cytopathic effect in human papillomavirus type 1–induced

9
Cytopathic Effect in Human Papillomavirus Type 1- Induced Inclusion Warts: In Vitro Analysis of the Contribution of Two Forms of the Viral E4 Protein Claire Rogel-Gaillard, Gerard Pehau-Arnaudet , Frans,:oise Breitburd, and Gerard Orth Unite INSERM 190, Unite des Papillomavims, Instit ut Pasteur, Paris, France Myrm.ecia warts induced by human papillomavirus type 1 (HP V1) are characterized by abundant eosinophilic inclu- sions associated with HPVl E4 gene products. The 111ajor BPVl E4 proteins are a 17-kilodalton (kDa) EI-E4 fusion protein and a 16-kDa species lacking the five El aminoacids and a few E4 residues. To study the contribution of E4 pro- teins to the formation of myrmecia inclusions, we used a previously designed transient ex pression system in the rabbit VX2 -R keratinocyte line. We find that the EI-E4 and an E4 protein without the El residues (E4-3200) form eosinophilic inclusions . Ultrastructural and immunoelectron microscopic studies show that the electron-dense, keratohyalin-like myr- rn ec ia inclusions are recognized by anti-E4 antibodies . They are associated with tonofilament bundles at their periphery in the cytoplasm or free of filaments in the nucleus. The EI-E4 inclusions formed in vitro are also homogeneously electron S ixty-six types of human papillomavirus (HPV) have been identified so far [1,2). Most of these small epitheliotropic DNA viruses induce benign cutaneous or genital tumors of keratinocytes. Certain HPV types are associated with premalignant lesions that may progress to malignancy under the influence of cofactors still little known [3,4). HPVs replicate exclusively in terminally differentiating nocytes of the le sions they induce. Vegetative viral DNA replIca- tion is triggered upon the onset of the differentiation process in the and is followed by the.synthes.is of .the L1 and L2 capsid protems and by . the particles super- ficial layers [5 -8). Viral multiplication IS associated With a cyto- pathic effect recognized by the presence of clear cells in skin warts and koilocytes in genital lesions [9). In HPV1-induced palmo-plan- tar warts, known al so as myrmecia or inclusion warts [10), the cytopathic effect is characterized by the presence of eosinophilic and electron-dense cytoplasmic and nuclear inclusions, usu ally referred to as keratohyalin [9) or keratohyalin-like granules [7 ,11, 12) . Fur- thermore, the lack of detection of keratins K1 and K10 [12) , filag- grin, and loricrin (our unpublished observations) indicates that the Manuscr ipt received Marc h 23, 1993; accepted for publication July 1 2, 1993. Presented in part at the 11 th Papi ll omavirus Workshop, Edinburgh, Scot- Jand , September 5 - 11 , 1992. Reprint requests to: Dr. Gerard Orth, Unite INSERM 190 , Unite des PapiJIornavirus, Institut Pasteur, 25 rue du Docteur Roux, 75724 Pans Cedex 15, France. Abbrevia tions : IF, intermediate filament; PFA, paraformaldehyde; TN- BS A, Tris-NaCI- bovine serum albumine buffer; RSV LTR: long terminal repeat of the Rou s sarcoma vims. dense, and are usually associated with tonofilaments at their periphery in the cytoplasm and free of filaments in the nu- cleus. The E4-3200 inclusions are exclusively cytoplasmic a nd heterogeneously electron dense, with a fibrillar structure made of enta ngled 10-nm filaments. The expression of either protein in VX2-R cells does not result in the collapse of the cytokeratin network, as shown by immunofluorescence doub le-labeling experiments. This is in contrast to data re- ported for the HPV16 EI-E4 protein. Our findings indicate that the EI-E4 protein by itself accounts for the formation of myrmecia inclusions, and suggest that the five N -termina l E 1 playa major role in the interaction ofE4 proteins With I11termed late filaments. Key words: intermedi ate fila- ment-associated protein/ cytokeratin/ myrmecia wart/immu- noelectron microscopy.] Invest D ennato l 1 01 :843 -8S1, 1993 keratinization process [13) is altered in infected cells. It was recently demonstrated that myrmecia inclusions contain the products of the H.PVl E4 reading frame [12,14). The HPVl E4 proteins con- stitute a family of polypeptides [1 2,14) that are ass umed to derive from an EI-E4 fusion protein [15). The E4 proteins are phosphory- lated [12,16) and are first detected as a doublet of 16 -17 kilodaltons (kDa) [12,15). The 17-kDa protein corresponds to the initial E1-E4 speCies whereas the 16-kDa protein lacks the fiveEl aminoacids and an as-yet undetermined small number ofE4 residues [15). The func- tion of the E4 proteins remains poorly understood although it has been suggested that E4 proteins behave like intermediate filament (IF)-associated proteins and cou ld disrupt and collapse the cytokera- tin network [17) . . We previ?usly designed an i/'l IJitro system allowing the tran- sient expressIOn of several forms of HPV1 E4 proteins in the rabbit VX2-R keratinocyte line [18). We found that the E1-E4 and an E4 protein close to the 16-kDa species (E4-3200) both form inclusions with distinct immunofluorescence patterns [18), suggesting that each protem cou ld have a spec ifi c contribution to the formation of myrmecia inclusions. Our aim has been to compare the E4 inclu- sions obtained i" IJ ilrO to those observed i" vivo by immunoelectron methods, to examine the effect of the E4 protein expressIOn on the keratm network by immunofluorescence double- labeling experiments. MATERIALS AND METHODS Expression Vectors, Cells, and Transfections The pE1 1\E4 and pE4- 3200 vectors were described previously [18] . The VX2-R ce ll line derived from a transplantable rabbit carcinoma was propagated as reported [19] . Cells were transfected, u S111g a calci um phosphate-DNA technique [20], in condi- 0022-202Xj93jS06.00 Copyright © 1993 by The Society for In vest i gative Dermatology, Inc. 843

