Herpes Simplex Virus Associated Erythema Multiforme(HAEM) is Mechanistically Distinct from Drug-InducedErythema Multiforme: Interferon-γ is Expressed in HAEMLesions and Tumor Necrosis Factor-α in Drug-InducedErythema Multiforme Lesions

Hisashi Kokuba,*‡ Laure Aurelian,†* and Joseph Burnett‡*Virology/Immunology Laboratories and †Departments of Pharmacology and Experimental Therapeutics, and ‡Dermatology3,The University of Maryland School of Medicine, Baltimore, Maryland, U.S.A.

Erythema multiforme follows administration of several of five (80%)], but lesional skin from drug-inducederythema multiforme patients was negative. Lesionaldrugs or infection with various agents, including herpesherpes simplex virus associated erythema multiformesimplex virus, a syndrome designated herpes simplexkeratinocytes also stained with antibody to trans-virus associated erythema multiforme. Lesional skinforming growth factor-β [14 of 23 (61%)] and cyclin-from 21 of 26 (81%) herpes simplex virus associateddependent kinase inhibitor waf [12 of 18 (67%)].erythema multiforme patients was positive for herpesStaining was also seen in keratinocytes from herpessimplex virus gene expression as evidenced by reversesimplex virus lesions [five of five (100%)], but not intranscriptase–polymerase chain reaction with primersnormal skin. By contrast, staining with antibody tofor DNA polymerase and/or immunohistochemistrytumor necrosis factor-α, another pro-inflammatorywith DNA polymerase antibody. Reverse tran-cytokine, was seen in seven of 11 (64%) drug-inducedscriptase–polymerase chain reaction and immuno-erythema multiforme patients, but not in herpeshistochemistry studies indicated that herpes simplex simplex virus or herpes simplex virus associated eryth-virus associated erythema multiforme lesional skin ema multiforme patients, and lesional keratinocytes

from 16 of 21 (76%) DNA polymerase positive herpes from drug-induced erythema multiforme patientssimplex virus associated erythema multiforme patients were negative for transforming growth factor-β andwas also positive for interferon-γ, a product of T cells cyclin-dependent kinase inhibitor waf. We interpretinvolved in delayed-type hypersensitivity (p < 0.0001 the data to indicate that herpes simplex virus associatedby Pearson correlation coefficient). Interferon-γ signals erythema multiforme pathology includes a delayed-were in infiltrating mononuclear cells and in intercellu- type hypersensitivity component and is mechanistic-lar spaces within inflammatory sites in the epidermis ally distinct from drug-induced erythema multiforme.and at the epidermis/dermis junction. Herpes simplex Key words: apoptosis/DTH/HSV antigen/HSV DNA/virus lesional skin was also positive for DNA poly- HSV DNA polymerase/HSV RNA/p21waf/PCR/TGFβ.

J Invest Dermatol 113:808–815, 1999merase [five of five (100%)] and interferon-γ [four

E 1998), but they are positive for viral DNA (Cao et al, 1989; Darraghet al, 1991; Aslanzadeh et al, 1992; Miura et al, 1992; Yokoi, 1995;Imafuku et al, 1997; Kokuba et al, 1998). Lesion development wasrecently associated with HSV gene expression (Imafuku et al, 1997;Kokuba et al, 1998). It has long been suspected that HAEM

rythema multiforme (EM) is a polymorphous, oftenrecurrent eruption composed of macules, papules,bullae and target lesions that are symmetrically distribu-ted and have a propensity for the distal extremities andoral mucosae. It can follow administration of several

drugs or infection with various agents, including herpes simplex pathogenesis includes an immune component that is directed tovirus (HSV). HSV is generally not isolated from herpes-associated HSV antigen(s) expressed in lesional skin (Shelley, 1967; HowlandEM (HAEM) lesions (Smith and McLaren, 1977; Kokuba et al, et al, 1984). Indeed, HSV-specific antibody responses were shown

to be higher in HAEM than HSV patients (Brice et al, 1993) andT cells were found to infiltrate HAEM lesions. Immunopathology

Manuscript received February 16, 1999; revised July 27, 1999; accepted was alternatively attributed to delayed-type hypersensitivity (DTH)for publication August 5, 1999. and cytotoxic T (CTL) responses (Shelley, 1967; Zaim et al, 1987;

Reprint requests to: Dr. Laure Aurelian, Virology/Immunology Labora- Malmstrom et al, 1990). T cells accumulating in HAEM lesionstories, University of Maryland School of Medicine, 10S. Pine Street,were primarily CD41 (Vβ21) cells that are principal responders toBaltimore, MD 21201, U.S.A. Email: laurelia@umaryland.eduHSV antigens in vitro (Malmstrom et al, 1990; Kokuba et al, 1998).Abbreviations: p21waf, cyclin-dependent inhibitor waf; Pol, DNA poly-

merase. This presentation is in stark contrast to the preponderance of non-T

0022-202X/99/$14.00 · Copyright © 1999 by The Society for Investigative Dermatology, Inc.

