In Three Types of Interface Dermatitis, Different Patterns of Expression of Intercellular Adhesion Molecule-l (ICAM-l) Indicate Different Triggers of Disease

Scott D. Bennion, Maljorie H. Middleton, * Kathleen M. David-Bajar, Sylvia Brice, * and David A. Norris*t Departments of Dermatology, Fitzsimons Army Medical Center and 'University of Colorado School of Medicine, and tDermatology

Service, Denver Department of Veterans Affairs Hospital, Denver; Colorado, U.S.A.

We found distinct patterns of intercellular adhesion molecule-l (ICAM-l) expression in three diseases characterized by interface dermatitis with mononu­clear infiltrates and keratinocyte cytotoxicity: lichen planus (LP), subacute cutaneous lupus erythematosus (SCLE), and erythema multiforme (EM). In LP, basal keratinocytes show strong ICAM-l expression associ­ated with a dermal infiltrate, but ICAM-l expression in the rest of the epidermis is minimal. In SCLE, there is diffuse epidermal ICAM-l expression, sometimes with accentuation on the cell surface of basal cells. In EM, there is strong basal cell expression of ICAM-l with evident cell surface accentuation, and also pockets of suprabasal expression with cell surface accentuation. These patterns are associated with dif­ferent factors that trigger cytokine release in differ­ent locations.

Both tumor necrosis factor-a (TNF-a) and inter­feron-l' (IFN-1') produce greater relative ICAM-l expression in basal keratinocytes than in more differ­entiated keratinocytes. In LP, the pure basal kerati­nocyte expression ofICAM-l appears to be caused by cytokines, predominantly IFN-l', released by dermal lymphocytes. The pattern of ICAM-l in SCLE corre­sponds to the pattern induced by ultraviolet radiation (UVR): diffuse epidermal ICAM-l expression, some-

Intercellular adhesion molecule-l (ICAM-l) (CD54) is a member of the immunoglobulin supergene family and is a single-chain glycoprotein weighing between 76 and 114 kD [1,2]. ICAM-l is the major ligand for leukocyte function associated antigen-l, an adhesion molecule found on all

leukocytes [3]. Both molecules are inducible by a number of cytokines and the affinity of leukocyte function associated antigen-l for ICAM-l can be increased by triggering numerous leukocyte

The views of the authors do not purport to reflect the views of the Department of the Army or the Department of the Defense.

Reprint requests to: Dr. David Norris, B144, Department of Derma to 1-ogy, University of Colorado Health Science Cente�, 4200 E. Ninth Avenue, Denver, CO 80262.

Abbreviations: EM, erythema multiforme; HSV, herpes simplex virus; LP, lichen planus.

times with basal accentuation. Some individuals are "responders" to TNF-a or UVR, showing high levels ofICAM-l expression following UVR or TNF-a stim­ulation in vitro or UVR stimulation in vivo. We pro­pose that the pattern ofICAM-l induction in SCLE is dependent on UVR-induced TNF-a release.

EM is associated with apparent latent Herpes sim­plex virus, and Herpes simplex virus (HSV)-infected keratinocytes show enhanced ICAM-l expression. We propose that in EM suprabasal ICAM-l expression may be induced directly by HSV infection or indi­rectly through TNF-a release induced by HSV reac­tivation.

Induction ofICAM-l within the epidermis is strat­ified and individually variable. Basal keratinocytes show maximal induction of ICAM-l expression due to innate sensitivity to TNF and IFN-l' stimulation, and to location adjacent to dermal sources of cyto­kines. Suprabasal ICAM-l can be induced by UVR and epidermal TNF-a release, and by factors such as viral infection. Different triggers of cytokine release and adhesion molecule induction may influence the different patterns of inflammation seen in diverse inflammatory skin diseases. Key words: TNF/IL-1Iinter­

feronllupus erythematosusllichen planus/erythema multi­forme. ] Invest Dermatol 105:71S-79S, 1995

cell-surface receptors [4 , 5] . This ligand/receptor pair is essential in many contact-dependent interactions between leukocytes and tis­sue cells, in specific antigen recognition by B and T lymphocytes, in superantigen stimulation ofT lymphocytes, in leukocyte migration into tissue, and in leukocyte-mediated cytotoxicity [6-8]. It has been proposed that this adhesion pair is crucial in directing activated leukocytes into tissue and in initiating and strengthening leukocyte attachment to parenchymal targets, facilitating specific recognition and subsequent cytotoxicity [9,10].

In contrast to endothelial cells, and other parenchymal cells, keratinocytes have very low constitutive expression of cell-surface ICAM-l [11]. The lack of expression of ICAM-l renders the resting epidermis resistant to interaction with leukocytes. Induction ofICAM-l expression in the epidermis is proposed as an important factor in the induction of leukocyte-dependent damage to kerati-

0022-202X/95/$09.50 • SSDI0022-202X(95)00184-M • Copyright © 1995 by The Society for Investigative Dermatology, Inc.

718

72S BENNION ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Figure 1. ICAM-1 staining in skin biopsies from LP, EM. and SCLE: immunohistochemical staining for ICAM-1 in biopsies from control normal skin (A). SCLE (B,C). EM (D), and LP (E,P). Sections were photographed at both medium (100X) and high power (400X).

VOL. lOS, NO. 1, SUPPLEMENT, JULY 1995

nocytes and melanocytes, especially immunologic cytotoxicity [9,12,13].

Induction of expression ofICAM-l on epidermal keratinocytes is found in many diverse inflammatory skin conditions including contact dermatitis, lichen planus (LP) , pemphigus, fixed drug reactions, erythema multiforme (EM), and subacute cutaneous lupus erythematosus (SCLE) [14-16]. It has been proposed by Nickolo1f et al that induction ofICAM-l may be a "final common pathway" in the pathophysiology of these inflammatory skin dis.,­eases [17].

