RAGE and HMGB1 Expression in Orbital Tissue Microenvironment in Graves' Ophthalmopathy

Graves' ophthalmopathy (GO) is a chronic autoimmune inflammatory disorder involving orbital tissues. A receptor for advanced glycation end products (RAGE) and its ligand high mobility group box 1 (HMGB1) protein trigger inflammation and cell proliferation and are involved in the pathogenesis of various chronic inflammatory diseases. This study was aimed to evaluate RAGE and HMGB1 expression in GO to determine its potential clinical significance. To the best of our knowledge, this is the first study showing RAGE and HMGB1 expression in orbital tissue using immunohistochemistry. Sections of orbital adipose tissue obtained from patients diagnosed with GO (23 patients; 36 orbits) and normal controls (NC) (15 patients; 15 orbits) were analyzed by immunohistochemistry for RAGE and HMGB1 expression. Expression profiles were then correlated with clinical data of the study group. RAGE and HMGB1 expression were elevated in GO patients in comparison with NC (p = 0.001 and p = 0.02, respectively). We observed a correlation between RAGE expression and occurrence of dysthyroid optic neuropathy (DON) (p = 0.05) and levels of TSH Receptor Antibodies (TRAb) (p = 0.01). Overexpression of RAGE and HMGB1 might be associated with GO pathogenesis. In addition, RAGE and HMGB1 proteins may be considered as promising therapeutic targets, but this requires further research.


Introduction
Graves' disease (GD) represents an autoimmune process in which circulating autoantibodies directed against thyrotropin receptor (TSHR)-TRAb (TSHR antibodies)-activate the thyroid gland, causing hyperthyroidism [1]. One of the extrathyroidal symptoms of GD is Graves' ophthalmopathy (GO), defined as a chronic autoimmune inflammatory disor-der involving orbital tissues [2]. Although patients with GO are mostly hyperthyroid, they can also be euthyroid or hypothyroid. Moreover, GO may be reported in Hashimoto's thyroiditis [3]. Cytokine production, inflammatory infiltration, and orbital fibroblast activity result in expansion and remodeling of extraocular tissues-mainly orbital adipose tissue and fibrous tissue of extraocular muscles. Edematousinfiltrative changes involving orbital soft tissues are observed in 25-50% of patients with GD [4]. Clinical manifestations of GO include lid retraction, double vision, soft tissue swelling, and erythema of the conjunctival and periorbital tissues. Increased intraocular pressure within the inflexible bony orbital walls can contribute to the development of proptosis and optic nerve compression, including dysthyroid optic neuropathy (DON). According to the European Group on Graves' Orbitopathy (EUGOGO), severity of GO is rated as mild, moderate-to-severe, and sight threatening (including DON and/or severe keratitis) [5].
Multiligand receptor for advanced glycation end products (RAGE) is suggested to initiate and amplify immune and inflammatory responses [6]. Increased levels of RAGE ligands in chronic disorders indicate that RAGE is involved in the pathogenesis of various inflammatory diseases [7]. Cellular stress causes the generation of RAGE ligands such as high mobility group box 1 (HMGB1) protein, S100 proteins, and nucleic acids, while prolonged hyperglycemia and inflammation induce the release of the ligands AGE and amyloid [8]. HMGB1 is one of the most significant members of the DAMP (damage-associated molecular patterns) family. DAMPs involve molecules released by dying or necrotic cells and can induce inflammation, cell proliferation, and migration [9]. HMGB1-RAGE interaction affects inflammation via the activation of proinflammatory transcription factor NF-κB (nuclear factor kappa B) which furthermore regulates RAGE [10].
In this study, we test the hypothesis that RAGE and HMGB1 are overexpressed in orbit tissue in GO and that expression patterns correlate with disease severity. To the best of our knowledge, this is the first study showing RAGE and HMGB1 expression in orbital tissue by immunohistochemistry.

Patients and Tissue
Collection. Archival tissue paraffin blocks were used for the studies. All orbital adipose tissue samples used in the study were collected during surgery for routine histopathological examination between 2016 and 2020. Tissues were obtained from 23 patients (18 females, 5 males, 36 orbits) diagnosed with GO who underwent transnasal endoscopic orbital decompression surgery in the Otolaryngology Department, Centre of Postgraduate Medical Education in Warsaw, Poland, or were derived from a tissue bank at the Department of Internal Medicine and Endocrinology, Medical University of Warsaw, Poland, from previous studies. Inclusion criteria for the study were in accordance with the EUGOGO guidelines: (1) moderate-tosevere GO or (2) sight-threatening GO. Patients with any of the following conditions were excluded from the study: (1) age under 18 years, (2) a history of eye surgery, (3) a previous history of metabolic diseases or other diseases that may affect orbital connective tissues, except for thyroid disease, and (4) pregnancy or lactation. As controls, orbital adipose tissue samples were obtained from 15 patients (15 orbits) operated on the orbit due to trauma at the Department of Maxillofacial Surgery, Medical University of Warsaw, Poland. The study was approved by Local Ethics . Tissues were fixed in 10% formaldehyde and embedded in paraffin. Sections were stained with hematoxylineosin (H+E) and evaluated by light microscopy before performing immunohistochemistry.

