Filopodia Initiate Choices Made by Sensory Neuron Growth Cones at Laminin/Fibronectin Borders in Vitro

The Journal of Neuroscience, October 1994, 74(10): 59595972

Filopodia Initiate Choices Made by Sensory Neuron Growth Cones at Laminin/Fibronectin Borders in vitro

Timothy M. Gomez and Paul C. Letourneau The University of Minnesota, Department of Cell Biology and Neuroanatomy, Minneapolis, Minnesota 55455

Localized expression of environmental cues is thought to One mechanismproposed to guide growth conesin vivo is the provide directional information to migrating neuronal growth restricted or graded expressionof growth-influencing molecules cones by enhancing or suppressing axon outgrowth over that may produce pathways of preferred migration. Consistent limited regions. To investigate how such a mechanism may with this proposal is the immunohistochemical localization of function in viva, we observed growth cones of embryonic both growth-promoting and growth-inhibiting macromolecules chick dorsal root ganglion neurons at a substratum border along many neuronal migratory routes (Palm and Furcht, 1983; between the extracellular matrix components laminin and Rogers et al., 1986; Riggott and Moody, 1987; Harris, 1989; fibronectin in vitro using time-lapse phase-contrast and in- Snow et al., 1990b). Further evidence for local regulation of terference reflection microscopy. We found that patterns of growth cone behavior along neuronal pathways is the presence laminin and fibronectin could locally promote or suppress of “decision points” where growth cones make stereotyped the direction of growth cone migration. While migrating on choices among several alternative routes (Tosney and Land- either laminin or fibronectin, at least 79% of growth cones messer,1985b; Caudy and Bentley, 1986; Bovolenta and Ma- changed their rate and/or direction of outgrowth upon con- son, 1987; Sretavan and Reichardt, 1993). At these decision tact with the alternative substratum, in a manner suggesting regions, changesin direction of outgrowth are often accompa- that growth cones were selecting one substratum over the nied by changesin growth cone morphology and/or rate of mi- other. Complex changes in growth cone behavior were ini- gration. The presenceofpathways for neurite outgrowth is clear- tiated by filopodial contact with the alternate substratum, ly demonstrated in the developing grasshopperlimb and CNS, suggesting that filopodia were providing intracellular signals where identified growth cones display a strong preference for to the growth cone. Using interference reflection micros- migration along certain neuronal and non-neuronal surfaces copy, we have found that selection of a substratum is in- (Raper et al., 1983; Caudy and Bentley, 1986; Kolodkin et al., dependent of the degree of close contact to the substratum. 1992).In further support of a role for local regulation of pathway We conclude that spatially localized ECM components can selection, many tissue culture studieshave shown that growth direct axon outgrowth by mechanisms based on intracellular conescan respond in specific ways to local variations in their signaling through growth cone filopodia. environment (Harrison, 1914; Letoumeau, 1975; Bonhoeffer [Key words: laminin, fibronectin, filopodia, growth cone, and Huf, 1982;Gundersen, 1987; Snow and Letoumeau, 1992). interference reflection microscopy, guidance] Several models have been presentedto explain growth cone steeringboth in vivo and at a substratum border in vitro. One The local environment of the neuronal growth cone is critical model proposesthat differential adhesionof filopodia may con- for directing neurite outgrowth. Extracellular matrix (ECM) gly- trol the direction of neurite outgrowth (Letoumeau, 1981; Bray, coproteins (Sanes,1989; Letoumeau et al., 1992) membrane- 1982). In this model, filopodia adhere to the substratum and associatedglycoproteins (Rutishauseret al., 1983; Changet al., exert tension, pulling the growth cone toward an areaof highest 1987; Lagenaurand Lemmon, 1987; Caroni and Schwab, 1988; adhesivity. Other models proposethat the filopodia are a scaf- Matsunaga et al., 1988) solublefactors (Gundersenand Barrett, fold to support veil extension, and subsequentasymmetric re- 1979; Lumsden and Davies, 1986; Placzek et al., 1990;O’Leary gression(Burmeister and Goldberg, 1988)or production (Gold- et al., 1991), and proteoglycans (Carbonetto et al., 1983; Tosney berg and Burmeister, 1986) of new areasof growth cone control and Landmesser,1985a; Stem and Keynes, 1987; Snow et al., in the direction of neurite outgrowth. Alternatively, studiesof 1990a; Tosney and Oakley, 1990;Cole and McCabe, I99 1)have growth cone guidance of pioneer neurons in grasshopperem- been shown to affect growth cone migration either positively or bryos have shownsteering is mediatedby the dilation of a single negatively and are likely to function in combination in vivo. filopodium in contact with a guidepost cell (O’Connor et al., 1990) in a processdependent on the accumulation of actin Received Dec. 3 1, 1993; revised Mar. 18, 1994; accepted Mar. 29, 1994. filaments (O’Connor and Bentley, 1993) and selectiveinvasion We thank Drs. Diane Snow and Maureen Condic for their insight during ex- of microtubules (Sabry et al., 1991). perimentation, and for critical comments on the manuscript. Portions ofthis work Consistent with each model is the initial exploration of the were presented in abstract form at the 199 I and 1992 annual meetings of the environment by filopodia. Recent reports indicate that these Society for Neuroscience. This research was supported by National Institutes of Health Predoctoral Training Grants EY07133 (T.M.G.) and HD19950 (T.M.G. exploratory filopodia, which greatly enhancethe samplingarea and P.C.L.). of a growth cone, are not only adhesive structures, but may Correspondence should be addressed to Timothy M. Gomez, University of Minnesota, Department of Cell Biology and Neuroanatomy, 32 1 Church Street, function as a sensoryeffector systemby expressingsecond mes- SE, 4-135 Jackson Hall, Minneapolis, MN 55455. senger-linked cell surface receptors (Davenport et al., 1993; Copyright 0 1994 Society for Neuroscience 0270-6474/94/145959-14$05.00/O O’Connor and Bentley, 1993). In this model, filopodial contact 5960 Gomez and Letourneau - Behaviors of Growth Cones at LM/FN Borders with environmental cues generates receptor-based second mes- entire coverslip and incubated for at least 6 hr at room temperature. senger cascades, which ultimately regulate cytoskeletal dynam- For some experiments the entire coverslip was subsequently blocked with 10 mg/ml Hgb. In most experiments, the first protein solution was ics and direct neurite outgrowth. Many growth-promoting mol- used at 50 pg/rnl and the second protein solution at 10 &ml. The order ecules have recently been shown to alter second messenger that LM and FN were applied to the coverslip was interchanged regu- systems (Bixby, 1989; Schuch et al., 1989; Doherty et al., 1991, larly. For control studies, LM/LM and FN/FN patterns were produced 1993; Davenport and Kater, 1992). in the same manner as described above, except equal concentrations of The glycoproteins laminin (LM) and fibronectin (FN) are per- protein were used in the first and second coatings. Characterization of the patterned substrata. The integrity of the LMI missive substrata for the development of the peripheral nervous FN patterns was tested both immunofluorescently and radiographically system, and the differential distribution of these two molecules (immunofluorescent labeling produced a more intense staining of LM in the periphery indicates