808

VOL. 113, NO. 5 NOVEMBER 1999 IFN-Γ IS EXPRESSED IN HAEM LESIONAL SKIN 809

Immunohistochemistry Staining with Pol antibody was as previouslyand CD81 T cells in drug-induced EM lesions (Margolis et al,described (Kokuba et al, 1998) using frozen sections fixed (4°C, overnight)1983). DTH is mediated by CD41 T helper (Th) cells and CTLswith 10% formalin and stored in 70% ethanol (at 4°C) prior to assay.are generally CD81. Because HSV-specific CTL are primarilyFormalin-fixed paraffin-embedded sections from the same tissues wereCD41 (Zarling et al, 1988), however, the type of response involvedused for staining with antibodies to IFN-γ, TNF-α, TGF-β, and p21waf.in HAEM immunopathology is still unclear. The studies described For TGF-β and p21waf antibodies, sections were blocked with normal goat

in this report indicate that HAEM lesions are positive for IFN-γ, serum (1:10 dilution in phosphate-buffered saline) and exposed to thea product of antigen-activated CD41 Th1 cells involved in DTH primary antibodies overnight at 4°C followed by a mixture of biotinylatedreactions (Feghali and Wright, 1977; Billiau et al, 1998), suggesting anti-rabbit and anti-mouse immunoglobulins for 30 min at room temper-

ature (LSAB kit, DAKO, Carpinteria CA). The color substrate wasthat HAEM pathology includes a DTH component. Drug-induced3-amino-9-ethylcarbazole. The reagents were supplied with the kit andEM is a mechanistically distinct condition in which keratinocytesused according to the manufacturer’s instructions. For TNF-α antibody,are positive for tumor necrosis factor-α, a sign of toxic injury.sections were exposed to the primary antibody for 30 min at roomtemperature and the a mixture of biotinylated anti-rabbit and anti-mouseMATERIALS AND METHODSimmunoglobulins (LSAB2 kit, DAKO) for 10 min at room temperature

Polymerase chain reaction (PCR) amplification DNA was extracted and the reaction developed with reagents supplied in this kit according tofrom formalin-fixed paraffin-embedded tissue sections and PCR amplified the manufacturer’s instructions. For IFN-γ antibody, sections were firstwith DNA polymerase (Pol) and control human β-globin primers. PCR trypsinized by exposure for 15 min at room temperature to 0.1% trypsinproducts were confirmed by hybridization with an internal oligonucleotide (type IX 15,550 Nα-benzoyl-L-arginine ethyl ester units, Sigma, St Louisprobe (Imafuku et al, 1997). MO) in calcium–magnesium-free phosphate-buffered saline (pH 7.4). The

trypsin reaction was stopped by exposure (5 min at room temperature) toSolution phase reverse transcriptase–PCR Solution phase reverse a 0.1 µg per ml solution of trypsin inhibitor (type I-S from soybean,transcriptase–PCR was as described by Imafuku et al (1997) with RNA Sigma). At this time the slides were exposed to IFN-γ antibody (1 h attreated (60 min, 37°C) with 1 U RNase-free Dnase per µl (Boehringer room temperature) followed (10 min at room temperature) by secondaryMannheim Biochemicals, Indianapolis, IN). Reverse transcriptase was on antibody and the other reagents provided in the LSAB2 kit (DAKO)total RNA (2 µg) with 200 U Moloney murine leukemia virus reverse according to the manufacturer’s instructions. All slides were counterstainedtranscriptase (Gibco BRL, Grand Island, NY) and 25 pmol of anti-sense with Mayer’s hematoxylin.primer. PCR was with 5 µl of reverse transcriptase product in 1 3 PCRbuffer II (Perkin-Elmer, Foster City, CA), 1.5 mM MgCl2, 200 µM dTNP,

Study groups Twenty-six patients with HAEM, defined as a cutaneous25 pmol sense and anti-sense primers and 5 U Taq DNA polymerase. Theeruption with target lesions having a prevalence for the extremities andreverse transcriptase products were held at 95°C (2 min), annealed at 55°Coral mucosa that follows a recurrent HSV episode were studied during the(1 min), extended at 72°C (1 min), and denatured at 95°C (1 min). Thirtyacute stage (2–10 d after lesion onset). Eight of these were also studied atcycles of amplification were used (Imafuku et al, 1997).5–6 mo after lesion resolution. Thirteen patients with drug-induced EMand five patients with HSV infection confirmed by virus isolation servedIn situ reverse transcriptase–PCR In situ reverse transcriptase–PCR,as controls and were studied at the time of acute disease (2–10 d afterwhich measures direct incorporation of digoxigenin-11-dUTP (DIG-lesion onset). All patients were HIV negative.dUTP), allows detection of PCR-amplified RNA in intact cells via cDNA

(Nuovo, 1993). Because problems related to RNA degradation wereRESULTSreported for formaldehyde-fixed paraffin-embedded tissues (Komminoth

et al, 1994), frozen sections fixed (4°C, overnight) with 10% formalin and HAEM lesional skin is positive for HSV RNA and proteinstored in 70% ethanol (at 4°C) were assayed as we have previously described We used PCR with Pol primers to examine the presence of HSV(Imafuku et al, 1997). Primers for Pol (Imafuku et al, 1997), interferon

DNA in HAEM lesional skin. Solution phase and in situ reverse(IFN)-γ, tumor necrosis factor (TNF) -α (Serrano et al, 1997), and thetranscriptase–PCR with Pol primers and/or immunohistochemistryHIV gag gene (Loche and Mach, 1998) used as a control were as described.with Pol antibody were used to examine HSV gene expression inFor cDNA synthesis, tissues were digested (15 min, 37°C) with proteinase

