Pemphigus Autoantibodies Induce Blistering in Human Conjunctiva

Abstract

Purpose: The autoimmune blistering skin disease pemphigus vulgaris (PV) is caused by autoantibodies against desmosomal adhesion molecules. Patients may suffer conjunctival involvement, yet the underlying mechanisms are largely unclear. We characterized human and murine conjunctiva with respect to the PV autoantigens, and evaluated the effects and mechanisms of PV autoantibodies applied to human conjunctiva ex vivo.

Methods: We obtained human conjunctiva specimens from surgical explants and established a short-term culture model to study the alterations induced by antibody fractions of PV patients (PV-IgG). Furthermore, we applied a mouse model depleted of the desmosomal cadherin desmoglein 3 (Dsg3), the primary autoantigen in PV. Murine and human conjunctiva also was used to analyze the expression pattern of desmosomal proteins by immunostaining and Western blotting.

Results: Human and murine conjunctiva samples expressed the majority of desmosomal molecules with an expression pattern similar to the epidermis. Interestingly, Dsg3 knock out animals frequently suffer eye lesions, histologically evident as microblisters in the eyelid epidermis and conjunctiva. Incubation of human specimens with PV-IgG for 12 hours caused blistering in the suprabasal layers of the conjunctiva as well as reduction of Dsg1 and Dsg3 protein levels. Furthermore, PV-IgG prompted activation of p38MAPK in the conjunctiva, which is a central pathomechanism leading to blistering in the epidermis.

Conclusions: PV-IgG leads to blister formation and p38MAPK activation in the conjunctiva and, thus, resembles the effects found in the epidermis. Our data indicate that the ocular involvement observed in PV patients is mainly based on conjunctival blistering.

Pemphigus vulgaris (PV) is a life-threatening autoimmune disease in which autoantibodies directed against the desmosomal adhesion molecules desmoglein (Dsg) 1 and 3 lead to flaccid blisters of the epidermis and the squamous epithelia of mucous membranes.¹ Blistering typically occurs in the suprabasal layers of the epidermis or mucosa, respectively.² Autoantibodies against Dsg3 only induce blistering in mucous membranes; most frequently in the epithelium of the oral cavity, whereas the presence of autoantibodies against Dsg3 and Dsg1 in addition leads to affection of the epidermis.¹ 

In general, binding of autoantibodies results in loss of cell cohesion due to compromised function of specific cell–cell contact structures, the desmosomes. The core of these complexes is composed of desmosomal cadherins, which facilitate homophilic and heterophilic binding to their counterparts on the adjacent cell. In total, 7 isoforms exist: Dsg1 to 4 and desmocollin (Dsc) 1 to 3.² Desmosomal cadherins are linked intracellularly to the intermediate filament cytoskeleton through the adapter proteins plakoglobin, desmoplakin, and several plakophilins.³ This composition is the molecular basis for strong tissue integrity, which is highly impaired in PV.⁴ Autoantibody-induced interference with Dsg interaction and subsequently altered intracellular signaling seem to be required for full loss of cell cohesion and blister formation.⁵ Here, an activation of p38MAPK appears to be central because inhibition of this kinase prevents loss of cell cohesion in cell cultures and murine PV models even under conditions in which Dsg3 interaction is blocked by autoantibody binding.^6–8 

An ocular involvement in PV also was reported to occur and was diagnosed most frequently as conjunctivitis,⁹ which was speculated to correlate with a disease severity¹⁰ or relapse.⁹ At first sight, conjunctivitis associated with PV appears to be surprising because the disease is strictly autoantibody-dependent, without any additional factors from the immune necessary.¹¹ However, it is not clear whether PV autoantibodies lead to blistering within the conjunctiva. Alternatively, superinfections of the skin secondary to split formation are common in PV and may be the primary cause for the conjunctivitis phenotype.¹² In the present study, we tested the hypothesis that, similar to the epidermis, the conjunctiva is affected directly by loss of desmosomal cohesion. We applied a Dsg3 knock out (KO) mouse model, which shows a PV phenotype. In addition, we incubated human conjunctiva compounds from surgical procedures or body donors with IgG fractions of pemphigus patients (PV-IgG). Additionally, we thoroughly characterized the expression pattern of desmosomal cadherins and other constitutive desmosomal proteins in human and murine conjunctiva, as to our knowledge only few data exist from bovine tissue.¹³ 

Materials and Methods

Antibodies and Test Reagents

Primary antibodies (Abs) were used for immunostaining and Western blotting (WB) as described in detail in the Table. 

