effect of strain on human keratinocytes in vitro

9
JOURNAL OF CELLULAR PHYSIOLOGY 173:64 – 72 (1997) Effect of Strain on Human Keratinocytes In Vitro TEIJI TAKEI, CARLOS RIVAS-GOTZ, CHRYS A. DELLING, JASON T. KOO, IRA MILLS, THOMAS L. MCCARTHY, MICHAEL CENTRELLA, AND BAUER E. SUMPIO* Department of Surgery, Yale University School of Medicine, New Haven, Connecticut Tissue expansion, a technique to enlarge the skin surface area with an expandable balloon, has been widely used in reconstructive surgery. Although the effect of tissue expansion on in vivo skin physiology and histology has been well docu- mented, it remains unclear whether keratinocytes or other cell types are responsi- ble for these changes. Therefore, we investigated the in vitro effect of cyclic (10 cycles/min, 150 mmHg) or constant (continuous, 150 mmHg) strain on human keratinocyte phenotype and relevant mechanosignaling pathways. Our results demonstrate that keratinocytes subjected to cyclic strain exhibit a significant (P õ 0.05) increase in cell proliferation (49.2 { 15.8%), DNA synthesis (37.7 { 4.5%), elongation (20.3 { 2.7%), and protein synthesis (17.9 { 6.6% increase) as compared with stationary controls. In contrast, keratinocytes subjected to constant strain were unaffected aside from a modest transitory increase in the proliferative rate. Keratinocytes subjected to cyclic strain aligned perpendicular to the force vector (24.2 { 1.67) as compared with stationary controls (40.4 { 2.27; the smaller degree indicates better alignment). We also report strain-induced reduction in the levels of cyclic adenosine mono phosphate (cAMP), protein kinase A (PKA), and prostaglandin E 2 (PGE 2 ) as compared with stationary controls (cAMP, 30 { 7.5%; PKA, 45 { 17%; PGE 2 , 58 { 4.3%; percent decrease vs. that of control). We conclude that direct application of cyclic strain on human keratinocytes modulates cell phenotype and cAMP-mediated signaling pathways in an inverse manner. Moreover, keratinocytes may play an important role in previously ob- served alterations in skin properties associated with tissue expansion and other strain-induced responses. J. Cell. Physiol. 173:64 – 72, 1997. q 1997 Wiley-Liss, Inc. Several studies have been conducted to explore the To test the hypothesis that keratinocytes are directly responsive to strain, we examined the ability of cyclic in vivo effects of tissue expansion on surrounding tissue and constant strain to influence keratinocyte prolifera- (Lorber and Milobsky, 1968; Francis and Marks, 1977; tion, orientation, protein synthesis, and alignment in Squier, 1980; Austad et al., 1986). Histological observa- vitro. In addition, we examined the effect of strain on tions after tissue expansion reveal a thinner dermis, a possible mediator mechanosignaling pathways includ- remarkable flattening of the basal cells (Austad et al., ing cAMP, PKA, and PGE 2 . These studies were per- 1982; Pasyk et al., 1982; Terracio and Borg, 1986; John- formed with an in vitro model of expansion (cyclic vs. son et al., 1993) and changes of keratinocytes from a constant strain) in which keratinocytes were cultured columnar to cuboidal morphology (Breidahl et al., 1989; on a flexible-bottom membrane that could be mechani- Johnson et al., 1993). In addition, epidermal hyperpla- cally deformed by vacuum (Johnson et al., 1988; Sum- sia has been reported with increased cell layers in basal pio et al., 1988; Gilbert et al., 1989; Knight et al., 1990). and suprabasal layers and with increased mitotic or labeling indices (Austad et al., 1982; Lew and Fuseler, MATERIALS AND METHODS 1991; Johnson et al., 1993). Exposure of keratinocytes to strain Skin tissue consists of dermis and epidermis, includ- ing different types of cells such as keratinocytes, mela- Human keratinocytes from neonatal foreskins were obtained as previously elsewhere (Eisinger, 1985). Skin nocytes, and fibroblasts. One problem with in vivo stud- ies is that the effects of implants on surrounding tissues or potential interactions between dermal and epider- mal layers cannot be eliminated in evaluating the Contract grant sponsor: NIH; Contract grant number: HL 47345; Contract grant sponsor: Department of Veterans Affairs Merit strain response. It has also been observed that surgery Review. itself, in the absence of any implant, may cause a hyper- *Correspondence to: Bauer E. Sumpio, M.D., Ph.D., Department plastic epidermis (Austad et al., 1982), whereas other of Surgery, Yale University School of Medicine, New Haven, CT studies have suggested that surgery has minimal ef- 06510. E-mail: [email protected] fects on an increase in surface area over the tissue expander (Vanderkolk et al., 1988). Received 22 January 1997; Accepted 28 May 1997 q 1997 WILEY-LISS, INC. 8925 560D / 8925$$560d 08-11-97 19:53:35 wlcpal W Liss: JCP

Upload: teiji-takei

Post on 06-Jun-2016

213 views

Category:

Documents


1 download

TRANSCRIPT

JOURNAL OF CELLULAR PHYSIOLOGY 173:64–72 (1997)

Effect of Strain on Human KeratinocytesIn Vitro

TEIJI TAKEI, CARLOS RIVAS-GOTZ, CHRYS A. DELLING, JASON T. KOO, IRA MILLS,THOMAS L. MCCARTHY, MICHAEL CENTRELLA, AND BAUER E. SUMPIO*

