Oral and skin keratinocytes are stimulatedto secrete monocyte chemoattractantprotein-1 by tumour necrosis factor-a andinterferon-g

Jie LiPaula M. FarthingMartin H. Thornhill

Oral Disease Research Centre,St Bartholomew’s and The Royal LondonSchool of Medicine and Dentistry, QueenMary and Westfield College, London, UK

Correspondence to:Martin H. ThornhillOral Disease Research Centre, Clinical SciencesBuilding, St Bartholomew’s and The Royal LondonSchool of Medicine and Dentistry, 2, Newark Street,London E1 2AD, UK

Accepted for publication October 28, 1999

Copyright C Munksgaard 2000J Oral Pathol Med . ISSN 0904-2512

Printed in Denmark . All rights reserved

438

Abstract: Monocyte chemoattractant protein-1 (MCP-1) is amember of the CC subfamily of chemokines. It is chemoattract-ant for monocytes, activated memory T cells and dendritic cells.We studied MCP-1 mRNA and protein production in cultured hu-man oral keratinocytes (OK) and skin keratinocytes (SK). MCP-1production was undetectable in resting keratinocytes. However,stimulation with tumour necrosis factor-a (TNF-a) or interferon-g (IFN-g) induced MCP-1 mRNA and protein synthesis in bothcell types. Together, TNF-a and IFN-g acted synergistically to in-crease MCP-1 production. In each case MCP-1 production wasgreater for SK than OK. The kinetics of IFN-g-induced MCP-1production were similar for both cell types, peaking around 72h. In contrast, TNF-a induced a more rapid increase in MCP-1production by SK than OK, peak production occurring after 24and 72 h, respectively. IL-1a and IL-4 did not induce MCP-1production. However, in combination with IFN-g, they induced asmall extra increase in MCP-1 production in SK but not OK.

Key words: chemokine; interferon-g; keratinocyte; MCP-1; oral;skin; TNF-a

J Oral Pathol Med 2000: 29: 438–44

Monocyte chemoattractant protein-1 (MCP-1), also known as mono-

cyte chemotactic and activating factor (MCAF), is a member of the

CC or b subfamily of chemokines. Human MCP-1 cDNA encodes a

99-amino acid precursor protein with a 23-residue signal peptide

that is cleaved to generate the 76-residue mature protein (1). MCP-

1 has been shown to selectively chemoattract and activate mono-

cytes both in vitro and in vivo. In addition, it stimulates monocyte

degranulation (1, 2). More recently, two further monocyte chemo-

attractants, MCP-2 and MCP-3, have been identified. These show

62% and 71% homology, respectively, with MCP-1 (3). Both have

monocyte chemotactic activity, but are less potent than MCP-1 (3,

4). More recently, MCP-1 has also been found to strongly chemoat-

tract T lymphocytes, particularly activated memory T cells (4, 5),

and transgenic mouse studies have shown that MCP-1 is also re-

sponsible for recruiting dendritic and Langerhans’ cells to skin (6)

Keratinocyte MCP-1 production

Although few cells produce MCP-1 constitutively, several cells

have been found to produce MCP-1 in response to cytokine stimula-

tion (7–10). In particular, skin keratinocytes have been shown to

produce MCP-1 following activation with interferon-g (IFN-g) (11).

Moreover, MCP-1 production has been demonstrated by dermal

keratinocytes in psoriasis (12), delayed type hypersensitivity reac-

tions (13), lichen planus (14) and wound healing (15, 16). In each

case MCP-1 production by dermal keratinocytes is thought to play

an important role in recruiting the focal accumulation of a mono-

nuclear cell dominated inflammatory infiltrate.

Mononuclear cell infiltrates are also a feature of a number of

oral mucosal conditions, including oral lichen planus (17) and peri-

odontal disease (18, 19). It is possible, as with the skin, that release

of MCP-1 by keratinocytes plays an important role in recruiting this

type of inflammatory infiltrate. It is also possible that keratinocyte

MCP-1 production is important in regulating Langerhans’ and den-

dritic cell traffic in the oral mucosa and skin (6, 11). The aim of this

study was, therefore, to investigate MCP-1 production by normal

human oral keratinocytes (OK), and for comparison skin (SK) kera-

tinocytes, and to investigate how MCP-1 production may be regu-

lated by pro-inflammatory cytokines.

