Aging-related changes and topology of adhesion responses sensitive to cycloheximide on collagen substrata by human dermal fibroblasts

EXPERIMENTAL

CELL

RESEARCH

186,158-168

(1990)

Aging-Related Changes and Topology of Adhesion Responses Sensitive to Cycloheximide on Collagen Substrata by Human Dermal Fibroblasts KELLY Department

of Molecular

Biology

S. FLICKINGER’

and Microbiology,

AND

Case Western

Reserve

LLOYD

A.

University,

CULP School

of Medicine,

Cleveland,

Ohio 44106

INTRODUCTION Human dermal fibroblasts (both papillary and reticular) were tested during in vivo or in vitro aging for their responses on collagen and/or fibronectin (FN) substrata, as well as on topologically mixed substrata. Cycloheximide treatments were used to evaluate whether receptors to these matrix molecules, mediating F-actin reorganization into stress fibers, are altered during aging processes. Late-passage (but not mid-passage) papillary and reticular cells from both an elderly male and a newborn infant spread effectively on collagen + FN but failed to generate stress fibers after lengthy pretreatment of cells with cycloheximide. In contrast, treatment with cycloheximide only when cells were reattaching was not inhibitory. None of the treatments had any effect on stress fiber formation of cells on FN-only substrata, demonstrating that drug sensitivity was specific for collagen responses. The inhibition could be reversed by rinsing cycloheximide out of cultures and could be prevented by prior growth of cells in ascorbate-supplemented medium to stimulate production/maturation of collagen and possibly collagen-specific receptors. Adjacent regions of coverslips were adsorbed with collagen and a proteolytic fragment of plasma FN lacking the collagen-binding domain but retaining other binding domains; cells bridging the interface were of special interest. When fragment F155 containing both the RGDS cell-binding and the heparinn-binding domains was tested in this paradigm, cells generated stress fibers continuous from the collagenfacing side into the Fl55-facing side of the same cell, consistent with the compatability of both collagen and FN receptors in generating the same stress fiber. However, Fl 10 lacking the heparinrI domain was incapable of facilitating stress fiber formation; fibers formed effectively on the collagen side and terminated abruptly at the collagen:FllO interface. These experiments demonstrate stringent regulation of where stress fibers begin, span, and terminate in the cytoplasm by matrix receptors at the cell’s undersurface and establish that there are alterations of collagen-specific receptors as a consequence of in vitro aging, but not of in vivo aging, in both papillary and reticular human dermal fibroblasts. 0 1990 Academic Press, Inc.

Our understanding of the molecular mechanisms by which fibroblasts from various animal sources adhere and respond to extracellular matrices has benefitted from complementary experimental approaches. For example, the structure and multiple binding activities of plasma and cellular fibronectins (pFN and cFN, respectively) are being analyzed, based on the premise that FNs mediate, at least in part, fibroblast responsesto collagen matrices [l-3]. A complementary approach involves biochemical and molecular biological analyses of receptors for matrix molecules [4-61. These receptors must communicate with a complex intracellular apparatus in order to achieve cell type-specific responses from cytoskeletal and other networks [ 71. There is evidence that some cells respond to collagen matrices by mechanisms independent of FN mediation. Some classesof CHO cell mutants were found defective in their adhesion processes on FN substrata but were found competent in adhesion on collagen [8]. This defect was subsequently linked to an altered protein kinase in the variant cells [9] and its effects upon a member of the integrin glycoprotein class on the cell surface analyzed [lo]. Alternatively, a BHK cell variant, defective in adhesion on FN, produced an altered heparan sulfate proteoglycan [ll], consistent with previous evidence that this membrane-intercalated class of proteoglycan plays some role in FN adhesion processes, including communication with the microfilament network in fibroblasts [ 12-141. Small dermatan sulfate proteoglycans inhibit fibroblast adhesion on FN substrata by interacting with the proteoglycan-binding domains in FNs but do not inhibit adhesion on collagen (even with the collagen postadsorbed with pFN) [15]. Contraction of three-dimensional collagen matrices by fibroblasts could be resolved into multiple phases [16] and shown to be independent of FN participation [17]. Such studies ultimately led to the identification of integrin glycoprotein receptors, containing (I! and ,8 subunits, that were specific for collagen and that failed to recognize FNs [18,191. ’ To whom

0014-4827/90 $3.00 Copyright 0 1990 by Academic Press, Inc. All rights of reproduction in any form reserved.

158

reprint

requests

should

be addressed.

