Laryngeal keratinocytes show variable inhibition of replication by TGF-/J

TERESA P. DiLORENZO* and BETTIE M. STEINBERG

Department of Otolaryngology and Communicative Disorders, Long Island Jewish Medical Center, New Hyde Park, NY 11042, USA

* Author for correspondence

Summary

The response of secondary human laryngeal epi-thelial cells to transforming growth factor-/* (TGF-/J)was investigated, and this response was comparedwith that of epithelial cells derived from virallyinduced laryngeal papillomas. In most cases, bothnormal laryngeal epithelial cells and those derivedfrom laryngeal papillomas exhibited growth inhi-bition in response to 10 ng ml"1 TGF-/J. Response wasnot a function of cell density or proliferation rate

when cells were in a low-calcium medium, but wasreduced in high calcium. Using keratinocytes de-rived from several different tissue explants, we foundthat cells grown from different explants showmarked variation in response to TGF-/3.

Key words: TGF-/3, laryngeal, keratinocytes, papilloma.

Introduction

Transforming growth factor-/} (TGF-/3) is a regulator of cellgrowth and differentiation that affects a number of differ-ent cell types and elicits a wide variety of cellular re-sponses (for review, see Sporn et al. 1987). TGF-/J stimu-lates the growth of fibroblastic cells (Leof et al. 1986). Incontrast, the proliferation of foreskin keratinocytes andmany other normal epithelial cell types is inhibited byTGF-/S (Moses et al. 1986; Shipley et al. 1986; Jetten et al.1986; Lin et al. 1987; Masui et al. 1986).

Little is known about the response of human laryngealepithelial cells to growth regulators such as TGF-/3. Thegoals of the present study were to investigate the responseof these cells to TGF-/3 and to compare their response tothat of epithelial cells derived from laryngeal papillomas.Papillomas of the larynx are benign neoplasms caused byone or more of the human papillomaviruses. The papillo-mas frequently recur and can be life-threatening becausethey obstruct the airway. Histologically, the papillomasexhibit hyperplasia of the spinous layer, leading to anoverall thickening of the epithelium. We were interestedin knowing whether the escape from normal growthcontrols exhibited by papillomas is due to an alteredresponse to TGF-/3.

Our findings show that, in most cases, both normallaryngeal epithelial cells and those derived from laryngealpapillomas exhibit growth inhibition in response to TGF-/S. However, cells derived from different tissue explantsshow varying levels of response to this growth regulator.

Materials and methods

CellsPrimary human laryngeal epithelial cells were cultured aspreviously described (Steinberg et al. 1982) from biopsy fragmentsof normal and papillomatous tissues.Journal of Cell Science 96, 115-119 (1990)Printed in Great Britain © The Company of Biologists Limited 1990

Assay for bromodeoxyuridine incorporationExperimental cultures were plated on glass coverslips in MCDB153 (Clonetics; Boyce and Ham, 1983) supplemented with epider-mal growth factor, retuioic acid, bovine pituitary extract, insulin,transferrin and hydrocortisone. All experiments were done in thiscomplete medium and in 0.1-0.2 mM calcium unless otherwisestated.

When cultures neared confluence, they were exposed tolOngml"1 (unless otherwise noted) TGF-/J (Collaborative Re-search) for 4 days and to 3 //gml"1 bromodeoxyuridine (BrdU) for24 h during the last day of exposure to TGF-/S. Control culturesreceived BrdU alone. Paired coverslips were always plated, fed,labeled and stained at the same time. The cells were fixed with70% ethanol at 4°C and stained for BrdU using an indirectimmunofluorescence technique. The primary anti-BrdU antibodywas purchased from Becton Dickinson and used as directed. Thesecondary antibody was purchased from Dako. Brightly fluor-escent nuclei were scored as positive. At least 10 fields werecounted per coverslip. The significance of the difference betweentreated and control means was determined by Student's <-test. Inmost cases, response was measured on a single set of coverslips foreach explant. When sufficient cells were available for duplicatecoverslips, they were in very good agreement.