possible roles in axon guidance (Rog- and FN than did the fluorochrome-conjugated forms of LM and FN, ers et al., 1986; Westerfield, 1987). Each of these molecules has allowing us to better visualize the patterns). For immunofluorescence, patterned substrata were fixed for 20 min with 4% paraformaldehyde/ multiple domains involved in cell attachment and/or neurite 0.4% glutaraldehyde in Ca*+/Mg*+-free PBS (CMF-PBS), pH 7.4, at promotion (Reichardt and Tomaselli, 199 1; Letourneau et al., 37°C. After fixation, the coverslips were rinsed with PBS and the sub- 1992). Numerous putative receptors for both LM and FN cell stratum patterns were immunofluorescently labeled by the following binding domains have been described, the most common of protocol. First, the patterns were blocked with 10% normal goat serum which are members of the pl family of integrins (reviewed in (NGS; GIBCO Laboratories, Grand Island, NY) in PBS for 30 min. The blocking solution was subsequently removed, and the following Buck and Horwitz, 1987; Hynes, 1992). LM and FN possess polyclonal primary antibodies were applied for 60 min: rat anti-LM, multiple binding sites for distinct integrins (Reichardt and To- 1: 100, and rabbit anti-M, 1: 100 (gjft of Dr. Sally Palm, University of maselli, 199 1). Motor neuron growth cones, which grow well Minnesota) in 10% NGS. The cov&slips were washed three times with on LM, are unable to extend on FN (Rogers et al., 1983) and PBS followed bv a second 30 min. 10% NGS block. Coverslios were then incubated with 1: 100 dilutions’of rhodamine-conjugated gdat anti- this inability may be one factor controlling guidance to their rat and fluorescein-conjugated goat anti-rabbit secondary antibodies peripheral targets (Westerfield, 1987). Sensory neuron growth (Cappel Laboratories, West Chester, PA) in 10% NGS for 60 min. The cones, on the other hand, extend neurites on both LM and FN coverslips were washed a final time with PBS and mounted over second and therefore may respond differently than motor axons to path- coverslips, using a carbonate-buffered glycerol medium containing a ways consisting of a patterned distribution of these two mole- reducing agent (p-phenylenediamine; Sigma Chemical Co.) to retard photobleaching. cules. The amount of LM and FN that binds to both blocked and unblocked In this report, we examine growth cones of chick dorsal root glass coverslips was determined radiographically using 3H-conjugated ganglion (DRG) neurons at a border between LM and FN, using forms of LM and FN, prepared in our laboratory according to the both phase-contrast and interference reflection microscopy methods of Jentoft (Jentoft and Dearborn, 1979) and Herbst (Herbst et al., 1988). The amount of 3H-LM and ‘H-FN binding to 12 mm round (IRM). The purpose of this study is to assess the behavior of coverslips was quantitated at each step in the substratum patterning growth cones at the point of initial contact with a new substra- protocol described above. Therefore, we measured the amount of 3H- tum, when the interactions of receptors and associated intra- LM and ‘H-FN binding to clean glass, glass saturated with LM or FN, cellular systems are changing. We examine the kinetics of neurite and glass subsequently blocked with 10 mg/ml Hgb. outgrowth, growth cone morphology, and directional changes Cell culture. Dorsal root ganglia (DRG) of the lumbar-sacral enlarge- ment were dissected from embryonic day 8-l 1 white Leghorn chicken in neurite outgrowth at a border between LM and FN. In ad- embryos and dissociated with 0.25% crude trypsin in CMF-PBS 7.89, dition, the role of filopodia in initiating changes in growth cone as described previously (Luduefia, 1973). Dissociated cells were sus- behavior is analyzed. IRM is used to address the role of close pended in serum-free medium consisting of 10 mM HEPES-buffered substratum contact in the behavioral responses seen at the LM/ F14 medium (GIBCO Laboratories, Grand Island, NY) supplemented with 2 mM L-glutamate, 5 @ml sodium selenite, 5 r&ml sodium py- FN border. ruvate, 5 mM phosphocreatine, 20 nM progesterone, 5 pg/ml insulin, 100 &ml transferrin (all from Sigma Chemical Co., St. Louis, MO), Materials and Methods and 15-50 rig/ml 2.5 S nerve growth factor (R&D systems, Minneapolis, Patterned substrata preparation. Acid-washed glasscoverslips were MN). An antibiotic/antimycotic solution (Sigma Chemical Co., St. Lou- mountedover 22 mm holes drilled into the bottom of 35 mm tissue is, MO) was also added. Approximately 1.2-l .5 x 1O4 cells were seeded culturedishes. Each dish wasUV sterilizedfor l-2 hr. Patternscon- onto each patterned coverslip in 1.O ml ofgrowth medium and incubated sistingof alternatingstripes of LM and FN wereprepared in the fol- in a humidified air chamber at 40°C for at least 6 hr. lowing manner.Strips of silicone2 mm x 20 mm were cut from a Videomicroscopy. Following incubation, a culture dish was placed on 2-mm-thick sheetof siliconeelastomer (Dow Corning93-500 Space an inverted microscope (IM-35, Carl Zeiss, Inc., Thomwood, NY, or gradeencapsulant, Dow ComingCorp., Midland, MI), sterilized in 70% Diaphot, Nikon Inc., Garden City, NY) under an air curtain incubator ethanol,and placedonto a glasscoverslip, creating’s tight seal.EHS (AS1 400, Carl Zeiss, Inc.), keeping the medium at a constant 40°C. tumorLM (generousgift of SallvPalm. Universitv of Minnesota1(Palm Growth cones were monitored by either phase-contrast optics, using a andFurcht;1983) an? FN, isoiatedf&m human-plasma(gene& gift Newvicon video camera (NC-65, Dage-MTI, Inc., Michigan City, IN), of Jim McCarthy, University of Minnesota) (McCarthy et al., 1986), or interference reflection microscopy (IRM), using an SIT camera (SIT- were diluted in PBS (pH 7.1) to a concentration of 10-50 &ml. To 66, Dage-MTI, Inc.). IRM was done on the IM-35 microscope using a locate the border between substratum proteins, 2.5-5% of the appro- 50 W HBO eoiilluminator. 546 ?nm orecision filter. Zeiss H-D reflector. priate fluorochrome-conjugated form of LM or FN was added (rho- pre- and postobjective pol&izers, 2 &housing, ani a Zeiss 63 x antiflex damine-LM or fluorescein-FN) to one or both solutions (conjugated by objective. To prevent UV photodamage, growth cones were exposed method of Harlow and Lane, 1988). Two hundred and fifty microliters intermittingly to light using a shutter (Uniblitz D122, Vincent Associ- of either LM or FN solution were applied to the exposed spaces of the ates, Rochester, NY). Immunofluorescence microscopy was performed coverslips around the silicone strips and incubated for 2-3 hr at room on the Diaphot microscope using a Zeiss 100 x Neofluor objective and temperature, then suctioned off, and the coverslips washed three times a Paultek ICCD camera (Paultek Imaging, Nevada City, CA). Image with PBS. To saturate any remaining protein binding sites, the coverslips enhancement and morphometric measurements were made using IMAGE were blocked with 10 mg/ml hemoglobin (Hgb) (Sigma Chemical Co., 1 software (Universal Imaging, Inc., West Chester, PA) run on a 486/ St. Louis, MO) in PBS for 2-3 hr at room temnerature. The Heb was 33 computer system (Gateway 2000, North Sioux City, SD). Images suctioned off, and the coverslips were again washed extensiveh with viewed with a video monitor (Trinitron, Sony Corp. of America, New PBS prior to the removal of the silicone strips. Without allowing the York, NY) were recorded every 30 set with an optical disc recorder plate to dry, 500 pl of the alternate protein solution was applied to the (TQ-2026F, Panasonic Industrial Comp., Secaucus, NJ). The Journal of Neuroscience, October 1994, 14(10) 5961