For WB experiments, horseradish peroxidase (HRP)–coupled goat anti-rabbit Abs, goat anti-mouse Abs, or donkey anti-goat Abs (all from Dianova, Hamburg, Germany) were applied as secondary Abs. Nitrocellulose membranes were developed using the ECL system (GE Healthcare, Solingen, Germany). For immunostaining, Cy3-labeled goat anti-mouse, goat anti-rabbit, or donkey anti-goat Abs served as secondary Abs (all from Dianova). For detection of human IgG, a Cy3-labeled goat anti-human Ab (Dianova) was used. 4′,6-Diamidino-2-phenylendole (DAPI) was added in the last 10 minutes of secondary antibody incubation to stain nuclei and thereby get insights into tissue morphology. 

Dsg3 Mice – Genotyping and Enucleation of Whole Eyes

Genotyping of progeny of heterozygous B6; 129X1-Dsg3^tm1Stan/J mice (Jackson Laboratory, Bar Harbor, ME, USA)¹⁴ was performed as described previously¹⁵ and PCR analyses were done after a well-established protocol.¹⁶ Homozygous Dsg3^+/+ (wildtype [WT]) and Dsg3^−/− (KO) as well as heterozygote Dsg3^+/− mice were evaluated daily with special emphasis on eye abnormalities. When an eye lesion was detected (only occurring in Dsg3^−/− animals), the animal and a WT littermate were euthanized and eyes together with eyelids were enucleated as described previously.¹⁷ Briefly, the eyelid was gently mobilized from the cranium by cutting the skin around the orbits. Afterwards, eyes were enucleated by gently pulling the lid while sharp dissection of the optic nerve was performed.^17,18 Whole eyes were embedded in tissue freezing medium (Leica, Nussloch, Germany) and cut with a cryostat (Vacutome; HM 500OM; Micron GmbH, Munich, Germany). Representative cryosections were used for hematoxylin and eosin (H&E) or immunostaining as described later. All animal experiments were approved by the Regierung von Oberbayern (Az. 55.2-1-54-2532-139-2014) and complied with the ARVO Animal Statement for the Use of Animals in Ophthalmic and Vision Research. 

Human Conjunctiva Acquisition and Processing

Bulbar conjunctiva was removed routinely during eye surgery and was used with informed written consent from the respective patients. This study was conducted according to the tenets of the Declaration of Helsinki. Samples were either directly embedded in tissue freezing medium for immunostaining, processed for WB or subjected to incubation with IgG fractions as follows. 

Samples were dissected under a binocular microscope and intermediately transferred to 96-well plates with prewarmed Dulbecco's modified Eagle's medium (DMEM; Life Technologies, Carlsbad, CA, USA) supplemented with 10% fetal calf serum (Biochrom, Berlin, Germany), 50 units/ml penicillin (AppliChem, Darmstadt, Germany), and 50 μg/ml streptomycin (AppliChem). Fetal calf serum–supplemented DMEM was chosen as culture media due to good evidence for growth of conjunctival cell lines^19,20 and the same medium is a standard medium in human keratinocytes⁸ in combination with human IgG-fractions. Samples were incubated for 12 hours in 5% CO₂ at 37°C under the respective conditions. Conjunctiva samples were freely floating in the medium. Probes were viable after incubation as revealed by MTT assay (Supplementary Fig. S1A; Supplementary Materials and Methods). Afterwards, samples were washed and embedded in tissue freezing medium. Accordingly, samples were cryosectioned and used for H&E or immunostaining as described below. Some samples were further lysed for WB. 

Purification of PV and Control IgG Fractions

Serum of PV patient was donated by Enno Schmidt (Department of Dermatology, University of Lübeck). Sera were used with patients' informed and written consent and with institutional approval of the ethics committee (# 07-180). Pemphigus vulgaris was diagnosed by clinical and histologic phenotype and by ELISA against Dsg1 (activity, 9263U/ml) and Dsg3 (activity, 18383U/ml). IgG-fractions of PV patients (PV-IgG) or healthy donors (control-IgG) were purified as described previously.^21–23 

H&E and Immunostaining

Cryosections were stained with H&E after standard procedures. For immunostaining, sections were fixed using 2% formalin solution (freshly prepared from paraformaldehyde) and subjected to the respective primary and secondary antibodies (Table). 