Involucrin and keratin stainsCultures to be stained for involucrin or keratins were plated onglass coverslips. For the involucrin stain, cells were fixed with 4 %formaldehyde in phosphate-buffered saline and stained for in-volucrin using the Vectastain ABC immunoperoxidase procedure(Vector Laboratories). The primary anti-involucrin antibody wasobtained from Biomedical Technologies, Inc. The peroxidasesubstrate used was diaminobenzidine.

For the keratin stain, cells were fixed with cold (-20°C)methanol and stained for keratins using an indirect immunoflu-orescence technique. The primary antibody used was a mixture ofAE1 and AE3 (Woodcock-Mitchell et al. 1982), which weregenerously provided by Dr T.-T. Sun. The secondary antibody waspurchased from Dako.

Results

An immunological method for the detection of incorpor-

115

Fig. 1. Indirect immunofluorescence staining, using an anti-BrdU antibody, of laryngeal epithelial cells labeled with BrdUfor 24h (x390). A. Bright field. B. Immunofluorescence of same

ated 5-bromodeoxyuridine (BrdU; Gratzner, 1982) can beused as a measure of cell proliferation. Davison et al.(1979) have reported that thymidine incorporation maynot be a valid indicator of keratinocyte proliferation. Theyfound that [3H]thymidine incorporation into DNA wasinconsistent with direct counts of cell number in studies onthe proliferation of epidermal keratinocytes. Their datasuggest that rapidly growing keratinocytes switch from asalvage pathway for DNA biosynthesis to de nouo nucleo-tide synthesis. Because of the limitation in the use of[3H]thymidine as an indicator of keratinocyte prolifer-ation, we used BrdU labeling as a measure of proliferationof normal laryngeal keratinocytes and those derived fromlaryngeal papillomas. Fig. 1 shows normal laryngeal epi-thelial cells that were incubated with BrdU for 24 h andthen stained for BrdU using the immunofluorescencetechnique described in Materials and methods. The cellshaving brightly fluorescent nuclei were in the S-phase ofthe cell cycle at some point during the 24-h labelingperiod.

This study was conducted using cells grown from bi-opsies of human laryngeal epithelium. All biopsies weretaken from either the true or false vocal cords, which arecovered by a non-keratinizing stratified squamous epi-thelium in vivo. The cultures derived from these biopsiescontained greater than 95 % epithelial cells, as determinedby morphology and staining with anti-keratin antibodies

Fig. 2. Cultured normal human laryngeal epithelial cells.A. Indirect immunofluorescent staining using anti-keratinantibodies AE1 and AE3 (x500). B. Phase-contrast micrographof live cells growing in MCDB 153 with 0.1 mM calcium (x80).C. Indirect immunoperoxidase staining using an anti-involucrinantiserum (xl90).

(Fig. 2A). The cells seen in Fig. 2B were cultured in theserum-free medium MCDB 153 containing 0.1 mM cal-cium. They exhibit the 'cobblestone' appearance typical ofepithelial cells in culture. When the laryngeal cells aregrown in 1 mM calcium, the individual cells show markedflattening, the culture stratifies, and some of the cellssynthesize involucrin (Fig. 2C), which is a precursor pro-

116 T. P. DiLorenzo and B. M. Steinberg

100

75

50

25

0"0 2 4 6

TGF-/3(ngmr')10

Fig. 3. Inhibition of DNA synthesis by TGF-/5 in cultures ofnormal human laryngeal epithelial cells. Cells from a singletissue explant were exposed to varying concentrations of TGF-/Jfor 4 days and to BrdU for 24 h during the last day of exposureto TGF-/3. Control cultures were exposed to BrdU alone.

tein of the keratinocyte's cornified envelope. The responseof laryngeal keratinocytes to high calcium concentrationis similar to that exhibited by keratinocytes derived fromhuman epidermis (Boyce and Ham, 1983). It should benoted that cells cultured from laryngeal papilloma bi-opsies have previously been shown to contain humanpapillomavirus DNA (Steinberg et al. 1983).