K (10 µg per ml) and treated (overnight, 37°C) with RNase-free DNase the same tissues (Imafuku et al, 1997; Kokuba et al, 1998). HIV(Boehringer Mannheim, 1 U per µl). cDNA was amplified by reverse gag primers were used as control. Pol DNA was seen in skin fromtranscriptase with the downstream primer. Tissues on the glass slides were five of five (100%) HSV patients and 21 of 26 (81%) HAEMincubated (42°C, 30 min) with 10 µl of reverse transcriptase solution [2 µl patients (Tables I and II). Pol RNA was seen in all HSV lesionaleach of 5 3 reverse transcriptase buffer and dNTP (stock solution is tissues and in six of eight frozen sections and 10 of 10 formalin-12.5 mM)], 1 µl dithiothreitol (0.1 M), 3.0 µl depC water, and 0.5 µl fixed sections of HAEM lesional skin, respectively, studied byprimer (stock solution is 25 µM), 20 µl RNase inhibitor (Promega, Madison,

solution phase and in situ reverse transcriptase–PCR (Tables I andWI) and 10 U Moloney murine leukemia virus reverse transcriptase perII). Normal uninvolved skin and lesional skin from drug-inducedµl (Gibco BRL) made according to the manufacturer’s protocol]. For inEM patients were negative (Table III). Amplification was not seensitu PCR, 25 µl of reaction mixture (1 3 buffer, 4.5 mM MgCl2, 200 µM

dNTP, 10 µM digoxigenin dUTP, 1 µM each of the downstream and with HIV primers. Twenty-one of 26 (81%) frozen and fixedupstream primers and 2.5 U Taq polymerase) was placed on sections, sections of HAEM lesions also stained with Pol antibody (Tables Icovered with a coverslip and overlaid with 1 ml of mineral oil preheated and II). Staining was intranuclear and restricted to the basal andto 82°C. After another denaturing step of 94°C for 3 min, 30 cycles were spinous cell layers of the epidermis (approximately 15–105 cellsaccomplished at 94°C, 55°C, and 72°C (1 min). Color substrate for per field) (Fig 1A). It was not seen with normal rabbit serum, nordigoxigenin detection was 0.45 mg per ml Nitro blue tetrazolium chloride, in normal (data not shown) or healed (Fig 1B) HAEM lesional0.18 mg per ml 5-bromo-4-chloro-3-indoyl-phosphate in Buffer 3 (Boehr-

skin (Table I). Only DNA-positive tissues were positive for viralinger Mannheim) and the slides were counterstained in neutral fast redRNA and protein (Tables I and II). The findings are consistent(Imafuku et al, 1997). They were read blindly by two dermatopathologists,with previous reports (Cao et al, 1989; Darragh et al, 1991;with similar results.Aslanzadeh et al, 1992; Miura et al, 1992; Yokoi, 1995; Imafuku

Antibodies Pol antibody was generated in rabbits with a peptide located et al, 1997; Kokuba et al, 1998).at residues 1216–1224 of the HSV-1 Pol protein and used as previouslydescribed (Kokuba et al, 1998). Rabbit anti-human IFN-γ antibody was HAEM and drug-induced EM lesions are positive for IFN-γfrom Serotec (Oxford, U.K.). It was used at 1:100 dilution according to and TNF-α, respectively Two series of experiments were donethe manufacturer’s instructions. Monoclonal antibody to transforming in order to examine whether the mononuclear cells in HAEMgrowth factor (TGF)-β (recognizes human TGF-β1 and TGF-β2) and lesional skin (Zaim et al, 1987; Malmstrom et al, 1990; Kokubapolyclonal antibody to human TNF-α [does not cross-react with TNF-β et al, 1998) are involved in cytokine-generating immune reactions.(Yu et al, 1994)] were from Genzyme Diagnostics (Cambridge MA). They

The focus was on IFN-γ and TNF-α, pro-inflammatory cytokineswere, respectively, used at 150 µg per ml and 1:250 dilution according tothat are produced by different cell types. IFN-γ is the product ofmanufacturer’s instructions. The monoclonal antibody to p21waf (cloneactivated T helper cells (effector Th1 subset) that cause DTH andEA10* ) was from Calbiochem (Cambridge, MA). It was used at 5 µg per

ml according to the manufacturer’s instructions. is present at DTH sites (Tsicopoulos et al, 1992). TNF-α is primarily

810 KOKUBA ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Table I. HSV antigen and cytokine expression in acute and healed HAEM lesional skina

Acute lesional skin Healed lesional skin

Patient Pol DNA/RNA/protein IFN-γ RNA/protein TGF-β Pol DNA/RNA/protein IFN-γ RNA/protein TGF-β

RL 1/1/1 ND/1 1 2/2/2 ND/2 2WS 1/1/1 1/1 1 2/2/2 2/2 2CI 1/1/1 1/ND 1 2/2/2 2/2 2AD 1/1/1 1/1 2 2/2/2 2/2 2BB 2/2/2 ND/2 2 ND ND/2 2MD 1/1/1 1/1 ND 2/2/2 2/2 NDSS 1/1/1 1/1 1 2/2/2 2/ND 1MC 2/2/2 2/2 2 2/2/2 2/2 2

aFrozen sections from acute (days 2–10 post onset) and healed (5–6 mo post-resolution) HAEM were studied for Pol DNA by PCR, Pol RNA by solution phase reversetranscriptase–PCR and Pol protein by immunohistochemistry with Pol antibody. Serial sections were studied for IFN-γ and TNF-α by in situ reverse transcriptase–PCR andstained with antibodies to IFN-γ, TNF-α, or TGF-β. All tissues were negative for TNF-α RNA and protein. ND 5 not done.

