Acitretin mitigates uroporphyrin-induced bone defects in congenital erythropoietic porphyria models

Congenital erythropoietic porphyria (CEP) is a rare genetic disorder leading to accumulation of uro/coproporphyrin-I in tissues due to inhibition of uroporphyrinogen-III synthase. Clinical manifestations of CEP include bone fragility, severe photosensitivity and photomutilation. Currently there is no specific treatment for CEP, except bone marrow transplantation, and there is an unmet need for treating this orphan disease. Fluorescent porphyrins cause protein aggregation, which led us to hypothesize that uroporphyrin-I accumulation leads to protein aggregation and CEP-related bone phenotype. We developed a zebrafish model that phenocopies features of CEP. As in human patients, uroporphyrin-I accumulated in the bones of zebrafish, leading to impaired bone development. Furthermore, in an osteoblast-like cell line, uroporphyrin-I decreased mineralization, aggregated bone matrix proteins, activated endoplasmic reticulum stress and disrupted autophagy. Using high-throughput drug screening, we identified acitretin, a second-generation retinoid, and showed that it reduced uroporphyrin-I accumulation and its deleterious effects on bones. Our findings provide a new CEP experimental model and a potential repurposed therapeutic.

Porphyrias are a group of inherited disorders due to defects in the heme biosynthetic pathway 1,2 . One such example is congenital erythropoietic porphyria (CEP), most commonly caused by loss of function mutation in uroporphyrinogen III synthase (UROS; Fig.S1), the enzyme that catalyzes the third step of the heme biosynthetic pathway 1,3 . CEP is rare, with ~ 250 cases reported to date 4,5 . It is autosomal recessive, and associated with reduced UROS activity (5% of normal) and consequent accumulation of uro/coproporphyrin-I (uro/copro-I) in bone marrow, erythrocytes, plasma, and increased uro/copro-I excretion in urine and stool 1,3,5,6 . CEP is characterized by severe photosensitivity, with skin fragility and blistering of sun-exposed areas 1,5,6 . Scaring due to secondary skin infections and bone resorption contribute to disfigurement of light-exposed areas 3 . Other clinical manifestations of this multisystem disease include chronic ulcerative keratitis, hemolysis, which may require repeated blood transfusions in severe cases, nonimmune hydrops fetalis, red urine since birth, erythrodontia and osteodystrophy 1,3,6,7 . Currently, there is no specific pharmacological treatment for CEP, with interventions being life-style-related (e.g. avoidance of sun) or complex procedures, including bone marrow transplantation 3,8,9 . Fluorescent porphyrin accumulation in porphyria causes organelle specific protein oxidation and aggregation through mechanisms that involve type-II photosensitive reactions and secondary oxidative stress [10][11][12][13][14][15][16][17] . We posit that porphyrin-mediated protein aggregation in CEP plays a major mechanistic role in tissue damage that involves accumulation of fluorescent uro/copro-I.
We demonstrated that uro-I injection of zebrafish larvae mimics features of CEP, including uro-I accumulation in bones and bone deformation, as judged by decreased vertebra and operculum volume. Uro-I treatment of an osteoblastic human osteosarcoma cell line, Saos-2, caused significant decrease in mineral matrix synthesis Figure 1. Zebrafish model of CEP develops bone phenotype resembling human disease. (A) 6dpf zebrafish larvae were injected with uro-I or vehicle and imaged by confocal microscopy at 7dpf. Porphyrin was detected only in the bones of uro-I-injected group. Arrowhead-operculum; box-vertebrae. (B) Larvae were treated as in (A) and injected with calcein prior to imaging. Arrowhead-operculum; arrow-4 th vertebra. (C) Quantification of bone volume in larvae from (B); bone volume was normalized to vehicle-injected larvae set to 100%. Symbols represent individual larvae (14-18/group) from 4-5 independent experiments. (D) Larvae were treated as in (A). At 7dpf bones were harvested and imaged by epifluorescence microscopy pre and post HCl bone demineralization, arrow-notochord. (E) Hydroxyapatite was incubated with calcein/uro-I/copro-I and imaged by epifluorescence microscopy. Scale bars: 200 µm (A-D); 50 µm (E). Three-dimensional image reconstruction (A, B) was performed using Imaris 3D software v7.7 (http:// imaris. oxinst. com/). **p < 0.01.