Material and methods

Cytokines

Recombinant human (rh) interferon-gamma (IFN-g) was obtained

from Biogen SA (Geneva, Switzerland). Rh tumour necrosis factor-

alpha (TNF-a) was a gift from Dr. B. A. Beutler (University of Texas

Health Sciences Centre, Dallas, TX, USA). Interleukin-1a (IL-1a) and

interleukin-4 (IL-4) were purchased from R&D System (Cowley, Ox-

ford, UK). Concentrations used were 0, 1, 10, 100, 250 and 1000 U/

ml of IFN-g and TNF-a; 0, 1, 10, 25 and 100 U/ml of IL-1a and IL-

4.

Keratinocyte culture

Keratinocytes were prepared from fresh biopsies of normal human

skin or oral mucosa using a modification (20) of the method de-

scribed by Rheinwald & Green (21). Skin keratinocytes were ob-

tained from three women and one man (age range 32–56 years),

using excess skin removed during breast reduction or apronectomy

procedures. Oral keratinocytes were prepared from normal non-

keratinised labial or buccal sulcular oral mucosa removed during

frenectomy, gingival surgery or from surgical flap margins from one

man and three women (age range 28–72 years). The biopsies were

439J Oral Pathol Med 29: 438–44

treated with 0.25% trypsin in Tris saline (Sigma, Poole, Dorset,

UK) at 37æC for 2 h to separate the epithelium/epidermis, and the

keratinocytes were dispersed with forceps to obtain a single cell

suspension. The cells were placed onto feeder layers of mitomycin

C-treated (Sigma) (10 mg/ml, 2 h, 37æC) mouse 3T3 fibroblasts and

cultured at 37æC in a humidified atmosphere containing 5% CO2 in

growth medium consisting of 3 parts Dulbecco’s modified Eagle’s

medium (Gibco, Paisley, Scotland, UK) and 1 part Hams F12 me-

dium supplemented with 10% foetal bovine serum, 10 ng/ml epider-

mal cell growth factor, 0.4 mg/ml hydrocortisone, 5 mg/ml transfer-

rin, 5 mg/ml insulin, 1.8¿10ª4 M adenine, 1¿10ª10 M cholera toxin,

50 U/ml penicillin, 50 mg/ml streptomycin (all from Sigma) and 0.25

mg/ml fungizone (Gibco). All experiments were performed on third

to fourth passage cultures. For experiments, the cells were re-plated

into serum free medium (Gibco) in the absence of feeder layers. The

cells were cultured until just confluent and were then treated for

various times with cytokines. The supernatants were collected and

stored at ª70æC until required for MCP-1 protein quantification and

the keratinocyte monolayers were lysed for RNA preparation. Each

experiment was repeated on three or more occasions. In addition,

experiments were replicated with cells from at least three different

donors. No significant difference in the results was found between

different donors. For kinetic studies samples were collected at 0, 4,

8, 24, 48 and 72 h following cytokine stimulation. Samples for other

studies were taken at the times indicated.

MCP-1 quantification

MCP-1 concentrations in keratinocyte culture supernatants were

measured using a commercial MCP-1 specific sandwich ELISA as-

say (R&D Systems). The detection limits of the assay were 3.13–

2000 pg/ml and samples were diluted 4–20 times to bring them into

the detection range.

Northern blot analysis

Total cellular RNA was extracted from monolayers of untreated and

cytokine-treated keratinocytes using the UltrasepcTM RNA isolation

system (AMS Biotechnology UK Ltd, Itmey, Oxford, UK). Northern

blot analysis was performed as decribed previously (22) using a

MCP-1 oligonucleotide probe cocktail (BPR 249; R&D Systems) con-

taining an equimolar mixture of 30 base length 5ø, mid and 3ø region

probes with double-end DIG labelling to detect MCP-1 specific

mRNA. The relative density of the signal from each band of the

Northern blots was quantified by scanning laser densitometry (LKB

Ultrascaner). The intensity of each band on the x-ray film was ex-

pressed as the area under the curve obtained from laser densitome-

Li et al.

try. Ethidium bromide staining of 18 S ribosomal RNA (rRNA) was

used to quantify the total amount of RNA loaded onto each lane

and to allow correction for any difference in the loading of each

lane.