We propose that the pattern of expression and the inducers of ICAM-l expression may vary in di1ferent skin diseases. We have identified three di1ferent patterns of ICAM-l expression in three diseases characterized by an interface dermatitis and keratinocyte cytotoxicity: LP, EM, and SCLE. These patterns can be explained by the stratified epidermal environment of induced ICAM-1 ex­pression and by di1ferent triggers of ICAM-l in the three diseases.

MATERIALS AND METHODS

Immunopathology Studies We obtained skin biopsies from four groups of six patients: nonnal controls (cosmetic plastic surgery), EM, LP, and SCLE. Punch biopsies were taken from lesional skin and also non­lesional skin of SCLE patients .

The tissue was snap-frozen and sectioned into 4-micron sections on a tissue-tek cryostat. The sections were air-dried and fixed in 100% acetone. These sections were washed three times in phosphate-bulfered saline (PBS) and incubated in blocking sera for 20 min . The tissue was incubated in an IgG1 monoclonal anti-human ICAM-l RRl/1 (graciously provided by Robert Rothlein, Boehringer-Ingelheim) for 30 min and washed in PBS. The sections were then stained ntilizing the avidin-biotin peroxidase complex immunoperoxidase method (ABC Vectastain, Vector Lab). The specificity of this antibody for rCAM-1 has been previously reported [2], and the usefulness in studying skin disease has been established [14]. Negative control sections used an IgGl isotype-matched control mouse anti-human IgA that does not bind normal keratinocytes.

Cell Cultures Human keratinocytes were cultured from neonatal fore­skins as previously described [13] in keratinocyte growth media (KGM, Clonetics, San Diego, CA), and studied at the time of third passage. Keratinocyte strains were usually obtained from a single foreskin source. Adult-derived keratinocytes were obtained from suction blister as described [18].

In some priming experiments, transfonned epithelial or keratinocyte cell lines were studied . KB and A431 cell lines were obtained from the American Type Culture Collection, and HaCaT cells were generously provided by Norbert Fusenig of the Deutsches Krebsforschungszentrum in Heidelberg, Gennany.

ICAM-l Expression Cells were cultured in six-well plates for 2 d followed by stimulation by 30 ng TNF-a/rul (Genzyme, 2 X 108 ng/mg, Cambridge , MA) or by 20 U IFN-')I (Genzyme, 2.5 X 107 U/mg). Cell-surface ICAM-l expression was measured by flow cytometry at the time of peak expression, which was 1 d following stimulation by IFN-')I and 3 d following stimulation by TNF-a [13].

Phototesting The effect of ultraviolet (UV) light on human skin and ICAM-l induction was carried out utilizing a Dennalight 2001 Sol 3 Solar Simulator (Dennalight Systems, Studio City, CA). This light source emits both UVB and UV A, but UVC and the lower range of the UVB spectrum are attenuated by filters; the emission spectrum of the source has been previously published [19]. The flux in the UVB spectrum was monitored with an International Light IL17001760D1791 radiometer (International Light, Newburyport, MA), and the time of exposure was adjusted to deliver 50 or 100 mJlcm2 of UVB. Five volunteers were exposed on the volar surface of the forearm to 50 m]lcm2 or to 100 m]lcm2 at the same site for three subsequent days or to 100 mJlcm2 on a single day. All five subjects were skin type II; the doses were roughly 1 and 2 times the minimal erythema dose (MED), respectively. Skin biopsies were taken from the volar forearm 24 h after the last UVR exposure and processed for ICAM-l immunoperoxidase staining.

Measuring ICAM-l Expression in Herpes Simplex Virus (HSV)­Infected Keratinocyte Cultures We utilized the technique previously described by Brice et al [20] for measuring the effects of HSV infection of keratinocyte cultures on rCAM-l expression. Human keratinocyte cultures were incubated with reagent grade acyclovir (Burroughs Wellcome, Gre­enville, NC) at 200 /LM to inhibit the pace of productive HSV infection.

EXPRESSION OF ICAM-l INDICATES DISEASE TRIGGERS 73S

Table I. ICAM-1 Patterns in Three Diseases with Keratinocyte Cytotoxicity

Disease Epidermal rCAM-I" Other Features

SCLE

EM

LP

Normal Skin

Dilfuse epidennal ICAM-l (1-2+) Sometimes basal layer accentuation

Dense basal layer ICAM-l (3-4+) Pockets of positive suprabasal staining (3-4+)

Intense basal layer ICAM-l (3- 4+)

Negative for epidermal ICAM-l

Faint leukocyte staining in the dermis

Intense leukocyte staining in the dermis

Intense leukocyte ICAM-l in the dermis

Mild endothelial cell rCAM-l

a Staining intensity: 1 +, taint staining, patchy; 2 +, more positive, more unifo.rril; 3+. more positive, uniform, cell surface; 4+, very positive, unifo� cell surface.

The cultures were infected at a multiplicity of infection of 1.0. At different times following infection, the cultures were harvested and ICAM-l expres­sion was measured quantitatively by fluorescence-activated cell sorter (FACS) analysis. These experiments were repeated four times. In addition,

at 24 h, ICAM-l expression induced by HSV infection was compared to that induced by IFN-')I versus medium control.

Priming of Cytokine Responses To determine whether pre-treatment ("priming") with small amounts of TNF-a would enhance the subsequent response to optimal doses of cytokine, keratinocytes were incubated with a single suboptimal dose ofTNF-a (the "priming" dose) for 3 d. This TNF-a was then washed away and 30 ng TNF-alml was added. and the cells incubated for an additional 3 d before assay for ICAM-l expression.