Department of Surgery, Yale University School of Medicine,New Haven, Connecticut

Tissue expansion, a technique to enlarge the skin surface area with an expandableballoon, has been widely used in reconstructive surgery. Although the effect oftissue expansion on in vivo skin physiology and histology has been well docu-mented, it remains unclear whether keratinocytes or other cell types are responsi-ble for these changes. Therefore, we investigated the in vitro effect of cyclic (10cycles/min, 150 mmHg) or constant (continuous, 150 mmHg) strain on humankeratinocyte phenotype and relevant mechanosignaling pathways. Our resultsdemonstrate that keratinocytes subjected to cyclic strain exhibit a significant(P õ 0.05) increase in cell proliferation (49.2 { 15.8%), DNA synthesis (37.7 {4.5%), elongation (20.3 { 2.7%), and protein synthesis (17.9 { 6.6% increase) ascompared with stationary controls. In contrast, keratinocytes subjected to constantstrain were unaffected aside from a modest transitory increase in the proliferativerate. Keratinocytes subjected to cyclic strain aligned perpendicular to the forcevector (24.2{ 1.67) as compared with stationary controls (40.4{ 2.27; the smallerdegree indicates better alignment). We also report strain-induced reduction inthe levels of cyclic adenosine mono phosphate (cAMP), protein kinase A (PKA),and prostaglandin E2 (PGE2) as compared with stationary controls (cAMP, 30 {7.5%; PKA, 45 { 17%; PGE2, 58 { 4.3%; percent decrease vs. that of control).We conclude that direct application of cyclic strain on human keratinocytesmodulates cell phenotype and cAMP-mediated signaling pathways in an inversemanner. Moreover, keratinocytes may play an important role in previously ob-served alterations in skin properties associated with tissue expansion and otherstrain-induced responses. J. Cell. Physiol. 173:64–72, 1997. q 1997 Wiley-Liss, Inc.

Several studies have been conducted to explore the To test the hypothesis that keratinocytes are directlyresponsive to strain, we examined the ability of cyclicin vivo effects of tissue expansion on surrounding tissueand constant strain to influence keratinocyte prolifera-(Lorber and Milobsky, 1968; Francis and Marks, 1977;tion, orientation, protein synthesis, and alignment inSquier, 1980; Austad et al., 1986). Histological observa-vitro. In addition, we examined the effect of strain ontions after tissue expansion reveal a thinner dermis, apossible mediator mechanosignaling pathways includ-remarkable flattening of the basal cells (Austad et al.,ing cAMP, PKA, and PGE2. These studies were per-1982; Pasyk et al., 1982; Terracio and Borg, 1986; John-formed with an in vitro model of expansion (cyclic vs.son et al., 1993) and changes of keratinocytes from aconstant strain) in which keratinocytes were culturedcolumnar to cuboidal morphology (Breidahl et al., 1989;on a flexible-bottom membrane that could be mechani-Johnson et al., 1993). In addition, epidermal hyperpla-cally deformed by vacuum (Johnson et al., 1988; Sum-sia has been reported with increased cell layers in basalpio et al., 1988; Gilbert et al., 1989; Knight et al., 1990).and suprabasal layers and with increased mitotic or

labeling indices (Austad et al., 1982; Lew and Fuseler, MATERIALS AND METHODS1991; Johnson et al., 1993). Exposure of keratinocytes to strainSkin tissue consists of dermis and epidermis, includ-ing different types of cells such as keratinocytes, mela- Human keratinocytes from neonatal foreskins were

obtained as previously elsewhere (Eisinger, 1985). Skinnocytes, and fibroblasts. One problem with in vivo stud-ies is that the effects of implants on surrounding tissuesor potential interactions between dermal and epider-mal layers cannot be eliminated in evaluating the Contract grant sponsor: NIH; Contract grant number: HL 47345;

Contract grant sponsor: Department of Veterans Affairs Meritstrain response. It has also been observed that surgeryReview.itself, in the absence of any implant, may cause a hyper-*Correspondence to: Bauer E. Sumpio, M.D., Ph.D., Departmentplastic epidermis (Austad et al., 1982), whereas otherof Surgery, Yale University School of Medicine, New Haven, CTstudies have suggested that surgery has minimal ef-06510. E-mail: [email protected] on an increase in surface area over the tissue

expander (Vanderkolk et al., 1988). Received 22 January 1997; Accepted 28 May 1997

q 1997 WILEY-LISS, INC.

8925 560D/ 8925$$560d 08-11-97 19:53:35 wlcpal W Liss: JCP

EFFECT OF STRAIN ON KERATINOCYTES 65

specimens were treated with antibiotics (1,000 units/ dt Å doubling time, CN Å cell number, n Å 0, 1, 2, 3,. . . , and t0 Å time zero.ml penicillin, 1 mg/ml streptomycin, and 2.5 mg/ml

Fungizone), the epidermal layer was trypsinized For morphological assessment, keratinocytes werestained with 1% crystal violet (Sigma, St Louis, MO)(0.1%), and keratinocytes were harvested. Keratino-

cytes were cultured in serum-free medium (Keratino- for 5 min after fixation in 3.7% formaldehyde (J.T.Baker Inc., Phillipsburg, NJ) for 10 min. Morphologicalcyte SFM, Gibco BRL, Grand Island, NY) containing

1.02 g/L L-glutamine, 0.2 ng/ml epidermal growth fac- changes were characterized by measurement of eachcell under microscopic magnification (140). At least 50tor, and 30 mg/ml bovine pituitary extract. Keratino-

cytes from passages 2–4 were used for this study. Kera- cells were randomly selected, and the ratio of the longaxis of the cell to the short axis was determined. Intinocytes (5,000 cells/cm2) were seeded equally on flex-

ible membranes of 6-well culture plates (29.45 cm2/ addition, the angle of the long axis of the cell withrespect to a tangential line of the well was measured;plate) coated with type I collagen (FLEX I, Flexcell

Corp., McKeesport, PA) and then were incubated at 07 indicates that cells align parallel to the margin ofthe well. Keratinocytes also were fixed with 3.7% form-377C in a 5% CO2 incubator for 48 hr. Keratinocytes

were then subjected to constant or cyclic strain, with aldehyde and stained with rhodamine-conjugated phal-loidin (Molecular Probe Inc., Eugene, OR) after 7 daysmedium changed every other day.