Upload: others

Post on 09-Nov-2021

4 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Cytopathic Effect in Human Papillomavirus Type 1–Induced

Cytopathic Effect in Human Papillomavirus Type 1-Induced Inclusion Warts: In Vitro Analysis of the Contribution of Two Forms of the Viral E4 Protein

Claire Rogel-Gaillard, Gerard Pehau-Arnaudet, Frans,:oise Breitburd, and Gerard Orth Unite INSERM 190, Unite des Papillomavims, Institut Pasteur, Paris, France

Myrm.ecia warts induced by human papillomavirus type 1 (HPV1) are characterized by abundant eosinophilic inclu­sions associated with HPVl E4 gene products. The 111ajor BPVl E4 proteins are a 17-kilodalton (kDa) EI-E4 fusion protein and a 16-kDa species lacking the five El aminoacids and a few E4 residues. To study the contribution of E4 pro­teins to the formation of myrmecia inclusions, we used a previously designed transient expression system in the rabbit VX2-R keratinocyte line. We find that the EI-E4 and an E4 protein without the El residues (E4-3200) form eosinophilic inclusions. Ultrastructural and immunoelectron microscopic studies show that the electron-dense, keratohyalin-like myr­rnecia inclusions are recognized by anti-E4 antibodies. They are associated with tonofilament bundles at their periphery in the cytoplasm or free of filaments in the nucleus. The EI-E4 inclusions formed in vitro are also homogeneously electron

Sixty-six types of human papillomavirus (HPV) have been identified so far [1,2). Most of these small epitheliotropic DNA viruses induce benign cutaneous or genital tumors of keratinocytes. Certain HPV types are associated with premalignant lesions that may progress to malignancy

under the influence of cofactors still little known [3,4). HPVs replicate exclusively in terminally differentiating ker~ti­

nocytes of the lesions they induce. Vegetative viral DNA replIca­tion is triggered upon the onset of the differentiation process in the suprabasall~yers and is followed by the.synthes.is of .the L1 and L2 capsid protems and by. the assemb~y o~ vlr~l particles mn~ore super­ficial layers [5 -8). Viral multiplication IS associated With a cyto­pathic effect recognized by the presence of clear cells in skin warts and koilocytes in genital lesions [9). In HPV1-induced palmo-plan­tar warts, known also as myrmecia or inclusion warts [10), the cytopathic effect is characterized by the presence of eosinophilic and electron-dense cytoplasmic and nuclear inclusions, usually referred to as keratohyalin [9) or keratohyalin-like granules [7,11,12) . Fur­thermore, the lack of detection of keratins K1 and K10 [12) , filag­grin, and loricrin (our unpublished observations) indicates that the

Manuscript received March 23, 1993; accepted for publication July 12, 1993.

Presented in part at the 11 th Papi llomavirus Workshop, Edinburgh, Scot­Jand, September 5 - 11 , 1992.

Reprint requests to: Dr. Gerard Orth, Unite INSERM 190, Unite des PapiJIornavirus, Institut Pasteur, 25 rue du Docteur Roux, 75724 Pans Cedex 15, France.

Abbreviations: IF, intermediate filament; PFA, paraformaldehyde; TN­BSA, Tris-NaCI-bovine serum albumine buffer; RSV LTR: long terminal repeat of the Rous sarcoma vims.

dense, and are usually associated with tonofilaments at their periphery in the cytoplasm and free of filaments in the nu­cleus. The E4-3200 inclusions are exclusively cytoplasmic and heterogeneously electron dense, with a fibrillar structure made of entangled 10-nm filaments. The expression of either protein in VX2-R cells does not result in the collapse of the cytokeratin network, as shown by immunofluorescence doub le-labeling experiments. This is in contrast to data re­ported for the HPV16 EI-E4 protein. Our findings indicate that the EI-E4 protein by itself accounts for the formation of myrmecia inclusions, and suggest that the five N -terminal E 1 al~in~acids playa major role in the interaction ofE4 proteins With I11termedlate filaments. Key words: intermediate fila­ment-associated protein/ cytokeratin/ myrmecia wart/immu­noelectron microscopy.] Invest Dennatol 1 01 :843 -8S1, 1993

keratinization process [13) is altered in infected cells. It was recently demonstrated that myrmecia inclusions contain the products of the H.PVl E4 op~n reading frame [12,14). The HPVl E4 proteins con­stitute a family of polypeptides [1 2,14) that are assumed to derive from an EI-E4 fusion protein [15). The E4 proteins are phosphory­lated [12,16) and are first detected as a doublet of 16 -17 kilodaltons (kDa) [12,15). The 17-kDa protein corresponds to the initial E1-E4 speCies whereas the 16-kDa protein lacks the fiveEl aminoacids and an as-yet undetermined small number ofE4 residues [15). The func­tion of the E4 proteins remains poorly understood although it has been suggested that E4 proteins behave like intermediate filament (IF)-associated proteins and cou ld disrupt and collapse the cytokera­tin network [17) . . We previ?usly designed an i/'l IJitro system allowing the tran­

sient expressIOn of several forms of HPV1 E4 proteins in the rabbit VX2-R keratinocyte line [18). We found that the E1-E4 and an E4 protein close to the 16-kDa species (E4-3200) both form inclusions with distinct immunofluorescence patterns [18), suggesting that each protem cou ld have a specific contribution to the formation of myrmecia inclusions. Our aim has been to compare the E4 inclu­sions obtained i" IJilrO to those observed i" vivo by immunoelectron micros~opic methods, ~nd to examine the effect of the E4 protein expressIOn on the keratm network by immunofluorescence double­labeling experiments.

MATERIALS AND METHODS

Expression Vectors, Cells, and Transfections The pE11\E4 and pE4-3200 vectors were described previously [18] . The VX2-R cell line derived from a transplantable rabbit carcinoma was propagated as reported [19]. Cells were transfected, uS111g a calcium phosphate-DNA technique [20], in condi-

0022-202Xj93jS06.00 Copyright © 1993 by The Society for Investigative Dermatology, Inc.