To determine the effect of TGF-/3 on the growth ofnormal human laryngeal epithelial cells, bromodeoxyuri-dine incorporation was measured in cultures exposed tolngml" 1 (40pM), S n g m r 1 (200pM), and lOngmT1

(400 PM) TGF-/S. The results for cells derived from a singletissue explant (Fig. 3) show that significant inhibition ofproliferation occurs at a TGF-/J concentration oflOngml"1. For this reason, lOngml"1 was the concen-tration used in all subsequent experiments.

We investigated whether cells at different densitieswould respond similarly to TGF-/3 (Table 1). Normallaryngeal epithelial cells derived from a single tissueexplant were seeded at 5 XlO3 cm"2 and allowed to growfor 3, 7 or 10 days before a 4-day exposure to TGF-/3. At 3days after seeding, cultures were sparse. At 7 days, theywere approximately half-confluent, and at 10 days theywere confluent. Cells were labeled with BrdU for 24 hduring the last day of exposure to TGF-f3. Labeled nucleiand total nuclei in 15 fields were counted, and labelingindices for control and TGF-/S-treated cultures were calcu-lated. Although the labeling index of both the control andtreated cultures varied widely depending on cell density,the ratio of the labeling index in treated and controlcultures remained fairly constant. Therefore, expressingthe effect of TGF-/J on BrdU labeling as a percentage of thecontrol eliminates any significant variability due toplating efficiency or cell number. In all subsequent exper-

iments cells were exposed to TGF-/3 when cultures werenearly confluent, to increase the number of cells per fieldand thus the accuracy of counts.

Note that although the labeling index is reduced ap-proximately fivefold in all cultures treated with TGF-/3(see Table 1), the cell number is not reduced proportion-ately. This would be expected in a culture where only asubpopulation of the cells are dividing, but where the non-dividing cells persist and are counted. This is what occurswhen keratinocytes are grown in culture. Upon plating,the cells form small colonies that spread outward from thecenter. As growth continues, cell division slows downwithin the central region of the colony and a proportion ofthe cells permanently withdraw from the cell cycle. Pro-liferation continues at the periphery of the colonies.Suppose 7 x 104 keratinocytes were seeded in each of twocultures, 5/7 of the cells in the control culture weredividing and 1/7 of the cells in the TGF-/5-treated culturewere dividing. After two doublings of the dividing sub-populations, 2.1 xlO6 cells would be counted in the controlculture and 9xl0 4 cells would be counted in the treatedculture. Since the non-dividing cells persist, cell number isreduced less than threefold, while the labeling index isreduced fivefold. In this study, we were interested in theeffect of TGF-/3 on the subpopulation of dividing cells, solabeling indices, rather than total cell counts, weremeasured.

Fig. 4A shows the effect of 10 ngml"1 TGF-/3 on bothnormal and papilloma-derived secondary laryngeal epi-thelial cultures. Cells from three of the four differentnormal explants and five of the six papilloma explantsshowed significant inhibition of proliferation, but theextent of inhibition varied from tissue to tissue. Wetherefore conclude that cells derived from both normallaryngeal epithelium and laryngeal papillomas respondsimilarly to TGF-/3. Most cultures examined showed inhi-bition of proliferation, although different tissue explantsshowed varying levels of response to TGF-/3, with somespecimens actually exhibiting a stimulation of prolifer-ation.