The Flexcell Stress Unit utilizes vacuum to deform of cyclic strain. Preparations were observed under afluorescence microscope (Olympus Optical C. Ltd.,a flexible bottom culture plate that is placed in a rubber

manifold, as described by Banes et al. (1985, 1990). Tokyo, Japan). All photographs were taken from theperiphery of well. The bottom lines of these photo-Briefly, vacuum that is controlled by solenoid valves

deforms the flexible membrane plate downward, graphs are parallel to the rims of the culture wells, andthe center of the well was located on the upper side ofthereby elongating cells attached to the upper side of

this membrane. The degree of deformation is regulated each photograph.by the level of vacuum; cells experience 24% strain in

Total protein assaythe periphery of the well, whereas the center is 1%(average is 10%) at 150 mmHg (Sumpio et al., 1988; Keratinocytes subjected to experimental regimens were

collected with a rubber policeman and homogenizedAwolesi et al., 1995).We also designed an apparatus based on the principle with a Dounce homogenizer (type A) for 10 strokes on

ice in extraction buffer containing 25 mM Tris HCl, 5of a pleural evacuation unit. The mechanical deforma-tion is controlled by a sealed 8-ft. water chamber, and mM EGTA, 0.7 mM CaCl2, 1 mM PMSF, 0.1 mM TLCK,

and 10 mM leupeptin. Total protein was assayed basedthe level of vacuum is monitored by a pressure gaugeand also verified with an 8-channel paper recorder on the Bradford (1976) method by using bovine serum

albumin (Pierce, Rockford, IL) as a standard.(Gould, Valley View, OH). The height of water, whichis regulated by a calibrated infusion pump, is propor-

Cyclic AMP assaytional to the level of vacuum. The upper space of thechamber, which is lower in pressure than atmosphere, Cyclic AMP (cAMP) assay was performed using a cAMP

radioimmunoassay kit (Biomedical Technologies Inc.,is connected to the manifold and causes the flexibleculture plate (FLEX I) to deform. This apparatus is able Stoughton, MA). Initially, keratinocytes were pre-

treated with 0.5 mM IBMX (3-isobutyl-1-methyl-xan-to produce a constant deformation (150 { 2.9 mmHg).thine; Sigma) for 5 min to prevent cAMP hydrolysis

Strain protocols and then stretched for 1, 2, 5, 10, and 30 min. ThecAMP was extracted from keratinocytes with 90%We studied two different patterns of strain (constant

or cyclic) and compared them with the stationary con- n-propanol and centrifuged with a Speedvac concentra-tor (Savant, Farmingdale, NY) for 45 min. The pelletstrol. For the constant strain experiments, the flexible

membranes were deformed and maintained with 150 were suspended with a calculated volume (1 ml/1,000cells) of buffer (0.05 M sodium acetate, 0.01% sodiummmHg of vacuum. For the cyclic strain experiments,

flexible membranes were subjected to 150 mmHg of azide, pH 6.2). Each sample was incubated with [I125]-labeled cAMP and anti-cAMP antibodies for 20 hr atvacuum for 3 sec alternating with 3 sec of relaxation

in the neutral position (10 cycles/min). Control cells 47C, and then antibody-bound cAMP was precipitatedby centrifugation at 2,000g for 20 min, and its radioac-were cultured on the same plates in the same incubator

but not subjected to strain. tivity was measured by a scintillation counter (LS 6000IC, Beckman, Fullerton, CA).

Cell proliferation and morphologyPKA activityKeratinocytes were incubated in the presence or ab-

sence (quiescent) of epidermal growth factor (Gibco Keratinocytes were collected by scraping and centri-fuged at 1,000g for 5 min. Pellets were dissolved inBRL) and pituitary extracts (Gibco BRL) and then sub-

jected to cyclic or constant strain for up to 7 days. The lysate buffer (25 mM Tris HCl, 5 mM EGTA, 0.7 mMCaCl2, 1 mM PMSF, 0.1 mM TLCK, and 10 mM leupep-cell number of an aliquot from each well (n Å 6) was

counted at each time point by using a Coulter Counter tin) and homogenized on ice. Samples were mixed withsolutions provided by the PKA assay kit (Pep TagZM (Coulter Corporation, Miami, FL). DNA synthesis

was assessed by [3H]-thymidine incorporation. The Assay, Promega, Madison, WI) and analyzed by frac-tionation on a 0.8% agarose (50 mM Tris-HCl, pH 8.0)[3H]-thymidine (1 mCi/well) was added to keratinocytes

2 hr prior to harvest. Disintegrations per minute (dpm) gel run at 100 V for 20 min. Phosphorylated bands weresolubilized according to the manufacturer’s instruc-were measured by a scintillation counter and expressed

as dpm/well. The calculation of doubling time is as fol- tions and quantitatively measured with a fluorescencespectrophotometer (Perkin-Elmer, Foster City, CA) bylows: dt Å tn/1 0 tn , CNn/1 Å 2 1 CNn, where t Å time,

8925 560D/ 8925$$560d 08-11-97 19:53:35 wlcpal W Liss: JCP

TAKEI ET AL.66

Fig. 2. Effects of strain on [3H]-thymidine incorporation in culturedkeratinocytes. a: Keratinocytes were subjected to cyclic strain (10cpm) for 15 min, 1, 4, 24 hr. After 24 hr strain, [3H]-thymidine incorpo-ration increased significantly (P õ 0.05) as compared with the time0 point. b: Keratinocytes were subjected to strain for 24 hr to investi-gate the effect of different strain regiments (cyclic or constant) on[3H]-thymidine incorporation in quiescent (without growth factors)or standard (with growth factors) medium. Cyclic strain induced asignificant increase in dpm (Põ 0.05) in both standard and quiescentmedia. Values are presented as mean { SEM, n Å 6. *P õ 0.05 vs.time 0 (a) or control (b, no strain).