ADHESION

PROCESSES

SENSITIVE

This laboratory has been investigating mechanisms by which human dermal fibroblasts, either as full-thickness populations or as separate reticular and papillary subpopulations [20], interact with defined matrices, as well as possible alterations of these processes during in vitro or in viuo aging of cells. Beyth and Culp [al] identified differences in the distributions of glycosaminoglycans in the substratum-attached material of full-thickness skin fibroblasts from a newborn infant or an elderly male, as well as from cells grown with ascorbate supplementation of the medium case’) to maximize collagen production/maturation. Analyses of proteoglycans in substratum-attached material confirmed a relative increase in heparan sulfate proteoglycans for in vitro-aged papillary and reticular cells and in the cells from an elderly male [22]. The functional consequences of these biochemical differences were then tested in adhesion experiments. Full-thickness dermal fibroblasts responded very differently to substrata coated, either separately or as mixtures, with a proteolytic fragment of pFN containing the RGDS (Arg-Gly-Asp-Ser)-dependent cell-binding domain or with the heparan sulfate-binding protein, platelet factor-4 [23] 9 ifferenees were also observed between asc+ or asc- ce in these experiments. Reticular cells were then tested with a panel of proteolytic fragments of pFNs and cFNs [24]. Fragment 155, retaining the RGDS cell-binding and heparinii-binding domains, provided the minimal information for complete F-actin stress fiber formation in fibroblasts from both a newborn infant and an elderly man [241. These anaiyses have now been extended to comparisons of the responses of papillary and reticular cells to with pFN alone, type I collagen alone, or collagen postadsorbed with pFN using in vitro- or in uivo-aged cells. The use of cycloheximide in several different protocols to inhibit protein synthesis has revealed functional differences in the collagen-specific adhesion responses during in uitro aging processes, as well as some interesting aspects of receptor topology in these responses. MATERIALS

AND

METHODS

Cells and growth cor,dttions. Human dermal fibroblasts for these studies have been described [21,22,24]. Separate populations of papillary and reticular dermal fibroblasts were obtained from the inner, upper aspect of the arm of two individuals-a newborn male infant (patient 5: abbreviated PAPS and RET, cells, respectively) and a 78year-old male (patient 8: abbreviated PAPs and RETs cells). All cells were free of Iviycophsma contamination and were grown in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with penicillin (250 units/ml), streptomycin (0.25 mg/ml), and 10% newborn calf serum in 10% CO,:humidified air at 37°C. In some cases, the medium was also supplemented with 50 fig/ml sodium ascorbate twice a week to maximize endogenous collagen production for at least two passages prior to and during experimental use 1231. Stock cultures were split at a 3:1 ratio. Patient 5 cells routinely began senescence at passages 233

TO

CYCLOHEXIMIDE

24 (60-65 population doublings) while patient 8 cells senesced at passages 18-19 (45-50 population doublings) [22]. Because of some cell death during subculturing and an increasing proportion ofnondividing cells in senescingpopulations, there is some discrepancy between population doubling and passage levels. Adhesion assay. Cells in stock cultures were grown as described above or, alternatively, pretreated with 2 pug/ml cycloheximide as described below. For experiments [23,24], fibroblast cultures at 6Q-80% confluence were rinsed three times with phosphate-buffered saline (PBS) and cells detached by addition of 0.5 mlkl EGTA in PBS by gentle shaking at 37°C for 30 min and a final pipetting of the cell suspension over the surface of the flask to suspend loosely adhering cells. The cells were pelleted by centrifugation, resuspended in adhesion medium (250 pg/ml heat-treated bovine serum albumin ic DMEM), and repelleted by centrifugation (repeated twice). The Bnaj. pellet of cells was resuspended in adhesion medium and 2.5 X lo4 c~ells inoculated into the wells of 24.well cluster dishes containing glass coverslips. For quantitation of attachment, cells were radioiabeled by [3H]thymidine incorporation as described previously [24]. The glass coverslips in wells had been coated with adhesion-promoting proteins whose purification and evaluation have been described previously [12, 151: (a) 100 pg/ml of rat tail type I collagen allowed to dry overnight at room temperature; (b) 100 @g/ml type I collagen overnight, followed by three PBS rinses and adsorption of 20 pg/ml human plasma fibronectin (pFN) for I h at 37°C; (c) 20 pg/ml human pFN for 1 h at 37°C; and (d) 250 fig/ml beat-treated bovine albumin at 37°C for 1 h (in all cases, cells failed to attach to this substratum). All coated substrata were rinsed three times with PBS; adhesion medium was added for 1 b at 37°C to guarantee coverage of any potential binding sites on the substratum with bovine albumin, and then cells were inoculated at the required concentration. Cells were incubated for 4 h (or in some cases an additional 14 b) and then assayed for their responses. In some cases, glass coverslips were coated in topologically separate regions with two different adhesive proteins [25]. One-half of the cower&p was dipped into a collagen solution (100 ,ug/ml) for I h at room temperature. After rinsing with PBS, the coverslip was inverted and the remaining uncoated half was dipped into a solution containing pFN or one of its proteolytic fragments. Thermoly§~~-venerated fragments of human pFN were generously donated by Dr. Luciano Zardi of Genoa, Italy, and have been shown to saturate the substratum with the concentrations used here [26]. F155 is derived speciically from the fl chain of pFN and spans the Arg-Gly-Asp-Ser (RGDS) ceil-binding domain and the heparinii-binding domain (but not the collagen-binding domain or the alternately spliced IIICS region of FNs); FllO is a further cleavage product of F155 containing the RGDS domair but not the Hepi, domain. Fluorescence assays. F-actin organization in cells was evaluated by staining with rbodamine-phalloidin 1243. Briefly, ceils on coverslips were fixed with 3.7% (v/v) paraformaldehyde in PBS (plus 1 rniW CaC& and MgQ) for 20 min at room temperature and permeabilized with 1 ml of 0.2% Triton X-100 in PBS (plus 1 mMCaC1, and MgC1.J for 1 min at room temperature. They were then treated with 250 ;ul of 2.0 units/ml rhodamine-phalloidin in PBS for 20 min at room temperature, rinsed three times with PBS, and mounted inverted on microscope slides in 50% (v/v) glycerol/PBS. For detection of FN or its proteolytic fragments, a goat polyclonal antiserum raised to human pFN was used along with an FITC-conjugated sheep anti-goat antibody as described previously [25]. Fluorescent cells were photographed on a Nikon Diaphot microscope equipped with epifluorescence illumination using Kodak 2475 recording film. Materials. Tissue culture cluster dishes (24-well) were purchased from Costar (Cambridge, MA); DMEM from GI NY); newborn calf serum from Biologos (Naperville, IL); 12-mm-diameter glass coverslips from Dynalab &o&ester, NY); rat tail coliagen type I as Vitrogen 100 from Collagen Corp. (Palo Alto, CA); so-