Cytokines Human recombinant TNF-a and IFN-')I from Genzyme were used throughout.

RESULTS

Three Distinct Patterns ofICAM-l Expression In the three different diseases studied, the expression ofICAM-1 in the epider­mis showed three di1ferent patterns as summarized in Table I and illustrated in Fig 1. In normal skin, ICAM-l expression is unde­tectable in the epidermis, and limited in dermal vessels (Fig 1A). In SCLE, ICAM-l expression was di1fuse throughout the epidermis (Figs 1B,C), occasionally with accentuation in the basal layer (Fig 1 C). In SCLE the staining was found to be either cytoplasmic (Fig 1B) or cell surface (Fig lC). Non-Iesional skin in SCLE did not demonstrate increased ICAM-1 expression either in the epidermis or dermis over the normal controls. EM typically demonstrated strong diffuse keratinocyte surface staining for ICAM-l in most areas with accentuation at the basal cell layer. The ICAM-l staining was not homogeneous with a few areas demonstrating little or no staining of the epidermis as illustrated in the central area of the epidermis pictured in Fig lD. In the papillary dermis there was also strong ICAM-l staining of the vessels (endothelial cells) and of the mononuclear cell infiltrate (Fig 1D). In LP keratinocyte cell­surface and cytoplasmic expression of ICAM were observed in a patchy distribution (+ 3-4) concentrated in the basal cell areas adjacent to the dense infiltrate of ICAM( +) lymphocytes in the upper and papillary dermis. There was significant strong increase in ICAM-l staining in the endothelial cells in these cases, especially around the papillary vessels; however, the endothelium staining for ICAM was variable (Fig 1E,F).

Preferential ICAM-1 Expression in Basal Keratinocytes Keratinocyte di1ferentiation is a potent influence on induction of ICAM-1. As illustrated in Fig lA, keratinocytes grown in low­calcium medium had the greatest induction ofICAM-l expression following stimulation with either TNF-a or IFN-I'. Cells grown in low-calcium medium have the least di1ferentiated phenotype. The FACS profiles of cells grown in medium containing di1ferent

74S BENNION ET AL

a 70

60

8 50 c:

§;::- 40 QI •

�� -i u 30 c::::' COl QI E 20 CD o

..J 10

0.03 0.15

mMCa++

c 200 -:-c:-::--.,.,....."..------0.03 mM Ca++

" .0 E �

z

" U

.. U

'00

1 0 1 0 100

200 �----------

.00

0.15 mM Ca++

n=6 � TNF � FN

TNF: 1_ ca va high ca: 1'=0.050

IFN: low.,. va high ca: p:O.015

IlgA) nolFN 20 U IFN

1000

IlgA) NolfN 20 U IfN

1.0

Log Mean B.B

14 .8 79.8

Co.Var. 0.40 0,31 0.26

log Mean Co, Vat. 9.7 0.4!

13.7 0.30

62.9 0.26

SLOe", 27,0

2B.O

37.9

SI.Oev, 28.7 25.9

35.B

o+-��----�--���--���

.. .0 E � z

.. U

1.0 10 100 1000

200 �--�-------

1.0

: � i.e .I >d\\ .00

I .... j\ \.\

�''''-... • 0 1 0 .00

Fluorescence

IlgA) no IFN 20 U IFN

100 0

log Mean Co.Vat.

7.9 0.40 14.2 0.28 49.1 0.25

St.Dey, 26.1 24.8

31.9

calcium concentrations and then stimulated with cytokines are shown in Fig 2B,C, allowing examination ofICAM-l distribution within the entire cell population. Neither the baseline ICAM-l expression nor the pattern of ICAM-l expression in cytokine stimulated cells is different in cells with calcium-altered differenti­ation phenotype. However, the absolute level of ICAM-l (log mean) is greater for both TNF-a (Fig 2B) and IFN-y (Fig 2q stimulation. The variability of response as measured by coefficient of variation and standard deviation is no different in the high- and low-calcium conditions.

Induction of ICAM-l Expression in Keratinocytes by TNF-a and UVR We have previously shown that ICAM-l induction by cytokines may be quite variable in different individu­als, and that some individuals may be "responders" and others "non-responders" [21]. In those studies, the responses to TNF-a

b

" .0 E � z

" U

.. .0 E � z

.. U

.. U

THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

200 --:-::-::---:-:---:-----__ 0.03 m M Ca++

.00

1 0 100

200 c' :-:-::---:-:-:::-------0.15 mM Ca ••

100

(lgA) no TNF 30 ng TNF lO ng TNF

. 000

IlgA) NoTNF 30 ngTNF 30 ng TNF

o +-��----�----��� __ ��� 1.0 10 100 1000

200 r-:-�:-:""""::-------

100

• 0 100

Fluorescence

IlgA) no TNF 30 ngTNF 30 ng TNF

1000

log Mean B.B 14.88 71.16 84.07

Co.Var. 0.40 0.31

0.30 0.2B

log Mean Co. Va,. 9.7 0.41

13.78 0.30 55.99 0.32 63.50 0.34

log Mean CO.Val. 7.9B 0.40

14.20 0.28 52.64 0.3' 5'.B2 0.2B

SI.Oe ... . 27.02

28.00 42.16 41.22

S1.0e ... , 2B.76 25.91 40,67 43.l5

SI.Dev. 26.09 24.89

38.81

36.16

Figure 2. Effect of keratinocyte differentiation on induction of ICAM-l expression. a) ICAM-l expression in keratinocytes grown in high- (1.0 mM). mediwn- (0.15 mM). or low- (0.03 mM) calcium concentration. The bars and brackets represent the mean :i: SEM of six diJferent experiments. Cells were stimulated with optimal TNF-a concen­tration for 3 d or optimal IFN-1' concentration for 1 d. Induced ICAM-l expression was greater in the less diJferentiated keratinocytes for both IFN-1' (p = 0.015) and TNF-a (p = 0.050). b) FACS profiles for keratinocyte cultures grown in three diJferent calcium concentrations. IgA indicates the anti-IgA isotype-matched negative control, whereas the other curves represent anti-ICAM-1 staining with interferon (IFN 20 U) and without interferon (No IFN). Log mean indicates a summation of the intensity of staining; Co.Var. and St. Dev. indicate the variability of staining. c) Similar experiment showing ICAM-l cell-surface expression after stimulation with TNF (TNF 30 ng in duplicate) or without TNF (No TNF).

and UVR were closely correlated in individual keratinocyte cul­tures.