was counted by a scintillation counter. The level ofFig. 1. Keratinocyte proliferation with exposure to strain. a: Kera-tinocytes were exposed to cyclic (10 cycles/min, or cpm) or constant PGE2 was calculated from the standard curve, whichstrain for 7 days, and the cell numbers were counted. Cyclic strain was determined from known concentrations of PGE2.increased cell number significantly as compared with constant strainand control (no strain). b: Keratinocytes were subjected to cyclic strain Statistical analysisfor 1, 2, or 7 days and then cultured under the stationary condition.Cells exposed to strain showed the enhanced proliferation through all The results are presented as the mean { SEM. Sig-experimental course. Values are presented as mean { SEM, n Å 6. nificance was assessed by one-way analysis of variance*P õ 0.05 vs. control, /P õ 0.05 vs. constant strain. with a multiple comparison method using a post hoc

test (Sigmastat, Jandel Scientific, San Rafael, CA).P õ 0.05 was considered significant.

using 568-nm excitation and 592-nm emission. The ra- RESULTStio of total PKA activity in strained keratinocytes rela- Cell proliferation and DNA synthesistive to time 0 was calculated.Keratinocyte proliferation was stimulated by cyclic

Assay for PGE2 strain (dt Å 28.1 { 7.6 hr) as compared with stationarycontrols (dtÅ 37.5{ 7.1 hr) or cells exposed to constantKeratinocytes were cultured for 48 hr prior to collect-

ing medium, followed by change to fresh medium. PGE2 strain (dt Å 36.4 { 8.3) (Fig. 1a). Keratinocyte pro-liferation in response to cyclic strain was pronouncedlevels in the culture medium were assayed by RIA (Ad-

vanced Magnetics Inc., Cambridge, MA). Samples were (49.2 { 15.8% increase; 124,960 { 2,240 vs. 186,480 {5,810 cells/well, cyclic strain vs. stationary control, re-diluted with ethyl acetate at a 1:3 ratio (pH 3–4) and

centrifuged to remove the upper organic phase. Then spectively) and sustained, whereas constant strain ledto a weaker and transient elevation in proliferation asthe pellet was dissolved in buffer containing 0.01 M

phosphate, 0.1% bovine gamma globulin, and 0.1% so- compared the elevation of stationary controls (Fig. 1a).Additional studies were conducted to determinedium azide, pH 7.0. Each sample was mixed with anti-

serum and then added to 125I-PGE2, magnetic goat anti- whether continuous cyclic strain was required to ob-serve stimulation of keratinocyte proliferation (Fig. 1b).rabbit by following the manufacturer’s instructions.

After centrifugation for 20 min at 1,000g, the pellet Keratinocytes subjected to 1–2 days of strain, followed

8925 560D/ 8925$$560d 08-11-97 19:53:35 wlcpal W Liss: JCP

EFFECT OF STRAIN ON KERATINOCYTES 67

TABLE 1. Mean measurements for morphological changes and statistical values (strain vs. control)1

Day 3 Day 5 Day 8

Orientation (degree)2 Control 44.2 { 2.3 42.4 { 2.2 40.4 { 2.2Strain 31.2 { 1.9* 29.9 { 2.0* 24.2 { 1.6*

Elongation (ratio)3 Control 1.92 { 0.05 2.19 { 0.07 2.12 { 0.07Strain 2.35 { 0.07* 2.52 { 0.08* 2.63 { 0.07*

1Keratinocytes were subjected to cyclic strain (10 cycles per min, 150 mmHg), as described in Materials and Methods.These values were measured three times and expressed as mean { SEM.2Degree represents the angle of the long axis of the cell to the tangent of the well.3Ratio is determined by measuring the long axis/the short axis.*P õ 0.05 is considered to be statistically significant vs. control (n Å 150).

by no strain, failed to show enhanced proliferation (Fig. induced a significant increase in total protein (17.9 {6.6% increase) as compared with control.1b). A regimen of 7 days of continuous cyclic strain

led to enhanced keratinocyte proliferation, whereas aCyclic AMP, PKA, and PGE2 levelsregimen of 1 or 2 days of cyclic strain demonstrated

more enhanced proliferation than that of control in the The levels of cAMP in keratinocytes exposed to 30min of strain were significantly lower (30 { 7.5% de-early time interval but drifted to control levels by day

7 (Fig. 1b). crease; 197 { 19.2 vs. 282 { 2.9 pmoles/ml) than thoseof control (Fig. 6a). PKA activity also decreased by 45%At least 24 hr of exposure to cyclic strain was neces-

sary to induce a significant increase in DNA synthesis, of control by 30 min of cyclic strain (0.55 { 0.18,changes in ratio) (Fig. 6b). This decrease was sustainedas measured by [3H]-thymidine incorporation (Fig. 2a).

After 24 hr of cyclic strain, a significant increase in for up to 60 min (data not shown). PGE2 levels in cyclicor constant strained keratinocytes declined on day 5 andincorporation of [3H]-thymidine (37.7 { 4.5% increase;

6,056 { 559 vs. 4,398 { 266 dpm/well, cyclic strain vs. were significantly lower (40 { 3.8% decrease; 4.63 {0.31 vs. 7.59 { 0.36 pg/ml) than that in control (Fig. 6c).stationary control, respectively) was observed in kera-

tinocytes subjected to the cyclic strain as comparedDISCUSSIONwith constant strain or stationary controls (Fig. 2b).

When experiments were performed under quiescent In the present study, we have demonstrated thatstrain, particularly cyclic strain, increases the prolifer-conditions for 24 hr (using media without growth fac-

tors) to synchronize the cell cycles, cyclic strain was ation and DNA synthesis of keratinocytes in vitro. Ourfindings are consistent with established clinical datamore potent than constant strain or control in inducing

DNA synthesis (Fig. 2b). (Bullough, 1972; Mackenzie, 1974; Machida et al.,1989; Baker, 1991). In in vivo studies, Gibson et al.