Uro-I impairs osteoblastic mineralization by aggregating matrix proteins, promoting ER stress and inhibiting autophagy.
To elucidate the molecular mechanism of uro-I-mediated bone damage, we used Saos-2 cells, a human osteosarcoma cell line with osteoblastic features 20 . Mineralization was stimulated by treating cells with a mineralization activation cocktail (MAC) and assayed by alizarin red S (ARS) staining. As expected, Saos-2 cells manifested a mineralization phenotype when cultured for 3 days in MACsupplemented medium, while in uro-I + MAC supplemented medium, mineralization decreased significantly (Fig. 3A, B). Uro-I also caused marked photosensitivity, leading to cell death when cells were not shielded from light (not shown).
Since fluorescent porphyrins cause protein aggregation or loss of antibody reactivity when tested by immunoblotting, we tested whether uro-I-mediated inhibition of Saos-2 mineralization led to aggregation of bone matrix proteins. Blotting Saos-2 cell lysates prepared from uro-I or vehicle treated cells using antibodies to fibronectin, osteonectin and type 1 pro-collagen showed a distinct loss of monomer for these proteins after uro-I treatment (Fig. 3C). We attribute the loss of antibody reactivity to epitope masking after uro-I binding and subsequent oxidation and aggregation, as shown previously for PP-IX 13 . The loss of matrix protein monomers and aggregation was verified by mass spectrometry (Fig.S3).
Given the effect of uro-I on protein aggregation, we tested whether uro-I treatment initiates unfolded protein response (UPR) and endoplasmic reticulum (ER) stress. We observed upregulation of BiP, consistent with UPR and ER stress 21 (Fig. 3C). Other ER stress markers, including PERK, IRE1α and ATF6, were likely oxidized and aggregated, as judged by monomer loss (Fig. 3C). Our findings suggest a non-canonical form of ER stress, which we have observed upon PP-IX accumulation 13 , that involves aggregation and possibly inactivation of ER resident proteins and chaperones.
Autophagy modulates exocytosis of hydroxyapatite crystals and thus plays an important role in bone mineralization by osteoblasts 22,23 . Since uro-I inhibited mineralization, we tested whether it also disrupted autophagy. As expected, MAC-treated Saos-2 cells showed increased LC3-II (Fig. 3D, left panel). Uro-I treatment also increased LC3-II levels (Fig. 3D, right panel). The likely explanation for the increased LC3-II is not increased autophagy but a slowing of autophagic flux, which could lead to stalling of exocytosis of mineral-loaded vesicles and decreased mineralization of bone matrix 24 .
To further characterize uro-I-mediated impairment of mineralization in Saos-2 cells, we performed gene expression analysis to probe for alterations in the stress response pathway. Genes that were differentially regulated two-fold or more after uro-I treatment were assessed further. Uro-I treatment increased HSPA5 (2.1x) and SOD3 (2.8x), while SERPINH1 decreased (3.3x) (Fig. 3E, left panel). Since HSPA5 encodes BiP, HSPA5 upregulation supports the BiP upregulation observed biochemically (Fig. 3C). SERPINH1, a collagen-specific chaperone, downregulation may account for collagen misfolding and aggregation. Because type 1 collagen is the most abundant protein in bone matrix 25 , we assessed COL1A1 expression and observed a 90% reduction in COL1A1 after uro-I treatment (Fig. 3E, right panel). This finding supports uro-I-induced loss of mineralization, since collagen serves as a matrix for mineral deposition.