843

Page 2: Cytopathic Effect in Human Papillomavirus Type 1–Induced

844 ROGEL-GAILLARD ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

"

c

• • ••

Figure L HPV1 .E4 inclusions in .wart se~tions (A,D) an~ in VX2-R cells transfected with the pEl 1\E4 (B,E) or pE4-3200 (C!F) vectors. Eosinophilic cytoplasmIc InclusIOns are observed 111 a section of an HPV1-l11duced wart (A) and In trans fcc ted VX2-R cells (B,C) after hematoxyll11 and eosin staining. The inclusions (arrows) are recognized by the G70S anti -E4 antiserum, using an immunoperoxidase test (D-F). Specimens were fixed in PFA (A-C) or Bouin's fluid (D-F). Magnification: bar, lS.S J.lm.

tions detailed previously [18]. For immunoprecipitation and electron micro­scopic experiments, cel ls were seeded in 60-mm dishes at a density of 3 X 10· cells per dish. For light and fluorescence microscopy, 1 X 10· cel ls were seeded on 30-mm glass coverslips in 3S-mm dishes. All experiments were performed 48 h after transfection.

Antisera The antiserum specific for HPVl E4 proteins (G70S) is a poly­clonal guinea pig antiserum raised against the 16-17-kDa E4 protein dou­blet purified from an HPV1-induced wart [1 2]. The polyclonal rabbit anti­serum (RAK) [21) and the AEl and AE3 monoclonal antibodies [22) raised against the keratins from human palm and sole were a generous gift from T.T. Sun. The G70S and RAK antisera were both used at a 1 : 100 dilution, unless specified otherwise, and the monoclonal antibodies were used undi­luted.

Radioimmunoprecipitation and Immunoblotting Experiments For immunoprecipitation experiments, the VX2-R cells were labeled with stabilized L[3SS)-cysteine (48 TBq/mM, O.S mCi per 60-mm dish; Amer­sham Ltd., UK) or with carrier-free 32P-orthophosphate (O.S mCi per 60-mm dish; Amersham) for 2 or 4 h. Cell extracts were prepared and immuno­precipitated with the G70S antiserum in conditions previously described l1 8). Immune complexes collected with protein A-sepharose were analyzed by electrophoresis in a 7.S-17.S% polyacrylamide gradient gel in the pres­ence of sodium dodecyl sulfate (SOS). Labeled proteins were detected by fluorography, using Autofluor (National Diagnostics, Somerville, NJ) and Hyper-Film Betamax (Amersham). Molecular weights were assessed by

comparison with I4C-labcled protein standards (Amersham) run in contigu­ous lanes.

For immunoblotting experiments, keratin extracts were prepared from VX2-R cells [1 9) and electrophoresed in an 8.S% polyacrylamide slab gel in the presence ofSOS. Polypeptides were either stained in the gel by Coomas­sie blue or transferred onto a nitrocellulose sheet (Hybond C; Amersham) by electroblotting. Blots were incubated successively with RAK antiserum (1: 200 dilution) or AE1 and AE3 monoclonal antibodies and anti-rabbit or anti-mouse IgG peroxidase-conjugated IgG (1 : 200 dilution; Cappel, USA). Peroxidase activity was detected in the presence of 0.03% H20 2 and 3,3'di­aminobenzidinc tetrachloride (O.S mg/ml) (Sigma Chemical Co., St. Louis, MO).

Light and Fluorescence Microscopy For hematoxylin and eosin stain­ing, plantar-wart specimens or VX2-R cells were fixed in 3% paraformalde­hyde (PFA) for 18 h or 20 min, respectively. Cells were washed in SO rnM ammonium chloride (NH.CI) -phosphate-bulfered saline (PBS) and per­meabilized as described [1 8). For immunoperoxidase experiments, wart specimens or VX2-R cells were fixed in Bouin's fluid for 48 h or 1 h. respectively. Oeparalfinized 7-J.lm tissue sections and fixed cells were incu­bated with the G70S antiserum for 30 min at 37"C, then with peroxidase­conjugated protein A (1 : 10 dilution) (Amersham) for 30 min at room tem­perature. The peroxidase activity was revealed as described above. Specimens were counterstaincd with hematoxylin and mounted in Eukitt. For immu­nofluorescence double-labeling experiments, VX2-R cell mono layers were

Page 3: Cytopathic Effect in Human Papillomavirus Type 1–Induced

VOL. 101, NO.6 DECEMBER 1993 HPV1 E4 PROTEINS AND MYRMECIA WART INCLUSIONS 845

Figure 2_ Ultrastructure (A) and immunoelectron microscopic identification (B,CJ ofHPVl E4 inclusions in a myrmecia wart. The myrmecia inclusions are homogeneously electron dense (A) and both the cytoplasmic (B) and the nu~lear (C) 1I1cluSlOns are labeled with the G705 anti-E4 antiserum, as revealed by IS-nm gold particle-conjugated protein A. Cytoplasmic inclusions are assocIated wIth abundant tonofi lament bundles at their periphery. For immunologic detection of E4 proteins, ultrathin sections of epon-araldite- embedded wart specimen were treated with sodium metaperiodate. Magnification: bar, 0.7 11m (AJ; 0.5 !-tm (B,C) .

fixed in acetone for 5 min at -20·C and air-dried. After hydration in PBS, cells were incubated successively with the RAK antiserum, a rhodamine­conjugated anti-rabbit antibody (1/200 dilution) (Immunotech S.A., Mar­seille, France), the G705 antiserum, and a fluorescein-conjugated anti­guinea pig antibody (1/100 dilution) (Biosys, Compiegne, France). All incubations were done at room temperature for 30 min and followed by extensive washings in PBS. Cells were mounted in Cytifluor (Agar Scien­tific Ltd., UK).

Electron Microscopy An HPV1-induced wart fragment was fixed at pH 7.2 in 1 % glutaraldehyde for 1 h at room temperature, post-fixed in 1 % osmium tetroxide for 1 hat 4· C, and embedded in an epon-araldite mixture.