Statistics

Differences between the results of experimental treatments were

evaluated by means of the two-tailed Student’s t-test.

Results

TNF-a and IFN-g induce MCP-1 production by oral

keratinocytes (OK) and skin keratinocytes (SK)

Neither oral nor skin keratinocytes produced MCP-1 constitutively.

However, stimulation with 250 U/ml TNF-a or IFN-g for 20 h in-

duced significant levels of MCP-1 production in culture super-

natants of both cell types, although the level produced by SK was

significantly higher than that produced by OK (P∞0.05) (Fig. 1).

Dose response and kinetic studies showed that peak production

of MCP-1 occurred after keratinoyctes were stimulated with 250–

1000 U/ml of TNF-a or 1000 U/ml of IFN-g (data not shown). For

IFN-g-stimulated keratinocytes, peak MCP-1 production occurred

Fig. 1. TNF-a and IFN-g induce MCP-1 production by oral keratinocytes(OK) and skin (SK) keratinocytes. OK and SK were incubated for 20 h withplain medium (BL), TNF-a (250 U/ml), IFN-g (250 U/ml), IL-1a (100 U/ml),IL-4 (10 ng/ml) or combinations of these cytokines. The MCP-1 concentrationin the culture supernatants was then determined by ELISA. For each pointthe standard deviation did not exceed 5% of the mean value. The resultsshown are representative of three similar experiments.

440 J Oral Pathol Med 29: 438–44

after 72 h (the last time point tested), and for TNF-a stimulated

keratinocytes peak production occurred between 24 and 72 h (Fig.

2). Similar concentration and kinetic studies following IL-1a and IL-

4 stimulation of OK and SK revealed no induction of MCP-1 produc-

tion (Fig. 1).

Effect of cytokine combinations on MCP-1 production

When used in combination, TNF-a and IFN-g had a strong syner-

gistic effect (P∞0.01) on MCP-1 production by both OK and SK (Fig.

1). Combinations of IFN-gπIL-1a and IFN-gπIL-4 also induced a

small extra increase in MCP-1 production by SK but not OK, when

compared with the effect of IFN-g alone (P50.05). However, TNF-

Fig. 2. Kinetics of MCP-1 production by keratinocytes stimulated with IFN-g or TNF-a. Oral keratinocytes (OK) and skin (SK) keratinocytes were incu-bated with a) 250 U/ml IFN-g or b) 250 U/ml TNF-a for different periods oftime. The MCP-1 concentration in the culture supernatants was then deter-mined by ELISA. For each point the standard deviation did not exceed 5%of the mean value. The results shown are representative of three similarexperiments.

Keratinocyte MCP-1 production

Fig. 3. The effect of different cytokines on keratinocyte MCP-1 mRNA syn-thesis. Total cellular RNA was extracted from oral keratinocytes (OK) andskin (SK) keratinocytes incubated for 18 h with plain medium (BL), TNF-a(250 U/ml), IFN-g (250 U/ml), IL-1a (100 U/ml), IL-4 (10 ng/ml) or combi-nations of these cytokines. MCP-1 mRNA was detected by Northern blotanalysis using a MCP-1 specific oligonucleotide probe. Ethidium bromidestaining of 18S rRNA was used to quantify the total amount of RNA loadedonto each lane.

aπIL-1a and TNF-aπIL-4 in combination had no significant syner-

gistic effect (Fig. 1).

Effect of cytokines on MCP-1 mRNA synthesis

Northern blot analysis showed that unstimulated OK and SK con-

tained little or no mRNA for MCP-1. However, stimulation with

TNF-a or IFN-g resulted in a large and rapid increase in MCP-1

mRNA levels (Fig. 3) that proceeded protein production. MCP-1

mRNA peaked after 4 h stimulation with IFN-g in both cell types

and after 4 h stimulation with TNF-a in OK and 24–48 h stimula-

tion in SK (Fig. 4).

When used in combination, TNF-a and IFN-g acted synergist-

ically to increase MCP-1 mRNA transcription in OK and SK. Combi-

nations of IFN-gπIL-1a and IFN-gπIL-4 also had a synergistic ef-

fect on MCP-1 mRNA level in SK but not in OK (Fig. 3).