Clinical Characteristics of Graves' Ophthalmopathy.
Ocular involvement, including GO severity and activity, was defined as absent, mild, moderate-to-severe, and sightthreatening-active or inactive, according to criteria reported in EUGOGO recommendations [5,11]. Patients were classified as "active" or "inactive" based on a Clinical Activity Score (CAS). Dysthyroid optic neuropathy (DON) was defined by visual dysfunction secondary to GO when other causes for visual impairment had been excluded. The diagnosis of DON was based on at least two criteria from the following: (1) reduced visual acuity (<1.0), (2) relative afferent pupillary defect, (3) reduced color vision (more than two errors in Ishihara plates), (4) optic disc swelling in the affected eye, and (5) magnetic resonance imaging of orbit showing apical crowding or optic nerve stretching. Corneal involvement (keratitis) was defined as absent or punctuate keratopathy/ulcer. Best-corrected visual acuity (BCVA) was examined using Snellen charts and expressed as a decimal fraction. Exophthalmos was measured by a Hertel exophthalmometer. The intraocular pressure was measured in the primary position using an applanation tonometer. All the ophthalmology examinations for all patients were carried out by the same ophthalmologist. Clinicopathological characteristics of Graves' ophthalmopathy (GO) patients are presented in Table 1.

Immunohistochemistry.
Paraffin sections of GO patients and NC patients were immunostained using NovoLink Polymer Detection Systems (Novocastra Laboratories, Newcastle, UK) and the following primary antibodies diluted 1 : 100 in Antibody Diluent (Dako): rabbit polyclonal anti-human HMGB1 (LS-C2691, LifeSpan BioSciences, Inc., Seattle, WA, USA) and mouse monoclonal anti-human RAGE (LS-B6042, LifeSpan BioSciences, Inc., Seattle, WA, USA). Deparaffinated and rehydrated sections were stained according to the manufacturer's instructions, as previously described [12]. The activity of endogenous peroxidase was blocked by Peroxidase Block (NovoLink Polymer Detection System; Novocastra Laboratories). Sections were incubated with primary antibodies overnight and were subsequently incubated with Post Primary (rabbit anti-mouse IgG; NovoLink Polymer Detection System; Novocastra Laboratories) which detected mouse antibodies and with Polymer (NovoLink Polymer Detection System; Novocastra Laboratories) to recognize rabbit antibodies and detect Post Primary and finally with 3,3′-diaminobenzidine chromogen (NovoLink Polymer Detection System; Novocastra Laboratories). Nonspecific binding of the primary antibodies and the polymer was eliminated by application of Protein Block (NovoLink Polymer Detection System; Novocastra Laboratories) before adding the primary antibodies. Sections were counterstained with

RAGE and HMGB1 Expression in Tissue
Sections. RAGE expression levels were elevated in GO tissues compared to those from NC (p = 0:001; Figure 1(a)). In GO tissues, RAGE was detected in the cytoplasm, and the expression was positive or heterogenous with staining intensities ranging from weak to strong. In contrast, RAGE expression was observed in only 25% of NC tissues and its staining intensity ranged from negative to weak (Figure 1(a)). HMGB1 was detectable in the nuclei and cytoplasm in all tissues of GO patients and NC (Figure 1(b)). Intensity of HMGB1 staining was evaluated as moderate or strong (Figures 1(a) and 1(b)). Differences between GO and NC in terms of HMGB1 expression were statistically significant (p = 0:02; Figure 1(a)).

RAGE Expression
Correlates with Disease Severity. In the GO cohort, we observed differences in RAGE positivity depending on occurrence of DON and TRAb levels ( Figure 2). GO patients with DON had stronger expression of RAGE (positivity "++") than tissues from patients without DON (p = 0:05) (Figure 2(a)). Moreover, RAGE staining correlated with TRAb levels in GO patients. Positivity "++" was characteristic for high levels of TRAb (p = 0:02; Figure 2(b)).
The correlation of HMGB1 expression levels in the GO cohort with clinical data revealed no statistically significant differences (p > 0:05). However, we observed the tendency between increased levels of HMGB1 expression in patients with sight-threatening GO with DON (data not shown).

RAGE and HMGB1 Expression in Inflammatory
Infiltrates. In GO tissues, we observed elevated numbers of inflammatory cells that expressed RAGE and HMGB1 proteins. Those polymorphonuclear and mononuclear cells were mainly localized in close proximity to the vessels (Figure 3).