Figure 1. An LM/FN patternthat was immunofluoreseentlylabeled for both LM (TRITC, A) andPN (FITC, B). LM andFN stainingare both adjacentand nonoverlapping.

Quantitativeanalysis. Growth coneswere scored for (1) crossingor perpendicular to it were recorded prior to, during, and after turningat the borderbetween LM andFN, (2)angle ofneuritic approach, contact with the alternative substratum. When using low-mag- and(3) rate of growthcone migration, according to the followingset of criteria: (1) a growthcone must have been observed from a positionat nification optics, a field containing neuronson both sidesof the least20 pm away from the borderand followedfor sufficienttime to substratum border was recorded for up to 24 hr. We analyzed assessits behaviorafter contactwith the border(typically growthcones growth cones migrating toward FN while elongating on LM wereobserved until theyreached a point atleast 20 pm across the border, (“LM-to-FN growth cones”), as well as growth conesmigrating or until they becamealigned parallel to the border);(2) growthcones toward LM while elongatingon FN (“IN-to-LM growth cones”). mustnot have contactedany interferingcells within 20 pm on either sideof the borderduring the entireprocess of crossingor turning;(3) Our data comprise observationsof over 100 growth conesat the anglea neuritemakes with respectto the borderwas measured from an LM/FN border. Nearly all growth conesat a border between the tip of the growthcone to a point 20 pm downthe neuriticshaft; (4) LM and FN exhibited behavioral changesupon contact with for growthcones that undergoan obviousbehavioral change upon con- the new substratum. At the LM/FN border, growth cones ex- tact with the alternativesubstratum, the rate andangle ofapproach were hibited changesin direction and rate of outgrowth, shape, fil- measuredprior to the beginningand after the completionof the change in behavior;(5) for thosegrowth cones that did not undergoany obvious opodial turnover, and axonal branching. Changesin rate and changein behavior,the rateand angle of approachwere measured before direction of outgrowth werethe mostconsistent, and were there- and after the growth conehad migratedwithin 10 pm of the border; fore quantified (below). As a control for any physical effects of and(6) whena branchoccurred that wasgreater than 20 pm from the a border, growth cone behaviors at an LM/FN border were border,each separate neurite was measured using the abovecriteria. The averagepixel intensity of the digitizedIRM imageswas quan- compared to responsesseen at borders between LM and LM, titatedusing IMAGE 1 software.The peripheryofa growthcone (filopodia or FN and FN (Fig. 2A,B). Importantly, growth cone behaviors not included)was traced by handto the proximalpoint where the growth were not influenced by the order that the proteins were applied coneconsolidated into neurite. to the coverslip (seeMaterials and Methods). Results Change in direction and rate of migration are correlated at an Prior to behavioral analysis,we examined the LM/FN patterns LM/FN border in two ways. First, by double immunofluorescentlabeling of the LM-to-FN growth cones.Sixty-two percent of the LM-to-FN substrata, we found the LM and FN stripes to be both adjacent growth cones(n = 74) crossedonto FN, 27% turned to remain and nonoverlapping (Fig. 1). Second, using tritiated forms of associatedwith LN, and 11% stopped at the border with FN LM and FN, we detected < 10% binding of the overlay protein for the duration of the observation period (averageobservation onto a substratum that had been previously saturated with LM time at border, 4 hr). For comparison, 97% of the LM-to-LM or FN and subsequentlyblocked with 10 mg/ml Hgb (Table 1). growth conescrossed the border (n = 30). Directional and elon- Theseresults demonstratethat we are able to generatepatterned gation rate changesfor 63 of the 74 growth conesobserved are substratasuitable for testing a growth cone’sresponse to a border summarized in Figure 2C. Seventy-eight percent of LM-to-F’N between different growth-promoting molecules. growth cones underwent significant behavioral changesupon We analyzed growth cone behaviors at an LM/FN border contact with FN, defined as a greater than 50% increase or between 6 and 36 hr in culture, using high- and low-magnifi- decreasein rate of migration and/or a greater than 20” change cation phase-contrastmicroscopy as well as IRM. We identified in direction of outgrowth. Only 12% of the LM-to-LM growth the border betweenLM and FN by preparingpatterned substrata conesunderwent changesin behavior by the samecriteria (n = with solutions that contained 5% fluorescent conjugatesof LM 26). and/or FN. For high-magnification studies, growth cones lo- The changesin direction and rate of migration exhibited by cated between 20 and 40 pm from a border and oriented nearly LM-to-FN growth coneswere closely correlated, as indicated

Table 1. Binding of 3H-FN and “H-LM is inhibited by prior treatment of substratum with LM or FN and Hgb

V-MI 6d-N WI Wml) 0 50 0 50 No block 100 25 f 0.1 100 33 ZlI5.3 10mg/ml Hgb 5.3 It 1.4 4.0 -t 0.3 10 + 1.2 6.0 + 0.9 (n = 6) Data are binding of 10 &ml )H-FN or )H-LM to blocked and unblocked glass coverslips. Twelve millimeter round coverslips were completely covered with protein solutions as described for pattern production in Materials and Methods. The amount of ‘H-FN and ‘H-LM that bound to the coverslips was counted at each step in the process. Numbers are expressed as percentage of binding to uncoated glass (+SEM). N = 6 for all ‘H-FN measurements and n = 3 for all ‘H- LM measurements, except where indicated. 5962 Gomez and Letourneau - Behaviors of Growth Cones at LM/FN Borders

Figure 2. Three-dimensional plots correlating changes in the rate and direction of growth cone migration. Rate changes are expressed as percentage change from the initial rate (deceleration expressed as negative); observations were binned and plotted along the x-axis (left). Changes in the direction of migration were binned and plotted on the y-axis (rig&). Growth cones that oriented to take a more direct path across the border were expressed as a positive value and growth cones that turned away or took a more indirect path across the border were expressed as a negative value. The number of observations appears on the z-axis. The darkened urea includes the limit of rate and direction changes we have defined as nonsignificant. A, LM-to-LM control. B, FN-to-F’N control. C, LM-to-FN growth cones. D, FN-to-LM growth cones. In the control plots (A and B), most of the observations fall within the center of the graph, indicating only moderate changes in behavior occurred. On the other hand, the LM-to-FN (C) and FW-to-LM (D) plots show that these growth cones underwent more significant changes in behavior upon contact with the alternative substratum. In addition, the observations fall into two opposing quadrants, indicating that directional and migration rate changes are linked; that is, growth cones that reorient toward the border tend to increase their rate of migration, whereas growth cones that turn away from the border tend to decrease their rate of migration. by clustering of observations into two opposing quadrants in filopodial contacts (Fig. 3F, arrows) were made with FN, and Figure 2C, that is, growth cones contacting FN either accelerate these filopodia appeared to fuse and rapidly expand into a new and orient more directly toward FN (positive response), or de- leading edge of the now reoriented growth cone. The rate of celerate and turn away from FN (negative response). Upon con- neurite outgrowth is plotted in Figure 3.J. On LM (Fig. 3,4-F), tact with FN, 37% of LM-to-FN growth cones underwent sig- this growth cone is predominately filopodial with few lamelli- nificant positive behavioral changes, whereas 4 1% underwent podia, whereas on FN (Fig. 3H,I) the growth cone appears flat- significant negative behavioral changes. An example of a pos- tencd and has fewer filopodia. During the process of orienting itive response of a growth cone to FN is shown in Figure 3. A toward FN, this growth cone produced a small branch on LM. few distant and transient filopodial contacts with FN were ob- Turning at a border occurred by either reorientation of the served early (Fig. 3B, arrow). Fifty minutes later several new primary growth cone or by branching. Figure 4 illustrates a

Figure 3. Time-lapse sequence of a growth cone that crosses from LM (below) onto FN (above). A-Z, The substratum border is indicated by arrowheads in all frames. Images are 10 min apart. The first observed filopodial contact with FN is at 10 min (B, arrow). This filopodial contact is retracted and the growth cone continues to migrate at the same rate (see .I) and direction until several new filopodial contacts with FN are made (F, black arrows). These filopodia appear to rapidly fuse to form a nascent neurite (F, G, white arrows), at the tip of which a new growth cone forms. The growth cone accelerates and expands as it crosses onto FN (G-Z). A branched growth cone develops on LM from a portion of the primary growth cone (F, G, open arrows). Once fully across the border, the growth cone decelerates and adopts a Battened morphology (I). Scale bar, 10 pm. J, Rate of migration of the growth cone in A-Z. Letters correspond with the images above. This growth cone increased its rate of migration from 0.39 Fm/min (points prior to position F) to 1.2 1 pm/min (points after position P’) upon contact with FN. 40 60 80 Minutes J 100’