Reiss and Sartorelli (1987) have shown that in theabsence of epidermal growth factor (EGF), the prolifer-ation of foreskin keratinocytes is not inhibited by TGF-/3,implying that actively growing cultures are most suscep-tible to the effects of TGF-/3. Although we saw no markeddifference in inhibition as a function of rate of prolifer-ation (Table 1), mitogenically stimulated cells might showan altered response to TGF-/3 compared to unstimulatedcells. Bovine pituitary extract and retinoic acid, like EGF,promote cell division of keratinocytes. The effect of TGF-/3on four secondary normal laryngeal cultures in the ab-

Table 1. Effect ofTGF-fi on BrdU incorporation in normal laryngeal epithelial cells at different cell densitiesCells cm"2 afterlabeling period Labeled nuclei* Total nuclei" Labeling index (%)

Control

A 4.3 XlO4

B 1.5xlO6

C 2.0x10*

+TGF-0

1.6 xlO4

5.8xlO4

1.3 x 10s

Control

427(28±12)

842(66±13)

252(17±9)

+TGF-0

31(2.1±1.7)

86(5.7±2.7)

27(1.8±2.9)

Control

557(37 ±14)

1997(133±9)

2694(173±20)

+TGF-/3

215(14±5)

753(60±ll)

1716(114±28)

Control

77

42

10

Cells were seeded at 5x 103 cm"3 and allowed to grow for 3 (row A), 7 (row B), or 10 (row C) days before a 4-day exposure to TGF-/9.labeled with BrdU for 24 h during the ftnal day of exposure to TGF-/3. Controls were exposed to BrdU alone.

* Values given are totals for 15 fields. NumberB in parentheses are mean±8.D. per field.

+TGF-/3

14

11

2

Cultures were

Laryngeal keratinocytes and TGF-fi 117

160

140

8 iooa?

3C

H 60

.340

20

0

*•

•

o

•

o*

o

oo

B

Fig. 4. Effect of TGF-/3 on DNA synthesis in different explantcultures of normal (O) and papilloma-derived ( • ) laryngealepithelial cells. Experimental cultures were exposed tolOngml"1 TGF-0 for 4 days and to BrdU for 24 h during thelast day of exposure to TGF-/3. Control cultures were exposed toBrdU alone. (*, difference between experimental and controlvalues was not statistically significant.) Column A, cells weregrown in the complete medium described in Materials andmethods. Column B, cells were grown in the absence of EGF,bovine pituitary extract and retinoic acid.

Table 2. Effect ofTGF-fi on BrdU incorporation inpapilloma-derived laryngeal epithelial cells under

various culture conditionsLabeled nuclei (% of control)

0.1 mM calcium 1 mM calcium

+EGF -EGF +EGF -EGF

Specimen 1Specimen 2

38*49*

16*40*

67t69t

65t59*

*P<0.003, compared to control without TGF-0.t.P<0.05, compared to control without TGF-0.$Did not differ significantly from control.

sence of EGF, bovine pituitary extract and retinoic acidwas determined (Fig. 4B). Under these conditions, three ofthe four cultures still showed significant inhibition inresponse to lOngrrJ-1 TGF-/S.

Calcium concentration is also known to affect growthand differentiation of keratinocytes (Boyce and Ham,1983; Rubin and Rice, 1986; Hennings et al. 1980). Table 2shows the effect of lOngmT1 TGF-£ on two differentlaryngeal papilloma specimens grown in 0.1 mM or lmMcalcium and with or without EGF. It appears that TGF-/3has a more inhibitory effect in the presence of 0.1 mMcalcium than in 1 mM calcium. The response of specimens1 and 2 to TGF-̂ 3 in 0.1 mM calcium is another example ofthe variability in the response to TGF-/3 seen with cellsderived from different primary cultures. Unfortunately,sufficient primary cells were not available to determine ifthere were any differences in EGF receptor expressionbetween the two specimens. As with normal laryngeal

Table 3. Lack of relationship between patient age,diagnosis or labeling index of control culture and the

effect of TGF-fi on BrdU incorporation in normallaryngeal epithelial cells

Age

778

35533

16536

•Did

Diagnosis

Subglottic stenosisLaryngeal stenosisVocal cord nodulesSubglottic stenosisLaryngeal papillomaVocal cord polypsSubglottic stenosisLaryngeal web

not differ significantly

Labeling index (%)of control

7512493347721546

from control

Labeled nuclei withTGF-0 (% control)

9194553576486*

157

keratinocytes (Fig. 4B), Table 2 also shows that EGF doesnot need to be present for papilloma-derived cells to showgrowth inhibition in response to TGF-/S.