Table II. HSV antigen and cytokine expression in HAEMlesional skina

Pol DNA/Patient protein IFN-γ TGF-β p21waf

13c 2/2 2 1 144b,c 1/1 ND 2 243b,c 1/1 1 1 142b,c 1/1 1 1 141 1/1 1 2 258 1/1 ND 1 157b,c 1/1 1 1 156b,c 1/1 1 2 155 1/1 2 2 253 1/1 ND 1 146c 2/2 1 1 245b,c 1/1 1 2 2

105 1/1 1 1 1103 1/1 1 1 1102 1/1 1 2 1111b,c 1/1 ND 1 1104 2/2 2 ND 154b,c 1/1 1 ND 2

aFormalin-fixed sections of HAEM lesional skin were studied for Pol DNA byPCR and Pol protein by immunohistochemistry. Serial sections were stained withantibodies to IFN-γ, TNF-α, TGF-β, p21waf, p53, c-myc, or bcl-2. All werenegative for TNF-α, p53, c-myc, and bcl-2.

bStudied for Pol RNA by in situ reverse transcriptase–PCR. All were positive.cApoptotic nuclei identified histologically.ND 5 not done

Table III. HSV antigen and cytokine expression indrug-induced EM lesional skina

Figure 1. Immunoperoxidase staining of frozen skin tissues withPatient Pol IFN-γ TGF-β p21waf TNF-αPol antibody. (A) Intranuclear staining (arrows) in HAEM lesional skin(day 2 after lesion onset) stained with Pol antibody (3100). Magnification4c 2 2 2 2 1(3300) shown in inset. (B) Healed HAEM lesional skin stained with Pol109c 2 2 2 2 1antibody does not reveal nuclear staining. Magnification (3200) shown in48b,c 2 2 1 2 2inset. Similar results were obtained for normal skin adjacent to the21b 2 2 1 2 1HAEM lesion.23c 2 ND 2 2 1

49 2 2 2 2 ND110b,c 2 1 2 2 1 produced by monocytes/macrophages (Feghali and Wright, 1997;106 2 2 2 2 2 Murphy, 1998). In a first series of experiments, HAEM lesional108b,c 2 1 1 2 2 skin from patients WS, CI, MD, AD, SS, and MC (Table I) was30 2 ND 2 2 1 used in in situ reverse transcriptase–PCR with IFN-γ and TNF-α50c 2 2 2 2 1 primers. As summarized in Table I, lesions from five of six of51b 2 2 2 2 2

these patients (83%) were positive for IFN-γ RNA, but allwere negative for TNF-α RNA. IFN-γ RNA signals were inaSerial sections of lesional skin were studied for Pol, IFN-γ, TNF-α, TGF-β, and

p21waf by immunohistochemistry and for IFN-γ and TNF-α RNA by in situ reverse mononuclear cells located at the epidermis/dermis junction andtranscriptase–PCR. were primarily cytoplasmic. Perinuclear signals were also observedbStudied for Pol RNA by solution phase reverse transcriptase–PCR. All were (Fig 2A, B). The signals were specific, as the tissues were DNase-negative.

treated prior to assay and signals were eliminated by treatment ofcApoptotic nuclei identified histologically.ND 5 not done. the tissues (2 h) with RNase. Weaker IFN-γ signals were also seen

VOL. 113, NO. 5 NOVEMBER 1999 IFN-Γ IS EXPRESSED IN HAEM LESIONAL SKIN 811

Figure 3. Immunoperoxide staining with IFN-γ and TNF-αantibodies. HAEM lesional skin was stained with IFN-γ antibody (3200)(A) or TNF-α antibody (3 200) (B) and normal skin was stainedwith IFN-γ antibody (3200) (C). TNF-α staining of the normal skinwas negative.

Figure 2. In situ reverse transcriptase–PCR of HAEM lesional skin.(A) IFN-γ primers reveal positive cytoplasmic signals in infiltrating cells inthe upper dermis and at the epidermis–dermis junction (arrow) (3100). (B) but not TNF-α (Fig 3B) antibody. IFN-γ staining was in mono-Magnification of area in panel A identified by arrow (3400). (C) In situ nuclear cells and inflammatory intercellular infiltrates located in thereverse transcriptase–PCR of HAEM lesional skin with TNF-α primers is epidermis and at the epidermis/dermis junction (Fig 3A). Normalnegative (3100). perilesional skin (Fig 3C) and healed HAEM lesional skin (Table I)

were negative. As summarized in Tables I and II, 16 of the 21Pol positive HAEM lesional skin tissues (76%) stained with IFN-γby in situ hybridization done as described by Raqib et al (1996).

