Inflammatory Variant of Epidermolysis Bullosa Acquisita With IgG Autoantibodies Against Type VII Collagen and Laminin α3 FREE

EPIDERMOLYSIS bullosa acquisita (EBA) is a subepidermal autoimmune bullous dermatosis with autoantibodies against type VII collagen.¹ Immunodeposition of IgG and blister formation occur at the level of the anchoring fibrils in the epidermal sublamina densa zone.² Clinically, 2 variants of EBA are recognized: (1) the classic type, with trauma-induced blisters; and (2) the inflammatory type, with acute blisters on erythematous bases clinically closely resembling bullous pemphigoid (BP).^{3 - 4} In addition, 2 other clinical variants with antibodies to type VII collagen exist: a subtype associated with systemic lupus erythematosus and a subtype mediated by IgA (linear IgA bullous dermatosis type 2).^{5 - 6}

Patients with anti–epiligrin cicatricial pemphigoid (CP) have extensive mucous membrane involvement and IgG autoantibodies that bind to the dermal side of 1-mol/L sodium chloride–split human skin.⁷ Epiligrin is identical to laminin 5, a heterotrimeric adhesion molecule (α3β3γ2)⁸ that is localized in the lamina densa and lower lamina lucida.⁹ Autoantibodies in patients with anti–epiligrin CP mostly bind to the α3 chain¹⁰ and less often to the γ2¹¹ or β3 chains^{12 - 13} of laminin 5.

We describe a patient with inflammatory EBA in whom we detected simultaneous IgG antibodies to type VII collagen and to the α3 chain of laminin 5.

A 64-year-old woman had a 3-week history of a progressive eruption of painful blisters and erythema on her hands, on her feet, and around her mouth. She complained of dysphagia due to involvement of the tongue and buccal mucosae. The blisters appeared spontaneously, and were not triggered by mechanical trauma. She had no itching, fever, or malaise. Her medical history included hay fever, for which she used cetirizine hydrochloride. The eruption did not respond to a short course of dapsone, 100 mg.

The results of a skin examination showed multiple serous vesicles with annular erythematous bases (Figure 1, B) distributed over her face, shoulders, arms, legs, neck, axillae, latera, vulva, and inguinal region. The vesicles evolved to polycyclic bullae and crusts. Vesicles on the face (Figure 1, A) dominated on the edematous eyelids; the palms and soles were covered with hemorrhagic bullae. There were no target lesions or milia. The Nikolsky sign was negative. The results of an examination of the mucous membranes revealed red conjunctivae; erosions with crusts in the antrum nasi; and vesicles on the palate, buccal mucosa, and tongue and in the vagina.

The results of viral cultures of lesional skin were negative for herpes simplex and varicella zoster. The test result for antinuclear antibody was negative, and the blood sedimentation rate was 80 mm in the first hour.

A biopsy specimen taken from the hand showed a subepidermal vesicle and a dense neutrophilic infiltrate in the superficial dermis. Direct immunofluorescence (IF) microscopy of perilesional skin of the back revealed linear deposition of IgG and complement C3 along the epidermal basement membrane zone (EBMZ).

Indirect IF on monkey esophagus showed anti–EBMZ IgG in the patient's serum at a titer of 160. Further characterization with class-specific monoclonal antibodies revealed the predominance of IgG1 and IgG4 autoantibodies (data not shown). No circulating IgA autoantibodies were detected. The patient's IgG bound to the dermal side of 1-mol/L sodium chloride–split skin (Figure 2, A). The differential diagnosis of disorders with dermal staining on indirect IF includes EBA,³ anti–epiligrin CP,⁷ and anti–105-kd¹⁴ or anti–200-kd¹⁵ pemphigoidlike disorders. A definitive diagnosis was made by identification of the autoantigens by immunoblot.

To further narrow the diagnosis, we immunomapped a vesicle in the perilesional skin biopsy specimen and found laminin 5 (using monoclonal antibody GB3) and type IV collagen (using polyclonal antibody; Silenus Laboratories, Hawthorn, Victoria, Australia) exclusively on the epidermal side (Figure 2, B). The in vivo immunomapping was consistent with a blister at the level of the sublamina densa zone, which is consistent with the diagnosis EBA, and not at the level of the lamina lucida or densa, which would occur in case of anti–epiligrin CP⁷ or anti–105-kd and anti–200-kd pemphigoidlike disorders.

The patient was treated with prednisolone, 80 mg/d. The lesions resolved promptly, and the dosage of prednisolone was decreased after 5 days to 30 mg/d and tapered further. In addition, the patient was treated with intravenous pulses of dexamethasone, 100 mg, on 3 subsequent days per month for 12 months. The lesions healed without scarring, alopecia, or milia formation. After 5 months, a mild relapse occurred on the palms and soles while she was taking prednisolone, 5 mg/d. A quick response was reached by increasing the prednisolone to 20 mg/d. After 12 months, oral prednisolone and intravenous pulse therapy were stopped entirely. No circulating anti–EBMZ autoantibodies were detectable at that time. The patient has remained in complete remission for more than 12 months.