Treatment of laryngeal epithelial cells with TGF-/3induced changes in cellular morphology. Treated cellsexhibited some increased spreading and flattening whencompared with untreated controls (data not shown). How-ever, this morphology does not indicate an effect of TGF-/3on differentiation, as control cultures and cultures treatedwith TGF-/3 for 4 days showed approximately equal num-bers of cells that were positive for involucrin when cul-tured in 1 mM calcium (data not shown).

Since normal laryngeal epithelial cells derived fromdifferent tissue explants showed a variable response toTGF-/J (Fig. 4), we investigated whether the degree ofresponse to TGF-/J was related to diagnosis or age of thepatient. Note that in no case was the biopsy taken at thesite of the lesion. For example, if a vocal cord polyp waspresent on the left side of the larynx, the biopsy was takenfrom the opposite side. Table 3 lists eight different tissuesin order of decreasing inhibition of DNA synthesis inresponse to TGF-/S. The age and diagnosis of the patientare also given. No obvious relationship was found betweeneither of these variables and the response to TGF- f3. Therewas also no correlation between the level of proliferationin the control cultures and inhibition by TGF-/3. This wasnot surprising in light of the data presented in Table 1.

Discussion

Using cells grown from several different tissue explants,we have found that, in most cases, TGF-/3 inhibits theproliferation of both normal laryngeal epithelial cells andepithelial cells derived from laryngeal papillomas. This isin agreement with previous reports that have demon-strated that TGF-/? inhibits the proliferation of a variety ofepithelial cell types, including human foreskin keratino-cytes (Moses et al. 1985) and human bronchial cells (Masuiet al. 1986). Moreover, there was no difference between thenormal cells and those derived from the virally inducedneoplasms. However, cells derived from different tissueexplants responded to varying degrees. All of the culturesstudied contained greater than 95% epithelial cells, asdetermined by morphology and staining with anti-keratinantibodies. Therefore, the variable response to TGF-/5 seenfrom culture to culture is not due to differences in thenumber of fibroblasts present in each culture. It alsocannot be due to variations in the ability of differentcultures to activate TGF-/3 (Lyons et al. 1988), becauseTGF-/3 as supplied by Collaborative Research is the active

118 T. P. DiLorenzo and B. M. Steinberg

form of the molecule. We do not believe that it reflectsexperimental error, because on those occasions whensufficient cells were available for duplicate coverslips,there was good agreement (data not shown).

Three structurally distinct cell-surface glycoproteinsthat specifically bind TGF-̂ 3 have been identified in avariety of cell types, and these three types of TGF-/3receptors bind ligand with different affinities (Massague,1985; Massague and Like, 1985; Cheifetz et al. 1986). Onelikely explanation for the variable response of laryngealepithelial cells to TGF-/3 is that cells derived from differentexplant cultures may possess variable numbers of TGF-/3receptors, or their receptors may bind TGF-/3 with differ-ent affinities. This hypothesis was not tested here, becausethe limited number of primary cells available from alaryngeal explant culture precludes performing both pro-liferation and receptor assays on the same sample.Methods are available for the expansion of human kera-tinocyte cultures to larger cell numbers (Green et al. 1979).However, all experiments described here were done onprimary cells that had been cultured for as short a time aspossible. This was done because extended time and pass-age in culture could result in the selection of a subpopu-lation of cells whose response to TGF-/3 would not berepresentative of the original explant culture. In mostcases, only a single doseofTGF-^dOngml"1) was used inthis study. Therefore, it is possible that differences inreceptor number and ligand affinity could account for thevariable response. Higher doses might have had an effecton the cultures that did not respond to lOngml"1 TGF-/3.However, the limited number of cells available did notallow multiple concentrations to be tested in most cases.