Morphology and alignment (1965) and Machida et al. (1989) described the effective-ness of repeated motions in maximizing skin surfaceElongation and alignment were observed in the pe-

riphery of the culture well, the area of greatest strain. area and demonstrated that cyclic loading (stretchingfollowed by relaxation) was more potent than constantKeratinocytes aligned perpendicular to the force vector

that was directed toward the center of well, whereas stretching to harvest extra skin. Lew and Fuseler(1991) also suggested that cell proliferation around im-keratinocytes in the center did not change and main-

tained the cobblestone appearance typical of keratino- plants might be controlled by appropriate pressurechanges in the expander. Other investigators have re-cytes. Cyclic strain caused a significant change in orien-

tation and elongation of keratinocytes exposed to strain ported that gentle massage of mouse skin induces atwofold increase of mitotic rate (Bullough, 1972) and aas compared with control (Table 1). There were signifi-

cant strain-induced shifts of cell orientation and elon- light brushing of mouse ear epidermis increases mitoticactivity 3.6 times higher than control within 24 hrgation, showing that mean values of orientation (Fig.

3a, arrow) shifted to lesser degrees, and long:short axis (Mackenzie, 1974), suggesting that repeated mechani-cal stimulation to the skin could cause epidermal hy-ratios shifted to greater values from day 3 to 7 (Fig.

3b). Keratinocytes remained elongated and aligned in perplasia. Although the Flexcell Stress Unit is not di-rectly related to the in vivo condition, cells attachedthe periphery of culture plates on day 7 (Fig. 4),

whereas keratinocytes returned to a normal random to the flexible membranes can experience mechanicalstress (Sumpio et al., 1988; Banes et al., 1990, 1985).distribution after the strain force was removed (2 days

of cyclic strain followed by 5 days of no strain). F-actin Thus, we are able to apply mechanical strain to kera-tinocytes by using this strain apparatus.in keratinocytes subjected to cyclic strain showed align-

ment perpendicular to the force vector. Keratinocytes Constant strain stimulated cell proliferation (P õ0.05) compared with control but was less potent thanwere also elongated in a similar pattern (Fig. 4). Thick

F-actin bundles were observed under the strain condi- cyclic strain. This increase was probably caused by theinitial stimulation of expansion. Continuous dynamiction, whereas those in control showed fine filaments

and no alignment. strain may be a crucial factor for enhanced cell growth(Figs. 1b, 2), which is consistent with the hypothesis

Total protein that signals for rapid cell division last as long as me-chanical stimulation continues (Austad et al., 1982;Total protein was higher in keratinocytes exposed

to cyclic strain than that of control or constant strain Wang et al., 1993). These mechanisms are still unclear,but our data indicate that exposure of keratinocytes to(15.3 { 5.3, 8.8 { 4.4% increase, respectively) (Table

2). Exposure of keratinocytes to cyclic strain for 5 days cyclic strain will activate DNA synthesis. This indica-

8925 560D/ 8925$$560d 08-11-97 19:53:35 wlcpal W Liss: JCP

TAKEI ET AL.68

Fig. 3. Quantitative measurements of morphological changes. a: Keratinocytes were subjected to cyclic strain during the same timeKeratinocytes were subjected to cyclic strain (10 cpm) for 3, 5, or 7 course. Cells elongated along the edge of culture plates, and this effectdays, and then orientation (the angle of the long axis of each cell with was most obvious on day 7. Cells, the number of cells; strain, 10 cpm;respect to a tangential line of the well) was measured as described in control, no strain; , mean value of control; F, mean value of cyclicMaterials and Methods. Keratinocytes progressively aligned perpen- strain. Values are presented as mean { SEM, n Å 150.dicular to the force vector, which was toward the centerof well. b:

tion also is supported by our finding that 24 hr of expo- depending on medium were observed (Fig. 2b), signifi-cant strain-induced changes were noted as comparedsure to cyclic strain significantly enhanced the [3H] in-

corporation into DNA as compared with the time 0 with control (stationary condition) both in standardand quiescence media. Prior to exposure to strain, cellpoint (Fig. 2a). Although differences in proliferation

8925 560D/ 8925$$560d 08-11-97 19:53:35 wlcpal W Liss: JCP

EFFECT OF STRAIN ON KERATINOCYTES 69

Fig. 4. Comparison of discontinuous and continuous strain. a: Kera- elongation was no longer apparent. b: Keratinocytes were subjectedtinocytes were subjected to cyclic strain (10 cpm) for 2 days and cul- to cyclic strain for 7 days (continuous); the photograph presented heretured under the stationary condition for more 5 days; the morphologi- was taken on day 7, which showed that cells remained oriented andcal changes were assessed on day 7. After cyclic strain was discon- elongated. Crystal violet stain. Scale bar Å 100 mm.tinued, keratinocytes returned to a nonspecific distribution and

cycle synchronization was achieved by incubation with progressively aligned perpendicular to the force vector.The same tendency occurred with regard to cell elonga-this quiescence medium for 24 hr, revealing no signifi-

cant increase in cell growth under stationary condition tion, which became more discernible in the late phase(days 5–7). In endothelial cells, these events occurred(data not shown). However, strain induced an enhance-

ment of cell growth in quiescence medium, suggesting mainly in the periphery of the culture well, which isdue to the mechanical characteristics of the flexiblethat strain-induced proliferation may not be due to the

effects of growth factors contained in standard medium. membrane. When this membrane is subjected to 150mmHg vacuum, there is a 24% strain in the peripheryOur study also showed that human keratinocytes