IFN-γ RNA was not seen in normal or healed HAEM lesional antibody and there was a very good correlation between stainingwith Pol and IFN-γ antibodies (p , 0.0001 by Pearson correlationskin, nor in HAEM lesional skin studied with TNF-α primers

(Fig 2C). In a second series of experiments, duplicate sections of coefficient). The tissues did not stain with TNF-α antibody.Staining with IFN-γ (Fig 4A) but not TNF-α (Fig 4B) antibodythe HAEM lesional skin tissues studied for Pol antigen (Tables I

and II) were stained with IFN-γ and TNF-α antibodies. Normal was also seen in lesional skin from all five HSV patients, suggestingthat IFN-γ presence in the skin is related to HSV antigen expression.perilesional skin and lesional skin from drug-induced EM patients

served as controls. HAEM lesional skin stained with IFN-γ (Fig 3A), By contrast to HAEM, only two of 10 (20%) drug-induced EM

812 KOKUBA ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Figure 4. Photomicrographs of HSV lesional skin. (A) Immuno-peroxidase staining with IFN-γ antibody evidences mononuclear andintercellular staining (arrows) (3200). (B) Immunoperoxidase staining withTNF-α antibody is negative (3100).

Figure 5. Immunoperoxidase staining of lesional skin from drug-induced EM with antibody to TGF-β, IFN-γ, or TNF-α. (A)TGF-β, (B) IFN-γ, (C) TNF-α. TNF-α staining is within keratinocytes,skin tissues stained with IFN-γ antibody (Fig 5B; Table III)mononuclear cells, and intercellular spaces in the upper dermis and at the(p , 0.0057 as compared with HAEM lesional skin by Fisher’sepidermis–dermis junction (3200).exact test), whereas a relatively high proportion [seven of 11 (64%)]

stained with TNF-α antibody (Fig 5C; Table III). We interpretthese data to indicate that: (i) mechanistically distinct processare involved in HAEM as compared with drug-induced EMinflammation, and (ii) a T cell component is involved in HAEM

kinase inhibitor p21waf is an effector of the TGF-β growth inhibitoryinflammation.signaling pathway (Datto et al, 1995; Ravitz and Wenner, 1997),serial sections were also stained with p21waf antibody. Staining wasKeratinocytes from HAEM, but not drug-induced EM

lesions, are positive for TGF-β and p21waf TGF-β is a seen in HSV (Fig 7D) and HAEM (Fig 7B, C) lesional skin, andthere was a good correlation between staining with TGF-β andcytokine that is upregulated in keratinocytes at sites of inflammation

(Kane et al, 1991) whereas it is virtually undetected in normal skin p21waf antibodies (p , 0.01 by Pearson correlation coefficient)(Table II). At the site of the HAEM lesion, p21waf staining was(Karonen et al, 1997). To examine whether TGF-β is expressed in

lesional skin from our patients, serial sections of the tissues studied intranuclear, consistent with the site of p21waf function (Fig 7B).In keratinocytes adjacent to the lesion the staining was cyto-for Pol and IFN-γ were stained with TGF-β antibody. Cytoplasmic

staining was seen in keratinocytes from 14 of 23 (61%) HAEM plasmic (Fig 7C).Normal skin (data not shown), healed HAEM lesional skin and(Fig 6A) and four of five (80%) HSV (Fig 6B) lesions. Normal

skin was negative (data not shown). Staining was not seen in most skin from drug-induced EM lesions did not stain (Tables I and III).The data suggest that HSV gene expression causes upregulation/[six of seven (86%)] of the healed HAEM lesional skin tissues

(Table II), nor in lesional skins from nine of 12 (75%) (Table III) activation of TGF-β and p21waf (directly or indirectly) in HSV andHAEM, but not in drug-induced EM lesions.drug-induced EM patients (Fig 5A). Because the cyclin-dependent

VOL. 113, NO. 5 NOVEMBER 1999 IFN-Γ IS EXPRESSED IN HAEM LESIONAL SKIN 813

Figure 6. Immunoperoxidase staining withTGF-β antibody. (A) HAEM lesional skin (3200).(B) HSV lesional skin (3100).

Figure 7. Photomicrograph of lesional skin (3200). (A) Hematoxylin and eosin staining of acute HAEM lesional skin. (B, C) Immunoperoxidasestaining of section in (A) with antibody to p21waf. The identity of the floating cells within bulla is unknown, but they demonstrate intranuclear signals(B, arrow). Cytoplasmic signals are seen in epidermis other than bulla (C, arrow). (D) Immunoperoxidase staining of HSV lesional skin with antibody top21waf. Cytoplasmic and intranuclear (arrow) signals are seen.

ingly demonstrate that tissues can retain restricted DNA sequencesDISCUSSIONfrom viral [namely, HTLV-1 (Pancake et al, 1996)] or bacterial

The salient feature of the data presented in this report is the [namely, Borrelia burgdorferi (Persing et al, 1994)] infectious agents,observation that HAEM lesions are positive for IFN-γ whereas while losing most of the genome. The expression of some viraldrug-induced EM lesions are positive for TNF-α, thereby providing genes in the absence of virus isolation suggests that promotermechanistic support for previous clinical conclusions that the two regulation differs from that operating within the context of viralconditions are distinct. The following comments seem pertinent infection. Indeed, DNA vaccine studies indicate that HSV geneswith respect to these findings. Our data are consistent with previous (namely, gB) are expressed at unexpectedly high levels in plasmid-reports that infectious virus is generally difficult to isolate from infected mice in the absence of virus specified proteins (namely,HAEM lesions (Smith and McLaren, 1977; Imafuku et al, 1997; ICP4) that are required for their expression in virus-infected cellsKokuba et al, 1998), but they retain and express viral DNA (McClements et al, 1996). It has long been suspected that thesequences (Yokoi, 1995; Imafuku et al, 1997; Kokuba et al, 1998). host immune response in HAEM is HSV-specific (Shelley, 1967;Because our studies were focused only on Pol, we do not exclude Howland et al, 1984). Previous studies had shown that T cells,the possibility that other viral genes are also expressed in HAEM primarily CD41 cells which respond to HSV antigens in vitro,

infiltrate HAEM lesions (Zaim et al, 1987; Malmstrom et al, 1990;lesions. Nonetheless, both experimental and human studies increas-