80-

60-

40-

100 150 200 250 Minutes The Journal of Neuroscience, October 1994, 14(10) 5965 typical response of a growth cone turning away from FN. This Response of growth cones encountering a border between LM growth cone made extremely long filopodial contacts with FN and FN at decreasing angles of contact (earliest contacts observed prior to Fig. 4B, not shown) and In an earlier report, we suggested that the angle at which a growth showed initial signs of turning when the body of the growth cone contacted a border between LM and FN affected the be- cone was greater than 40 pm away from the substratum border havior of an elongating neurite (Gomez and Letoumeau, 199 1). (Fig. 4B, arrow). As this growth cone approached the border it By examining fixed cultures, we found that for neurites that had underwent transient decreases in its rate of migration (Fig. 4J, turned at a border, the portion of the neurite proximal to the arrows) which were often associated with periods of apparently border was more likely to make a ~40” angle with the border. heightened filopodial sampling of FN (Fig. 40,G). The periods Conversely, for neurites that had crossed a border, the portion of heightened filopodial sampling were followed by further de- of the neurite proximal to the border was more likely to make viation away from FW and an increased rate of neurite out- a >40” angle with the border. These results suggested that elon- growth. This growth cone eventually oriented parallel to the gating neurites have a greater probability of crossing a border borderline, while maintaining filopodial contact with FN. if they contact the border at angles approaching perpendicular. FN-to-LM growth cones. Many FN-to-LM growth cones un- This conclusion, however, was not supported by video records derwent behavioral changes similar to those of the LM-to-FN of living growth cones. Many neurites that contacted a border growth cones. Fifty-nine percent of FN-to-LM growth cones (n at a near perpendicular angle did not cross the border, but turned = 66) crossed onto LM, 32% turned at the border to remain as much as 90” to remain associated with the same substratum. associated with FN, and 9% stopped at the border with LM for Upon further examination of the video records we confirmed the remainder of the observation period (average observation that the angle measurements made of fixed neurites did not time at border, 6.5 hr). For comparison, 94% of the FN-to-FN reliably reflect the actual angle at which a growth cone had growth cones crossed the border (n = 18). Directional and elon- encountered a border. In other words, the angle of initial contact gation rate changes in response to contact with LM for 50 of of a growth cone with a border is not necessarily maintained as the 66 growth cones observed are summarized in Figure 20. the growth cone continues to migrate. Using the same criteria described above, 80% of the FN-to-LM In order to better understand how growth cones approaching growth cones compared to 26% of the FN-to-FN growth cones LM/F’N borders at different angles respond to a substratum underwent significant changes in behavior upon contact with change, we correlated the angle of approach with the crossing the substratum border. As with the LM-to-FN growth cones, or turning behavior of more than 120 growth cones (Table 2). changes in rate and direction of outgrowth were correlated, as Both LM-to-FN and FN-to-LM growth cones turned at the indicated by the clustering of observations into two opposing border with the alternate substratum regardless of the angle of quadrants in Figure 20. Upon contact with LM, 28% of growth approach (average angle of approach, 6 1” and 64”, respectively). cones underwent significant positive behavioral changes, where- Along with those that turned, 14 growth cones stopped at the as 52% underwent significant negative behavioral changes. border for the remainder of the observation period (data not included in Table 2). Although we saw fewer neurites approach Changes in growth cone behavior are initiated byjilopodia the border at the most indirect angles (n = 19 at 5 30”), among The behavioral changes exhibited by the growth cones in Figures the growth cones observed we saw no correlation between the 3 and 4 occurred after only filopodial contact with FN, when angle of approach and the probability of turning. the bodies of these growth cones were on LM, many micrometers from the substratum border. Analysis of the high-magnification records indicates that the LM-to-FN growth cones that crossed Changes in growth cone-substratum close contacts do not onto FN altered their behavior at an average distance of 7 It_ predict the substratum of preferred outgrowth 1.5 pm from the border (n = 14). LM-to-m growth cones that The extent to which a growth cone adheres to different ECM turned at the border with FN began turning when the body of molecules may be one factor that influences its pathway selection the growth cone was at an average of 13 f 4 km from the border (Letoumeau, 1975). To test this hypothesis, we examined growth (n = 11). Similarly, for FN-to-LM growth cones, the average cones at the border between LN and FN using time-lapse in- distance to the border at the point the behavioral change began terference reflection microscopy (IRM). As previously shown was 6 + 1.3 pm for those that crossed onto LM (n = 17) and (Harris, 1973; Izzard and Lochner, 1976; Letoumeau, 1979), 13 & 3.1 pm for those that turned away from LM (n = 9). These the dark regions of an IRM image represent areas of close mem- results show that filopodia initiate behavioral changes by the brane apposition and greater adherence to the underlying sub- growth cone, suggesting that filopodia may be transducing en- stratum, whereas lighter areas represent regions of greater sep- vironmental cues into intracellular signals. aration and reduced adherence. c Figure 4. Time-lapse sequence of a growth cone on LM (above) that turned at the border with FN (below). A-Z, The substratum border is indicated by arrowheads in C-I (the border is below the field in A andB). Framesare 30 min apart,except for I andH, whichare 10min apart.The field of view hasbeen shifted many times to keepthe growthcone in view. A, Forward-projectingfilopodia extend greater than 40 pm from the distal extent of the growthcone body andcontact FN (not seen).B, Newfilopodia and veil protrudelaterally from the growthcone (arrow), whichresults in a slight turn to a less direct approach toward the border (compare the angle the growth cone makes with respect to the x-axis in A vs C). The rate of migration of this growth cone decreased (see .T) at times of apparent increased filopodial sampling of FN in D and G (arrows), after which the growth cone turned further (E, I). In G (arrow) a veil extends toward FN, but it is later retracted (H). Scale bar, 10 pm. J, Rateof migration of the growth conein A-Z. Letterscorrespond with the images above. Overall this growth cone underwent only a small decrease in its rate of migration (0.54 rmlmin prior to point D. 0.41 pm/min after point D). However, for brief periods larger drops in this growth cone’s rate of migration (arrows) occurred and these rate changes were associated with periods of heightened filopodial sampling of FW (0, G). 5966 Gomez and Letourneau - Behaviors of Growth Cones at LMlFN Borders

IV J” I II n:::: m 40 0 ;:;: ;.;: - :::: :::: -.-. :.:. u&l 30

1::: .:.: :::: 20 :::: :::: -.*. D :::: :::: :::: -20 86 4 ::::IU 1.1.*.a. -10 I 1.:.

-30 ’ ; = ;’ I I I I I I I I I I ' 0 0 8 16 24 32 40 48 56 64 72 80 88 Minutes

Figure 5. Time-lapse IRM sequence of a growth cone that crosses from LM (upper left) onto FN (lower right). A-F, The substratum border in this sequence projects vertically (arrowheads indicate border) at the point this growth cone crosses it. Images are approximately 15 min apart. After The Journal of Neuroscience, October 1994, 14(10) 5967

Table 2. Crossing/turning response of growth cones approaching a substratum border at different angles of approach

LM-to-LM FN-to-l3 LM-to-m FN-to-LM Angle Cross Turn Cross Turn Cross Turn Cross Turn 10” 1 0 0 0 3 0 1 1 20" 1 1 0 0 4 2 0 1 30" 1 0 1 0 3 2 1 1 40" 1 0 1 0 1 1 5 5 50” 3 0 2 0 I 2 4 1 60" I 0 3 1 4 2 7 3 IO” 5 0 4 0 5 6 5 0 80” 8 0 3 0 9 0 5 5 90” 2 0 4 0 10 5 11 4 Totals 29 1 17 1 46 20 39 21

The angle that a growth cone approached and contacted a border did not influence its behavior. Numbers indicate the number of growth cones observed to cross or turn at a border at a given angle of approach. Growth cones that stopped at the border are not included in this data set. On LM-to-LM and FN-to-FW control patterns, a large majority of growth cones crossed the border at all angles of approach. On the other hand, 30% of the LM-to-FN and 35% of the FN-to- LM growth cones turned at the border with the alternative substratum, at various angles of approach. However, no consistent significant difference in the percentage of growth cones that turned at a border was seen at decreasing angles of approach.