We next asked whether acitretin can protect from the effects of uro-I, by treating Saos-2 cells with uro-I in the presence of acitretin. Although acitretin did not prevent uro-I-mediated loss of mineralization (Fig. 3F), it blunted the ER stress response by reducing BiP level and normalized the autophagic flux by reducing LC3-II (Fig. 3G, H). Acitretin also downregulated SOD3 3.8-fold, thereby suggesting that acitretin mitigates the oxidative stress caused by uro-I (Fig. 3I). Upregulation of COL1A1 (1.7x) and SERPINH1 (5.4x) was also observed.
Taken together, our data demonstrate that acitretin mitigates uro-I-mediated proteotoxicity and oxidative stress. However, under the conditions tested, acitretin did not rescue the impairment of mineralization caused www.nature.com/scientificreports/ www.nature.com/scientificreports/ www.nature.com/scientificreports/ by uro-I treatment of Saos-2 cells. A possible explanation for why mineralization was not normalized by acitretin is that ER stress and autophagy pathways need to be normalized in order for cells to have their mineralization ability restored. Alternatively, acitretin may act differently on various cell types which is one major advantage offered by the in vivo zebrafish system.

Discussion
Uro-I is a fluorescent porphyrin capable of types I/II-photosensitized reactions [26][27][28] , which explains the observed photosensitivity in CEP, damage to digits and facial features 3 . However, light is unlikely to reach deep internal tissues, which are also affected in CEP. Of note, we observed uro-I-mediated protein aggregation and decreased mineralization in the dark. Previous studies had also reported dark effects of porphyrins. For example, uro-I increased collagen biosynthesis in human skin fibroblasts 29 , and inhibited erythrocytic uroporphyrinogen decarboxylase activity 30 . A 2-hit model could explain light-independent porphyrin-mediated protein aggregation and proteotoxicity whereby, in absence of light, a secondary oxidant source (eg. inflammatory cells) causes protein oxidation followed by porphyrin binding to oxidized protein, yielding protein aggregates 11,12 (Fig. 4). CEP is frequently associated with superinfections and osteolysis 3,31 . Hence, infiltrating immune cell-generated oxidants might serve as a secondary source of oxidant, leading to uro-I mediated protein aggregation in internal organs such as bones. Additionally, uro-I might generate oxidants by acting as a substrate for the ferredoxin/ ferredoxin:NADP + oxidoreductase system 32 . Although ferredoxin/ferredoxin:NADP + oxidoreductase are commonly associated with hepatic microsomes, they are also expressed in osteoblasts 33 and could metabolize uro-I to generate oxidants in the absence of light. The differences in charge and polarity of uro-I and PP-IX might explain the striking difference in their tissue localization. Retro-orbitally injected PP-IX accumulated in zebrafish liver 10 , while uro-I accumulated preferentially in bone (Fig. 1). Of note, liver cancer cell lines do not uptake uroporphyrin (unpublished data), possibly www.nature.com/scientificreports/ due to its high negative charge that prevents traversing the cell membrane 34 . Based on our data, we propose that negatively charged uro-I binds to Ca 2+ in hydroxyapatite (Fig. 4) and thus bone and Saos-2 cells are affected by uro-I. This association with bone matrix causes uro-I to have a different protein aggregation signature compared to PP-IX, which is primarily internalized. PP-IX aggregated intracellular proteins such as keratins and glyceraldehyde 3-phosphate dehydrogenase 13,15,35 , whereas uro-I affected extracellular bone matrix proteins (Fig. 3). Oxidants such as singlet oxygen, a major oxidant produced by photosensitive reactions 26 , have extremely small intracellular diffusion distance (10-20 nm) and lifetime (10-40 ns) [36][37][38] . Binding of uro-I to bone matrix causes a 'sensitizer-acceptor' coupling, as observed for other diffusible oxidants 39,40 , and greatly increases the oxidation efficiency and specificity. Of note, oxidized fibronectin reduces mineralization of rat calvarial osteoblasts in vitro 41 . The high selectivity of uro-I localization to bone matrix might provide a pathway to develop photodynamic therapeutic agents for bone cancers such as osteosarcoma.