For immunoblot analysis, we used cell extracts of cultured human keratinocytes, dermal extract, and purified human laminin 5, as previously described.^{10 ,16} For immunoprecipitation studies, we used extracts and medium from normal human keratinocytes radiolabeled with sulfur 35–methionine (New England Nuclear, Boston, Mass), as described previously.¹⁰

Indirect IF studies with patient and control sera on skin samples deficient in an EBMZ molecule ("knockout substrates") were performed on laminin 5–deficient skin of a patient with Herlitz junctional epidermolysis bullosa (due to a LAMB3 mutation) (serum dilution, 1:40) and on type VII collagen–deficient skin of a patient with mutilating recessive dystrophic epidermolysis bullosa (serum dilution, 1:5) using a mixture of fluorescein isothiocyanate–conjugated monoclonal anti–human IgG1 and IgG4 subclass antibodies (Zymed, San Francisco, Calif) (dilution, 1:20).¹⁷

Direct IEM for in vivo–bound IgG was performed on a frozen specimen from nonlesional skin of the patient using en bloc preembedding on 7-µm-thick cryosections with immunoperoxidase-labeled rabbit polyclonal anti–human IgG (Dako, Glostrup, Denmark) or with silver-enhanced, 1-nm immunogold label with gold-conjugated goat anti–human IgG (Zymed).¹⁸

Immunoblot analysis on dermal extract showed IgG bound to a 290-kd band (Figure 3, lane a) with the same mobility as the EBA antigen (type VII collagen) recognized by control serum from a patient with EBA (Figure 3, lane b). Surprisingly, using keratinocyte extract, the IgG reacted by immunoblot also with 3 bands (Figure 3, lane c) with similar mobilities as the autoantigens bound by control anti–epiligrin CP serum (Figure 3, lane d), namely, the unprocessed (200 kd), processed (165 kd), and minor (145 kd) forms of the α3 chain of laminin 5. This reactivity to laminin 5 was confirmed by immunoblot using purified human laminin 5 as the substrate (not shown). Remarkably, no IgG reactivity to laminin 5 (Figure 3, lane e) was identified in immunoprecipitation studies on extracts (Figure 3, lane f) and on medium (data not shown) of biosynthetically radiolabeled human keratinocytes that contained immunoreactive laminin 5 and laminin 6. Immunoblot reactivity to laminin α3, on the other hand, was relatively strong. These results suggested that the patient's IgG recognized a cryptic epitope on the molecule.

To confirm by other means the simultaneous presence of autoantibodies against laminin 5 and type VII collagen, we performed indirect IF on skin substrates lacking EBMZ molecules.¹⁷ The patient's IgG bound to the EBMZ of skin specimens lacking either laminin 5 (Figure 4, D) or type VII collagen (Figure 4, A). The binding to these knockout substrates thus revealed the dual presence of autoantibodies reacting with type VII collagen and with laminin 5, respectively. As controls, serum IgG of a patient with anti–epiligrin CP bound to substrate lacking type VII collagen (Figure 4, C) but not to substrate lacking laminin 5 (Figure 4, F). Conversely, serum IgG of a patient with classic EBA bound to substrate lacking laminin 5 (Figure 4, E) but not to substrate lacking type VII collagen (Figure 4, B).

The occurrence of combined IgG directed to 2 molecules in the lower EBMZ lead us to examine the in vivo deposition of IgG at the ultrastructural level. Direct IEM using the immunoperoxidase label revealed in vivo deposition of IgG in the upper and lower surface of the lamina densa and in the sublamina densa zone (data not shown). Direct IEM using the silver-enhanced, 1-nm immunogold label showed particles clustered on electron-dense globules (IgG aggregations) along anchoring fibrils in the sublamina densa zone and occasionally in the lower lamina lucida (Figure 5). The ultrastructural localization of in vivo IgG deposits agrees with the diagnosis of EBA in combination with anti–laminin 5 autoantibodies.

A final diagnosis of EBA was made based on antibodies against type VII collagen and the level of blister formation that was in the sublamina densa zone. On clinical grounds, we initially considered the diagnosis linear IgA bullous dermatosis (linear IgA bullous dermatosis type 1) because of the circinate pattern of vesiculation (Figure 1, B), and CP because of the mucous membrane involvement. But mucosal lesions are not a specific symptom and may also be present in patients with EBA, even if the mucosal lesions are nonscarring.¹⁹ Our patient had the inflammatory variant of EBA, which is often accompanied by mucosal lesions.³ A diagnosis of anti–epiligrin (laminin 5) CP was also considered,²⁰ since circulating antibodies against laminin 5 were demonstrated in our patient. The diagnosis of anti–epiligrin CP was, however, rejected, since the level of blistering was not through the lower lamina lucida⁷ but deeper in the EBMZ, below the lamina densa (Figure 2, B).