Our results confirm a previous report that DRG growth cones contact can be sufficient to initiate changesin growth cone be- express closer contacts when migrating on FN than on LM haviors (seeFig. 3) close filopodial-substratum apposition is (Gundersen, 1988). Our new finding using time-lapse IRM is not sufficient to reorient the growth cone. Our resultsusing IRM that contacts of a singlegrowth conechange rapidly as the growth suggestthat there is no direct correlation betweenthe degreeof cone passesfrom one substratum onto another, such that a substratum contact and choice of substratum. growth cone in contact with both substrata can show two patterns of contact. FN-to-LM growth conesthat orient toward and Discussion accelerate onto LM simultaneously lose many close contacts, Previous studiesof the effects of patterned substrataon neurite acquiring an IRM pattern typical for migration on LM. Con- outgrowth (Hammarback et al., 1985;Gundersen, 1987; Walter versely, as shown in Figure 5, LM-to-IN growth cones that et al., 1987; Lemmon et al., 1992) did not assessthe detailed orient toward and accelerateonto FN increasetheir total area behavior of growth conesand filopodia at substratum borders. of closecontact and acquire an IRM pattern typical of migration In the present study, we analyzed the behaviors of individual on FN. Note that the changein the IRM pattern is coincident dorsal root ganglion neuron growth cones at a choice point with the LM/FN border, when this growth cone is spanningthe betweentwo growth-promoting substrata,LM and FN. We draw border between LM and FN (Fig. 5E). Quantitative analysis of three conclusionsfrom this study. First, growth conesreadily the average pixel intensity of IRM images from the sequence detect a change in ECM substrata, and respond in a manner presentedin Figure 5A-F revealsboth a decreasein the average that we interpret as an active choice to either remain on their pixel intensity and an increasedrate of migration as this growth original substratum,or to crossonto the alternative substratum. cone crossesonto FN (Fig. 5G). Growth conescrossing from Second, changesin growth cone behavior are initiated by filo- FN onto LM showed an average increasein pixel intensity of podia that extend forward and contact the alternate substratum. 16 k 5% (n = 5) while those crossingfrom LM onto FN de- Third, selection of one substratum over the other is not corre- creasedby an average of 11 + 2% (n = 5). lated with the degreeof growth cone-substratum contact. Growth conesthat turn at a border with FN or LM typically make many filopodial contacts on both substrata.The filopodial Growth cone behavior changesat an LM/FN border contacts seenover both LM and FN often appear similar under We found that 79% of all DRG growth cones changed their IRM optics, although differencesmay be difficult to resolve. In behavior upon encountering an LM/FN border. To distinguish one instance, however, we observed an LM-to-IN growth cone this behavior from random or stochasticchanges (Katz et al., that turned at the border with FN, despite making many very 1984) or from potential physical effects of the borders, we an- stable and close filopodial contacts onto the FN-treated sub- alyzed growth cone behavior at FN/FN and LM/LM borders. strata (Fig. 6). This observation indicates that while filopodial Only 26% of growth cones at an FN/FN and 12% of growth