The management of CEP is challenging, with current therapeutic options focusing on bone marrow/hematopoietic stem cell transplantation 3 , and by avoidance of sun and light exposure, including the use of protective clothing [42][43][44][45] . There are also potential experimental therapeutic approaches including gene therapy 46 , proteasomal inhibitors 47,48 , iron chelation 49,50 and phlebotomy 51 . Most recently, the repurposed use of ciclopirox, an approved broad-spectrum antifungal agent, showed promising results in the treatment of CEP using a mouse model 52 . However, there are limitations to these approaches, such as complications from transplantation and neurotoxic side effects of proteasome inhibitors 53 . Currently there are no known pharmaceuticals that act by clearance of uro-I, and in this regard acitretin provides a novel approach. Acitretin might also act as an antioxidant (Fig. 4) due to its hyperconjugated nucleophilic double bonds. Thus, through a combination of destabilizing uro-I-bone matrix interaction and antioxidant activity, acitretin could ameliorate CEP manifestations (Fig. 4). Acitretin also offers a drug repurposing advantage since it is already approved for psoriasis treatment 54 .

Materials and methods
Zebrafish experiments and cell culture. Zebrafish (Danio rerio) experiments were conducted using ABxTL hybrid and NHGRI-1 wild type zebrafish lines. All animal procedures were approved by the Rutgers University Institutional Animal Care and Use Committee (protocol number PROTO201900147) and performed in compliance with federal guidelines and the standards of the NIH Guide for the Care and Use of Laboratory Animals 55 , the Rutgers University IACUC Policy Handbook and the Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines.

Uro-I solution preparation and treatment of zebrafish larvae and Saos-2 cells.
Uro-I (uroporphyrin-I dihydrochloride; Frontier Scientific, Catalog#:U830-1) was initially resuspended in 0.1 M NaOH and the pH was adjusted to neutral using 0.2 M Na 2 HPO 4 . Six days post fertilization (dpf), ABTL zebrafish larvae were injected via the retro-orbital route with approximately 3nL of 7.2 mM uro-I solution and control larvae were injected with vehicle (0.1 M NaOH in 0.2 M Na 2 HPO 4 ). After injection, larvae were immediately transferred to Petri dishes wrapped with heavy duty aluminum foil and kept in a dark incubator, at 28.5 °C for 24 h. Where indicated, 7 dpf larvae were injected with approximately 2 nL of 0.2% w/v calcein (Sigma, Catalog#:C0875) 2 h prior to imaging.
Saos-2 cells were plated in 12-well plates (1.5 × 10 5 cells/well) and allowed to attach overnight. Cells were then treated with uro-I (144 µM final concentration) or vehicle in medium containing MAC for 3 days. Experiments were conducted in a dark room and cells were kept shielded from light in a tissue culture incubator.
Confocal microscopy imaging and quantification. Seven dpf ABxTL zebrafish larvae were anesthetized with tricaine-S (Syndel) and immobilized in 0.5% low melt agarose. Fluorescent z-series were captured using an Olympus FV500 confocal microscope (10X objective, confocal aperture of 300 µµ) with an optical thickness of 10 µm and z-step size of 10 µm. Calcein was excited with a 488 nm argon laser and emission was captured between 505 and 525 nm. Porphyrin was excited with a 405 nm laser diode and emission was captured above 560 nm. Three-dimensional image reconstruction and quantification of fluorescent signal and bone volume were performed using Imaris 3D visualization and analysis software v7.7 (Bitplane).