(Fig. 4) and F-actin (Fig. 5) align perpendicular to the vs. less than 1% in the center of the flexible well (Sum-pio et al., 1988; Awolesi et al., 1995). Our results repre-direction of force, which is consistent with previous

findings that longitudinal strain causes alignment of sent the average of one total well, thus experimentalvalues may be underestimated. The strain dependenceother cells such as vascular endothelial cells, cardiac

cells, and fibroblasts (Sumpio et al., 1988; Terracio et needs to characterized in another study.Once the mechanical stimulus was removed, the ker-al., 1988). Although these findings are common re-

sponses observed in different cell types, diversities atinocytes returned to normal distribution (randompattern), as seen in Figure 4. In contrast, keratinocytesamong cell types have been noted (Sumpio et al., 1988;

Du et al., 1995). Therefore, individual characteristics remained elongated and aligned under the continuousstrain condition. It is of interest that other investiga-of each cell type need to be addressed independently.

In our studies, alignment was more prominent with tors have reported that expanded skin gradually re-turns to the previous condition histologically followingexposure to cyclic strain than that of control (24.2 {

1.67 vs. 40.4{ 2.27, respectively). Keratinocytes became removal of implanted tissue expander (Gibson, 1977;

8925 560D/ 8925$$560d 08-11-97 19:53:35 wlcpal W Liss: JCP

TAKEI ET AL.70

Fig. 5. Keratinocytes were subjected to cyclic strain for 7 days and ment perpendicular to the force vector (arrow) and cells were elon-stained with rhodamine phalloidin, as described in Materials and gated. Control cells (stationary) showed fine filaments and nonspecificMethods. F-actin in keratinocytes under cyclic strain revealed align- distribution of F-actin (n Å 6). Scale bar Å 20 mm.

TABLE 2. Total protein levels in human keratinocytes under ble intracellular signaling pathways for the observeddifferent conditions1

effects of strain on keratinocytes, we analyzed cAMP,PKA, and PGE2. A role of protein synthesis throughDay 3 Day 5the cAMP pathway has been reported by Kollros et al.

Control 4.92 { 0.07 7.59 { 0.31 (1987). They noted that elevated cAMP levels inhibitedConstant strain 5.33 { 0.26 7.93 { 0.36net protein synthesis, especially collagen, in rabbit ar-Cyclic strain 5.57 { 0.41 8.93 { 0.30*terial smooth muscle cells that were subjected to cyclic

1Keratinocytes were subjected to constant or cyclic strain (10 cycles per min, strain. An increase in the levels of cAMP in skin fibro-150 mmHg) or cultured in the stationary condition, as described in Materialsand Methods. Total protein values (mg/plate) were done in triplicate and are blasts also resulted in a decrease of the net collagenexpressed as mean { SEM. amount, implying that collagen and protein synthesis*P õ 0.05 is considered to be statistically significant vs. control (stationary) orconstant strain. are inversely related to cAMP level. Although more

studies need to be performed, it could be postulatedthat a decrease in cAMP and PKA levels might induceprotein synthesis. In the present study, the level ofCurtis and Seehar, 1978). This finding is confirmingcAMP in keratinocytes subjected to cyclic strain wasevidence that continuous strain is required for thelower than that of control but also stayed at the sus-strain-induced alignment and elongation.tained level, as shown in Figure 6a, probably becauseAnother finding of tissue expansion is an increase inkeratinocytes were incubated with IBMX to inhibit en-protein production (Johnson et al., 1988; Knight et al.,dogenous phosphodiesterase activity, so that cAMP1990). We also have demonstrated a significant in-continued to accumulate in control cells during the ex-crease in protein production by keratinocytes subjected

to cyclic strain for 5 days (Table 2). To investigate possi- periment. However, the amount of cAMP in keratino-

8925 560D/ 8925$$560d 08-11-97 19:53:35 wlcpal W Liss: JCP

EFFECT OF STRAIN ON KERATINOCYTES 71

production of PGE2, and a decrease in PKA. Althoughdirect association between two events has not been de-scribed and responses of PGE2, cAMP, and PKA arerelatively earlier than that of protein synthesis, this invitro finding may suggest a close relation with the netgain of protein production that is usually observed aftertissue expansion in vivo (Johnson et al., 1988; Knightet al., 1990).

Although these factors may influence protein synthe-sis, it is still unclear whether mediators such as PGE2or cAMP are major determinants. Other factors includ-ing transforming growth factor b (Miyazono and Hel-din, 1992; Schmid et al., 1993) and epidermal growthfactor (Boonstra et al., 1995) also have been suggestedto play an important role in regulating cellular mecha-nism. In addition, several other hypotheses need to beaddressed. One possibility is that mechanical strainmight directly influence intracellular organelles, suchas microfilaments, and stimulate cell growth (Wessellset al., 1971; Curtis and Seehar, 1978). Electromicro-scope studies (Pasyk et al., 1982) have demonstratedlarger bundles of tonofilaments and cytoskeletal fila-ments of epithelial cells in expanded skin. In addition,the orientation of F-actin filaments changes in culturedendothelial cells subjected to strain (Iba and Sumpio,1991). We also observed that F-actin in human kera-tinocytes aligned under strain condition (Fig. 5), im-plying that the alignment or qualitative changes ofthese filaments may be responsible for signals to thenucleus resulting in an enhanced proliferation. Ingberet al. (1994) suggested a tensegrity model that explainsvarious cellular responses such as morphogenesis,growth, and migration through an intracellular cy-toskeletal network. Another hypothesis has focused onthe intercellular junction or biomechanical change ofthe surface membrane. This hypothesis is based on thefact that intercellular spaces change after tissue expan-