The precise mechanism by which TGF-/3 inhibits epi-thelial cell growth is largely unknown, and the intracellu-lar signaling mechanism activated by TGF-/3-receptorinteraction has not yet been elucidated. However, anotherpossible explanation for the variable response of laryngealkeratinocytes to TGF-/3 is that there are variations fromspecimen to specimen that occur distal to TGF-/3 ligand-receptor binding.

Calcium concentration is known to have an effect on theproliferation of human keratinocytes in culture. Boyce andHam (1983) have shown that the growth rate of humanepidermal keratinocyte cultures increases between0.03 mM and 0.3 mM calcium, and that between 0.3 mM and1 mM growth rate is not changed. Similarly, we have foundthat the proliferation of human laryngeal keratinocytes isincreased when the calcium concentration is raised from0.1 mM to lmM (data not shown). Table 2 indicates thatTGF-/3 has a more inhibitory effect on laryngeal keratino-cytes in the presence of 0.1 mM calcium than in 1 mMcalcium. It may be that those cells that continue to divideeven in the sub-optimal calcium concentration of 0.1 mMare somewhat more sensitive to the effect of TGF-/J than isthe general population. It is also possible that calciumplays some role in the regulation of receptor expression orthe intracellular signaling mechanism triggered byTGF-/S-receptor interaction.

Although further work must be done to determine thecause or causes for the variability in response of laryngealkeratinocytes to TGF-/3, the data reported here do indicatethat, when working with tissue explants, conclusionscannot necessarily be drawn on the basis of the study of asingle explant culture. It is important to repeat exper-iments with samples from different sources before con-clusions are drawn.

This work was supported by grant 2PO1NS19214 from theNational Institute of Neurological and Communicative Disorders

and Stroke (now 8-POl DC00203 from the National Institute onDeafness and other Communication Disorders). We thank DrsAllan Abramson and Mark Shikowitz for the generous contri-bution of tissue specimens.

References

BOYCB, S. T. AND HAM, R. G. (1983). Calcium-regulated differentiation ofnormal human epidermal keratinocytes in chemically denned clonalculture and serum-free Berial culture. J. invest. Derm. (Supplement)81, 33-40.

CHEIFETZ, S., LIKE, B. AND MASSAGUB, J. (1986). Cellular distribution oftype I and type II receptors for transforming growth factor-/). J. bwl.Chem. 261, 9972-9978.

DAVISON, P., LIU, S. AND KARASKK, M. (1979). Limitations in the use of[3H]thymidine incorporation into DNA as an indicator of epidermalkeratinocyte proliferation in vitro. Cell Tiss. Kinet. 12, 605-614.

GRATZNEB, H. G. (1982). Monoclonal antibody to 5-bromo- and 5-iododeoxyuridine: a new reagent for detection of DNA replication.Science 218, 474-475.

GREEN, H., KEHINDK, O. AND THOMAS, J. (1979). Growth of culturedhuman epidermal cells into multiple epithelia suitable for grafting.Proc. natn. Acad. Sci. U.S.A. 76, 5665-5668.

HENNINOS, H., MICHAEL, D., CHENG, C, STEINERT, P., HOLBROOK, K. ANDYUSPA, S H. (1980). Calcium regulation of growth and differentiationof mouse epidermal cells in culture. Cell 19, 245-254.

JETTEN, A. M., SHIRLEY, J. E. AND STONER, G. (1986). Regulation ofproliferation and differentiation of respiratory tract epithelial cells byTGF/3. Expl Cell Res. 167, 639-549.