160

FLICKINGER FIX AND STAIN

EGTA I

t I 0 hr

t -18 hr

II

Ill

1 -18 hr

+cyclo

+cyclo

-18 hr

V

I -18 hr

-cycle

t

,

0 hr

+4 hr

t

+ ,

, +cyc10

0 hr

+cyclo

+ , +4 hr

+ , +cyclo

I

I

I +4 hr

0 hr

-18 hr

IV

+ ,

t

+ ,

+4 hr

+cyclo

0 hr

, +4 hr

FIX AND STAIN + I

-cycle +I8

hr

FIG. 1. Cycloheximide treatment protocols. Human dermal fibroblasts were treated with 2 pg/ml cycloheximide (Cycle) in various order-of-addition experiments either prior to or after they were subcultured (EGTA) for experiments and prior to fixation and staining with rhodamine-phalloidin to evaluate F-actin organization (FIX AND STAIN). In protocol I, cells were not treated at all with cycloheximide while in protocol II cycloheximide was included only during the 18 h prior to EGTA-mediated subculture from stock cultures into experimental wells coated with test proteins. In protocol III, the drug was included in the medium only during the 4 h of cell incubation in coated wells. In protocol IV, the drug was used for an 18-h pretreatment as well as during the 4-h experimental period in coated wells. Protocol V included the extensive drug treatment of protocol IV and was followed by three PBS rinses of the cells in coated wells and an 18-h period in adhesion medium without cycloheximide in order to permit recovery of protein synthesis.

dium ascorbate, cycloheximide, BSA, Triton X-100, and EGTA from Sigma Chemical CO. (St. Louis, MO); paraformaldehyde from Fisher Scientific Co. (Fairlawn, NJ); rhodamine-phalloidin from Molecular Probes, Inc. (Junction City, OR); [methyZ-3H]thymidine from New England Nuclear Corp. (Boston, MA); and FITC-conjugated sheep anti-goat antibody kindly provided by Dr. Abram Stavitsky of this department. RESULTS

Cycloheximide has been used previously to inhibit endogenous production of matrix molecules while studying adhesion responses of cells to exogenously provided matrices [24,35]. Therefore, cycloheximide was used under several experimental conditions to identify adhesion processes that may be inhibitable in particular dermal fibroblast populations. These protocols are described in Fig. 1. The control protocol (I) did not involve cycloheximide treatment at all, while protocol II involved an 18h drug pretreatment prior to EGTA-mediated detachment of cells from stock cultures and their inoculation into coverslip-containing wells in medium without drug.