814 KOKUBA ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Kokuba et al, 1998). Because HSV-specific responses, including upregulation/activation is in response to HSV antigen expressionand/or IFN-γ production (Nickoloff et al, 1989). The exact roleCTL are primarily CD41 (Schmid, 1988; Zarling et al, 1988;

Aurelian et al, 1991), however, the immune mechanism responsible of TGF-β and p21waf in HAEM pathogenesis is unclear. Earlyin the inflammatory process TGF-β could be involved in thefor HAEM pathology (DTH or CTL) remained unclear. We find

that HAEM lesional skin is positive for IFN-γ, both as determined recruitment, activation (McCartney-Francis and Wahl, 1994), andsequestration (Barker et al, 1991) of immature leukocyte popula-by in situ reverse transcriptase–PCR and immunohistochemistry.

RNA signals were in mononuclear cells accumulating at the tions, whereas at later times it could suppress leukocyte activity,thereby contributing to lesion resolution (McCartney-Francis andepidermis/dermis boundary; staining was both intracellular and in

intercellular spaces at the site of mononuclear infiltration, below Wahl, 1994). p21waf, a cell cycle regulator involved in apoptosis,may be induced by TGF-β (Datto et al, 1995; Iavarone andthe area where Pol was found. Double staining with antibodies to

IFN-γ and various T cell subsets was not done because these Massague, 1997). Apoptotic nuclei were histopathologicallydetected in some HAEM (Table II) as well as drug-induced EMantibodies generally stain frozen sections which do not retain tissue

integrity, and we wanted to define the site of IFN-γ localization. lesions (Table III), however, and there was no apparent relationshipbetween their detection and the upregulation p21waf. Furthermore,It is well known, however, that IFN-γ is a product of antigen-

activated T cells; the production of which is now classically seen whereas p21waf functions in the nucleus (el-Deiry et al, 1995), itwas primarily intracytoplasmic in both HAEM and HSV lesions.as a hallmark of CD41 T helper 1 (Th1) type cells involved in

inflammatory reactions with a DTH character (Feghali and Wright, Nonetheless, the presence of IFN-γ, TGF-β, and p21waf in HAEM,but not drug-induced EM lesions, provides the mechanistic support1997; Billiau et al, 1998). Because: (i) IFN-γ is present at DTH

sites (Tsicopoulos et al, 1992), (ii) it was present in HSV lesional for prior clinical and histopathologic observations that these areseparate conditions (Howland et al, 1984; Zaim et al, 1987; Assierskin, and (iii) there was a strong correlation (p , 0.0001) between

IFN-γ presence and HSV gene expression in HAEM lesions, et al, 1995).we conclude that HAEM immunopathology includes a DTHcomponent to HSV antigens in the skin. Consistent with thisinterpretation, IFN-γ was not seen in healed HAEM lesions that We thank Dr. Catharine Lisa Kauffman for help with interpretation of thewere negative for HSV antigens and had previously been shown dermatopathology data and Dr. Thomas Horn for help with the recruitment ofto be free of T cell infiltration (Kokuba et al, 1998). We do specimens. Dr. Kokuba was supported by a grant from the Albert Shapiro Fundnot exclude the role of HSV genes other than Pol in HAEM for Dermatologic Research. These studies were supported by grant AR42647 fromimmunopathology, however, and final conclusions must await the the National Institute of Arthritis and Musculoskeletal and Skin Diseases, NIH.establishment from HAEM lesions, of T cell clones that are specificfor Pol and/or other HSV proteins and produce IFN-γ.

What is the role of IFN-γ in HAEM lesions? IFN-γ could REFERENCESactivate macrophages and keratinocytes to release a wide spectrum Abbas AK, Murphy KM, Sher A: Functional diversity of helper T lymphocytes.of cytokines and inflammatory molecules that amplify the cutaneous Science 383:787–793, 1996

Aslanzadeh J, Helm KF, Espy MJ, Muller SA, Smith TF: Detection of HSV-specificinflammation and increase the accumulation of circulating leuko-DNA in biopsy tissue of patients with erythema multiforme by polymerasecytes (Barker et al, 1991; Nacy and Maltzer, 1995). It could alsochain reaction. Br J Dermatol 126:19–23, 1992be involved in amplifying the immune process by inducing CD41

Assier H, Bastuji-Garin S, Revuz J, Roujeau J-C: Erythema multiforme with mucousT cell accumulation in the skin (Colditz and Watson, 1992), membrane involvement and Stevens–Johnson syndrome are clinically different

disorders with distinct causes. Arch Dermatol 131:539, 1995favoring the differentiation of Th1 cells while inhibiting theAurelian L, Smith CC, Wachsman M, Paoletti E: Immune responses to herpesgeneration of Th2 cells (Abbas et al, 1996) and by upregulating

simplex virus in guinea pigs (footpad model) and mice immunized withHLA class I and II expression, thereby enhancing the antigen- vaccinia virus recombinants containing herpes simplex virus glycoprotein D.presenting capability of keratinocytes (Basham et al, 1984). Finally, Rev Infect Dis 13(Suppl. 11):S924–S934, 1991 (review)