Transfected VX2-R cell monolayc[s wcre rinscd in 0.1 M phosphate buffer, pH 7.2, three times for 5 min and fixed in I % glutaraldehyde for 1 h at room temperature. After washing three times for 5 min in phos­phate buffer and once in 50 mM NH.Cl for 30 min at 4·C, cell mono­layers were washed overnight at 4· C in phosphate buffer and treated with 1% osmium tetroxide for 1 h at 4 · C. The cells were scraped with a rubber policeman, pre-embedded in 2% agar in 0.1 M cacodylate buffer, pH 7.2, as described [23). Agar blocks were progressively dehydrated and embedded in an epon-araldite mixture. VX2-R cells were also processed for embedding in LR White [24) and Lowicryl HM20 [25) resins. Fixa­tion was done in a mixture of 1 % glutaraldehyde and 0.2% picric acid for 15 min before LR White embedding or in a mixture of 4% paraformalde-

Page 4: Cytopathic Effect in Human Papillomavirus Type 1–Induced

846 ROGEL-GAILLARD ET AL

E1- E4 ""­E4-3200'-

a b

2h

35 5

c a

2h

b c a b

--

4h

32p

Figure 3_ Expression and phosphorylation of HPVl E4 proteins in VX2-R cells. VX2-R cells transfected with pEl1\E4 (aJ or pE4-3200 (bJ and mock­transfected ce lls (c) were labeled with [35S) cysteine or [32P) orthophosphate for 2 h or 4 h. Protein extracts were immunoprecipitated with the G705 anti-E4 antiserum and separated in an SDS-polyacrylamide gradient gel. Labeled EI-E4 and E4-3200 proteins are indicated.

hyde (PFA) and 0.5% glutaraldehyde for 30 min before Lowicryl embed­ding. After washing three times fo r 5 min in phosphate buffer, once In

50 mM NH.Cl for 30 min at 4 · C, and overnight in phosphate buffer, the cells were collected and embedded in 2% agar, as described above. The blocks were progressively dehydrated and embedded in ice-cold LR White resin (Agar Scientific Ltd.) (24) or treated in 0.5% uranyl acetate for 30 nl1n at 4· C, before progressive dehydration and embedding at - 30· C in HM20 Lowicryl resin (Balzers Union, Boiziau Distribution, Selles sur Cher, France) [25).

Ultrathin sections were cut with an ultramicrotome (Reichert OMU2) and mounted on 200-mesh carbon-coated nickel grids. Sections stained with aqueous solutions of 2% uranyl acetate for 60 mi11 and lead citrate for 1 min were observed in a Philips EM 201 transmission electron micro­scope.

Immunoelectron Microscopy Ultrathin sections of VX2-R cells and wart tissue embedded in epon-araldite were treated in a saturated aqueous solution of sodium metaperiodate for 60 min (26). Sections of cells embed­ded in LR-white or Lowicryl resins were processed without pretreatment. Sections were incubated in 0.5% bovine serum albumin diluted in 20 mM T ris!HCl, pH 7.4, and 0.5 M NaCI (TN-BSA) for 20 min, followed by a 60 min-incubation in G705 antiserum diluted in TN-BSA. After washing four times for 3 min and twice for 30 min in TN-BSA, sections were incubated for 30 min in 15 nm gold particle-conjugated protein A (Amersham), washed twice for 3 min and once for 20 min in TN-BSA, then for 20 min in distilled water. After fixation in t % glutaraldehyde for 10 min and washing five times in distilled water for 5 min, sections were stained in aqueous solutions of2% uranyl acetate fo r 60 min and in lead citrate for 1 min. All steps were performed at room temperature.

RESULTS

Immunoelectron Microscopic Localization of HPVl E4 Proteins to Myrmecia Inclusions HPV1-induced myrmecia are characterized by the presence of irregularly shaped eosinophilic inclusions, mostly cytoplasmic, that are found as early as in the parabasal cell layers (Fig 1A) [7,10 -12] . Inclusion-containing cells are labeled by an antiserum raised against the 16 - 17-kDa HPV1 E4 proteins [12]. which stains both the cytoplasm and the inclusions (Fig 1D). In contrast to granules found in the normally differentiat­ing cells running amid productively infected wart cells [7,10-12], these inclusions are not stained by anti-fi laggrin and anti-loricrin antibodies (data not shown). Ultrastructural studies show that the cytoplasmic myrmecia inclusions are homogeneously electron dense and associated at their periphery with tonofilament .bundles (Fig 2A) . Free tonofilament bundles are also detected , but m lesser amounts than in the normally differentiating wart cells, as previ­ously described [7] . Round electron-dense inclusions free of tonofi­laments are also observed in the nuclei of some cells (Fig 2A) [7] . Immunogold-Iabeling experiments show that both th e cytoplasmic (Fig 2B) and the nuclear (Fig 2C) inclusions are recognized by the

THE JOUrtNAL OF INVESTIGATIVE DERMATOLOGY

anti-E4 antiserum. This clearly demo nstrates that the HPV1 E4 gene products, although predominantly localized in the cytoplasm, may be found in the nucleus also. The to nofilaments found asso­ciated ,:ith the .inclusions or free in the cytoplasm are not stained by the antl-E4 antIserum (FIg 2B), but the dIffus.e cytoplasmic labeling observed by ll ght mICroscopy unmunoperoxldase experiments (Fig 1D) is not detected either.