We':selected five subjects whose cultured keratinocytes were either responders or non-responders in vitro to TNF-a as judged by level of induced ICAM-l expression. In Table n the expression of ICAM-l of cultured keratinocytes from responders and non­responders to TNF-a is compared to the induction in vivo of ICAM-1 expression by UVR in the same adult subjects. There is a positive correlation between the increase in ICAM-1 expression in response to TNF-a and the in vivo response to UVR as demon­strated by ICAM-l staining in the epidermis. The non-responders had minimal ICAM-l expression on their cultured keratinocytes when exposed to TNF-a and no ICAM-l expression in their epidermis on exposure to UVR. This comparison clearly demon­strates the differences between non-responders and responders:

VOL. lOS, NO. 1, SUPPLEMENT, JULY 1995

Table II. ICAM-l Induction in Keratinocytes and Epidermis by UVR and TNF-a

UV Irradiation In Vivo TNF-a In Vitro 100 50 100 mJlcm2

Subjects" ICAM-l 0 (1 d) (3 d) (3 d)

Nonresponders MM 22b OC 0 0 0 SS 40 0 0 0 0

Responders

JB 67 0 3+ 2+ 2+ SB 107 1+ 2+ 3+ .3+ DN 64 0 1+ 2+ 2+

• All five subjects were skin type II, with similar minimal erythemia doses. In all 50 mJ/cm2 and 100 mJ/cm2 represented approximately 1 and 2 minimal erythemia doses, respectively.

b Tn vitro increase in ICAM-l (log mean fluorescence) afier TNF-a stimulation of keratinocyres for three days.

'Degree of diffuse epidermal staining for ICAM-1 of biopsies ofUV-irradiated skin.

responders did up-regulate ICAM-l expression on keratinocytes in vitro after TNF-a stimulation and in vivo after UVR exposure. All five subjects were Caucasians with type II and similar minimal erythema doses on phototesting .

The immunohistochemical staining in vivo and the FACS profiles are illustrated in Fig 3 for one nonresponder (MM) and three nonresponders (DN. ]B. and SB). In all subjects. control non­irradiated skin had negative epidermal ICAM-l staining. Multiple sequential daily irradiation produced more epidermal damage than a single daily dose. but did not necessarily produce greater ICAM-l expression (Table 11).

Induction of Expression ICAM-l by HSV As shown in Table III, there is a small increase in ICAM-l expression in keratinocytes infected with HSV. The fold increase in cell-surface ICAM-l was 1.65 for HSV infection and 9 when the HSV -infected cells were exposed to IFN-y. Similar levels of ICAM-l induction were seen out to 40 h (data not shown).

Priming with TNF Significandy Enhances Subsequent Re­sponse to TNF The expression of ICAM-l on human kerati­nocytes in culture was evaluated after priming and challenge doses of TNF-a. Priming with suboptimal doses of TNF-a produced small increases in ICAM-l expression when no TNF-a was used as challenge (Fig 4). However. significant enhancement of ICAM-l expression was seen with priming with 0.03. 0.3. or 3.0 ng/ml of TNF-a when the cells were subsequently challenged with 30 ng/ml ofTNF.

Priming with TNF-a was also tested in cells challenged with exposure to UVR. In this experiment. there were small increases in ICAM-1 expression (Fig 5) when irradiated cells were primed with suboptimal TNF-a (0.3 and 3.0 ng/ml).

Lack of Priming with interleukin 1 (IL-l) in Keratinocytes and Epithelial Cell Strains A direct comparison of IL-1 responses of normal human keratinocytes to the responses of three transformed keratinocyte or epithelial cell lines is shown in Table IV. As previously described.; KB cells have greater baseline ICAM-l expression and show a substantial induction of ICAM-1 expression by IL-l (1.0 ng/ml). This level of IL-1 (1.0 ng/ml) will induce significant responses such as IL-8 production in keratino­cytes. but not increased ICAM-1 expression (data not shown). However, pretreatment of the four cell lines or strains with suboptimal IL-1 fails to induce priming of any of the cell lines. in distinction to the priming seen with TNF (Figs 4 and 5). In

* Middleton MH. Arend wp. Norris DA: High intracellular interleu­kin-I receptor antagonist (icIL-1ra) levels are correlated with low expres­sion of intercellular adhesion molecule-l (ICAM-1) in keratinocytes and fibroblasts (abstr). ] Invest Dermotol 102:524. 1994.

EXPRESSION OF ICAM-1 INDICATES DISEASE TRIGGERS 75S

addition. IL-1 at 0.1 ng/ml does not prime for subsequent responses to TNF-a or UVR (data not shown).

DISCUSSION

The diversity of ICAM-1 staining patterns in various cutaneous diseases is relatively remarkable considering the fairly limited repertoire of cytokines that are known to up-regulate the expres­sion of ICAM on keratinocytes. We found that the pattern of ICAM-1 expression in the epidermis was quite different in LP, EM. and SCLE. even though all are associated with keratinocyte cyto­toxicity and a mononuclear infiltrate. To further describe the different patterns of ICAM-l in the suprabasallayer in SCLE and EM. we further investigated the roles of the apparent trigger factors UVR and HSV infection on the induction oflCAM-1 expression in human keratinocytes .