Discussion

Inflammatory mucocutaneous diseases are characterised, by the ac-

cumulation of different populations of leukocytes in the tissues. In

conditions such as lichen planus, lichenoid reactions, psoriasis,

graft-versus-host disease (17) and periodontal disease (18, 19),

monocytes together with T lymphocytes accumulate in the skin or

oral mucosa. These leave the circulation to accumulate in the con-

nective tissue and may invade the epidermis or epithelium. This is

441J Oral Pathol Med 29: 438–44

Fig. 4. Kinetics of MCP-1 mRNA synthesis in keratinocytes stimulated withTNF-a or TNF-g. OK and SK were stimulated with 250 U/ml IFN-g (a) or250 U/ml TNF-a (b) for different periods of time. Total RNA was extractedand MCP-1 mRNA was detected by hybridisation with a MCP-1 specificoligo-probe. The relative density of the signal from each lane was determinedby scanning laser densitometry and plotted as a percentage of the maximalMCP-1 mRNA signal.

a multi-stage, highly selective process in which chemokines and

adhesion molecules work together to ensure that only selected sub-

populations of leukocytes leave the circulation and are directed to

migrate to the affected area (23).

Margination and rolling of leukocytes along the endothelial lin-

Li et al.

ing of blood vessels is the first step in this process. Endothelial cell

expression of selectin family adhesion molecules, such as E-selectin,

is important in allowing leukocytes to make this initial loose contact

(24). However, to turn this into firm adhesion, the leukocyte needs

to be activated so that it can also bind the adhesion molecule inter-

cellular adhesion molecule-1 (ICAM-1) on the endothelial cell sur-

face. The interaction with ICAM-1 results in firm adhesion and

transmigration into the surrounding tissues.

Recent studies have shown that this activation is caused by

chemokines, such as MCP-1, increasing the avidity of leukocyte inte-

grin binding for ICAM-1 (25, 26). The wide diversity of chemokines

and the different patterns of chemokine receptor expression on

leukocytes combine to form a highly selective system for ensuring

that only particular sub-populations of leukocytes are able to mi-

grate into the tissues. The precise combination of chemokines re-

leased determine the nature of the infiltrate that develops. The im-

portance of MCP-1 in promoting the selective adhesion and trans-

migration of monocytes via this mechanism was recently

demonstrated in in vitro experiments on vascular endothelium

under flow conditions (26). Memory T cells (4, 5), dendritic cells and

Langerhans’ cells (6) also express MCP-1 receptors (27, 28), and it

is likely that MCP-1 is involved in the firm adhesion and trans-

migration of these cells as well. Clearly, the release of MCP-1 by

epidermal or oral keratinocytes will, therefore, have an important

influence over the nature of the inflammatory infiltrate recruited

from the circulation.

Once in the tissues, the distribution of MCP-1 receptors on the

leukocyte surface enables them to detect the surrounding concen-

tration gradient of MCP-1 and directs their migration toward the

area of highest concentration. The ability of keratinocytes to release

MCP-1 will set up a concentration gradient within the connective

tissue, directing leukocyte migration towards the epithelial or epi-

dermal basement membrane. This may well explain the sub-epi-

thelial band-like distribution of mononuclear cells in lichen planus

and the accumulation of leukocytes in the sub-junctional epithelial

connective tissue in periodontal disease.

In lichen planus, some leukocytes cross the basement membrane

and invade the epidermis or epithelium, while in periodontal disease

large numbers of leukocytes migrate across the junctional epithel-

ium into the gingival crevice. The induction or expression of ICAM-

1 on keratinocytes may be important in permitting this invasion

(20, 29–31). However, the keratinocyte chemokine environment may

determine which cells are able to do this by selectively activating

specific sub-populations of leukocytes in much the same way as

chemokines determine which cells interact with ICAM-1 at the endo-

thelial cell surface.

The specific pattern of chemokines released by keratinocytes

442 J Oral Pathol Med 29: 438–44

will determine the pattern of the resulting inflammatory infiltrate.

Epidermal keratinocytes have been found to be capable of produc-

ing the CXC chemokines IL-8 (22, 32, 33), growth regulated peptide-

a (GRO-a) (34), IP10 (35) and MIG (14), and the CC chemokines

RANTES (Regulated and normal T-cell expressed and secreted) (22),

and MCP-1 (11). Normal oral keratinocytes have previously been

found to produce IL-8 and RANTES (22), and mRNA for IL-8, MCP-

1 and GRO-g has been detected in a human embryonal palatal mu-

cosa cell line and an oral cancer cell line under differing conditions

(36). In this study we have demonstrated that normal oral keratino-

cytes can also produce MCP-1 and confirmed that dermal keratino-

cytes can do so.