AND

CULP

Protocol III utilized cycloheximide treatment only during the 4-h experimental period while protocol IV utilized treatment throughout. In some cases, reversal of the cycloheximide effect was tested (V) by rinsing cells three times with adhesion medium at the end of the 4-h period and then with a further incubation for 18 h in medium without the drug. In all cases, these treatments had no detectable effects on the attachment levels of either papillary or reticular cells over the time intervals described below and did not cause irreversable damage to cells (data not shown). These conditions were tested with RET5 cells at a middle passage level and well before in. vitro aging processes had made noticeable changes in the cultures. As shown in Fig. 2A using protocol I of Fig. 1, cells established an extensive array of F-actin stress fibers within 4 h on a type I collagen substratum postadsorbed with pFN. The same was true when these cells were challenged with a type I collagen coating alone (i.e., no pFN postadsorbed to the collagen layer) (Fig. 2B). Resorting to the most stringent cycloheximide treatment, such as protocol IV of Fig. 1, failed to alter the stress fiber arrays in these cells either on a collagen:pFN substratum (Fig. 2C) or on a collagen-only substratum (Fig. 2D). Similarly, protocols II and III had no detectable effect on Factin reorganization in these cells and identical results were obtained with PAP, cells at a middle-passage level (data not shown). In contrast, late-passage RET5 cells entering senescence at passage 22 displayed sensitivity to cycloheximide treatments. The control treatments under protocol I with a collagen:pFN substratum (Fig. 3A) or with collagen-only substratum (Fig. 3B) revealed very similar stress fiber patterns to those observed in the middle-passage cells above. The 18-h cycloheximide pretreatment of protocol II reduced the density of stress fibers in virtually all cells on either collagen:pFN or collagen-only substrata (open arrows in Figs. 3C and 3D, respectively), demonstrating that supplementation of the collagen layer with exogenous pFN could not overcome the deficiency in stress fiber formation. Therefore, it is unlikely that deprivation of the cell’s ability to synthesize cellular FNs during the pretreatment period had led to these changes. Treatment of cells under the more stringent protocol IV resulted in the elimination of all stress fibers from cells, generating focal pools of F-actin in the cytoplasm (hooked arrows in Figs. 3E and 3F). Even without stress fibers, these cells demonstrated effective cytoplasmic spreading over the substratum, presumably by other cytoskeletal networks; intermediate filament and microtubule networks were unaffected by these treatments, consistent with their possible roles in stress fiber-independent spreading mechanisms (data not shown). This result was true for both collagen:pFN (Fig. 3E) and collagen-only (Fig. 3F) substrata, demonstrating again in-

ADIIESION

PROCESSES

SENSITIVE

TO

CYC~O~~~I~IDE

161

FIG. 2. F-actin reorganization in middle-passage RET5 cells. Passage 13 RET5 cells were treated, as described in Fig. 1, as follows: (A) protocol I on a collagen:pFN substratum; (B) protocol I on a collagen-only substratum; (C) protocol IV on a coliagen:pFN substratum; and (D) protocol IV on a collagen-only substratum. In all cases, rhodamine-phalloidin-stainable stress fibers composed of F-actin are observed througbout the cytoplasm of all cells (e.g., black arrows in A-D). Magnification (using a 100X objective with oil immersion), X270.

dependence of the effect from endogenous cellular FN synthesis. Protocol III resulted in no reduction in stress fiber formation (not shown), indicating that the 4-h reattachment period is not particularly sensitive for inhibition of protein synthesis in these processes and that a lengthy pretreatment period is required to deplete the protein cofactor(s). The same results ‘were obtained for PAPS cells. These results indicate that one or more of the receptor molecules that recognize collagen type I or pFN in latepassage cells might be present in limiting amounts or exbibit a short half-life, therefore becoming depleted in cells deprived of continuous protein synthesis during eycloheximide treatment. To resolve whether collagen and/or pFN receptors are involved, late-passage RET5 cells were tested on substrata coated with pFN alone and with cells that had not been supplemented with ascorbate, thereby minimizing the maturation of their endogenous collagen products (see below as well). In this case, both protocols II and IV minimally perturbed the ability of these late-passage cells to reorganize F-actin into

stress fibers (data not shown). Therefore, collagen-specific receptors are uniquely sensitive to inhibition of protein synthesis. This experiment also estab;blishesthat tbe effect cannot be directed at a ~~7~os~~leta~protein per se, since stress fiber formation occurs perfectly well in response to FN receptor activities Middle passage RET, cells were then tested on coverslips coated on two adjacent regions of their surface with two different adhesive proteins. ne-half of the coverslip was adsorbed with eollag adsorbed with pFN or with a 155 or F110--see the collagen-bind cell-secreted collagen can bind to turn) but that contained otber [24]. The interface between regions could be readily staining with a lyclonal a stains snly the s tratum po derived molecules (see Fig. 4). face are of ~a~~~~~a~ interes

162

FLICKINGER

AND

CULP

FIG. 3. F-actin reorganization in late-passage RET, cells. Passage 22 RET5 cells were treated by the protocols described in Fig. 1 as follows: (A) protocol I on a collagen:pFN substratum; (ES) protocol I on a collagen-only substratum; (C) protocol II on a collagen:pFN substratum; (D) protocol II on a collagen-only substratum; (E) protocol IV on a collagen:pFN substratum; and (F) protocol IV on a collagen-only substratum. Excellent stress fibers composed of F-a&in are observed in both control experiments (black arrows in A and B) while stress fibers are far fewer in C and D where shorter linear bundles are observed (at the open arrows). In E and F, F-actin is not observed in stress fibers but is condensed into focal pools in the cytoplasm of cells (curved black arrows); however, cell spreading is still excellent even in the absence of stress fiber formation. Magnification (using a 100X objective with oil immersion), X270.

their undersurface adheres to collagen while the remaining portion adheres to the pFN representative. This experimental paradigm tests the specificity of a collagen-only substratum to promote F-actin reorganization when compared to proteolytic fragments of pFN

containing selected binding domains. It also tests whether the collagen receptor and the FN receptor can generate continuous stress fibers in different regions of the same cell, thereby displaying functional interchangeability.