ICAM-l expression is induced in many cell types by the cyto­kines TNF-a and TNF-f3. IL-l. and IFN-y. In keratinocytes. ICAM-l expression is constitutively suppressed by the upstream promoter region [22]. and can be induced by TNF-a and TNF-f3. IFNy. and UVR, but not by IL-1 [11.13.21.23]. Injection ofIL-l intradermally induces ICAM-1 expression in vivo, presumably by release ofTNF-a [24]. Our own studies indicate that high levels of intracellular IL-lra in human keratinocytes may act to down­regulate baseline ICAM-1 expression and actually block subsequent IL-1 induction ofICAM-l expression.;

In LP. the basal cells of the epidermis. especially those adjacent to the lymphocytic infiltrate in the papillary dermis. are heavily adorned with ICAM-1 [14]. The cytokines IFN-y and TNF produced by the dermal mononuclear infiltrate in lichenoid derma­titis have been implicated in previous studies as inducers of epidermal damage [25.26]. The basal keratinocytes are particularly susceptible to ICAM-l induction by these cytokines . as demon­strated in these and previous studies [27]. Studies of cloned T cells also have shown that T cells can secrete lymphokines in a focused directional fashion [28]. This would allow dermal lymphocytes in LP to "target" basal layer keratinocytes by secreting lymphokines in a directed fashion. This may explain the highly localized and intense ICAM-l staining we found in the basal keratinocytes in LP. The failure to block this reaction with anti-TNF raises doubt about the role of TNF-a in lichenoid reactions [25]. The lack of suprabasal ICAM-1 staining in LP is most consistent with leuko­cyte-derived cytokines. but not epidermal TNF as inducers of ICAM-1 expression.

EM exhibits a totally different pattern ofICAM-1 expression than LP. involving the entire epidermis as demonstrated in Fig ID. It frequendy shows foci of ICAM-l expressed on the surface of keratinocytes in the mid and upper epidermis. Brice and colleagures demonstrated HSV protein [29] and genome [30] in the epidermis in EM lesions. but not in areas of repeated EM between episodes. The diffuse binding of the probe for HSV in those studies is similar to the diffuse ICAM-l staining found in our EM biopsies. In our ill vitro studies ofHSV effect on ICAM-l expression. we found that HSV infection could induce modest expression of ICAM-1 on keratinocytes (Table 111). In addition. several studies have dem­onstrated that HSV-infected keratinocytes can produce interferons [31.32] and also TNF-a [33]. which might induce keratinocytes to up-regulate ICAM-1 on their cell surfaces in an autocrine or paracrine fashion. Another cytokine present in keratinocytes in high quantity is IL-l [34-36]. It is reasonable to conclude that viral damage and subsequent degeneration of keratinocytes may release IL-1 in quantities great enough to induce increased ICAM-1 expression indirecdy through increased inflammation. However. as noted earlier. IL-1ra is also present within keratinocytes [34.37] and may act to down-regulate the effect ofIL-l locally.

In biopsies of lesional skin from SCLE. keratinocytes showed either a diffuse cell membrane ICAM-1 staining or a diffuse cytoplasmic staining (Fig IB,C). These patterns were similar to the patterns seen in UVR-induced ICAM-l in patients who were determined to be TNF-a responders by ill vitro testing of their keratinocytes (Fig 2. DN.]B. and SB). The strong coincidence of

76S BENNION ET AL

Q; .0 E ::J Z

Q; U

Q; .0 E ::J Z

Q; U

Q; .0 E ::J Z

Q; .0 E ::J Z

Q; U

300 ,-----------------log mean IgA 15.07 Nocytoklne 25,90 MM

200

'00

0 1.0

Fluorescence

TNFa 46,05

TNFa 50.78

1000

JOO ,.------;;------------ Log mean tgA 9.47 ON

200

'00

0 1.0 ,0 '00

Fluorescence

No cytoklne 1 7.74 TNFa 83.47 TNFa 81.84

300 ,.------------------ log mean �A 10.3 J6

200

100

Fluorescence

300 S6

200

100

0 1.0 10 '00

Fluorescence

No cytoklne 20. 3 4 TNF. 87.6 TNFe 86.19

log mean 19' 7.75 NocVloklne 35.39 TNFa 138.5 TNFa 146.0

1000

THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

MM

JB

Figure 3. Response to TNF in vitro correlates with response to UVR in vivo. Compariso n ofICAM-l induction in vitro by TNF-� with ICAM-l induction i,t vivo with UVR in four of the five subjects described in Table U. The ICAM-l induction in vitro was measured by FACS analysis. In this analysis, an isotype-matched negative control (JgA = anti IgA) antibody binding is compared with baseline cultures (no cytokine) and duplicate cultures stimulated with cytokine (TNF-�). These data are shown in the left column as FACS tracings and ICAM-l expression (log mean). In the right column the immunohistochemical staining for ICAM-l is presented for each subject shown in the left column. In vitro responders were defined as those who increased ICAM-l expression greater than a 50 increase in log mean after TNF stimulation (see Table U). Responders in vitro to TNF-� showed increased epidermal ICAM-l in vivo following UVR.