All reported cases of anti–epiligrin CP have had immunoreactants deposited on the dermal side of sodium chloride–split skin,²¹ corresponding to the location of laminin 5 in the upper lamina densa.²² The exclusive dermal binding of our patient's IgG to split skin did not allow us to conclude that 2 sets of circulating antibodies to laminin 5 and type VII collagen were present. This could only be demonstrated by immunoblot and by indirect IF using substrates deficient in selected basement membrane molecules. The predominance of IgG4 autoantibodies in our patient agrees with what is found in most patients with anti– epiligrin CP²³ or EBA.²⁴ Furthermore, the positive labeling of the lower lamina lucida, upper lamina densa, or sublamina densa by direct IEM in the nonlesional skin of our patient suggests that autoantibodies against laminin 5 and type VII collagen were deposited in vivo. The small amount of label in the lamina densa after preembedding immunogold labeling may be attributed to limited penetration of the gold particles into the dense tissue of the lamina densa.²²

The identity of the anti–laminin 5 epitope recognized by the IgG of this particular patient was not determined. It is different from the laminin 5 epitopes previously reported since such anti–epiligrin CP sera immunoprecipitate laminin 5.²³ Despite these findings, our patient's IgG reacted strongly with the denatured laminin α3 chain by immunoblot analysis. Most patients with anti–epiligrin CP have autoantibodies that react with the laminin α3 chain that is present in laminin 5 (α3β3γ2) and laminin 6 (α3β1γ1).²⁵ However, the α chains of laminin 5 and laminin 6 are translated from distinct LAMA3 transcripts²⁶ and, therefore, may contain unique epitopes. Since the serum of the presented case neither reacted with laminin 5 nor laminin 6 in immunoprecipitation studies on keratinocyte medium, it remains obscure whether the IgG recognized an epitope on the α3 chain of either or both laminins. In any event, the binding of the patient's IgG by indirect IF to skin substrate lacking type VII collagen and in vivo IgG deposition in the upper lamina densa in the patient's nonlesional skin suggest that the laminin α3 epitope was not hidden in tissue.

The significance of the anti–laminin 5 antibodies and their possible involvement in the pathogenesis of blister formation in this patient remain obscure. The low titer of circulating anti–laminin 5 antibodies in our patient with EBA may represent a secondary reaction to the pathogenic effects of anti–type VII collagen antibodies in which cryptic laminin 5 epitopes have been uncovered in lesional skin. On the other hand, the laminin 5 antibodies may have been the initiating factor followed by an anti–type VII collagen response. This phenomenon of intermolecular epitope spreading to closely located antigens has been recognized in immunobullous disease.²⁷ Laminin 5 and type VII collagen molecules are intimately connected in that the β3 chain of the former binds the NC-1 domain of the latter.^{28 - 29} The complete remission with nonscarring outcome in our patient would be somewhat atypical for anti–epiligrin CP, which normally compromises the skin and mucous membranes.^{11 ,20}

The presence of autoantibodies to more than 1 antigen is not rare in immunobullous skin diseases. Interestingly, in a patient with anti–laminin 5 CP, a second set of IgG antibodies against 180-kd BP antigen was recently found.³⁰ And in patients with EBA, a second set of IgG antibodies has been reported against 180-kd BP antigen,³¹ against 230- and 180-kd BP antigens,³² and against the epidermal side of salt-split skin.³³ Recently, Chan et al³⁴ reported a remarkable case of bullous systemic lupus erythematosus with IgG and IgA autoantibodies against the α3 chain of laminin 5 (and possibly against laminin 6) and against 230-kd BP antigen in addition to those against type VII collagen.

In conclusion, we characterized skin immunodeposits and serum autoreactivity of an adult with acute inflammatory EBA who had circulating IgG against type VII collagen and the laminin chain α3.

Accepted for publication September 14, 1999.

Presented in part at the Joint Society Meeting of the European Academy of Dermatology and Venereology and the Groupe d'Immuno-dermatologie, Société Française de Dermatologie, Nice, France, October 7, 1998.

We thank Guerrino Meneguzzi, PhD, for providing monoclonal antibody GB3, and Siep Noorman for the clinical photography.

Corresponding author: Marcel F. Jonkman, MD, PhD, Department of Dermatology, Groningen University Hospital, Hanzeplein 1, 9713 E2 Groningen, the Netherlands (e-mail: m.f.jonkman@derm.azg.nl).