ADHESION

PROCESSES

SENSITIVE

TO

C~~~O~EXI~I~E

163

PIG. 4. Cells bridging a topologically mixed substratum. Glass coverslips were coated on one-half of their surface wii:h collagen (COLE) and then the remainder was coated either with pFN proteolytic fragment F-155 (A and C) or with pFN fragment FlLO (B and D)--see Materials and Methods for details of these fragments. Passage 13 RET, cells were inoculated under protocol I conditions. After 4 h, coverslips were fixed and stained with both a goat polyclonal anti-human pFN antiserum (plus an FITG rabbit anti-goat IgG antibody) and with rhodamine-pballoidin as described under Materials and Methods. The same fields on coverslips (A and C; B and D) were photographed using a fluorescein-specific filter (A and B) and a rhodamine-specific filter (C and D). In A and B, the collagen and F155 or FllO interface is very evident with the anti-FN staining of the substratum (note that the collagen-only substratum fails to stain with anti-FN); in C and D, this interface is d.enoted with a dashed white line. Stress fibers are evident in cells on both sides of the collagen:F155 interface (e.g., at the white arrows in 6) while cells bridging the interface contain stress fibers that also extend continuously across the interface (e.g., at the white arrowhead in C). In contrast, stress fibers o:n the collagen:FllO substratum are only evident in cells on the collagen side (white arrows in D) which terminate abruptly at the COLE:FllO interface (dashed wbite line in ID); cells on the FllO side display only short bundles of F-actin (e.g., at the open arrowhead in D). Magnification (using a 20~ objective under dry conditions), X54.

In Figs. 4A and 4C, thermolysin-generated fragment F’155 of pFN containing both the RGDS-dependent cellbinding domain and the C-terminal hepariqI-binding domain was teste against collagen using protocol I. s of the substratum (anti-pFN stainCells on both regi ing in Fig. 4A, white arrows in Fig. 4C stained with rhodamine-phalloidin), as well as cells bridging the interface (i.e., witb one portion of their undersurface adherent on 155 while another portion was adherent on collagen as shown in Fig. 4A with anti-pFN staining), were fully capable of forming continuous stress fibers (at the white arrowhead in Fig. 46). Therefore, both the collagen receptor and the pFN receptors could function to generate the same stress fiber in different regions of the same cell. This result was also observed at a collagen interface with intact pFN, as well as with mid-passage

RET5 cells using protocols ) consistent with the data of Fig. 2. Late-passage cells usi and IV and bridging the interface fail stress fibers on the collagen side but did representative side (data not shown). In contrast, cells bridging the i tween collagen and pFN frag C-terminal heparin~l-bindin the RGDS cell-binding different result. As shown effectively on the eollag extend rig& up to the i the FllO-adherent side of the cell’s cytoplasm (white arrows in Fig. 4D). On the FPlO side ofthe same,cell, stress n&es of F-actin are fibers are absent and only sma 4D). These results observed (open arrowhead in