Table.

 

Overview Primary Antibodies

 

Western Blotting

Western blotting was performed using standard procedures.²³ Briefly, tissue was minced and subjected to SDS-based lysis buffer. After clearing by centrifugation to remove tissue debris, lysates were collected, protein concentration was normalized, and subjected to gel electrophoresis and Western blotting. 

Image Acquisition and Processing

Hematoxylin and eosin images were taken using a DMC2900 camera mounted on DMi8 inverted microscope (both Leica Microsystems, Wetzlar, Germany). Further, images of immuno-stained probes were acquired using a Leica SP5 confocal microscope with the 20 × HCX PL APO 0.7 IMM objective. Images were processed further using Photoshop CS5 (Adobe Systems, San Jose, CA, USA). 

Results

Dsg3 KO Mice Develop Conjunctival Disruption

Mice with genetic Dsg3 ablation have a PV phenotype with lesions in the oral and esophageal mucous membranes with typical suprabasal blistering.^14,24 Further, some of the mice have an eye involvement,¹⁴ which to our knowledge has not been characterized before. In our study, approximately half (43%) of Dsg3 KO mice showed eye symptoms during the observation period, such as reddened conjunctiva, small bleedings at the inner corner of the eye, and clotted eyelids (Fig. 1A, right; average age of eye symptom occurrence, 12.87 ± 1.5 weeks), whereas their WT littermates did not have any eye involvements during the observation period (Fig. 1A, left). Genotypes of the mice were checked by genotyping (data not shown) and WB (Fig. 1B). To investigate the reason for the pathologically altered eyes in Dsg3 KO mice, whole eyes including the globe and eyelids were harvested and H&E staining was performed. The eyes from Dsg3 KO mice showed no massively altered histology compared to those from Dsg3 WT mice (Fig. 1C). Especially, we did not find signs of infections. We further analyzed cornea, and bulbar, fornix, and tarsal conjunctivae in more detail as these areas express Dsg3 in WT mice (Supplementary Fig. S2). Cornea and conjunctiva showed normal morphology in Dsg3 WT mice (Fig. 1D, left). In contrast, morphology was changed in all three areas in Dsg3 KO mice. The stratified epithelium of the cornea was separated in the superficial layers (Figs. 1D, 1Db, arrows). Further, conjunctiva, which consists of 2 to 4 layers at the bulbar and 4 to 7 layers at the tarsal side,¹⁷ was regular and unperturbed in WT animals but appeared severely altered in KO animals (Figs. 1D, 1Dd, 1Df, arrows). Specifically, the layer structure appeared to be disrupted with the superficial layers often being detached or absent. 

Figure 1

 

Dsg3 KO mice display eyes lesions with a phenotype similar to PV. (A) In contrast to their WT littermates (left), Dsg3 KO mice have macroscopically visible eye involvement with bulbar redness, clotted eyelids, and small bleedings at the inner eye angle (right, arrows). Figures are representative of five different animal pairs. (B) Western blotting experiment confirming animal Dsg3 WT or KO genotype; α-tubulin was used as loading control. (C) H&E staining of cryo-sections from whole mice eyes including the cornea and whole conjunctiva (bulbar and tarsal part). Areas for magnifications were depicted and are shown in (D). (D) H&E of cornea (a, b), bulbar conjunctiva with only 2 to 4 cell layers (c, d) and fornix conjunctiva with tarsal conjunctiva with 4 to 7 layers (e, f) show intact epithelia in WT animals, while in Dsg3 KO animals gaps and blistering occurs in all areas (arrows). Figures are representative of five different animal pairs.

These data demonstrated that loss of Dsg3 results in conjunctival alterations histologically reminiscent of lesions in PV. Inflammatory reactions were not detectable; thus, it is assumable that this phenotype is a direct effect of the loss of Dsg3 and not a secondary effect of inflammation. 