VOL. lOS, NO. 1, SUPPLEMENT, JULY 1995

Table m. Herpes Simplex Infection Stimulates Low­Level ICAM-l Expression

Experimental Percent Log Mean Fold Condition Positive Fluorescence Increase

Negative Control 0 lS.7 Anti-lgA HSV infected

ICAM-l control 4.9 22.7 1.0 Anti-ICAM-l HSV-

ICAM-l: HSV induced 21.2 37.5 �.65 Anti-ICAM-l

HSV+ Positive control 7S.6 202 9.0

ICAM-l: IFN-y induced Anti-ICAM-l HSV+ IFN-y+

the ICAM-1 patterns in SCLE and in the UVR-irradiated skin of responders supports the hypothesis that UVR induces ICAM-1 expression in highly photosensitive SCLE patients . It also supports the hypothesis that UVR susceptibility, perhaps mediated through TNF-a release and response, is an important determinant of who will develop SCLE. We have previously shown that the UVR and TNF-a induction of ICAM-1 expression in keratinocytes are linked, and that susceptibility to UVB and TNF-a correlates in cell strains from di1ferent individuals [21]. We have also shown that

UVR induces TNF-a release from keratinocytes, but in different levels from di1ferent individuals [21]. This work extends these observations to link ICAM-1 expression in SCLE lesions to TNF-a

dependent induction by UVR. Priming ofkeratinocyte response by prior exposure to low levels of TNF-a or UVR may be one mechanism by which low-level repeated exposures to stimuli (such as UVR or latent HSV) may potentiate ICAM-1 induction within the epidermis to subsequent stimulus. Trefzer et al [38] have shown

3 c: 3 ., � " =;: i:i ClC Eo 01:::-.E CI .. �

70

60

50

40

30

20

10

0 o 30

Challenge (ng TNF/ml)

Day -3 • Priming dose of TNF

Day 0 • Challenge dose of TNF (30 ng/ml

Day +3 • Measure ICAM-1 expression

Priming dose of TNFa

o O.Onglml [J 0.03 nglml B .0.3 nglml � 3.0nglml

Figure 4. Suboptimal TNF "primes" keratinocytes for subsequent response to TNF. The effect of priming of keratinocyte cultures with suboptimal levels of TNF-a on the subsequent response to TNF-a is illustrated. The ICAM-l expression of cultured human keratinocytes is expressed as delta log mean fluorescence (stimulated values minus negative control) . In this experiment, cells were prime d on day -3 with one offour doses ofTNF, an d then challenged on day 0 with 0 or 30 ng/ml TNF. On day 3, the ICAM-l expression was measured. Significant (p < 0.01) priming of response was seen with TNF primary dose of 0.03, 0.3, and 3.0 ng/m!.

EXPRESSION OF ICAM-l INDICATES DISEASE TRIGGERS 77S

.. u c: .. u "' � 0 � c: �;:" E . en=< 0« -0 �::::.. OJ c

o 60 Challenge (UVR, mJ/sq.cm.)

Day -2 - Priming dose of TNF

Day 0 - Challenge dose of UVR (60 mJ/sq.cm).

Day +3 - Measure ICAM-1 expression

Priming dose otTNFa

o O.Ong/ml � 0.3 nglml 9 3.0 ng/ml

Figure 5. Priming with TNF enhances response to UVR. The effect of priu1ing with suboptimal doses of TNF-a is shown on subsequent ICAM-l expression following exposure to UVR. Cells were primed on day -2 with three doses ofTNF-a and then challenged with UVR on day O. ICAM-l expression was measured by FACS on day +3. TNF at 0.3 and 3.0 ng/ml enhanced subsequent response to UVR.

that TNF-a and UVR both induce increases in TNFR mRNA, associated with enhanced TNF-a dependent responses. Krutmann and Grewe also propose that UVR can induce IL-1-mediated priming through upregulation ofIL-l type I receptors [39]. In our hands, lL-1 will not prime keratinocytes to subsequent increases in ICAM-1 expression in response to IL-1 [39].

We propose that the epidermis is a stratified environment with respect to ICAM-1 induction (Fig 6). The basal keratinocytes are exposed to synergistic stimulation by TNF plus IFN--y [17] andlor TNF and histamine [40], and are intrinsically the most responsive of the keratinocytes in induction ofICAM-1 expression. However, in some diseases , suprabasal ICAM-l expression is also induced by exogenous factors such as UVR. Release of TNF-a and lL-1 by UVR may induce ICAM-1 expression in the suprabasal epidermis, but the response appears to be complex and individually variable. Our work suggests that low-level TNF-a release (by UVR or HSV

Table IV. IL-l PrUning ofKeratinocyte for Induction of ICAM-l Expression by IL-l

ICAM-l Expression (log mean)

IL-l Cell Strain No IL-l Stimulation

or Line Stimulation Stimulationa After Priming"

HI{ se 13 13 HaCaT 6 6 5 A431 4 20 12 KB 23 72 59

• (L-t stimulation (1.0 ng/mI) at day 0 followed by measurement of ICAM-I expression at day 1. This level of !L-l induces keratinocytes to produce IL-B (data not shown).

b Priming with !L-l (0.1 ng) at day -1, followed by stimulation with !L-t (I ng) at day 0, and measurement of ICAM-t expression at day 1. Day 1 is the point of maximum stimulated ICAM-l expression.

, Log mean ICAM-l by FACS analysis. The data in this table represent a single experiment performed on the same day comparing all four cell types. The lack of priming by !L-l has been verified in multiple experiments using priming for IL-l response and also priming by IL-l for TNF-a- and UVR-induced I CAM-l expression.

78S BENNION ET AL

Keratinocyte Terminal Differentiation

Papillary Dermis

UVB

TNF .. prl .... lo, TNF ,--

UVB

Max i m al IL-1 ra

IL-1 IL-1 ra

IL·' prl .... 1o, Il·1 r...,on ..