164

FLICKINGER

confirm the requirement for both the RGDS cell-binding and heparan sulfate-binding domains in the FNs for formation of stress fibers in these human dermal fibroblasts as shown previously [23, 241, as well as the ability of a collagen substratum to satisfy these requirements in the absence of any exogenous FN added to the system. They also illustrate the stringent requirement that the topology of FN and collagen receptors play in cell responses and that the nonpermissive substratum represented by FllO cannot transmit a dominant signal through the cytoplasm that inhibits stress fiber formation on the collagen-adherent side of the’same cell. The issue was then addressed whether in uiuo-aged cells displayed cycloheximide sensitivity throughout their entire in vitro lifespan or only as a consequence of entering the senescent period. When middle-passage RET8 cells were examined in the experimental paradigm II of Fig. 2, the same results were obtained-protocols through IV failed to alter the stress fiber distributions in cells (data not shown). However, these same cells grown into late passage levels became sensitive (Fig. 5). With either protocols II (Fig. 5B) or IV (Fig. LX), stress fibers became nonexistent and focal pools of F-actin were observed (at the hooked arrows in Figs. 5B and 5C) in well-spread cells under conditions where control cells (Fig. 5A) displayed an excellent array of stress fibers. These results were obtained whether or not the collagen layer was postadsorbed with pFN and did not occur on a pFN-only substratum. Therefore, the sensitivity of collagen receptor functions in either papillary or reticular dermal fibroblasts can be specifically ascribed to in vitro aging and not to in vivo aging processes. Experiments were then designed to test the potential reversal of the cycloheximide effect. In the first case, protocol IV was followed by the rinsing of the cycloheximide out of the culture with several PBS rinses, followed by an 18-h incubation in the absence of the drug (protocol V of Fig. 1). As shown in Fig. 6 for late-passage RET5 cells, protocol IV led to abolition of stress fibers as determined in low-magnification micrographs of these cultures (Fig. 6A, for example, at the white hooked arrow) or a high-magnification micrograph of one cell (Fig. 6B, white hooked arrow). The 18-h recovery period of protocol V permitted cells to reestablish a stress fiber network (white arrows in both the low-magnification micrograph in Fig. 6C and in a high-magnification micrograph in Fig. 6D). These results indicate that cycloheximide treatment does not lead to irreversible damage of cells and that the appropriate collagen-dependent machinery can be regenerated upon recovery of protein synthesis. A second approach was tested for overcoming cycloheximide sensitivity. In this case, cells were pregrown for two passages in medium containing sodium ascorbate in order to maximize their ability to make their own mature collagen matrix (and possibly collagen-depen-

AND

CULP

FIG. 5. F-actin reorganization in late-passage RET* cells. Passage 19 RET8 cells were treated as described in the legend to Fig. 3 and as follows (in all cases on a collagen:pFN substratum): (A) protocol I; (B) protocol II; and (C) protocol IV. Stress fibers are evident throughout the cytoplasm in A (e.g., at the white arrow) while only focal pools of F-actin are observed in B and C (e.g., at the curved arrows). Magnification (using 100X objective with oil immersion), X270.

dent receptors as well) [23]. When ascorbate-grown and late-passage RET5 cells were tested (Fig. 7), protocol IV failed to prevent stress fiber formation either on a collagen-only substratum (arrow in Fig. 7C) or on a pFN-only substratum (arrow in Fig. 7D) when compared to the un-

ADHESION

PROCESSES

SENSITIVE

TO

C~C~~~EXI~I~~

165

FIG. 6. Reversability of cycloheximide effect. Passage 22 RET5 ceils were treated under protocol IV (A and B) or protocol V (C and D) conditions (see Fig. 1) on a collagen:pFN substratum in all cases. After fixation and staining, the coverslips were photographed either with a 20X objective under dry conditions (72X for A and C) or with a 100X objective under oil immersion (260X for B and D). Focal pools of F-actin are denoted in A and B with curved white arrows while stress fibers are denoted in C and D with single white arrows.

treated controls (arrows in Fig. 7A and B, respectively). This contrasted with the data in Fig. 3 for the same ceils grown in medium without ascorbate. Therefore, stimulation of collagen production/maturation processes in late-passage cells with ascorbate leads either to increased levels of collagen receptors and/or their reduced turnover such that they are no longer sensitive to protocols II or IV. In addition to this finding with ascorbate-stimulated cells, another variation was obtained using protocol III in which cycloheximide is introduced only during the 4h reattachment period. Either on collagen-only (Fig. 7E) or on pFN-only (Fig. 7F) substrata, stress fibers were much longer and thicker (e.g., the open arrow in Fig. 7F) or they occurred in dense swirls in the cytoplasm (thick arrow in Fig. 7E); also, cells had spread much more extensively than in the untreated controls. This indicates that there is some negative regulation of stress fiber formation in these cells such that short-term treatment with cycloheximide inhibits the synthesis of some protein(s) that restricts stress fiber development.

The principal findings of these stu ies can be summarized as follows. Lengthy treatments of late-passage papillary or reticular cells with c~c~~beximide leads to their inability to generate F-a&in stress fibers while they continue to spread their cytoplasm quite successfuully over a collagen substratum. This inability is not due to irreversible damage to cells since they regenerate stress fibers within an 18-h recovery period without inhibitor. Cycloheximide treatment only during the 4-h reattachment period was insufficient to inhibit stress fiber formation, demonstrating that only longer treatment periods could deplete cells of the a~~~o~r~ate receptor and may be consistent with endocytotic recycling of collagen CHO cells [27]. These results als a limiting size to the pool of co11

FLICKINGER

AND

CULP

COLLAGEN

FIG. 7. Reversability of cycloheximide effects in ascorbate-supplemented cellsPassage 22 RET5 cells, pregrown for two passages in complete medium supplemented with sodium ascorbate as explained under Materials and Methods, were treated as follows (in all cases, the medium contained ascorbate): (A) protocol I on a collagen-only substratum; (B) protocol I on a pFN-only substratum; (C) protocol IV on a collagenonly substratum; (D) protocol IV on a pFN-only substratum; (E) protocol III on a collagen-only substratum; and (F) protocol III on a pFNonly substratum. Stress fiber examples are denoted with single arrows in A-D. In contrast, protocol III treatments led to much greater spreading of cells over both substrata, along with denser “swirls” of stress fibers in some cells (bold arrow in E) or thicker and longer stress fibers in other cells (open arrow in F). Magnification (using a 100X objective with oil immersion), X270.