Barcellos-Hoff MH: Latency and activation in the control of TGF-beta. J Mammaryby increasing the susceptibility of keratinocytes to lysis by activatedGland Biol Neopl 1:353–363, 1996PBL (Kalish, 1989), IFN-γ could also be involved in cell cyto-

Barker JNWN, Mitra RS, Griffiths CEM, Dixit VM, Nickoloff BJ: Keratinocytestoxicity. as initiators of inflammation. Lancet 337:211–214, 1991Significantly, HSV and HAEM lesions were free of TNF-α, an Basham TY, Nickoloff BJ, Merigan TC, Morhenn VB: Recombinant gamma

interferon induces HLA-DR on cultured human keratinocytes. J Invest Dermatolinflammatory cytokine produced by monocytes/macrophages in an83:88–92, 1984antigen-unrelated process (Feghali and Wright, 1997; Murphy,

Beutler B: TNF immunity and inflammatory disease: lessons of the past decade.1998). By contrast, IFN-γ was not seen in drug-induced EM lesions J Invest Med 43:227–235, 1995and they did not show any evidence of antigen-driven immune Billiau A, Hermans H, Vermeire K, Matthys P: Immunomodulatory properties of

interferon-gamma. An update. Ann N Y Acad Sci 856:22–32, 1998response. Keratinocytes from a relatively high proportion of theBrice SL, Stockert SS, Bunker JD, Bloomfield D, Huff JC, Weston WL: Comparisondrug-induced EM lesions were positive for TNF-α, its release a

of the immune response in patients with herpes labialis and those with herpessign of toxic injury to the cells. TNF-α was also present at associated erythema multiforme. Arch Dermatol Res 285:193–195, 1993inflammatory sites located at the epidermis/dermis junction, pre- Cao M, Xiao X, Egbert B, Darragh TM, Yen TSB: Rapid detection of cutaneous

herpes simplex virus infection with polymerase-chain reaction. J Invest Dermatolsumably produced by monocytes/macrophages (Beutler, 1995;92:391–392, 1989Feghali and Wright, 1997; Murphy, 1998). TNF-α induces skin

Colditz IG, Watson DL: The effect of cytokines and chemotactic agonists on theinfiltration by neutrophils in preference to lymphocytes (Colditz migration of T lymphocytes into skin. Immunology 76:272–278, 1992and Watson, 1992), its presence in drug-induced EM lesions Darragh TM, Egbert M, Berger TG, Yen TSB: Identification of herpes simplex virus

DNA in lesions of erythema multiforme by the polymerase chain reaction.consistent with the preponderance of mononuclear cells other thanJ Am Acad Dermatol 24:23–26, 1991T cells in infiltrates from these lesions (Margolis et al, 1983). In

Datto MB, Li Y, Panus JF, Howe DJ, Xiong Y, Wang XF: Transforming growthaddition to their difference in terms of IFN-γ presence, HAEM factor beta induces the cyclin dependent kinase inhibitor p21 through a p53and drug-induced EM lesions differed in the expression of TGF-β independent mechanism. Proc Natl Acad Sci USA 92:5545–5549, 1995

el-Deiry WS, Tokina T, Waldman T, et al: Topological control of p21/Waf/CIP1and p21waf, both of which were identified by immunohistochemicalexpression in normal and neoplastic tissues. Cancer Res 55:2910–2919, 1995staining in a relatively high proportion of HAEM lesions, but were

Feghali CA, Wright TM: Cytokines in acute and chronic inflammation. Front Bioscnot seen in drug-induced EM lesions. TGF-β immunoreactivity 2:d12–d26, 1997could reflect de novo synthesis, or release from a latent complex, in Howland WW, Golitz LE, Huff JC, Weston WL: Erythema multiforme: Clinical,

histopathologic and immunologic study. J Am Acad Dermatol 10:438–446, 1984which the latency associated protein and latent TGF-β bindingIavarone A, Massague J: Repression of the CDK activator Cdc25A and cell cycleprotein mask epitopes of mature TGF-β (Barcellos-Hoff, 1996).

arrest by cytokine TGF-β in cells lacking the CDK inhibitor p15. NatureRNA studies are needed in order to differentiate between these 387:417–422, 1997two interpretations. Immunoreactivity with both antibodies was Imafuku S, Kokuba H, Aurelian L, Burnett J: Expression of herpes simplex virus

DNA fragments located in epidermal keratinocytes and germinative cells isalso seen in HSV lesional skin, suggesting that TGF-β and p21waf

VOL. 113, NO. 5 NOVEMBER 1999 IFN-Γ IS EXPRESSED IN HAEM LESIONAL SKIN 815

associated with the development of erythema multiforme lesions. J Invest inhibition by recombinant gamma interferon is associated with increasedproduction of transforming growth factor alpha in keratinocytes cultured fromDermatol 109:550–556, 1997

Kalish RS: Non-specifically activated human peripheral blood mononuclear cells are psoriatic lesions. Br J Dermatol 121:161–174, 1989Nuovo GJ: Reverse transcriptase in situ PCR with direct incorporation of digoxigenin-cytotoxic for human keratinocytes in vitro. J Immunol 142:74–80, 1989