Basal Zone

IL- 1 Histamine IFN-g TNF-B TNF-a TNF-a

� � Macrophage Mast Cel l T Lymphocyte

Figure 6. The epidermis is a stratified environment with respect to ICAM-l induction. The greatest level of ICAM-l induction is seen in basal keratinocytes stimulated with IFN-'Y or TNF-a. UVR induces TNF-a release in the suprabasal epidermis and also lL-l a and IL-lra release from damaged keratinocytes in the suprabasal epidermis. These cytokines might regulate suprabasal or basal ICAM-l expression directly or by priming the target cells for subsequent responses to cytokines. However, in human keratinocytes, iclL-lra appears to limit IL-l-induced ICAM-l expression.

infection) may "prime" the epidermis to express significant ICAM-l with continued suboptimal exposure.

rhe functional significance of clliferential expression of ICAM-1 within the levels of the epidermis is not yet known. In the three diseases considered here, cytotoxic damage of basal cells is seen, but more extensive suprabasal epidermal cytotoxic damage can also be seen in EM and in SCLE. This suggests that ICAM-1-positive suprabasal keratinocytes are rendered susceptible to immunologic damage, perhaps through ICAM-1-mediated interaction of kerati­nocytes with activated cytotoxic leukocytes.

T1lis work was supporred by Grants R01 AR26427 atld T32 AR0741 1 .froln tile Natiollal Institutes of Hen/tit.

REFERENCES

1 . Staunton DE, Marlin SD, Stratowa C, Dustin ML, Springer TA: Primary structure of ICAM-I demonstrates interaction between members of the immunoglobulin and integrin supergene families. Cell 52 :925-933 , 1 988

2. Rothleiu R, Dustin ML, Marlin SD, Springer TA: A human intercellular adhesion molecule (ICAM-l) distinct &om LFA-1.] Imnrunol 137:1270-1274,

1986 3. Makgoba MW, Sanders ME, Ginther Luce GE, Gugel FA, Dustin ML, Springer

TA, Shaw S: Functional evidence tbat intercellular adhesion molecule 1 (ICAM-l ) is a ligand for LFA-l dependent adhesion in T cell-mediated cytotoxicity. Eur] ImmutlOI 18 :63 7-640, 1988

4. Figdor CG, van Kooyk Y, Keizer GD: On the mode ofaetion ofLFA-l . Immllnol Today 1 1 :277-280, 1 990

5. van Kooyk Y, van de Wiel-van Kemenade P, Weder P, Kuijpers TW, Figdor CG: Enhancement of LFA-l-mediated cell adhesion by triggering through CD2 or CD3 or T lymphocytes . Natll" 342:81 1-813, 1989

THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

6. Boyd A W, Wawryk SO, Burns GF, Fecondo ]V: Intercellular adhesion molecule 1 (ICAM-l) has a central role in cell-cell contact dependent mediated immune mechanism. Proc Natl Atad Sci USA 85:3095-3099, 1988

7. .Lanier LL, Ruitenberg JJ, Phillips JH: Functional and biochemical analysis of CD1 6 antigen on natural killer cells and granulocytes. ] I,"",uooI 141:3478-3485 , 1988

8. Makgoba MW, Sanders ME, Ginther Luce GE, Dustin ML, Springer TA, Clark EA, Mannori P, Shaw S: ICAM-l a ligand for LFA-l-dependent adhesion by B, T and myeloid cells. Nature 3 3 1 : 86-9 1 , 1 988

9. Norris DA: Cytokine modulation of adhesion molecules in the regulation of immunologic cytotoxicity of epidennal targets.] Invest DermatoI 9 5 : 1 1 1 S-120S,

1 990

10. Norris DA: Immunological cytotoxicity of cutaneous cellular targets. In: Notris DA (ed.). Immu"" Mechanisms in Cutaneous Disease. Marcel Dekker, Inc., New York, 1 989, pp 123-166

1 1 . Dustin ML, Singer RH, Tuck DT, Springer TA: Adhesion of T lymphoblasts to epidermal keratinocytes is regulated by interferon-gamma and is mediated by intercellular adhesion molecule-l (ICAM-l) . ] Exp Med 167:1323-1 340, 1988

12. Yohn lJ, Critelli M, Lyons MB, Norris DA: Modulation of melanocyte intercel­lular adhesion molecule-l by immune cytokines.] Invest Dermato! 95:233-237, 1990

1 3 . Norris DA, Lyons MB, Middleton MH, Yohn lJ, Kashihara Sawami M: Ultraviolet radiation can either suppress or induce expression of intercellular adhesion molecule 1 (ICAM-l) on the sumce of cultured human keratino­eytes.] Invest Dermatol 95:132-1 38, 1 990

14. Vejlsgaard GL, Ral£kiaer E, Avnstorp C, Czajkowski M, Marlin SD, Rothlein R: Kinetics and characterization of intercellular adhesion molecule-l (ICAM-l) expression on keratinocytes in various inflammatory skin lesions and malignant cutaneous lymphomas. ] Am Atad Dermatol 20:782-790, 1989

15. Lewis RE, Buchsbaum M, Whitaker D, Murphy GF: Intercellular adhesion molecule expression in the evulving human cutaneous delayed hypersensitivity reaction. ] Invest Dermatol 93:672- 677, 1989

16. Shiohara T: The lichenoid tissue reaction. An immunological perspective. Am ] DermatopathoI 1 0:252-256, 1988

17. NickololfBJ, Griffiths CE, Barker ]N: The role of adhesion molecules, chemo­tactic mctors, and eytokines in inflammatory and neoplastic skin diseases: 1 990 update. ] Invest Dermato! 2 0 : 1 5 1 S-1 57S, 1990

18. Furukawa F, Hu1fJC, Weston WL, Norris DA: Serwn-free serial culture of adult human keratinocytes from suction-blister roof epidennis . ] Invest Dermato' 89:460-463, 1987