Since the effect was not observed on a pFN-only substratum but was observed on collagen f pFN, it can be ascribed to a collagen-specific receptor and not to the

integrin or heparan sulfate proteoglycans important in FN-mediated adhesion responses. Therefore, these two integrin classes display significantly different metabolic

ADHESION

PROCESSES

SENSITIVE

properties during in. vitro aging of dermal fibroblasts, as shown by the experiments reported here. Similarly, the effect cannot be ascribed to depletion of some microfilament protein component in the cytosol since stress fibers form effectively on pFN-only substrata; the topology experiments with cells bridging the interface between collagen and a pFN proteolytic fragment on substrata provide further support for this argument. The candidate receptor molecule depleted by lengthy cycloheximide treatment is probably the collagen-specific integrin molecule, consistent with previous studies identifying a specific integrin involved in collagen adhesion processes that does not participate in FN adhesion processes [IO, 18, 191. This depletion effect was not observed in mid-passage cells, consistent with a phenomenon associated with in vitro aging, but was observed under the same conditions in both patient 5 and patient 8 cells, demonstrating that it is not associated with in uiuo aging processes. Therefore, in vitro aging of cells apparently leads to limiting amounts of this receptor and/or increased turnover of receptor requi.ring a greater continuity of protein synthesis. The reversability of the cycloheximide effect using cells continuously grown in medium containing ascorbate, which stimulates production/maturation of collagen and possibly collagenspecific receptor levels [3,29-311, is probably more consistent with the former possibility than the latter. In this regard, it has also been shown that FN-specific receptor levels increase coordinately with increased cFN production by human lung fibroblasts stimulated with transforming growth factor-0 [32,33]. There is an alternative explanation for the cycloheximide-sensitive process in aging cells-namely, alteration of a post-translational process that specificahy affects collagen receptor activity and not FN receptor activity (e.g., phosphorylation [tp]); however, the inability to demonstrate the effect in asc+ cells runs counter to this possibility. Direct examination of the levels of collagen-specific integrin levels in these cells and its turnover properties under various physiological conditions should answer this question [18,19]. The results of the topology experiment in which different regions of the substratum are coated with differing adhesion proteins are significant in a number of contexts. First, these experiments confirm the results of Hall et al. [24] and others [34, 351 that both the RGDS-dependent cell-binding and heparinii-binding domains are minimal requirements from FN on the substratum in order to generate transmembrane signals for successful formation of F-actin stress fibers. F155 contained this sufficient information while FllO, lacking the heparinii-binding domain, was insufficient. Second, cells bridging the collagemF155 interface were able to generate continuous stress fibers across the interface. This indicates that the same stress fiber can transmem-

TO

167

CYCLOHEXIMIDE

brane link collagen receptors at its one end with FN receptors at its other end. Therefore, Qmologous receptors at both ends of the fiber are not a stringent requirement for these F-actin structures to form. This is also consistent with recent descriptions sf both FN and vitronectin receptors in the same focal hesive contacts of fibroblasts [36] and endothelial ce 1371 on serumcoated substrata. Third, the inability to form stress fibers in one region of t 11does not inhibit formation of stress fibers in ano region of the same cell on a permissive substratu erefore, matrix receptors provide the dominant mechanism in their initiation and maintenance. Fourth while stress form effectively on the permissive side of cells bri a collagemFl10 interface, they terminate precisely at the interface, undoubtedly because of insufficient receptor organization on the FlIO-facing portion of the cell’s u.ndersurface. This leads to the most interesting question of the molecular mechanisms by which these stress fibers terminate at the interface; perhaps, some unique cytoplasmic signal is generated at this region of the cell or, alternatively, some specialized adhesive contact on the collagen side of the interface. Additional experiments must be performed to address this important issue These studies were supported The authors extend appreciation for contributing the proteolytic Alan Tartakoff of the Department magnification objectives in some

1.

Yamada,

K. M. (1983)

2.

Ruoslahti,

E. (1988)

3.

McDonald,

4.

Buck,

5.

Ruoslahti,

Annu. Annu.

J. A. (1988)

C., and Horwitz,

by an NIH research grant (AG02921). to Dr. Euciano Zardi of Genoa, Italy, fragments of fibronectin and to Dr. of Pathology for the use of lowof these fmorescence analyses.

Reu. Biochem.

Anna

Biochim.

5?‘,375.