Characterization of the Expression Pattern of Desmosomal Molecules in Human and Murine Conjunctiva

In pemphigus patients an eye involvement, most commonly diagnosed as conjunctivitis, is described.^9,25 However, the underlying mechanisms are not yet clear. We next clarified the expression pattern of desmosomal molecules in human and murine conjunctiva. Human specimens were acquired from routine surgical procedures. A typical surgical procedure included the implantation of epibulbar drainage devices in glaucoma patients. The bulbar conjunctiva was opened, the device was implanted, and excessive conjunctiva was removed when closing the wound. Specimens were processed for immunostaining (Fig. 2A) or WB experiments (Fig. 2B). Expression of Dsg1-3 and Dsc2, and Dsc3 was clearly detectable in human conjunctiva (Figs. 2A, 2B). All isoforms depicted a membrane-localized staining (Fig. 2A). Similar to the epidermis,² Dsg3 and Dsc3 expression was higher in the basal layers, whereas Dsg1 was enhanced in the superficial layers of the conjunctiva (Figs. 2A, 2Aa, 2Ac, 2Af). Expression of Dsg2 was not as prominent. Nevertheless, Dsg2 expression was visible which is different from human epidermis, where Dsg2 could be found in hair follicles only.¹⁵ Specific Dsc1 staining was absent. The desmosomal plaque proteins plakoglobin (PG) and desmoplakin (DP) were uniformly expressed throughout all layers (Figs. 2A, 2Ag, 2Ah). Similarly, the adherens junction protein E-cadherin (E-Cad) generally was localized to the membrane of all layers (Figs. 2A, 2Ai). Unspecific staining was excluded by secondary antibody controls (Supplementary Fig. S3A). The expression of these molecules also was confirmed by WB of whole conjunctiva samples (Fig. 2B). The PV autoantigens Dsg1 and Dsg3 as well as PG and DP also were present in murine conjunctiva (Supplementary Fig. S2). 

Figure 2

 

Characterization of human conjunctiva with respect to desmosomal proteins. (A) Western blotting of human conjunctiva showing the expression pattern of the desmosomal cadherins (Dsg1-3 and Dsc1-3) as well as the desmosomal plaque proteins PG and DP and the adherens junction protein E-Cad. Dsg1 and 3 as well as Dsc 2 and 3 are highly expressed, whereas expression of Dsg2 is lower and nearly absent in case of Dsc1. Both plaque proteins as well as E-Cad also were clearly expressed. (B) Western blotting of human conjunctiva specimens confirm the results from immunostaining (n = 4).

Pemphigus Autoantibodies Induce Blistering in Human Conjunctiva

We next took human conjunctiva samples into culture to directly study the effects of pemphigus autoantibodies. Each specimen was separated into two pieces and incubated for 12 hours with control-IgG (from healthy volunteers) or PV-IgG, respectively. The effective dose of PV-IgG was determined in previous studies and again reevaluated by dispase-based keratinocyte dissociation assays in HaCaT keratinocytes (Supplementary Material and Methods). This particular PV-IgG was set to a 1:50 dilution (Supplementary Fig. S1B) and compared to an equal dose of control-IgG. Under control conditions, bulbar conjunctiva showed normal morphology in H&E staining (Fig. 3A, left). In contrast, specimens incubated with PV-IgG displayed suprabasal blistering typical for PV in epidermis and mucous membranes (Fig. 3A, right).^1,26,27 To prove that PV-IgG is capable of binding to the conjunctiva, we performed immunostaining using a fluorescently-labeled antibody against the Fc-domain of human IgG (gah-cy3). In control-IgG incubations, no staining of the conjunctiva was observed, demonstrating the absence of unspecific IgG binding in our model system (Fig. 3B, upper row). However, PV-IgG incubation revealed membrane-coupled staining surrounding the blisters (Fig. 3B, lower row). Taken together, these data show that PV-IgG bind to human conjunctiva and are capable of inducing blistering. 

Figure 3

 

PV-IgG induced blistering in the human conjunctiva model. (A) H&E staining of human conjunctiva specimens. Controls show an intact stratified epithelium, whereas pronounced blistering occurs after 12 hours of PV-IgG incubation. (B) Staining of human IgG-antibodies shows no binding of the control-IgG to the human conjunctiva specimens. In contrast, PV-IgG binding was identified by a membrane-localized deposition. DAPI was used as overview for tissue morphology. White rectangles identify magnified areas. *Blister areas (n = 4).