Kane CJM, Hebda PA, Mansbridge JN, Hanawalt PC: Direct evidence for spatial II-dUTP. Protocol and applications. Biochemica 11:4–6, 1993Pancake BA, Wassef EH, Zucker-Franklin D: Demonstration of antibodies toand temporal regulation of transforming growth factor β1 expression during

cutaneous wound healing J. Cell Physiol 148:157–173, 1991 HTLV-I tax in patients with cutaneous T cell lymphoma, mycosis fungoides,who are seronegative for antibodies to the structural proteins of the virus.Karonen T, Jeskanen L, Keski-Oja J: Transforming growth factor β1 and its

latent form binding protein-1 associate with elastic fibres in human dermis: Blood 88:3004–3009, 1996Persing DH, Rutledge BJ, Rys PN, et al: Target imbalance: disparity of Borreliaaccumulation in actinic damage and absence in anetoderma. Br J Dermatol

137:51–58, 1997 burgdorferi genetic material in synovial fluid from Lyme arthritis patients. J InfectDis 169:668–672, 1994Kokuba S, Imafuku S, Huang S, Aurelian L, Burnett J: Expression of a herpes

simplex virus (HSV) gene and the qualitative nature of the HSV-specific T Raqib R, Ljungdahl A, Lindberg AA, Andersson U, Andersson J: Local entrapmentof interferon γ in the recovery from Shigella dysenteriae type I infection. Gutcell response are associated with the development of erythema multiforme

lesions. Br J Dermatol 138:952–964, 1998 38:328–336, 1996Ravitz MJ, Wenner CE: Cyclin dependent regulation during G1 phase and cellKomminoth P, Adams V, Long AA, et al: Evaluation of methods for hepatitis C virus

detection in archival liver biopsies. Pathol Res Pract 190:1017–1025, 1994 cycle regulation by TGFβ. Adv Cancer Res 71:165–207, 1997Schmid DS: The human MHC-restricted cellular response to herpes simplex virusLoche M, Mach B: Identification of HIV-infected seronegative individuals by a

direct diagnostic test based on hybridization to amplified viral DNA. Lancet type 1 is mediated by CD41, CD8- T cells and is restricted to the DR regionof the MHC complex. J Immunol 140:3610–3616, 1988ii:418–421, 1998

Malmstrom M, Ruokonen H, Konttinen YT, et al: Herpes simplex virus antigens Serrano D, Ghiotto F, Roncella S, et al: The patterns of IL-2, IFN-γ, IL-4 and IL-5gene expression in Hodgkin’s disease and reactive lymph nodes are similar.and inflammatory cells in oral lesions in recurrent erythema multiforme.

Immunoperoxidase and autoradiographic studies. Acta Derm Venereol (Stockh) Haematologica 82:542–549, 1997Shelley WB: Herpes simplex virus as a cause of erythema multiforme. JAMA70:405–410, 1990

Margolis RJ, Tonnesen MG, Harrist TJ, Bhan AK, Wintroub BV, Mihm MC, Soter 201:153–156, 1967Smith EB, McLaren LC: Attempt to recover simplex virus from the skin sites ofNA: Lymphocyte subsets and Langerhans cells indeterminate cells in erythema

multiforme. J Invest Dermatol 81:403–406, 1983 recurrent infection. Int J Dermatol 16:748–757, 1977Tsicopoulos A, Hamid Q, Varney V, Ying S, Moqbel R, Durham SR, Kay AB:McCartney-Francis NL, Wahl SM: Transforming growth factor β: a matter of life

and death. J Leukocyte Biol 55:401–409, 1994 Preferential messenger RNA expression of Th1-type cells (IFN-γ1, IL-21)in classical delayed type (tuberculin) hypersensitivity reactions in human skin.McClements WL, Armstrong ME, Keys RD, Liu MA: Immunization with DNA

vaccines encoding glycoprotein D or glycoprotein B, alone or in combination J Immunol 148:2058–2061, 1992Yokoi K: Erythema multiforme and HSV. Jpn J Dermatol 105:1661–1664, 1995induces protective immunity in animal models of herpes simplex virus-2

disease. Proc Natl Acad Sci USA 93:11414–11420, 1996 Yu F, Itoyama Y, Kiar J-I, Fujihara K, et al: TNF-β produced by human Tlymphotropic virus type I infected cells influences the proliferation of humanMiura S, Smith CC, Burnett JW, Aurelian L: Detection of viral DNA within skin

of healed recurrent herpes simplex infection and erythema multiforme lesions. endothelial cells and fibroblasts. J Immunol 152:5930–5938, 1994Zaim MT, Giorno RC, Golitz LE, Kunke KS, Huff JC: An immunopathologicJ Invest Dermatol 98:68–72, 1992

Murphy KM: T lymphocyte differentiation in the periphery. Curr Opin Immunol study of herpes-associated erythema multiforme. J Cutan Pathol 14:257–262, 198710:226–232, 1998

Nacy CA, Maltzer MS: T cell-mediated activation of macrophages. Curr Opin Zarling JM, Moran PA, Brewer L, Ashley R, Corey L: Herpes simplex virus (HSV) -specific proliferative and cytotoxic T cell responses in humans immunized withImmunol 3:330–335, 1995

Nickoloff BJ, Mitra RS, Elder JT, Fisher GH, Voorhes JJ: Decreased growth an HSV type 2 glycoprotein subunit vaccine. J Virol 62:4481–4485, 1988