c severalfilopodial contacts with F’N(B, C, arrows),this growthcone orients toward the border(C, D) andaccelerates (see G) as it crossesonto the F%treated surface. Whenthis growthcone was spanning the border,it sometimesexhibited two patternsof closecontacts, which changed at points coincidentwith the border(E). This growthcone continued to migrateonto FN, at an elevatedrate of outgrowth,and altered morphology. Scale bar, 5 pm. G, Rateof migration(line) and average IRM imageintensity change (bar) of the growthcone in A-F. Letters correspondwith the images above.The averagepixel intensityof growthcone images was normalized and presented as the differencebetween subsequent image measurements. At the point this growthcone crosses the borderonto FN, it simultaneouslyincreases its rate of migrationand extent of closecontact (lower image intensity). 5969 Gomez and Letourneau * Behaviors of Growth Cones at LM/FN Borders The Journal of Neuroscience, October 1994, 14(10) 5969 cones at an LM/LM border exhibited behavioral changes that DRG neurons and their growth cones may also become het- we defined as significant. Thus, we conclude that most DRG erogeneous, as the result of interactions with components of our growth cones respond to a change in the ECM substratum, in vitro system. For example, we add NGF but not other neu- whether they are extending originally on FN or LM. In addition, rotrophins to our culture media. Since DRG neurons express there is an important correlation between changes in growth different arrays of trk receptors (Carroll et al., 1992) this could cone orientation and rate of migration (Fig. 2). Growth cones lead to differential induction of cellular factors that influence that crossed a border increased their rate of migration, while neurite outgrowth on LN or FN. In addition, DRG growth cones growth cones that turned away from a border decreased their may contact other neurons, Schwann cells, fibroblasts, and ECM rate of migration. Less then 2% of the growth cones exhibited components and these interactions may modulate the expres- opposing behaviors (e.g., oriented toward and decelerated, or sion or activity of growth cone surface components (reviewed turned and accelerated). Importantly, changes in growth cone in Damsky and Werb, 1992; Ginsberg et al., 1992; Hynes, 1992; behavior were not brief responses, but rather they were sustained Schweighoffer and Shaw, 1992). changes that significantly altered growth cone migration. Our results indicate that local discontinuities of two ECM Growth cone steering at a border between LM and FN oc- components can direct DRG neurite outgrowth and can affect curred by several different processes, all of which have been growth cone velocity and morphology. This is in contradiction observed in other systems. In most cases, growth cones altered to previous studies suggesting that ECM components (reviewed their direction by localized changes in protrusion of cytoplasm in Reichardt and Tomaselli, 199 1) and cell adhesion molecules toward filopodia and veil (e.g., Fig. 4B), as has been described (Lemmon et al., 1992) are permissive for neurite outgrowth, in detail for growth cone advance in vitro (Bray and Chapman, but do not provide instructive information to direct neurite 1985; Goldberg and Burmeister, 1986) and in vivo (O’Connor outgrowth. Our findings also show that DRG growth cones can et al., 1990). Less frequently, we observed growth cone steering respond differently to the same change in ECM substratum, by the selective regression of a veil (Fig. 4G), a process that suggesting that patterns of LM and FN could function as in- occurred only when growth cones turned away from a border. structive pathways in vivo. Further, many of the behavioral The selective regression of veils and growth cones has been changes we observed at LM/FN borders have been reported at described previously on patterned substrata (Burmeister and characteristic positions along many neuronal migratory path- Goldberg, 1988) and upon contact of motor growth cones with ways in vivo (Raper et al., 1983; Tosney and Landmesser, 1985b; sclerotome cells in vitro (Oakley and Tosney, 1993). Lastly, Caudy and Bentley, 1986; Eisen et al., 1986; Godement et al., growth cone steering occasionally resulted from the rapid ex- 1990). Therefore, differentially distributed ECM molecules may pansion of a single filopodium into a nascent neurite. This pro- account for some of the behavioral changes observed in vivo. cess, similar to that described in the grasshopper limb (O’Con- nor et al., 1990) occurred only when filopodia contacted a FilopodiaJirst detect a change in EC44 substratum preferred substratum. This process is shown in Figure 3, where Filopodia have been shown to play an important navigational two filopodia apparently fused (Fig. 3F,G) to form a nascent role for growth cones both in vitro (Kapthammer et al., 1986; neurite, at the tip of which a new growth cone formed. Bandtlow et al., 1990; Letoumeau et al., 1990, 1991) and in Although most DRG growth cones detect a substratum change, vivo (Raper et al., 1983; Bastiani et al., 1984; Bentley and To- and choose to migrate on LM or FN, our results do not indicate roian-Raymond, 1986; O’Connor et al., 1990; Letoumeau et that DRG growth cones comprise a single population that pre- al., 1991; Chien et al., 1993). As a growth cone approaches a fers FN or LM. This is unlike other studies, where the behavior border, filopodia extend as much as 50 Km, and interact with of migrating cells or growth cones was interpreted as a consis- the alternate ECM molecule (Figs. 3-6). Such interactions pro- tently observed preference for one substratum versus another duced behavioral changes in the parent growth cone while it (Hammarback et al., 1985; Gundersen, 1987; Walter et al., 1987; was still on its original substratum. At high magnification, we Calof and Lander, 199 1). Perhaps cultured DRG neurons com- measured the distance between the growth cone body and the prise a heterogeneous population, including some neurons with substratum border, when a change in behavior was initiated. neurites that prefer FN at an LM/FN border, and other neurons Growth cones that crossed from LM onto FN or vice versa whose growth cones prefer LM. Such preferences may be in- initiated the behavior change (an increased rate of migration trinsic to cultured DRG neurons, or the differences may arise and/or a reorientation toward the border) at an average distance as a result of the in vitro conditions. Subsets of DRG growth of 7 pm from the border, whereas growth cones that turned to cones may express distinct combinations of cell surface recep- remain on their original substratum began turning at an average tors. Differential expression of receptors for adhesion molecules distance of 13 Irrn from the border. Importantly, filopodial con- and trophic factors may mediate growth cone migration along tact with the alternative substratum always preceded a change the different pathways taken by the central and peripheral axons in growth cone behavior. It is unknown why growth cones that of each DRG neuron (Mu et al., 1993). Additional evidence turn at a border initiate behavior changes at a greater distance suggests that cutaneous afferents and muscle afferents have dis- than growth cones that cross a border; however, it is possible tinct pathfinding abilities (Lance-Jones and Landmesser, 1981; that directional changes are initiated more quickly than rate Scott, 1988; Maisonpierre et al., 1990; LoPresti and Scott, 1992). changes. DRG neurons may also differ because the expression of surface How could filopodial contacts change growth cone behavior? receptors may change from the time of growth cone formation, One possibility is that filopodia have ECM receptors that are through pathfinding, and, finally, after reaching a target (Hall et activated by binding to LM or FN and generate transmembrane al., 1987; Cohen et al., 1989; Lefcort et al., 1992). Since DRG signals that pass to the growth cone to change motile behavior. neurons are not synchronized in these events at ElO, when they Several receptors for ECM and cell adhesion molecules have are cultured, these differences may contribute to heterogeneity been localized to growth cones, including some that are con- of DRG growth cones. centrated on filopodia (Van den Pol et al., 1986; Letoumeau 5970 Gomez and Letourneau - Behaviors of Growth Cones at LM/FN Borders and Shattuck, 1989). The integrin family of heterodimeric cell adhesive molecule, FN, and a less adhesive molecule, LM, we surface receptors are a particularly important class of ECM re- have found that selection of one substratum over another does ceptors. Integrins containing the pl subunit, some ofwhich bind not necessarily involve achieving the highest degree of growth LM and FN (reviewed in Reichardt and Tomaselli, 199 1; Hynes, cone-to-substratum contact. This conclusion is based on two 1992; Letourneau et al., 1992; Reichardt, 1992), are expressed behaviors. (1) Some growth cones oriented toward and accel- by DRG growth cones and are often concentrated at the tips of erated onto FN, coming into closer contact with the substratum filopodia (Letourneau and Shattuck, 1989). Although at least (Fig. 5), as indicated by an 11 -t 2% decrease in the average four different /31-containing integrins have been implicated in pixel intensity of the IRM images. Other growth cones oriented DRG growth cone migration on LM and FN (Reichardt, 1992; toward and accelerated onto LM, decreasing contact with the Tomaselli et al., 1993), these receptor subtypes have not been substratum, indicated by a 16 f 5% increase in the average shown to be differentially expressed by subpopulations ofgrowth pixel intensity of the IRM images. (2) Growth cones turned at cones or filopodia. the border with FN to remain on LM (Fig. 6), despite having Recent pharmacological findings support a sensory role for many close filopodial and lamellipodial contacts with FN. adhesive interactions of growth cone filopodia. Several cell ad- Neuronal growth cones (Fig. 5E), as well as non-neuronal hesion molecules and ECM binding integrin receptors have been cells (i.e., Schwann cells, fibroblasts, and neuroblastoma cells, shown to promote neurite outgrowth by activating second mes- not shown), spanning an LM/FN border exhibited a precise senger systems that affect cytoskeletal polymerization and cell change in the pattern and degree of close contact, indicating that behavior (reviewed in Damsky and Werb, 1992; Hynes, 1992; as a growth cone migrates onto a new substratum it concomi- Hynes and Lander, 1992). In particular, integrin activation has tantly reorganizes its associations with that substratum. Perhaps been implicated in the fluctuation of intracellular messengers along with local differences in the involvement of adhesive re- such as calcium (Jaconi et al., 199 l), CAMP (Nathan and San- ceptors, there are local differences in transmembrane signaling, chez, 1990), and pH (Schwartz et al., 1989, 199 l), as well as which underlie the ability of filopodia contacts to regulate growth protein kinase C activation (Bixby, 1989) and tyrosine phos- cone steering. phorylation (Guan et al., 199 1; Kornberg et al., 1991). There- fore, activation of integrin receptors on the tips of filopodia Conclusions could produce changes in diffusible second messengers that af- Presented with a choice between LM and FN, DRG growth fect the body of a growth cone. Recently, growth cone filopodia cones exhibit behavioral changes indicating that they prefer one were directly shown to possess the necessary signal transduction substratum over another. Further, DRG growth cones appear mechanisms to allow autonomous responses to environmental heterogeneous in their preference for LM or FN, indicating that stimuli (Davenport et al., 1993). subpopulations of DRG neurons may exist. Behavioral changes How may filopodia-based intracellular signaling result in began after filopodial contact with the alternative ECM com- growth cone steering? Perhaps forward-projecting filopodia that ponent, indicating that filopodia provide sensory input that reg- cross a border transmit new signals to the growth cone, while ulates growth cone migration. Finally, using IRM we conclude more lateral filopodia continue to interact with the original sub- that substratum preference is not dependent on achieving the stratum. In an environment of spatially segregated extracellular highest degree of growth cone-substratum close contact. Taken ligands, brief asymmetries of intracellular signals may arise within together, this study presents evidence that localized distribution the growth cone. Thus, a few filopodia, interacting with a new of ECM components can direct DRG neurite outgrowth in a substratum, may produce intracellular signals that are sufficient- manner that may be determined by the receptors and associated ly strong or amplified, so as to dominate the continued signals second messenger systems expressed by the growth cone and its received from the lateral filopodia. It would be interesting to filopodia. know whether receptor distribution and/or transmembrane signals are different for forward-projecting versus lateral-projecting References filopodia. Bandtlow C, Zachleder T, Schwab ME (1990) Oligodendrocytes arrest neurite outgrowth by contact inhibition. J Neurosci 10:3837-3848. Bastiani M J, Raper JA, Goodman CS ( 1984) Pathfinding by neuronal Does adhesion play a role in growth cone steering? growth cones in grasshopper embryos. III. Selective affi&iy of the G growth cone for the P cells within the A/P fascicle. J Neurosci 4:23 1l- It has been suggested that guidance of growth cones could be 2328. due to local variations in substratum adhesivity, possibly acting BentleyD, Toroian-RaymondA (1986) Disorientedpathfinding by to stabilize protrusions and provide anchorage for mechano- pioneerneurone growth cones deprived of filopodiaby cytochalasin chemical force (Harris, 1973; Letourneau, 1975, 1979; Ham- treatment.Nature 323:7 12-7 15. Bixby JL (1989) Proteinkinase C is involved in lamininstimulation marback et al., 1988;. O’Connor et al., 1990). However, other of neuriteoutgrowth. Neuron 3:287-297. studies indicate that growth cone-to-substratum adhesion does BonhoefferF, Huf J (1982) In vitro experimentson axon guidance not determine substratum preference (Gundersen, 1987, 1988; demonstratingan anterior-posteriorgradient on the tectum.EMBO Calof and Lander, 199 1; Lemmon et al., 1992). Our results using J 1:427431. time-lapse IRM argue that receptors for LM and FN do not BovolentaP, MasonCA (1987) Growth conemorphology varies with positionin the developingmouse visual pathway from retinato first function equally in mediating growth cone-substratum inter- targets.J Neurosci7:1447-1460. actions. Instead, it appears that the adhesive nature of receptors Bray D (1982) Filopodialcontraction and growthcone guidance. In: may be only one factor in the control of growth cone migration. Cellbehavior (Bellairs R, CurtisA, DunnG, eds),pp 299-317. Cam- IRM is a reliable indicator of cell-substratum adhesion (Har- bridge:Cambridge UP. Bray D, ChapmanK (1985) Analysisof microspikemovements on ris, 1973; Izzard and Lochner, 1976; Gundersen, 1988; Calof the neuronalgrowth cone. J Neurosci5:3204-32 13. and Lander, 199 1; Lemmon et al., 1992). By observing growth Buck CA, Horwitz AF (1987) Cellsurface receptors for extracellular cones using time-lapse IRM at the border between a highly matrix molecules. Annu Rev Cell Biol 3: 179-205. The Journal of Neuroscience, October 1994, 14(10) 5971