MCP-1 mRNA and protein production was low or undetectable

in skin or oral keratinocytes under resting conditions. However, the

pro-inflammatory cytokines TNF-a and IFN-g were able to induce

MCP-1 production in both cell types in a time and dose dependent

manner. Together they had a synergistic effect on the production of

both MCP-1 mRNA and protein. Barker et al. (11) have previously

reported MCP-1 production by skin keratinocytes following IFN-g

stimulation. They also noted synergy between TNF-a and IFN-g for

MCP-1 production, although in their study TNF-a alone had no

effect.

In our experiments, IL-1a and IL-4 had no effect on MCP-1 pro-

duction although in combination with IFN-g they did induce a small

extra increase in production by skin keratinocytes but not oral kera-

tinocytes. Barker et al. (11) also found IL-1 had no effect on MCP-1

production by skin keratinocytes. In contrast, Bickel et al. (36) re-

ported increased levels of MCP-1 mRNA in a human embryonal

palatal mucosa cell line and an oral cancer cell line stimulated with

IL-1a. However, this response was only seen under certain con-

ditions of serum exposure that differed between the two cell lines.

Our experiments, and those of Barker et al., were performed in

serum-free medium on normal skin and oral keratinocytes, and this

probably explains the difference between our observations.

TNF-a and IFN-g are produced by activated Th1 lymphocytes.

TNF-a may also be produced by keratinocytes both from skin (37)

and oral mucosa (38). TNF-a and IFN-g from both of these sources

are likely to stimulate keratinocytes to secrete MCP-1. In turn this

would recruit further mononuclear cells that would secrete more

TNF-a and IFN-g that would further enhance MCP-1 production.

The normal resting concentration of TNF-a and IFN-g in the tissues

is very low. However, at sites of inflammation the concentration

rises to quite high levels. The concentrations used in this study

(0–1000 U/ml) fall within the range of tissue concentrations where

immunomodulatory effects occur and are widely used by investi-

gators studying the effects of these cytokines in vitro.

As well as recruiting a mononuclear inflammatory cell infiltrate

Keratinocyte MCP-1 production

it seems likely that MCP-1 would recruit antigen presenting den-

dritic cells to the area, including some with a Langerhans’ cell

phenotype (6, 11). This would be consistent with the observed in-

crease in both dendritic (39, 40) and Langerhans’ cell (40) number

and turnover (41) in lichen planus and similar conditions.

In conclusion, we have shown that the pro-inflammatory cyto-

kines TNF-a and IFN-g can induce oral and skin keratinocytes to

produce the chemokine MCP-1 that is chemotactic for monocytes,

activated T cells and dendritic cells. Further studies will be needed

to determine what role MCP-1 plays in recruiting mononuclear cells

and dendritic cells in periodontal disease and mucocutaneous con-

ditions such as lichen planus.

References

1. Yoshimura T, Yuhki N, Moore SK, Appella E, Lerman MI, Leonard EJ.Human monocyte chemoattractant protein-1 (MCP-1). FEBS Lett 1989;244: 487–93.

2. Zachariae CO, Anderson AO, Thompson HL, et al. Properties of mono-cyte chemotactic and activating factor (MCAF) purified from a humanfibrosarcoma cell line. J Exp Med 1990; 171: 2177–82.

3. Van Damme J, Proost P, Lenaerts JP, Opdenakker G. Structural and func-tional identification of two human, tumor-derived monocyte chemotacticproteins (MCP-2 and MCP-3) belonging to the chemokine family. J ExpMed 1992; 176: 59–65.

4. Taub DD, Proost P, Murphy WJ, et al. Monocyte chemotactic protein-1(MCP-1), -2, and -3 are chemotactic for human T lymphocytes. J ClinInvest 1995; 95: 1370–6.

5. Carr MW, Roth SJ, Luther E, Springer TA. Monocyte chemoattractantprotein 1 as a T-lymphocyte chemoattractant. Proc Natl Acad Sci USA1994; 91: 3652–6.