In vivo and in vitro binding of uro-I. Six dpf ABxTL zebrafish larvae were injected with uro-I or vehicle (as described above) and 24 h later were euthanized by tricaine-S overdose on ice bath. Bones were harvested as previously described 57 . Briefly, soft tissue was removed by incubating larvae with Accumax solution (Millipore-Sigma, Catalog#:A7089) under vigorous shaking. Bones were collected using a 70 µm cell strainer, followed by demineralization with 1.2 M HCl. Fluorescent images were captured prior to and after the demineralization step using a Zeiss Axio Imager M2 fluorescence microscope. Porphyrin signal was captured using the red fluorescent channel.
10 mg hydroxyapatite (Acros Organics, Catalog#:1306-06-5) was incubated with 1 mM uro-I, 1 mM copro-I (coproporphyrin-I dihydrochloride, Frontier Scientific, Catalog#:C654-1), vehicle or 0.2% calcein for 30 min in the dark and vortexed every five minutes. Samples were washed and imaged by epifluorescence microscopy as described above. Six dpf NHGRI-1 zebrafish larvae were injected retro-orbitally with approximately 2nL of a solution of uro-I (10 mM) and calcein (0.2% w/v). Immediately after injection, larvae were transferred to a 96-well test plate (two larvae in 50 µL of E3 medium/well), including the DMSO control wells. Larvae were kept in the dark, at 28.5 °C. After 24 h, they were anesthetized with tricaine-S, centrifuged at 500xg for two minutes and imaged using the ImageXpress Micro Cellular Imaging and Analysis System (Molecular Devices). Positive hits were selected based on visual identification of calcein signal increase and porphyrin signal decrease compared to DMSO-treated larvae. Compounds in test wells that met the inclusion criterion were tested individually in the same manner as described above ( Figure S2).

Acitretin validation and treatment.
A dose-response curve with acitretin (Selleck Chemicals, Houston, TX) was conducted (0.5-12.5 µM) and 10 µM was observed to yield consistent results, without being toxic to zebrafish larvae. Validation and characterization of acitretin as a potential treatment for CEP was performed. Six dpf ABxTL zebrafish larvae injected with uro-I were immediately transferred to 10 cm plastic dishes containing 10 µM acitretin or DMSO in E3 medium (prophylaxis protocol, Fig. 2A), and incubated for 24 h in the dark at 28.5 °C. Porphyrin binding to bones and bone volume were analyzed by confocal microscopy as described above. Porphyrin excretion into the medium was quantified. Uro-I-injected larvae were transferred to 96-well plates, one larva/well, 100µL of 10 µM acitretin or DMSO/well. Medium was collected after 24 h and porphyrin was quantified as described previously 13 . Etretinate (Selleck Chemicals, Houston, TX) and tretinoin (Selleck Chemicals, Houston, TX) treatment was performed as described for acitretin.
In addition to being used as prophylaxis, we evaluated whether acitretin had a therapeutic effect. Six dpf ABxTL zebrafish larvae were injected with uro-I and 24 h later they were transferred to E3 medium containing 10 µM acitretin or DMSO. Porphyrin binding, excretion and bone volume were analyzed as described above. For Saos-2 cells, they were treated with 10 µM acitretin or DMSO in medium containing MAC and uro-I.

Alizarin Red S (ARS) staining and quantification. Cell mineralization was quantified by ARS (Sigma
Aldrich St. Louis, MO) staining as described previously 58 with minor modifications. Briefly, cells were fixed with 100% ethanol at 37 °C for 1 h, stained with 40 mM (pH4.2) ARS solution for 20 min in an orbital shaker. Cells were washed and ARS was extracted by incubation of fixed cells with 10% (v/v) acetic acid, followed by scraping, incubation of suspension (85 °C, 10 min), centrifugation and neutralization of supernatant with 10% (v/v) ammonium hydroxide. ARS standard curve (from 2-0.02 mM) and samples were transferred in triplicate to a 96-well plate and absorbance was measured at 405 nm.