Transient Expression of Two HPVl E4 Proteins (EI-E4 and E4-3200) in the VX2-R Keratinocyte Line We have previ­ously developed an ill lI itro system allow ing the transient expression ofHPV1 E4 proteins differing by their N -terminal end in the rabbit VX2-R cells [18] . To analyze the contribution of different E4 pro­teins to th e formation of myrmecia inclusions, we have used vectors expressing the E1-E4 protein (pE1 I\E4) and an E4 protein lacking the five E1 amino acids (pE4-3200) under the control of the long terminal repeat of the Rous sarcoma virus (RSV L TR) [18] . The E1-E4 fusion protein corresponds to the 17 -kDa species expressed itl vivo [15]. The E4-3200 protein starts at the E1-E4 hybrid alanine [27] and is close to the in vivo 16-kDaE4 species [15] . The E1-E4 and the E4-3200 proteins expressed in VX2-R cells have a size of 17 kDa and of about 16 kDa, respectively, as demonstrated by radioimmunoprecipitation experiments (Fig 3). It has been reported that th e 17-kDa [12,16] and the 16-kDa [12] E4 proteins are phos­phorylated ill villo. In our ill vitro system, the E1-E4 protein was found significantly phosphorylated after a 2-h labeling. In contrast, the E4-3200 protein was only very weakly phosphorylated after a 4-h labeling (Fig 3).

Both proteins expressed in VX2-R cells form multiple rounded eosinophilic granules (Fig IB,C) resembling myrmecia inclusions (Fig 1A). In pEl l\E4 vector-transfected cells, some inclusions coin­cide with th e nucleus (Fig 1B). Inclusions are recognized by the anti-E4 antiserum, as shown by immunoperoxidase experiments (Fig 1E,F).

Formation of Nuclear and Tonofilament-Associated Cyto­plasmic Granules by the EI-E4 Protein Ultrastructural study ofVX2-R cells expressing the E1-E4 protein shows the presence of prominent homogeneously electro n-dense inclusions in the cyto­plasm and in the nucleus (Fig 4A,B) . Cytoplasmic (Fig 4C,E) and nucl ear (Fig 4D) inclusions are recognized by the anti-E4 antiserum, as shown by immunogold labeling. Similar results were obtained when using a milder fixation and an embedding procedure involv­ing an LR White resin (Fig 4F) designed for immulloelectron mi­croscopic detection of antigens [24]. Some cytoplasmic inclusions are associated with tonofilament bundles at their periphery (Fig 4B,E,F), as observed ill villo (Fig 2A,B) , whereas others are free of detectabl e filaments (Fig 4C) like nuclear inclusions (Fig 4D). Free tonofilament bundles are also found in the cytoplasm (Fig 4A,E,F). Tonofilament bundles are not stained by the anti-E4 antiserum, even when associated with the E4 inclusions (Fig 4E,F).

Distinct Localization and Ultrastructure of the E4-3200 Granules Ultrastructural study of VX2-R cells transfected w ith pE4-3200 shows the prese nce of heterogeneously electron-dense inclusions to be exclusively cytoplasmic (Fig SA - C). In less elec­tron-dense inclusions, entangled fil am ents associated with elec­tron-dense deposits are observed (Fig SC). All inclusions are recog­nized by the anti-E4 antiserum (Fig SD,E). Strikingly, whereas the tonofi lament bundles associated with the periphery of the inclu­sions or free in the cytoplasm are not stained by the anti-E4 anti­serum (Fig SD,E), th e intertwined filaments inside the inclusions are labeled by the anti-E4 antiserum (Fig SE). These labeled fila­ments have a diameter of about 10 nm, as expected for keratin IFs [28]. and some of them are in continuity with tonofilament bundles (Fig SE) .

Persistence of a Cytokeratin Network in VX2-R Cells E x ­pressing the HPVl E4 Proteins It has been reported that the expression of the HPV16 E1-E4 protein in epithelial cells resulted

Page 5: Cytopathic Effect in Human Papillomavirus Type 1–Induced

VOL. 101, NO.6 DECEMBER 1993 HPV1 £4 PROTEINS AND MYRMECIA WART INCLUSIONS 847

Figure 4. Ultrastructure (A,B) and in1tnunoe!ectron microscopic identification (C-F) ofEl-E4 inclusions. VX2-R cells were transfected with pEIIIE4 and embedded in an epon-araldirc mixture (A -E) or ill LR-Whirc resin (F). Note the cytoplasmic and nuclear homogeneously electron-dcnse inclusions (A,B) ,irnilar to myrmccia inclusions. Inclusions are labeled with the G705 anti-E4 antiserum, as revealed by IS-nm gold particle-conjugated protein A (C-F). The nuclear (D) and some rounded cytoplas mic (C) inclusions arc not associated with tonofilament bundles. Magnification: bar, 0.5 Jl!l1 .

Page 6: Cytopathic Effect in Human Papillomavirus Type 1–Induced

848 ROGEL-GAILLARD ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Figure 5. Ultrastructure (A-C) and immunoelectron microscopic identification (D,E) of E4-3200 inclusions. VX2-R cells were transfected with the pE4-3200 vector and embedded in a mixture of epon and araldite (A-D) or in Lowicryl resin (E) . Inclusions are labeled with the G70s anti-E4 antiserum, as revealed by is-nm gold particle-conjugated protein A (D,E) . Note the heterogeneous structure of inclusions (A-E), the fibrillar structure of some of them (C,E) and the labeled entangled filaments of about 10 nm in a less dense inclusion (E) . Magnification : bar, 0.5 Jim.

Page 7: Cytopathic Effect in Human Papillomavirus Type 1–Induced

VOL. 101 , NO. 6 DECEMBER 1993

RK5 C

RK6,10 _

RK14-RK15,16_

act-

RK19_

CB

.. ,

RAK

.: , -'

, . ) , , .... .