6. Nakamura K, Williams IR, Kupper TS. Keratinocyte-derived monocytechemoattractant protein 1 (MCP-1): analysis in a transgenic model dem-onstrates MCP-1 can recruit dendritic and Langerhans cells to skin. JInvest Dermatol 1995; 105: 635–43.

7. Strieter RM, Wiggins R, Phan SH, et al. Monocyte chemotactic proteingene expression by cytokine-treated human fibroblasts and endothelialcells. Biochem Biophys Res Commun 1989; 162: 694–700.

8. Larsen CG, Zachariae CO, Oppenheim JJ, Matsushima K. Production ofmonocyte chemotactic and activating factor (MCAF) by human dermalfibroblasts in response to interleukin 1 or tumor necrosis factor. BiochemBiophys Res Commun 1989; 160: 1403–8.

9. Kristensen MS, Deleuran BW, Larsen CG, Thestrup-Pedersen K, PaludanK. Expression of monocyte chemotactic and activating factor (MCAF)in skin related cells. Cytokine 1993; 5: 520–4.

10. Rollins BJ, Pober JS. Interleukin-4 induces the synthesis and secretion ofMCP-1/JE by human endothelial cells. Am J Pathol 1991; 138: 1315–9.

11. Barker JN, Jones ML, Swenson CL, et al. Monocyte chemotaxis and activ-ating factor production by keratinocytes in response to IFN-gamma. JImmunol 1991; 146: 1192–7.

12. Gillitzer R, Wolff K, Tong D, et al. MCP-1 mRNA expression in basal

443J Oral Pathol Med 29: 438–44

keratinocytes of psoriatic lesions. J Invest Dermatol 1993; 101: 127–31.13. Yu X, Barnhill RL, Graves DT. Expression of monocyte chemoattractant

protein-1 in delayed type hypersensitivity reactions in the skin. Lab In-vest 1994; 71: 226–35.

14. Spandau U, Toksoy A, Goebeler M, Brocker EB, Gillitzer R. MIG is adominant lymphocyte-attractant chemokine in lichen planus lesions. JInvest Dermatol 1998; 111: 1003–9.

15. Engelhardt E, Toksoy A, Goebeler M, Debus S, Brocker EB, Gillitzer R.Chemokines IL-8, GROalpha, MCP-1, IP-10, and MIG are sequentiallyand differentially expressed during phase-specific infiltration of leuko-cyte subsets in human wound healing. Am J Pathol 1998; 153: 1849–60.

16. Gibran NS, Ferguson M, Heimbach DM, Isik FF. Monocyte chemoattract-ant protein-1 mRNA expression in the human burn wound. J Surg Res1997; 70: 1–6.

17. Porter SR, Kirby A, Olsen I, Barrett W. Immunologic aspects of dermaland oral lichen planus: a review. Oral Surg Oral Med Oral Pathol OralRadiol Endod 1997; 83: 358–66.

18. Zappa U, Reinking-Zappa M, Graf H, Espeland M. Cell populations andepisodic periodontal attachment loss in humans. J Clin Periodontol 1991;18: 508–15.

19. Johannessen AC, Nilsen R, Kristoffersen T, Knudsen GE. Variation inthe composition of gingival inflammatory cell infiltrates. J Clin Peri-odontol 1990; 17: 298–305.

20. Li J, Mahiouz KL, Farthing PM, Haskard DO, Thornhill MH. Heteroge-neity of ICAM-1 expression, and cytokine regulation of ICAM-1 express-ion, in skin and oral keratinocytes. J Oral Pathol Med 1996; 25: 112–8.

21. Rheinwald JG, Green H. Serial cultivation of strains of human epidermalkeratinocytes: the formation of keratinizing colonies from single cells.Cell 1975; 6: 331–44.

22. Li J, Ireland GW, Farthing PM, Thornhill MH. Epidermal and oral kera-tinocytes are induced to produce RANTES and IL-8 by cytokine stimula-tion. J Invest Dermatol 1996; 106: 661–6.

23. Imhof BA, Dunon D. Leukocyte migration and adhesion. Adv Immunol1995; 58: 345–416.

24. Luscinskas FW, Ding H, Tan P, Cumming D, Tedder TF, Gerritsen ME.L- and P-selectins, but not CD49d (VLA-4) integrins, mediate monocyteinitial attachment to TNF-alpha-activated vascular endothelium underflow in vitro. J Immunol 1996; 157: 326–35.