Discussion
In our study, we demonstrated significant differences between GO and NC tissues for RAGE and HMGB1 expression. Enhanced expression of RAGE occurs in conditions of inflammatory mediators and ligands for RAGE accumulation [8]. Literature data indicate that the inflammatory process comprising orbital tissues underlies the pathogenesis of GO [14,15]. Immune cell infiltration as well as cytokine production activates orbital fibroblasts and GAG synthesis resulting in adipose tissue expansion. Yoon et al. have demonstrated that mRNA and protein levels of the proinflammatory cytokines TNF-α and IL-1β are increased in GO tissues compared to healthy controls [16]. Furthermore, our preliminary study has shown that a variety of proteins and cytokines involved in inflammation, autoimmunity, hypoxia, and fibrosis are expressed in GO orbital adipose tissue microenvironment, including TGF-β, TLR-4 (Toll-like receptor 4), HIF-1α (hypoxia-inducible factor-1α), NF-κB (nuclear factor kappa B), and IL-17 [14].
Current evidences point out that HMGB1 protein is involved in numerous chronic inflammatory and autoimmune diseases including rheumatoid arthritis, atherosclerosis, and systemic lupus erythematosus (SLE) [17,18]. Released HMGB1 interacts with its cell-surface receptor-RAGE, activating the main signaling pathways responsible for the pathogenesis of these diseases. HMGB1 is involved in the activation of innate and adaptive immunity, development of inflammation, and increased production of cytokines. In patients with rheumatoid arthritis, the production of HMGB1 and the number of cells secreting HMGB1 at specific inflammation areas are elevated [19,20]. Experimental models of arthritis showed inhibitory effects of anti-HMGB1 antibodies on the development of synovial inflammation. HMGB1 is also found in SLE patients' plasma, and after being released from apoptotic cells, HMGB1 is bound to nucleosomes [21]. In turn, autoantibodies directed against nucleosomes and double-stranded DNA are characteristics for SLE. Our previous study demonstrated the expression of RAGE and HMGB1 in epithelial cells of sinonasal mucosa samples obtained from patients with chronic rhinosinusitis or in epithelial cells of middle ear cholesteatoma [12,22]. We observed a strong correlation between the disease severity and RAGE expression. Studies have shown that HMGB1 is involved in the pathogenesis of autoimmune thyroiditis and other chronic diseases by augmenting inflammation signaling and inflammatory infiltration [22][23][24]. Han et al. demonstrated that HMGB1 and its receptors are associated with the inflammatory mechanisms of GO and blocking of the HMGB1 pathway can be utilized to treat GO patients [25]. The authors demonstrated higher gene expression levels of RAGE and TLRs and higher mRNA and protein levels of HMGB1 in GO tissues compared to non-GO tissues. Moreover, blocking of HMGB1, RAGE, and TLR caused diminished production of proinflammatory cytokines confirming the participation of HMGB1 in GO inflammation. Plasma levels of HMGB1 were shown to correlate with the clinical activity score (CAS), indicating a valuable biomarker of disease activity. Being coherent with previous studies, we confirm the expression of RAGE and HMGB1 in GO tissues using immunohistochemistry. We observed a strong correlation between DON and RAGE expression in GO tissues. In addition, in GO tissues we demonstrated the presence of RAGE and HMGB1 positive inflammatory cells which were closely located to  Mediators of Inflammation the vessels. We hypothesize that those cells may also play an important role in driving the inflammatory process. According to our knowledge, this is the first study that has shown the presence of RAGE and HMGB1 positive inflammatory cells in the GO adipose tissue microenvironment. Peng et al. demonstrated that HMGB1 and RAGE expression on monocytes isolated from PBMCs was higher in patients with autoimmune thyroid diseases (AITD) compared to those isolated from healthy controls [26]. Their findings suggested a vital role of RAGE and HMGB1 in the pathogen-esis of AITD. RAGE participates in inflammatory cell recruitment and chronic inflammation development, which are considered crucial aspects of GO pathogenesis [25,27].
Pathogenesis of GO is closely associated with the presence of TRAb which involves autoantibodies directed against TSHR expressed not only on thyrocytes but also on orbital tissues. TRAb levels correlate with the clinical activity of GO [28]. Diana et al. reported a correlation between TRAb and the degree of ocular changes in GD [29]. However, the relationship between TRAb level and HMGB1 expression

Mediators of Inflammation
was not observed by us. In turn, Han et al. demonstrated a strong correlation between plasma level of HMGB1 and TRAb as well as CAS. Literature data report a correlation between RAGE levels in the serum and disease activity of other autoimmune diseases [30]. Studies analyzing the involvement of RAGE and HMGB1 in the pathogenesis of various inflammatory diseases are accompanied by experimental studies showing anti-RAGE/HMGB1 antibodies that prevent chronic inflammation [31,32].
In summary, we demonstrated for the first time by using immunohistochemistry that RAGE and HMGB1 are overexpressed in GO patients in comparison with NC. We also observed a correlation between RAGE expression and occurrence of dysthyroid optic neuropathy (DON) or elevated levels of TSH Receptor Antibodies (TRAb) in GO patients. Overexpression of RAGE and HMGB1 might be associated with GO pathogenesis.

Data Availability
The research and clinical data used to support the findings of this study are included within the article.

Conflicts of Interest
The authors declare that they have no conflict of interest.