AEI AE3

Figure 6. Immunoblot analysis ofVX2-R keratins. Keratins extracted from VX2-R cells [1 9] and separated by SOS. polyacrylamide gel electrophoresis were either stained in the gel by Coomassie blue (CB) or processed for jJ1lIIlunoblotting experiments. VX2-R keratins were stained by the anti· Im­man keratin antiserum RAK [21] and identified according to the established nomenclature [31] , by using monoclonal AE1 and AE3 antil:iodies [22]. Immune complexes were detected with peroxidase. labeled anti·rabbit IgG (RAK) or anti·mouse IgG (AEI. AE3). Note that RAK antiserum recognizes all VX2-R keratins except RK 19.

in the rapid collapse of the cytokeratin network into a unique cyto­plasmic inclusion stained both by anti-keratin and anti-E4 antibod­ies [29]. This prompted us to study the effect of HPVl E4 protein expression on the VX2-R cytokeratin network. By using the AEI and AE3 monoclonal antibodies [22] in Western blot experiments, VX2- R cells were found to express the keratin pairs specific of basa l (RKS, RKI4/RKI5) and hyperproliferative (RK6, RKI6) kerati­nocytes, as well as the mesothelial RK1 9 keratin (Fig 6) [30,3 1] . A band with a mobili ty close to that of the RK10 was also observed with the AEI antibody, whereas RKI was not detected with the AE3 a ntibody (Fig 6). The RAK antiserum [21] recognizes all the VX2-R cytokeratins except the RK19 (Fig 6) and stains the keratin network of acetone-fixed VX2-R cells (Fig 7B,D).

Immunofluorescence double-labeling experiments were per­formed using the anti-E4 antiserum and the RAK antiserum. The £1-E4 and E4-3200 inclusions stained by the anti-E4 antiserum (Fig 7A,C) [18] are not stained but only delineated by the RAK anti­serum (Fig 7B,D) . The cytokeratin network remains detectable in £4 inclusion-containing cells (Fig 7B,D), in agreement with elec­cron microscopic observation of preserved tonofilament bundles ill vivo (Fig 2) and ill lIitro (Figs 4,5) .

DISCUSSION

HPV1 E4 proteins are found exclusively in terminally differentiat­ing infected keratinocytes of myrmecia warts and are associated wi th the cytoplasmic and nuclear eosinophilic and electron-dense inclusions pathognomonic of HPVl infection ([7 ,12,14], this paper). The most abundant of the multiple E4 proteins found in lIitlO are an E1-E4 species of 17 kDa and a protein of 16 kDa that lacks [he five E 1 aminoacids and a small but unknown number of N -ter­minal E4 residues [15]. W e expressed in the rabbit VX2-R keratino­eyte line the 17-kDa and a 16-kDa E4 protein starting at the E1-E4 hybrid alanine residue (E4-3200) , close to but not identical to the ;1/ vivo 16-kDa species. Both E4 proteins form inclusions, but with distinct immunofluorescence patterns [1 8] in culture conditions where keratinocytes do not terminally differentiate.

In this paper, we find that the EI-E4 and E4-3200 proteins differ by their phosphorylation state and that the inclusions they form are both eosinophilic but distinct by their localization and ultras truc­[Uca l features (Table I). The deletion of the five E1 amino acids does Dot eliminate any of the phosphorylation sites previously identified [16). It has to be assumed that some conformationa l changes render

HPVl E4 PROTEINS AND MYRMEClA WART INCLUSIONS 849

Figure 7. lm~nunofluo rescence double.labeling experiments. VX2.R cells transfected with pE1 1\E4 (A,B) or pE4-3200 (C,D) were fixed in cold acetone. T he E4 proteins were stained with the G705 anti·E4 antiserum and fluorescein·conjugated secondary antibodies (A,C). Note that E4 inclusions (a rrows) appear ~s annular structures surrounding a bright core for El-E4 (A) and as round b.nght granules for E4-3200 (C), as previously described [1 8]. The cytokeratm network was stained with the RAK antiserum and rhoda­mine·conjugated secondary antibodies (B,D) . Note that the cytokeratin network IS stdl detectable in cells expressing E4 proteins (B,D) and that the RAK .antlserum stains only the periphery of the E4 inclusions (arrows). Magl1lncatJon: bar, 15.5 pm.

these motifs l:ss accessible. The E1 -E4 protein by itself reproduces the cytoplasmIC and nucl ear E4 inclusions found illlli[Io. The nuclear and some of the cytoplasmic EI-E4 inclusions are free of tonofi la­ment bundles, .s~ggesting that this protein has the capacity to self­aggregate. Stnklllgly, the N-termina1 end (MADNKA) of the E1-E4 protem corresponds to a conserved motif (MADXXA), also found at the N-ter~llnal end of the HPV6, 11, and 16 EI-E4 pro­tems [32] . ThIS motif might confer some specific properties to these E4 gene products.

The HPVl EI-E4 protein is cytoplasmic and nuclear, w hereas the E4-3200 protein is exclusively cytoplasmic. Moreover, the putative full. length HPVl E4 protein (E4-3181), with the N -terminal aminoacids (MLCLPL) correspond ing to E4 sequences spliced out in EI-E4 transcripts, is essentially nuclear [1 8]. The distinct N-ter­minal end of these three E4 proteins cou ld be involved in their different localizations, either directly by a translocation mechanism or indirectly by a cytoplasm ic anchoring. The N-terminal ends of the E I-E4 and E4-3181 proteins are not obviolls nuclear-targeting

Page 8: Cytopathic Effect in Human Papillomavirus Type 1–Induced

850 ROGEL-GAILLARD ET A L THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Table I. Properties of EI -E4 and E4-3200 Proteins Expressed in VX2-R Cell s

Proteins

Species Size (kDa) Phosphorylation

E1-£4 17 Yes

E4-3200 16 Very weak

sequences [33,34]. Worth stressing is that the three E4 proteins share the motif RRPSTTPNSQDRGRPRR (nt 3277 to 3328) [35], which corresponds to a bipartite consensus motif, comprising two basic amino acids, a 10-amino acid spacer region, and a five­amino acid basic cluster containing at least three basic residues [34]. This motif has been recently proposed to be involved in nuclear translocation of proteins [34] but it remains to be determined whether this sequence is functional in E4 proteins and whether its activity can be modulated by cytoplasmic anchoring. Different in­teractions of E4 proteins with the keratin IFs could account for the variable extents of cytoplasmic anchoring [33 ,34] .