PV-IgG Induces p38MAPK Signaling and Depletion of Dsg3 and Dsg1 in Human Conjunctiva

Activation of p38MAPK and its target MK2 is central in pemphigus pathogenesis^6,7 and is detectable in patient skin.^28,29 Thus, we checked the phosphorylation of p38MAPK and MK2,³⁰ which correlate with activation state, in control and PV-IgG–treated samples (Figs. 4A, 4B). Under control conditions, staining for the phosphorylated form of both molecules was virtually absent (Figs. 4A, 4B; upper row), whereas a clear increase was seen after incubation of PV-IgG (Figs. 4A, 4B; lower row). Activation was found predominantly at cell borders (Figs. 4A, 4B; arrows). We previously observed p38MAPK activation in the skin of Dsg3 KO animals²⁴ and, thus, investigated the activation of p38MAPK and MK2 in eyes of Dsg3 WT and KO mice. Activation of p38MAPK and less pronounced of MK2 was observed in Dsg3 KO animals, whereas no change of the total protein expression was detectable (Fig. 4C). Interestingly, no difference in activation of p38MAPK and MK2 was visible between lesional and nonlesional areas of the mice cornea and conjunctiva (Fig. 4C), indicating that the p38MAPK activation observed here is not a secondary result in response to lesion formation. Cross-reactivity of the secondary Ab was excluded by secondary antibody controls (Supplementary Figs. 3B–D). 

Figure 4

 

PV-IgG as well as loss of Dsg3 induce conjunctival activation of p38MAPK. (A) p-p38MAPK (p-p38) staining of human conjunctiva demonstrates nearly no activation under control conditions, while a clear increase was detectable following PV-IgG incubation. (B) Similar to p-p38MAPK, p-MK2 staining shows a membrane-localized staining only under PV-IgG conditions, but not in controls. (A, B) White lines depict basement membrane localization. *Blistering areas (n > 3). (C) p-p38 and p-MK2 staining of mice eyes demonstrates an activation of p-p38 and less pronounced of pMK2 in both lesional (arrows) and nonlesional areas of Dsg3 KO animals compared to WT littermates (n > 3). Expression pattern of total protein for p38 and MK2 were similar in Dsg3 WT and KO animals.

In the epidermis, depletion of the autoantibody antigens Dsg1 and Dsg3 are hallmarks of PV and a result of p38MAPK signaling.^31–33 Therefore, we examined the expression pattern of Dsg1 and 3 in controls and PV-IgG–treated samples. Compared to controls, PV-IgG–incubated specimens showed reduced Dsg1 signal with a predominant reduction of the membrane-staining (Fig. 5A). This phenomenon was particularly evident in cells surrounding the blisters (Fig. 5A, lower panel, blister marked by asterisk). Interestingly, Dsg3 also appeared depleted; however, mainly at the blister floor and the basal layers of the conjunctiva (Fig. 5B, arrows), which is similar to human epidermis.³² To confirm these results, we performed WB-experiments under the same conditions (Fig. 5C). In this approach, depletion was clearly detectable for Dsg1, but only slightly for Dsg3 (Fig. 5C). 

Figure 5

 

PV-IgG induce Dsg1 and Dsg3 depletion in human conjunctiva. (A) Compared to controls, PV-IgG-treated conjunctiva displays a reduction of Dsg1 staining. (B) In contrast to membrane-localized Dsg3 staining with a high expression in basal layers in controls, PV-IgG–treated specimens show loss of staining intensity particular pronounced in basal layers (white arrows). (A, B) White lines depict basement membrane localization, white rectangles areas of magnification. *Blistering areas (n > 3). (C) Western blotting of human conjunctiva specimens show a clear reduction of Dsg1 levels after PV-IgG treatment, while only a slight decrease was detectable for Dsg3; α-tubulin was used as loading control (n > 3).

Taken together, these data indicated that the mechanisms of blister formation in conjunctiva resembled those found in the epidermis. 

Discussion

Our present study demonstrated that loss of Dsg3 function, either by genetic ablation or by binding of PV autoantibodies, leads to conjunctival blistering. Loss of conjunctival integrity was not dependent on inflammatory reactions and, thus, may be the reason for the eye involvement observed in PV patients. 