Reu. CelEBio2.

A. F. (1987)

E., and Pierschbacher, R. L. (1987)

52,761.

Reu. Biochern. Annu. M.D.

(1987)

6.

Juliano,

7.

Burridge, K., Fath, K., Kelly, T., Nuckolis, (1988) Annu. Rev. Cell Biol. 4,487.

8.

Harper, 136.

9.

Cheung, E., Brown, iol. 130,118.

P. A., and Juliano,

Biophys.

R. L. (1981)

P. J., and Juliano,

P. A., and Juliano,

4,183.

Reu. Cell A.

238,491.

7,261. C., and Turner, Arab-e

(London)

R. E. (1987)

6. 290,

J. Cell. Phys-

10.

Harper,

11.

Couchman, J. R., Austria, R., Woods, (1958) J Cell. Physiol. 136,226.

12.

Laterra, 112.

13.

Woods, A., Couchman, 260,10872.

14.

Culp, L. A., Laterra, J., Lark, PI/I. W., Beyth, R. J., and Tobey, S. L. (1986) fn Functions of the Proteoglycans (Evered, D., and Whelan, J., Eds.), CIBA Symposium No. 124, pp. 158-183, Wiley, London.

J., Silbert,

R. L. (1981)

3,179.

Viol.

Science

J. E., and Gulp,

9. Cell Viol. 91,647. A., and

L. A. (1983)

J. R., and Hook,

M. (1985)

Ifughes,

R. C.

J. Cell Biol. J. Biol.

9

Chen.

168 15.

FLICKINGER Lewandowska, K., Choi, H. U., Rosenberg, Gulp, L. A. (1987) J. Cell Biol. 105,1443.

L. C., Zardi,

16.

Nishiyama, T., Tominaga, N., (1988) Collagen Rel. Res. 8,259.

17.

Schafer, I. A., Shapiro, A., Kovach, M., Lang, R. B. (1989) Exp. Ce22 Res. 183,112.

18.

Wayner,

19.

Wayner, E. A., Carter, T. J. (1988) J. Cell Biol.

20.

Harper,

21.

Beyth,

22. Kent, Ageing

E. A., and Carter,

W. G. (1987)

K.,

G. (1979) L. A. (1985)

W. M., Funderburg, Deu. 33,115.

F. M.,

L., and

Hayashi,

T.

31. 105,1873. Kunicki,

204,526. Ageing

and Gulp,

Deu. 29,151.

L. A. (1986)

Let. 237,35.

Received June 19,1989 Revised version received

August

23,1989

32. 33.

Me&.

23. Beyth, R. J., and Culp, L. A. (1984) Exp. Cell Res. 155,537. 24. Hall, M. D., Flickinger, K. S., Cutolo, M., Zardi, L., and Culp, L. A. (1988) Exp. Cell Res. 179,115. 25. Tobey, S. L., McClelland, K. J., and Culp, L. A. (1985) Exp. Cell Res. 158,395. 26. Lewandowska, K., Kaetzel, C. S., Zardi, L., and Culp, L. A. (1988) FEBS

27. 28. 29. 30.

C., and Fratianne,

R. S., and

Science Me&.

and

J. Cell Biol.

W. G., Piotrowicz, 107, 1881.

R. A., and Grove, R. J., and Gulp,

Nakajimi,

AND

34.

CULP Bretscher,

M. S. (1989)

165,320. 35. Woods, A., Couchman, 36. 37.

EMB,O

J. 8,134l.

Brown, P. J., and Juliano, R. L. (1987) Exp. Cell Res. 171,376. Pinnell, S. R. (1979) J. Invest. Dermatol. 73,485. Booth, B. A., Polak, K. L., and Uitto, J. (1980) Biochim. Biophys. A 607,145. Grinnell, F., Fukamizu, H., Pawelek, P., and Nakagawa, S. (1989) Exp. Cell Res. 181,483. Roberts, C. J., Birkenmeier, T. M., McQuillan, J. J., Akiyama, S. K., Yamada, K. M., Chen, W.-T., Yamada, K. M., and McDonald, J. A. (1988) J. Biol. Chem. 263,4586. Keski-Oja, J., Raghow, R., Sawdey, M., Loskutoff, D. J., Postlethwaite, A. E., Kang, A. H., and Moses, H. L. (1988) J. Biol. Chem. 263,311l. Izzard, C. S., Radinsky, R., and Culp, L. A. (1986) Exp. Cell Res. J. R., Johansson, S., and Hook, M. (1986) EMBO J. 5,665. Singer, I. I., Scott, S., Kawka, D. W., Kazazis, D. M., Gailit, J., and Ruoslahti, E. (1988) J. Cell Biol. 106,217l. Dejana, E., Colella, S., Conforti, G., Abbadini, M., Gaboli, M., and Marehisio, P. C. (1988) J. Cell Biol. 107, 1215.