Fig. 6. Effects of cyclic strain on cAMP (a), PKA (b), and PGE2 (c). sion (Pasyk et al., 1982), suggesting that structurala: Keratinocytes were subjected to cyclic strain (10 cpm) or stationary or molecular changes of cell membrane may have acondition for 1, 2, 5, 10, and 30 min. The cAMP levels were assayed at important role of proliferation. It had been previouslyeach time point, as described in Materials and Methods. Experimental

noted that keratinocytes become cytostatic when theyvalues remained lower than that of control. b: Keratinocytes weresubjected to cyclic strain for 15 and 30 min. The ratio of PKA activity reach an optimal density or increased the surface areato control (time 0 point) was measured. The level of PKA in cells large enough to relax tension (Squier, 1980; Austad etsubjected to strain showed a gradual decrease during the time course al., 1982, 1986).of strain. c: Keratinocytes were subjected to cyclic strain for 5 days,

In summary, our in vitro studies demonstrate thatand then PGE2 measurements were performed, as described in Mate-rials and Methods. PGE2 levels in cells subjected to strain significantly keratinocytes subjected to cyclic strain exhibit stimula-declined on day 5, whereas control levels remained stable. Values are tion of proliferation, elongation, alignment, and proteinmeans { SEM, n Å 6. *P õ 0.05 vs. control (a,c: no strain) or time 0 synthesis. In addition, we present evidence of a possible(b). Units of these levels are: cAMP, pmoles/ml/50,000 cells; PKA,

mediator pathway because cyclic strain of keratino-total PKA activity of strain/that of time 0; PGE2, pg/ml.cytes caused reductions in levels of cAMP, PKA, andPGE2. Our findings also support the hypothesis thatkeratinocytes may play a key role in the observed ef-cytes exposed to strain did not increase, suggesting an

inhibition of cAMP production rather than an increase fects of tissue expansion and other strain-dependentresponses present in skin.in turnover.

PGE2 measurements revealed a decrease on day 5,LITERATURE CITEDin cyclic or constant strain, indicating that strain may

downregulate PGE2 synthesis and lead to a reduction Austad, E.D., Pasyk, K.A., McClatchey, K.D., and Cherry, G.W. (1982)Histomorphologic evaluation of guinea pig skin and soft tissue afterof cAMP. Serum-free keratinocyte medium is fatty acidcontrolled tissue expansion. Plast. Reconstr. Surg., 70:704–710.deficient (Gamer et al., 1996), which could explain a

Austad, E.D., Thomas, S.B., and Pasyk, K. (1986) Tissue expansion:decrease in lipid metabolism (Oku, 1996). Therefore, in Divided or loan. Plast. Reconstr. Surg., 78:63–68.vitro values of our studies may underestimate those Awolesi, M.A., Sessa, W.C., and Sumpio, B.E. (1995) Cyclic strain

upregulates nitric oxide synthase in cultured bovine aortic endothe-changes observed in in vivo studies. Our finding alsolial cells. J. Clin. Invest., 96:1449–1454.was supported by a previous report that the level of

Baker, S.R. (1991) Fundamentals of expanded tissue. Head NeckPGE2 is inversely related to the protein synthesis level 13:327–333.(Kollros et al., 1987). Our findings in the present study Banes, A.J., Gilbert, J., Taylor, D., and Monbureau, O.A. (1985) A new

vacuum-operated stress-providing instrument that applies static orappear to be associated with a decrease in cAMP, less

8925 560D/ 8925$$560d 08-11-97 19:53:35 wlcpal W Liss: JCP

TAKEI ET AL.72

variable duration cyclic tension on compression to cells in vitro. J. Lansman, J.B., Hallam, T.J., and Rink, T.J. (1987) Single strain-activated ion channels in vascular endothelial cells as mechano-Cell. Sci., 75:35–37.

Banes, A.J., Link, J.R., Gilbert, J.W., Tay, R.T., and Monbureau, O. transducers. Nature, 325:811–813.Lew, D., and Fuseler, J.W. (1991) The effect of stepwise expansion on(1990) Culture cells in a mechanically active environment. Am. Bio-

tech. Lab., 8:12–22. the mitotic activity and vascularity of subdermal tissue and inducedcapsule in the rat. J. Oral Maxillofac. Surg., 49:848–853.Berridge, M.J., Dawson, R.M.C., Downes, C.P., Heslop, J.P., and Ir-

vine, R.F. (1983) Changes in the levels of inositol phosphates after Lorber, M., and Milobsky, S.A. (1968) A straining of the skin in vivo.A method of influencing cell division and migration in the rat epider-agonist-dependent hydrolysis of membrane phosphoinositides. Bio-

chem. J., 212:473–482. mis. J. Invest. Dermatol., 51:395–402.Machida, B.K., Liu-Shindo, M., Sasaki, G.H., Chandrasoma, P., andBoonstra, J., Rijken, P., Humbel, B., Cremers, F., Verkleij, A., and

van Bergen en Henegouwen, P. (1995) The epidermal growth factor. Rice, D.H. (1989) Immediate vs chronic tissue expansion. Paperpresented at the Fifth International Symposium on Facial PlasticCell. Biol. Int., 19:413–430.

Bradford, M.M. (1976) A rapid and sensitive method for the quantita- and Reconstructive Surgery, Toronto, Canada.Mackenzie, I.C. (1974) The effects of frictional stimulation on mousetion of microgram quantities of protein utilizing the principle of

protein-dye binding. Anal. Biochem., 72:248–54. ear epidermis. J. Invest. Dermatol., 63:194–198.Miyazono, K., and Heldin, C.H. (1992) Structure, function and possi-Breidahl, A.F., Judson, R.T., and Clunie, G.J.A. (1989) Review of kera-

tinocyte culture techniques: Problems of growing skin. Aust. N.Z. ble clinical application of transforming growth factor-b. J. Derma-tol., 19:644–647.J. Surg., 59:485–497.

Bullough, W.S. (1972) The control of epidermal thickness. Br. J. Der- Morris, C.E., and Sigurdson, W.J. (1989) Strain-inactivated ion chan-nels coexist with strained-activated ion channels. Science, 243:807–matol., 87:187–199.