Eye Involvement in PV May Be Based on Conjunctival Blistering

An eye involvement in PV patients is repeatedly described in the literature.^9,25,34–36 Case reports range from severe phenotypes with cicatrization, corneal ulceration, blepharitis and corneal perforation,³⁷ bilateral eyelid involvement and lid margin erosions,³⁸ to only mild conjunctivitis.³⁵ Overall, conjunctivitis was the most frequently established diagnosis and, therefore, was the basis for the present study.⁹ Indeed, the majority of desmosomal molecules are found in human and murine conjunctiva, showing an expression pattern similar to the epidermis. Thus, all primary targets of PV autoantibodies are present in conjunctival tissue and affection is likely. Indeed, we showed for the first time to our knowledge that PV-IgG induces conjunctival blistering. This is in agreement with some studies presenting single biopsy results with conjunctival blistering and membrane-bound PV-IgG similar to our model.^25,34 Typically for PV, blistering in conjunctiva occurs in the suprabasal layers, which fits well to the findings of case reports^25,34 and in the epidermis.¹ Thus, conjunctival blistering apparently is the primary cause for the eye involvement observed in pemphigus patients, which may be partly aggravated by superinfection. Most case reports state that an immunosuppressive treatment improved the clinical phenotype, which may indicate that the symptoms of “conjunctivitis” are directly associated with autoantibody-mediated blistering and not with pronounced infection.³⁹ This is further supported by our ex vivo model, in which no immune reaction is present, and by the mouse model, in which loss of Dsg3 alone is sufficient to impair conjunctiva integrity. Nevertheless, other parameters also may contribute to the eye involvement. Patients with ocular pemphigus often suffer from dry eye symptoms,⁴⁰ which may be attributed to Meibomian gland dysfunction. Interestingly, Meibomian glands also bear desmosomes (unpublished data) and, thus, also may be affected by pemphigus autoantibodies. Ocular involvement is not a very common feature in PV, which may be a result of missing or reduced mechanical stress the tissue is exposed to, at least compared to the epidermis or oral mucous membranes.²⁵ Nevertheless, in our model blistering was evident without mechanical manipulation of the tissue. Thus, it is possible that mild eye involvement is underdiagnosed in patients, which also is supported by some authors.^25,34 

Mechanisms Underlying Blister Formation are Similar in Conjunctiva and Epidermis

In our conjunctiva model, a depletion of Dsg1 and 3 was observed which is a well known phenomenon in epidermal PV and correlates with loss of intercellular adhesion, as blocking of the depletion recovers cell adhesion in several models.^{21,31–33,41–43} Fitting to observations in the epidermis, Dsg3 depletion predominantly occurs in the basal layers of the conjunctiva.³² This may contribute, together with the different expression patterns of Dsg1 and Dsg3, to the suprabasal localization of the blistering.^16,44 Depletion of Dsg3 was only slightly detectable by WB-experiments, which may be explained by the observation that depletion (as indicated by immunofluorescence) was found primarily in the basal conjunctiva and not throughout all layers. Nevertheless, Dsg1 depletion was readily evident by immunoblotting, indicating a participation of Dsg1 autoantibodies in conjunctival blistering. However, eyelid involvement, but not conjunctivitis was reported up to now in pemphigus foliaceus, another form of pemphigus in which autoantibodies against Dsg1 only occur.⁴⁵ Taken together, the data from Dsg3 KO animals further showed that loss of Dsg3 alone is sufficient to cause conjunctival blistering similar to autoantibody fractions targeting Dsg1 and 3. 

A central pathomechanism in PV is the activation of p38MAPK.^6,7,28 In our human conjunctiva model, p38MAPK as well as its known downstream target in PV, MK2,^8,30 were activated after PV-IgG treatment. Interestingly, this activation still was present after 12 hours of incubation in our model and is known to be rapidly activated after PV-IgG incubation.^21,46 Activation of p38MAPK was observed mainly at cell borders fitting to the observation that Dsg3-attached p38MAPK activity is increased after incubation with pathogenic Dsg3-antibodies.²¹ Furthermore, it is known that activation of p38MAPK and desmoglein depletion are linked processes in response to PV-IgG,³³ which seems also to be true in conjunctiva. These data also are supported by the observation of p38MAPK and MK2 activation in Dsg3 KO animals. Taken together, these data suggested that in the conjunctiva the same mechanisms are important for blister formation as in the epidermis. 

Acknowledgments

The authors thank Sabine Mühlsimer, Renate Scheler, Andrea Wehmeyer, and Cathleen Plietz for excellent technical assistance. 

Supported by Deutsche Forschungsgemeinschaft DFG SP1300/1-3. 

Disclosure: F. Vielmuth, None; V. Rötzer, None; E. Hartlieb, None; C. Hirneiß, None; J. Waschke, None; V. Spindler, None