19. Yohn lJ, Lyons MB, Norris DA: Cultured human melanocytes &om Black and White donors have different sunlight and ultraviolet A radiation sensitivities.] Invest Dermatol 99:454 -459, 1 992

20. Brice SL, Norris DA, Weston WL, Tedder DG, Lyons MB, Furukawa F, Huff JC: Development of an in vitro keratinocyte model for use in the study ofHSV specific cytotoxicity. ] Invest DermatoI 90:629 - 633, 1988

21. Middleton MH, Noms DA: Cytokine induced ICAM-l expression in human­keratinocyte strains is highly variable in keratinoeyte strains &om diJferent donors. ] Invest Dermatol l 04:489-496, 1995

22. Cornelius LA, Taylor JT, Degitz K, Li LJ, Lawley 1], Caughman SW: A 5' portion of the ICAM-l gene confers tissue-specific diJferential expression levels and cytokine responsiveness. ] Invest Dermatol l 00:753-758, 1 993

23 . Dustin ML, Rothlein R, Bhan AK, Dinarello CA, Springer TA: A natural adherence molecule (ICAM-l): induction by IL-l and interferon gamma, tissue distribution, and function. ] Im",unol 137:245-254, 1986

24. Groves RW, Ross E, Barker ]N, Ross JS, Camp RD , MacDonaid DM: Effect of in vivo interleukin-l on adhesion molecule expression in normal human skin.

] Invest DermatoI 98:384-3 87, 1 992 25. Shiohara T, Nickoloff BJ, Moriya N, Gotoh C, Nagashima M: In vivo elfects of

interferon-gamma and anti-interferon-gamma antibody on the experimentally induced lichenoid tissue reaction. Br] DermatoI 1 1 9 : 199-206, 1988

26. Shiohara T, Moriya N, Tanaka Y, Arai Y, HayakawaJ, Chiba M, Nagashima M: Immunopathologic study of lichenoid skin diseases: correlation between HLA-DR-positive keratinoeyte. or Langerhans cells and epidermotropic T cells. ] Am Acad Dermatol 1 8(part 1):67-74, 1988

27. Ka..hihara Sawami M, Norris DA: The state of diJferentiation of cultured human ker.tinoeytes detemrlnes the level of intercellular adhesion molecule-l (rCAM-l) expression induced by gamma interferon. ] Invest D",,,,atol 98:741-747, 1 992

28. Poo WJ, Conrad L, Janeway CA Jr: Receptor-directed focusing of lynlphokine rel�ase by helper T cells. Nature 332:378-380, 1 988

29. Orton PW, Hufl'jC, Tonnesen MG, Weston WL: Detection ofa herpes simplex viral antigen in skin lesions of erythema multiforme . Ann Intem Med 101 :48-50, 1984

30. Brice SL, Krzemicn D, Weston WL, Hufl'jC: Detection of herpes simplex virus DNA in cutaneous lesions of erythema multiforme. ] Invest Dermatol 93:1 83-187, 1989

31. Torseth JW, Nickolofl'Bj, Basham TY, Merigan TC: Beta interferon produced by keratinocytes in human cutaneous infection with herpes simplex virus. ] I,yat Dis 155:641-648 , 1987

32. Torseth jW, Mengan TC: Significance of local gamma interferon in recurrent herpes simplex infection. ] Infect Dis 1 53 :979 -984 , 1986

33 . Goldfeld AE, Maniatis T: Coordinate viral induction of tumor necrosis mctor alpha and interferon beta in human B cell. and monocytes. Proc Nat! Acad Sci USA 86:1490-1494, 1 989

34. Bigler CF, Norris DA, Weston WL, Arend WP: lnterleukin-l receptor antag­onist production by human keratinocytes. ] Invest Dmnatol 98:38 -44, 1992

VOL. 105. NO. I. SUPPLEMENT. JULY 1 995

35. Mizutani H. Black R. Kupper TS: Human keratinocytes produce but do uot process pro-interleukin-I (IL-l) beta. DiJferent strategies of IL-l production and processing in monocytes and keratinocytes. ] Clin Invest 87:1066-1071.

1991 36. Blanton RA. Kupper TS. McDougall]K. Dower S: Regulation ofintedeukin 1 and

its receptor in humart keratinocytes. Proc Nat! Acad Sci USA 86:1273-1277. 1989

37. Haskill S. Martin G. V.nLe L. Morns J. Peace A. Bigler CF. Jalfe GJ. Hammerberg C. Sporn SA. Fong S. et al: eDNA cloning of an intracellular form of the human interleukin 1 receptor antagonist associated with epithe­lium. Proe Nail Acad Sci USA 88:3 681-3685. 1991

EXPRESSION OF JCAM-l INDICATES DISEASE TRIGGERS 79S

38. TrelZer U. Brockhaus M. Lotscher H. Parlow F. Budnik A. Grewe M. Christoph H. Kapp A. Schopf E. Luger TA. et al: The 55-kD tumor necrosis !actor receptor on human keratinocytes is regulated by tumor necrosis factor-alpha and by ultraviolet B radiation. ] CIi,. Invest 92:462-470. 1993

39. Krutmann J. Grewe M: Involvement of cytokines. DNA damage and radical formation in ultraviolet radiation-induced modulatinn of intercellular adhesion molecule-l (JeAM-i) expression. ] Invesl Dermalol (in press)

40. Mitra RS. Shimizu Y. NickololfBJ: Histamine and cis-urocanie acid augment tumor necrosis factor-alpha mediated induction nf keratinocyte intercellular adhesion molecule-l expression. ] Cell Physio/ 1 56:348 -3 57. 1993