The association of tonofilament bundles with the periphery of EI-E4 and E4-3200 inclusions, as well as the peripheral immuno­fluorescence staining observed with an anti-keratin antiserum, sug­gest that both proteins are IF-associated proteins. Colocalization of E4 proteins and clumped tonofi lament bundles has been observed recently in keratinized epithelium formed by grafting HPVI6-con­taining keratinocytes [36] . In our study, only entangled filaments of E4-3200 inclusions, which for some of them appear to be in the continuation of neighboring tonofilaments, are stained by anti-E4 antibodies. If present inside the EI-E4 inclusions, such filaments would not have been observed because of the homogeneously dense structure of the inclusions. It has to be kept in mind, however, that the E4-3200 protein is only close to the in vivo I6-kDa species, but

. not identical. The lack of staining of HPVI E4 inclusions by the RAK anti-keratin antiserum suggests that interaction with E4 pro­teins masks keratin epitopes or destroys them by disorganizing the tonofilament bundles. It cannot be excluded, however, that the filaments correspond to keratin species not recognized by the RAK antiserum or that E4 protein-associated fi laments correspond to IFs other than keratins [28].

The expression of HPV1 E4 proteins in VX2-R cells does not result in the total collapse of the cytokeratin network, in contrast to the transient expression of HPV16 EI-E4 protein in different epi­thelial cell lines [29] . In transfected VX2-R cells and in HPVI­infected wart cells expressing E4 proteins, free tonofilament bun­dl es are observed, together with inclusion-associated fil aments. HPVI E4 proteins might interact with specific subset(s) of keratins, resulting in the collapse of the corresponding network(s) . For in­stance, it has been shown that in epidermolytic hyperkeratosis dis­ease, the keratins Kl and KI0 are selectively aggregated in cells still preserving K5 and KI4 keratin filament bundles [37] . The keratins KI and KI0 are not detected in wart cells containing E4 proteins [12] and in differentiating skin keratinocytes of transgenic mice expressing the HPVI early region [38]' whereas other keratin sub­sets, such as hyperproliferative keratins K6 and K16, are found [38] . It remains to be determined whether the lack of detection of kera­tins KI and KI0 is due to some epitope masking in collapsed kera­tins or to a specific down-regulation of their expression.

As for the functions of E4 proteins, it is now admitted that IF­associated proteins have a great potential as regulators of IF supra­molecular organization and functions [39) but the functions of IFs other than structural and mechanical are only beginning to be ex­plored [40,41) . By interfering with such functions and, as they are also found in the nucleus, by interacting with some additional cellu­lar mechanisms, E4 proteins are likely to have a prominent and pleiotropic role in the life cycle of the virus.

Inclusions

Association to Localization Tonofilaments Ultrastructure

Cytoplasm Yes Myrmecia-like Nucleus No Myrmecia-like Cytoplasm Rare Heterogeneous, fibrillar

We thallk O. Croissa llt Jor itltrodtlcitlg tiS to wart tllcrastmclure. We thank also P. Flomolll Jor her help ill pe10rmillg ittlltHlllohislochemical experiwems atld I-G. Bettichotl Jo r Jruitful adlliceJor the lise OJittltlllltloelectroll microscopy tcciltliqlles. We aregratejid to T. -T. SlmJor providillg liS with AE1, AE3, atld RAK atllibodies.

CRG was sl,'pported by the IIIStitll t Natiollal de 10 Recherche AgrollOltJique as all Attache Scietltifique CotJtractuel.

REFERENCES

1. Dc Villiers EM: Heterogeneity of the human papillomavirus group.] Vi",1 63:4898 - 4903, 1989

2. Tawheed AR, Beaudenon S, Favre M, Orth G: Characterization of human papil­lomavlnls type 66 from an InvaSIve ca rcliloma of the uterine cervix. ] elin MicrobioI 29:2656-2660, 1991

3. O rth G: Epidermodysplasia verrucifonnis. In: Sa lzman NP, Howley PM (eds.). The Papovaviridae vol 2; The Papillolllavi",ses. Plenum Press, N ew York, 1987, pp 199 -244

4. Zur Hausen H: Papilloma viruses in anogenital cancer as a model to understa.nd rhe role of viruses in human ca ncers. Ca ll"r Res 49:4677 - 4681, 1989

5. Epstein WL, Fukuyama K: Human wart virus proliferation and its in.f1uence On keratinization in epidermal cells. In: Prunieras M (cd.) . Biomedical Aspects oj Hllmall Wart Vi",s lrifectioll. Fondation Mcrieux,. Lyon, France, 1975, pp 7 -30

6. Orth G, Breltburd F, Favre M, CroISsant 0: Paptllomavimses: possible rol e in human cancer. Cold Sl'rillg Harbor Cell Prolif 4:1043-1068, 1977

7. Croissa nt 0, Breitburd F, Orth G: Specificity of cytopathic effect of cutaneous human papillomavintses. Clill Demlato/3(4):43 -55 , 1985

8. Stoler MH, Whitbeck A, W olinsky SM, Broker TR, Chow LT, Howett MK, Kreider JW: Infectious cycle of human papillomavirus type 11 in human fore­skin xenografts in nude mice.] Viro/64:3310-3318, 1990

9. Lever WF, Schaumburg-Lever G (cds .) : Histopathology oj the Skill , 7th editioll. JB Lippincott Company, Philadelphia, 1990, pp 399 - 425 .