LKOF, E. B., PROPER, J. A., GOUSTIN, A. S., SHIPLEY, G. D., DICORLETO, P.E. AND MOSES, H. L. (1986). Induction of c-sis mRNA and activitysimilar to platelet-derived growth factor by transforming growth factor/): a proposed model for indirect mitogenesis involving autocrineactivity. Proc. natn. Acad. Sci. U.SA. 83, 2453-2467.

LIN, P., LIU, C., TSAO, M.-S. AND GRBHAM, J. W. (1987). Inhibition ofproliferation of cultured rat liver epithelial cells at specific cell cyclestages by transforming growth factor-/). Bwchem biophys Res.Commun. 143, 26-30.

LYONS, R. M., KESKI-OJA, J. AND MOSES, H. L. (1988). Proteolyticactivation of latent transforming growth factor-/? from fibroblast-conditioned medium. J. Cell Bwl. 106, 1659-1665.

MASSAGUE, J. (1986). Subunit structure of a high affinity receptor fortype /? transforming growth factor. Evidence for a disulflde-linkedglycosylated receptor complex. J. biol. Chem. 260, 7059-7066.

MASSAGUE, J. AND LIKE, B. (1986). Cellular receptors for type p"transforming growth factor: ligand binding and affinity labeling inhuman and rodent cell Iine3. J. biol. Chem. 260, 2636-2645.

MASUI, T., WAKEFIELD, L. M., LECHNER, J. F., LAVECK, M. A., SPORN, M.B. AND HARRIS, C. C. (1986). Type fl transforming growth factor U theprimary differentiation-inducing serum factor for normal humanbronchial epithelial cells. Proc. natn. Acad. Sci. U.S.A. 83, 2438-2442.

MOSES, H. L., TUCKER, R. F , LEOF, E. B , COFFEY, R. J. JR, HALPER, J.AND SHIPLEY, G. D. (1985). Type-/? transforming growth factor is agrowth stimulator and a growth inhibitor. In Growth Factors andTransformation, Cancer Cells 3 (ed. J. Feramisco, B. Ozanne and C.Stiles), pp. 65-71. Cold Spring Harbor, New York: Cold Spring HarborPress.

RBISS, M. AND SARTORELU, A. C. (1987). Regulation of growth anddifferentiation of human keratinocytes by type f3 transforming growthfactor and epidermal growth factor Cancer Res. 47, 6706-6709.

RUBIN, A. L. AND RICE, R. H. (1986). Differential regulation by retinoicacid and calcium of transglutaminases in cultured neoplastic andnormal human keratinocytes. Cancer Res. 46, 2356-2361

SHIPLEY, G. D., PITTELKOW, M. R., WILLE, J. J JR, SCOTT, R. E. ANDMOSES, H. L. (1986). Reversible inhibition of normal humanprokeratinocyte proliferation by type fi transforming growth factor-growth inhibitor in serum-free medium. Cancer Res. 46, 2068-2071.

SPORN, M. B., ROBERTS, A. B., WAKEFIELD, L. M. AND DE CROMBRUGCHE,B. (1987) Some recent advances in the chemistry and biology oftransforming growth factor-beta. J. Cell Biol. 105, 1039-1045.

STEINBERG, B. M., ABRAMSON, A. L. AND MEADE, R. P. (1982). Culture ofhuman laryngeal papilloma cells in vitro. Otolaryngol. Head NeckSurg. 90, 728-735.

STEINBERG, B. M., TOPP, W. C., SCHNEIDER, P. S. AND ABRAMSON, A. L.(1983). Laryngeal papillomavirus infection during clinical remission.New Engl. J. Med. 308, 1261-1264.

WOODCOCK-MITCHELL, J., EICHNER, R., NELSON, W. G. AND SUN, T.-T.(1982) Immunolocalization of keratin polypeptides in humanepidermis using monoclonal antibodies. J. Cell Biol. 96, 580-588.

(Received 8 November 1989 - Accepted 1 February 1990)

Laryngeal keratinocytes and TGF-fi 119