25. Campbell JJ, Hedrick J, Zlotnik A, Siani MA, Thompson DA, Butcher EC.Chemokines and the arrest of lymphocytes rolling under flow conditions.Science 1998; 279: 381–4.

26. Gerszten RE, Garcia-Zepeda EA, Lim YC, et al. MCP-1 and IL-8 triggerfirm adhesion of monocytes to vascular endothelium under flow con-ditions. Nature 1999; 398: 718–23.

27. Neote K, Digregorio D, Mak JY, Horuk R, SchalL TJ. Molecular cloning,functional expression, and signaling characteristics of a c-c chemokinereceptor. Cell 1993; 72: 415–25.

28. Charo IF, Myers SJ, Herman A, Franci C, Connolly AJ, Coughlin SR.Molecular cloning and functional expression of two monocyte chemoat-tractant protein 1 receptors reveals alternative splicing of the carboxyl-terminal tails. Proc Natl Acad Sci USA 1994; 91: 2752–6.

29. Dustin ML, Singer KH, Tuck DT, Springer TA. Adhesion of T lympho-blasts to epidermal keratinocytes is regulated by interferon-g and ismediated by intercellular adhesion molecule-1 (ICAM-1). J Exp Med 1988;167: 1323–40.

30. Nickoloff BJ, Griffiths CEM, Barker JNWN. The role of adhesion mol-ecules, chemotactic factors, and cytokines in inflammatory and neoplas-tic skin disease – 1990 update. J Invest Dermatol 1990; 94: 151S-7S.

Li et al.

31. Moughal NA, Adonogianaki E, Thornhill MH, Kinane DF. Endothelialcell leukocyte adhesion molecule-1 (ELAM-1) and intercellular adhesionmolecule-1 (ICAM-1) expression in gingival tissue during health and ex-perimentally-induced gingivitis. J Periodont Res 1992; 27: 623–30.

32. Larsen CG, Anderson AO, Oppenheim JJ, Matsushima K. Production ofinterleukin-8 by human dermal fibroblasts and keratinocytes in responseto interleukin-1 or tumour necrosis factor. Immunology 1989; 68: 31–6.

33. Barker JNWN, Jones ML, Mitra RS, et al. Modulation of keratinocyte-derived interleukin-8 which is chemotactic for neutrophils and Tlymphocytes. Am J Pathol 1991; 139: 869–76.

34. Richmond A, Thomas HG. Melanoma growth stimulatory activity: iso-lation from human melanoma tumors and characterization of tissue dis-tribution. J Cell Biochem 1988; 36: 185–98.

35. Kaplan G, Luster AD, Hancock G, Cohn ZA. The expression of a gammainterferon-induced protein (IP-10) in delayed immune responses in hu-man skin. J Exp Med 1987; 166: 1098–108.

36. Bickel M, Nothen SM, Freiburghaus K, Shire D. Chemokine expressionin human oral keratinocyte cell lines and keratinized mucosa. J Dent Res1996; 75: 1827–34.

37. Kock A, Schwarz T, Kirnbauer R, et al. Human keratinocytes are asource for tumor necrosis factor a: evidence for synthesis and release

444 J Oral Pathol Med 29: 438–44

upon stimulation with endotoxin or ultraviolet light. J Exp Med 1990;172: 1609–14.

38. Yamamoto T, Osaki T, Yoneda K, Ueta E. Cytokine production by kera-tinocytes and mononuclear infiltrates in oral lichen planus. J Oral PatholMed 1994; 23: 309–15.

39. Regezi JA, Daniels TE, Saeb F, Nickoloff BJ. Increased submucosal factorXIIIa-positive dendrocytes in oral lichen planus. J Oral Pathol Med 1994;23: 114–8.

40. Akasu R, From L, Kahn HJ. Lymphocyte and macrophage subsets inactive and inactive lesions of lichen planus. Am J Dermatopathol 1993;15: 217–23.

41. Walton LJ, Macey MG, Thornhill MH, Farthing PM. Intra-epithelial sub-populations of T lymphocytes and Langerhans cells in oral lichenplanus. J Oral Pathol Med 1998; 27: 116–23.

Acknowledgements

The authors acknowledge the support received for this study from a RoyalSociety Grant.