10. Lyell A, Miles JAR: The myrmecia: a study of inclusion bodies in warts. Br M ed] 1:139 -143, 195 1

11. Laurent R, Agache P, Coumc-Marquet J : Ultrastructure of clear cells in human viral warts.] Clltall PathoI 2: 140 - 148, 1975

12. Breitburd F, Croissant 0, Orth G: Expression of human papillomavirus type 1 E4 gene products in warts . In: Steinberg BM, Brandsma JL, Taichman LB (eds.). Callcer Cells, Vol. 5. Papillomavi",ses. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York, 1987, pp 115 - 122

13. Fuchs E: Epidermal differentiation: the bare essentials. ] Cell Bioi 111 :2807-2814,1990

14. Doorbar J, Campbell D, Grand RJA, Gallimore PH: Identification of the human papilloma virus-l a E4 gene products. EMBO] 5:355 - 362, 1986

15. Doorb.r J, Evans HS, Coneron I, Crawford LV, Gallimore PH: Analysis of HPV -1 E4 gene expression using epilope-defined antibodies. EMBO] 7:825-833, 1988

16. Grand RJA, Doorbar J, Smith KJ, Coneron I, Gallimore PH: Phosphorylation of the human papilloma virus type 1 E4 proteins in vivo and in vitro. Virology 170:201 -213, 1989

17. Doorbar J: An emerging function for E4 proteins. Papillomavi",s Rep 2:145 -147, 1991

18. Rogel-Gaillard C, Breitburd F, Orth G: Human papillomavirus type 1 E4 proteins differing by their N-terminal ends have distinct cellular localizations when transiently expressed in vitto. ] ViroI66:816-823, 1992

19. Georges E, Breitburd F, Jibard N, Orth G: Two Shope papillomavirus-associatcd VX2 carcinoma cell lines with different levels of keratinocyte differentiation and transplantability.] ViroI55:246-250, 1985

20. Chen C, Okayama H: High effi ciency transformation of mammalian cells by plasmid DNA. M ol Cell Bioi 7:2745 -2752, 1987

21. Sun TT, Green H: Immunofluorescent staining of keratin fibers in culrured cells. Cell 14:469-476, 1978

22. Sun TT, Eichner R, Schermer A, Cooper D, Nelson WG, Weiss RA: Classifica­tion, expression, and possible mechanisms of evolution of mammalian epithe­lial keratins: a unifying model. In : Levine A, Topp W, Van de W oude G, WatsonJD (cds.). Ca ll cer Cells, Volt, Th e TratlSfo mred Phellotype. Cold Spring Harbor Laboratory , Cold Spring Harbor, New York, 1984, pp 169- 176

23. Whitehouse R, Bcnichou J C, Ryter A: Procedure for longirudinal orientation of rodshaped bacteria and the production of a high cell density of procaryotic aod

Page 9: Cytopathic Effect in Human Papillomavirus Type 1–Induced

VOL. 101, NO.6 DECEMBER 1993

24.

eucaryotic cells in thin seccions for electron microscopy. Bioi Ce1/30: 155 - 158, 1977

Horvat R, Hovorka A, Dekan G, Poczewski H, Kerjascki D: Endothelial cell membranes contain podocalyxin- the m~or sialoprotein of visceral glomeru­lar epithelial cells.] Cell Bioi 102:484-491, 1986

25. Benichou JC, Frehel C, Ryter A: Improved sectioning and ultrastructure ofbacte­ria and animal cells embedded in Lowicryl.] EleelrO,. Mierose Teclr 14:289-297, 1990

26. Bendayan M, Zollinger M: Ultrastructural localization of antigenic sites on osmium-fixed tissues applying the protein A-gold technique.] Hisloelrem Cylo­chem 31:101-109,1983

27.

28·

29.

palermo-Dilts DA, Broker TR, Chow L T: Human papilloma virus type 1 pro­duces redundant as well as polycistronic mRNAs in plantar warts.] Virol 64:3144-3149,1990

Steinert PM, Roop DR: Molecular and cellular biology of intermediate filaments . Annu Rev Biochelll 57:593-625, 1988

Doorbar], Ely S, Sterling], Mclean C, Crawford L: Specific interaction berween J-lPV-16 El-E4 and cytokeratins results in collapse of the epithelial cell inter-mediate filament nerwork. Nalure 352:824-827,1991 .

30. Moll R, Franke WW, Schiller DL: The caralog of human cytokeratins: patterns of expression in normal epithelia, tumors and cultured cells. CeIl31:11-24, 1982

Schermer A, Galvin S, Sun TT: Differentiation-related expression of a major 64K corneal keratin in vivo and in culture suggests limballocation of corneal epithe­lial stem cells.] Cell Bioi 103:49 - 62, 1986

31.

32. Chow LT, N asseri M, Wolinsky SM, Broker TR: Human papillomavirus types 6

HPVl E4 PROTEINS AND MYRMECIA WART INCLUSIONS 851

and II mRNAs from genital condylomata acuminata .] ViroI61:2581-2588, 1987

33. Silver PA: How proteins enter the nucleus. Cell 64:489-497, 1991 34. Dingwall C, Laskey RA: Nuclear targeting sequences-a consensus? Tm,ds Bio­

diem Sci 16:478-481, 1991 35. Danos 0, Katinka M, Y.niv M: Human papillomavirus 1a complete DNA se­

quence: a novel rype of genome organization among papovaviridae. EMBOJ 1:231 - 236,1982

36. Sterling JC, Skepper)N, Stanley MA: Immunoclectron microscopicallocaliza­tton of human paplliomavlrus type 16 L 1 and E4 proteins in cervical keratino­cytes cultured ill vivo. ] /rUlest Dermatol 100: 154 -158, 1993

37. Ishida-Yamamoto A, McGrath ]A, Judge MR, Leigh 1M, Lane EB, Eady RAJ: Selective involv,ement of keratins KI and Kl0 in the cytoskeletal abnormaliry of ep.dermolyttc hyperkeratosis (bullous congenital ichryosiform erythro­derma).] Iflvest Dennolo/99:19 - 26, 1992

38. Tinsley]M, Fisher C, Searle PF: Abnormalities of epidermal differentiation asso­ciated with ~xpression of the human papillomavirus rype 1 early region in transgemc rmce.] Cell ViroI73:1251-1260, 1992

39. Foisner R, Wiehe G: Intermediate filament-associated proteins. CllrrOp Cell Bioi 3:75-81,1991

40. Steinett PM, Liem RKH: Intermediate filament dynamics. Cell 60:521-523, 1990

41. Marugg RA: Transient storage of a nuclear matrix protein along intermediate­type filaments during mitosis: a novel function of cytoplasmic intermediate filaments .] Slnlet Bioi 108:129-139, 1992