Burmeister DW, Goldberg DJ (1988) Micropruning: the mechanism changes and effects of the cell substratum attachment antibody (CSAT). of turning of Aplysia growth cones at substrate borders in vitro. J J Cell Biol 104:623-634. Neurosci 83151-3159. Hammarback JA, Palm SL, Furcht LT, Letoumeau PC (1985) Guid- Calof AL, Lander AD (199 1) Relationship between neuronal migra- ance of neurite outgrowth by pathways of substratum-adsorbed lam- tion and cell-substratum adhesion: laminin and merosin promote inin. J Neurosci Res 13:213-220. olfactory neuronal migration but are anti-adhesive. J Cell Biol 115: Hammarback JA, McCarthy JB, Palm SL, Furcht LT, Letoumeau PC 779-794. (1988) Growth cone guidance by substrate-bound laminin pathways Carbonetto S, Gruver MM, Turner DC (1983) Nerve fiber growth in is correlated with neuron-to-oathwav adhesivitv. Dev Biol 126:29- culture on fibronectin, collagen and glycosaminoglycan substrates. J 39. Neurosci 3:2324-2335. Harlow E, Lane D (1988) Antibodies: a laboratory manual. Cold Spring Caroni P, Schwab ME (1988) Antibody against myelin-associated in- Harbor, NY: Cold Spring Harbor Laboratory. hibitor of neurite growth neutralizes nonpermissive substrate prop- Harris WA (1973) Location of cellular adhesions to solid substrata. erties of CNS white matter. Neuron 1:85-96. Dev Biol 35:83-96. Carroll SL, Silos-Santiago I, Frese SE, Ruit KG, Milbrandt J, Snider Harris WA (1989) Local positional cues in the neuroepithelium guide WD (1992) Dorsal root ganglion neurons expressing trk are selec- retinal axons in embrvonic Xenoaus brain. Nature 339:2 18-22 1. tively sensitive to NGF deprivation in utero. Neuron 9:779-788. Harrison RW (1914) The reaction of embryonic cells to solid struc- Caudy M, Bentley D (1986) Pioneer growth cone steering along a series tures. J Exp Zoo1 17:521-544. of neuronal and non-neuronal cues of different affinities. J Neurosci Herbst TJ. McCarthv JB. Tsilibarv EC. Furcht LT (1988) Differential 6:1781-1795. effects of laminin: intact type IV collagen, and specific domains of Chang S, Rathjen FG, Raper JA (1987) Extension of neurites on axons type IV collagen on endothelial cell adhesion and migration. J Cell is impaired by antibodies against specific neural cell surface glyco- Biol 106: 1365-l 373. proteins. J Cell Biol 104:355-362. Hynes RO (1992) Integrins: versatility, modulation, and signaling in Chien C, Rosenthal DE, Harris WA, Holt CE (1993) Navigational cell adhesion. Cell 69: 1 l-25. errors made by growth cones without filopodia in the embryonic Hynes RO, Lander AD (1992) Contact and adhesive specificities in Xenopus brain. Neuron 11:237-25 1. the associations, migrations, and targeting of cells and axons. Cell 68: Cohen J, Nurcombe V, Jeffrey P, Edgar D (1989) Developmental loss 303-322. of functional laminin receptors on retinal ganglion cells is regulated Izzard CS, Lochner LR (1976) Cell-to-substrate contacts in living by their target tissue, the optic tectum. Development 107:381-387. fibroblasts: an interference reflexion study with an evaluation of the Cole GJ, McCabe CF (199 1) Identification of a developmentally reg- technique. J Cell Sci 2 1: 129-l 59. ulated keratan sulfate proteoglycan that inhibits cell adhesion and Jaconi MEE, Theler JM, Schlegel W, Appel RD, Wright SD, Lew PD neurite outgrowth. Neuron 7:1007-1018. (199 1) Multiple elevations of cytosolic-free Ca2+ in human neutro- Damskv CH. Werb Z (1992) Sianal transduction bv intearin recenters phils: initiation by adherence receptors of the integrin family. J Cell for extracellular matrix: cooperative processing of ext;acellular in- Biol 112:1249-1257. formation. Curr Opin Cell Biol 4:772-78 1. Jentoft N, Dearborn D (1979) Labeling of proteins by reductive meth- Davenport RW, Kater SB (1992) Local increases in intracellular cal- vlation usina sodium cvanoborohvdride. J Biol Chem 25443594365. cium elicit local filopodial responses in Helisoma neuronal growth KaplhammerjP, Grunewald BE, Raper JA (1986) The selective in- cones. Neuron 9:405416. hibition of growth cone advancement of specific neurites in culture. Davenport RW, Dou P, Rehder V, Kater SB (1993) A sensory role J Neurosci 6~2527-2534. for neuronal growth cone filopodia. Science 36 1:I2 l-724. Katz MJ, George EB, Gilbert LJ (1984) Axonal elongation as a sto- Doherty P, Ashton SV, Moore SE, Walsh FS (199 1) Morphoregulatory chastic. walk. Cell Motil 4:35 l-370. activities of NCAM and N-cadherin can be accounted for by G pro- Kolodkin AL. Matthes DJ. O’Connor TP. Pate1 NH. Admon A. Bentlev tein-dependent activation of L- and N-type neuronal Caz+ channels. D, Goodman CS (1992) Fasciclin IV: sequence, expression, and Cell 67:21-33. function during growth cone guidance in the grasshopper embryo. Neuron 9:831-845. Doherty P, Singh A, Rimon G, Bolsover SR, Walsh FS (1993) Thy- 1 Komberg LJ, Earp HS, Turner CE, Prockop C, Juliano RL (1991) antibody-triggered neurite outgrowth requires an influx of calcium Signal transduction by integrins: increased protein tyrosine phos- into neurons via N- and L-tvneI_ calcium channels. J Cell Biol 122: phorylation caused by clustering of 0, integrins. Proc Nat1 Acad Sci 181-189. USA 88:8392-8396. Eisen JS, Myers PZ, Westerheld M (1986) Pathway selection by growth Lagenaur C, Lemmon V (1987) An Ll -like molecule, the 8D9 antigen, cones of identified motoneurones in live zebra fish embryos. Nature is a potent substrate for neurite extension. Proc Nat1 Acad Sci USA 320~269-271. 84:7753-7757. Ginsberg MH, Du X, Plow EF (1992) Inside-out integrin signalling. Lance-Jones C, Landmesser L (198 1) Pathway selection by chick lum- Curr Opin Cell Biol 4:766-77 1. bosacral motoneurons during normal development. Proc R Sot Lond Godement P, Salaun J, Mason CA (1990) Retinal axon pathfinding [Biol] 214:1-18. in the optic chiasm: divergence of crossed and uncrossed fibers. Neu- Lefcort F, Venstrom K, McDonald JA, Reichardt LF (1992) Regu- ron 5:173-186. lation of expression of fibronectin and its receptor, a5@1, during de- Goldberg DJ, Burmeister DW (1986) Stages in axon formation: ob- velopment and regeneration of peripheral nerve. Development 116: servations of Aplysia axons in culture using video-enhanced contrast- 767-782. differential interference contrast microscopy. J Cell Biol 103: 192 l- Lemmon V, Burden SM, Payne HR, Elmslie GJ, Hlavin ML (1992) 1931. Neurite growth on different substrates: permissive versus instructive Gomez TM, Letoumeau PC (199 1) Sot Neurosci Abstr 17. influences and the role of adhesive strength. J Neurosci 12:8 18-826. Guan JL, Trevithick JE, Hynes RO (1991) Fibronectin/integrin in- Letoumeau PC (1975) Cell-to-substratum adhesion and guidance of teraction induces tyrosine phosphorylation of a 120 kDa protein. Cell axonal elongation. Dev Biol 44:92-101. Regul 2:95 l-964. Letoumeau PC (1979) Cell-substratum adhesion of neurite growth Gundersen RW (1987) Response of sensory neurites and growth cones cones, and its role in neurite elongation. Exp Cell Res 124:127-138. to patterned substrata of laminin and fibronectin in vitro. Dev Biol Letoumeau PC (198 1) Immunocytochemical evidence for colocali- 121:4231131. zation in neurite growth cones of actin and myosin and their rela- Gundersen RW (1988) Interference reflection microscopic study of tionship to cell-substratum adhesions. Dev Biol 85: 113-122. dorsal root growth cones on different substrates: assessment ofgrowth Letoumeau PC, Shattuck TA (1989) Distribution and possible inter- cone-substrate contacts. J Neurosci Res 21:298-306. actions of actin-associated proteins and cell adhesion molecules of Gundersen RW, Barrett JN (1979) Neuronal chemotaxis: chick dorsal- nerve growth cones. Development 105:505-519. root axons turn toward high concentrations of nerve growth factor Letoumeau PC, Shattuck TA, Roche FK, Takeichi M, Lemmon V (1990) Science 206:1079-1080. Nerve growth cone migration onto Schwann cells involves the cal- Hall DE, Neugebauer KM, Reichardt LF (1987) Embryonic neural cium-denendent adhesion molecule. N-cadherin. Dev Biol 138:430- retinal cell response to extracellular matrix proteins: developmental 442. - 5972 Gomez and Letourneau * Behaviors of Growth Cones at LM/FN Borders

