al 1988) also had no effect on dendritic growth. β1 integrins that have been demonstrated to bind solely to the RGD motif include 5β1 and 5β1 (Hynes 1992); thus, our data would suggest that these specific integrin heteodimers are not involved in the regulation of dendritic growth. Additional integrin subunits known to be expressed by neurons and/or neuronal cell lines (see Venstrom and Reichardt 1993; 1995) that have not been tested in our experimental paradigm include 2, 3, 6, and 8. There is also the possibility that sympathetic neurons express a novel subunit on dendrites.
In summary, our data clearly indicate that integrin(s) of the β1 subfamily play a critical role in the development of dendrites. Since the β1 integrin subunit also figures prominently in interactions between axons and the ECM, we hypothesize that differences in axonal and dendritic responses to the ECM result from differential dendritic and axonal distribution of integrin subunits. Rigorous examination of this hypothesis will require the production of antibodies that inhibit the function of specific rat integrin integrin subunit(s).
This work was supported by a grant from the National Science Foundation (IBN 9319710) to D. Higgins and by a Predoctoral Graduate Research Fellowship (18-87-13) from the March of Dimes Birth Defects Foundation to P. Lein. We thank Ann Marie Hedges for her assistance with photographic work.