Curtis, A.S.G., and Seehar, G.M. (1978) The control of cell division 809.Myer-Franke, A., Kaplan, M.R., Pfrieger, F.W., and Barres, B.A.by tension or diffusion. Nature, 274:52–53.

(1995) Characterization of the signaling interactions that promoteDu, W., Mills, I., and Sumpio, B.E. (1995) Cyclic strain causes hetero-the survival and growth of developing retinal ganglion cells in cul-geneous induction of transcription factors, AP-1, CRE binding pro-ture. Neuron, 15:805–819.tein and NF-kB, in endothelial cells: Species and vascular bed diver-

Oelsen, S., Clapham, D.E., and Davies, P.F. (1988) Haemodynamicsity. J. Biomech., 28:1485–1491.shear stress activates a K/ current in vascular endothelial cells.Eisinger, M. (1985) Cultivation of normal human epidermal keratino-Nature, 331:169–170.cytes and melanocytes. In: Methods in Skin Research. D. Skerrow

Pasyk, K.A., Austad, E.D., McClatchey, K.D., and Cherry, G.W. (1982)and C.J. Skerrow, eds. John Wiley & Sons, New York, pp. 191–Electron microscopic evaluation of guinea pig skin and soft tissues212.expanded with a self-inflating silicone implant. Plast. Reconstr.Francis, A.J., and Marks, R. (1977) Factors responsible for dermallySurg., 70:37–45.induced epidermal hyperplasia. Arch. Dermatol. Res., 258:275–280.

Schmid, P., Cox, D., Bilbe, G., Mcmaster, G., Morrison, C., Stahelin,Gibson, T. (1977) The physical properties of skin. In: ReconstructiveH., Luscher, N., and Seiler, W. (1993) TGF-bs and TGF-b type IIPlastic Surgery. J.M. Converse, ed. W.B. Saunders, Philadelphia,receptor in human epidermis: Differential expression in acute andVol. 1, pp. 70–77.chronic skin wounds. J. Pathol., 171:191–197.Gibson, T., Kenedi, R.M., and Craik, J.E. (1965) The mobile microar-

Squier, C.A. (1980) The straining of mouse skin in vivo: Effect onchitecture of dermal collagen: A bioengineering study. Br. J. Surg.,epidermal proliferation and thickness. J. Invest. Dermatol., 74:68–52:764–769.71.Gilbert, J.A., Banes, A.J., Link, G.W., and Jones, G.L. (1989) Surface

Sumpio, B.E., Banes, A.J., Buckle, M., and Johnson, G. (1988) Alter-strain of living cells in a mechanically active in vitro environment. ation in aortic endothelial cell morphology and cytoskeletal proteinIn: ANSYS Conference Proceedings. D.E. Dietrich, ed. Swanson synthesis during cyclic tensional deformation. J. Vasc. Surg., 7:130–Analysis System, Houston, PA, pp. 13.2–13.7. 138.Iba, T., and Sumpio, B.E. (1991) Morphological response of human Terracio, L., and Borg, T.K. (1986) Alteration in cell distribution andendothelial cells subjected to cyclic strain in vitro. Microvasc. Res., cytoskeleton during linear strain in vitro. J. Mol. Cell. Cardiol.,42:245–254. 18:329.Ingber, D.E., Dike, L., Hansen, L., Karp, S., Liley, H., Maniotis, A., Terracio, L., Miller, B., and Borg, T.K. (1988) Effects of cyclic mechani-

McNamee, H., Mooney, D., Plopper, G., Sims, J., and Wang, N. cal stimulation of the cellular components of the heart. In Vitro(1994) Cellular tensegrity: Exploring how mechanical changes in Cell. Dev. Biol., 24:53–58.the cytoskeleton regulate cell growth, migration, and tissue pattern Vanderkolk, C.A., McCann, J.J., Mitchell, G.M., and O’Brien, B.M.during morphogenesis. Int. Rev. Cytol., 150:173–224. (1988) Changes in area and unexpanded axial pattern skin flaps in

Johnson, P.E., Kernahan, D.A., and Bauer, B.S. (1988) Dermal and pigs. Br. J. Plast. Surg., 41:284–293.epidermal response to soft tissue expansion in the pig. Plast. Re- Wang, N., Butler, J.P., and Ingber, D.E. (1993) Mechanotransductionconstr. Surg., 81:390–397. across the cell surface and through the cytoskeleton. Science,

Johnson, T.M., Lowe, L., Brown, M.D., Sullivan, M.J., and Nelson, 260:1124–1127.B.R. (1993) Histology and physiology of tissue expansion. J. Derma- Wessells, N.K., Spooner, B.S., Ash, J.F., Fradley, M.O., Luduena,tol. Surg. Oncol., 19:1074–1078. M.A., Taylor, E.L., Wrenn, J.T., and Yamada, K.M. (1971) Microfil-

Knight, K.R., McCann, J.J., Vanderkolk, C.A., Coe, S.A., and aments in cellular and developmental processes. Science, 171:135–O’Brien, B. (1990) The redistribution of collagen in expanded pig 143.skin. Br. J. Plast. Surg., 43:565–570. Wu, J., Dent, P., Jelinek, T., Wolfman, A., Weber, M.J., and Sturgill,

Kollros, P.R., Bates, S.R., Mathews, M.B., Horwitz, A.L., and Glagov, T.W. (1993) Inhibition of the EGF-activated MAP kinase signalingS. (1987) Cyclic AMP inhibits increased collagen production by cycli- pathway by adenosine 3*,5*-monophosphate. Science, 262:1065–

1069.cally strained smooth muscle cells. Lab. Invest., 56:410–417.

8925 560D/ 8925$$560d 08-11-97 19:53:35 wlcpal W Liss: JCP