Letoumeau PC, Kater SB, Macagno ER, eds (199 1) The nerve growth Rogers SL, Edson KJ, Letourneau PC, McLoon SC (1986) Distribution cone. New York: Raven. of laminin in the developing peripheral nervous system of the chick. Letoumeau PC, Condic ML, Snow DM (1992) Extracellular matrix Development 113:429435. and neurite outgrowth. Curr Opin Gen Dev 2:625-634. Rutishauser U, Grumet M, Edelman GM (1983) Neural cell adhesion LoPresti P, Scott SA (1992) Cutaneous and muscle afferents in vitro: molecule mediates initial interactions between spinal cord neurons different requirements for survival factors. Sot Neurosci Abstr 18: and muscle cells in culture. J Cell Biol 97: 145-l 52. 919. Sabry JH, O’Connor TP, Evans L, Toroian RA, Kirschner M, Bentley Ludueiia MA (1973) The growth of spinal ganglion neurons in serum- D (199 1) Microtubule behavior during guidance of pioneer neuron free medium. Dev Biol 33:470-476. growth cones in situ. J Cell Biol 115:38 l-395. Lumsden AG, Davies AM (1986) Chemotropic effect of specific target Sanes JR (1989) Extracellular matrix molecules that influence neural epithelium in the developing mammalian nervous system. Nature development. Annu Rev Neurosci 12:49 l-5 16. 323:538-539. Schuch U, Lohse MJ, Schachner M (1989) Neural cell adhesion mol- Maisonpietre PC, Belluscio L, Squint0 S, Ip NY, Furth ME, Lindsay ecules influence second messenger systems. Neuron 3: 13-20. RM, Yancopoulos GD (1990) Neurotrophin-3: a neurotrophic fac- Schwartz MA, Both G, Lechene C (1989) Effect of cell spreading on tor related to NGF and BDNF. Science 247: 1446-l 45 1. cytoplasmic pH in normal and transformed fibroblasts. Proc Nat1 Matsunaga M, Hatta K, Nagafuchi A, Takeichi M (1988) Guidance Acad Sci USA 86:4525-4529. of optic nerve fibres by N-cadherin adhesion molecules. Nature 334: Schwartz MA, Ingber DE, Lawrence M, Springer TA, Lechene C (199 1) 62-64. Multiple integrins share the ability to induce elevation of intracellular McCarthy JB, Hagen ST, Furcht LT (1986) Human fibronectin con- pH. Exp Cell Res 195:533-535. tains distinct adhesion- and motility-promoting domains for meta- Schweighoffer T, Shaw S (1992) Adhesion cascades: diversity through static melanoma cells. J Cell Biol 102: 179-l 88. combinatorial strategies. Curr Opin Cell Biol 4:824-829. Mu X, Silos SI, Carroll SL, Snider WD (1993) Neurotrophin receptor Scott SA (1988) Skin sensory innervation patterns in embryonic chick genes are expressed in distinct patterns in developing dorsal root hindlimbs deprived of motorneurons. Dev Biol 126:362-374. ganglia. J Neurosci 13:40294041. Snow DM, Letoumeau PC (1992) Neurite outgrowth on a step gradient Nathan C, Sanchez E ( 1990) Tumor necrosis factor and CD 1l/CD 18 of chondroitin sulfate proteoglycan (CS-PG). J Neurobiol 23:322- (0,) integrins act synergistically to lower CAMP in human neutrophils. 336. J Cell Biol 111:2171-2181. Snow DM, Steindler DA, Silver J (1990a) Molecular and cellular Oakley RA, Tosney KW (1993) Contact-mediated mechanisms of characterization of the glial roof plate of the spinal cord and optic motor axon segmentation. J Neurosci 13:3773-3792. tectum: a possible role for a proteoglycan in the development of an O’Connor TP, Bentley D (1993) Accumulation of actin in subsets of axon barrier. Dev Biol 138:359-376. pioneer growth cone filopodia in response to neural and epithelial Snow DM, Lemmon V, Canino DA, Caplan AI, Silver J (1990b) Sul- guidance cues in situ. J Cell Biol 123:935-948. fated proteoglycans in astroglial barriers inhibit neurite outgrowth in O’Connor TP, Duerr JS, Bentley D (1990) Pioneer growth cone steer- vitro. Exp Neurol 109: 11 l-130. ing decisions mediated by single filopodial contact in situ. J Neurosci Sretavan DW, Reichardt LF (1993) Time-lapse video analysis of ret- 10:3935-3946. inal ganglion cell axon pathfinding at the mammalian optic chiasm: O’Leary DD, Helliier CD, Kutka L, Lopez ML, Missias A, Reinoso BS growth cone guidance using intrinsic chiasm cues. Neuron 10:761- (199 1) A target-derived chemoattractant controls the development 777. of the corticopontine proiection by a novel mechanism of axon tar- Stem CD, Keynes RJ (1987) Interactions between somite cells: the geting. Development [Suppl] 2: 123-l 30. formation and maintenance of segment boundaries in the chick em- Palm SL. Furcht LT ( 1983) Production of laminin and fibronectin bv bryo. Development 99:26 l-272. Schwannoma cells: ‘cell-protein interactions in vitro and production Tomaselli KJ, Doherty P, Emmett CJ, Damsky CH, Walsh FS, Rei- localization in peripheral nerve in vivo. J Cell Biol 96: 12 18-1226. chardt LF (1993) Expression of 01 integrins in sensory neurons of Placzek M, Tessier LM, Jesse11T, Dodd J (1990) Orientation of com- the dorsal root ganglion and their function in neurite outgrowth on missural axons in vitro in response to a floor platederived che- two laminin isoforms. J Neurosci 13:4880-4888. moattractant. Development 110: 19-30. Tosney KW, Landmesser LT (1985a) Development of the major path- Raper JA, Bastiani M, Goodman CS (1983) Pathfinding by neuronal ways for neurite outgrowth in the chick hindlimb. Dev Biol 109: 193- growth cones in grasshopper embryos. I. Divergent choices made by 214. the growth cones of sibling neurons. J Neurosci 3:20-30. Tosney KW, Landmesser LT (1985b) Specificity of early motoneuron Reichardt LF (1992) Adhesive interactions that regulate neuronal be- growth cone outgrowth in the chick embryo. J Neurosci 5:2336-2344. havior. In: Sensory neurons: diversity, development, and plasticity Tosney K, Oakley R (1990) The perinotochordal mesenchyme acts as (Scott SA, ed), pp 2 15-24 1. New York: Oxford UP. a barrier to axon advance in the chick embryo: implications for a Reichardt LF, Tomaselli KJ (199 1) Extracellular matrix molecules general mechanism of axonal guidance. Exp Neurol 109:75-89. and their receptors: functions in neural development. Annu Rev Neu- Van den Pol AN, di Porizio U, Rutishauser U (1986) Growth cone rosci 14:531-570. localization of neural cell adhesion molecule on central nervous sys- Riggott MJ, Moody SA (1987) Distribution oflaminin and fibronectin tem neurons in vitro. J Cell Biol 102:228 l-2294. along peripheral trigeminal axon pathways in the developing chick. Walter J, Henke FS, Bonhoeffer F (1987) Avoidance of posterior tectal J Comp Neurol258:580-596. membranes by temporal retinal axons. Development 10 1:909-9 13. Rogers SL, Letoumeau PC, Palm SL, McCarthy J, Furcht LT (1983) Westerfield M (1987) Substrate interactions affecting motor growth Neurite extension by peripheral and central nervous system neurons cone guidance during development and regeneration. J Exp Biol 132: in response to substratum-bound fibronectin and laminin. Dev Biol 161-175. 98:2 12-220.