Peroxidasin mediates bromination of tyrosine residues in the extracellular matrix.

Peroxidasin is a heme peroxidase that oxidizes bromide to hypobromous acid (HOBr), a powerful oxidant that promotes the formation of the sulfilimine cross link in collagen IV in basement membranes. We investigated whether HOBr released by peroxidasin leads to other oxidative modifications of proteins, particularly bromination of tyrosine residues in peroxidasin-expressing PFHR9 cells. Using stable isotope dilution LC/MS/MS, we detected the formation of 3-bromotyrosine, a specific biomarker of HOBr-mediated protein modification. The level of 3-bromotyrosine in extracellular matrix proteins from normally cultured cells was 1.1 mmol/mol tyrosine and decreased significantly in the presence of the peroxidasin inhibitor, phloroglucinol. A negligible amount of 3-bromotyrosine was detected in peroxidasin-knockout cells. 3-Bromotyrosine formed both during cell growth in culture, and in the isolated decellularized extracellular matrix when embedded peroxidasin was supplied with hydrogen peroxide and bromide. The level of 3-bromotyrosine was significantly higher in extracellular matrix than intracellular proteins, although a low amount was detected intracellularly. 3-Bromotyrosine levels increased with higher bromide concentrations and decreased in the presence of physiological concentrations of thiocyanate and urate. However, these peroxidase substrates showed moderate to minimal inhibition of collagen IV cross-linking. Our findings provide evidence that peroxidasin promotes the formation of 3-bromotyrosine in proteins. They show that HOBr produced by peroxidasin is selective for, but not limited to, the cross-linking of collagen IV. Based on our findings, the use of 3-bromotyrosine as a specific biomarker of oxidative damage by HOBr warrants further investigation in clinical conditions linked to high peroxidasin expression.


Introduction
Peroxidasin is a heme peroxidase that is widely distributed throughout the animal kingdom (1)(2)(3). In mammals, its main known physiological function is to catalyze the formation of an essential sulfilimine bond (S=N) between specific methionine and hydroxylysine residues in collagen IV (4). This intermolecular bond is crucial for the functionality and integrity of the basement membrane, a specialized form of the extracellular matrix (ECM). The mechanism involves the oxidation of bromide to hypobromous acid (HOBr), which then promotes the cross link between specific methionine and hydroxylysine residues in the non-collagenous domain (NCD) to join two protomers of collagen IV (5). Other pathophysiological roles of peroxidasin have been suggested, such as antimicrobial defense (6,7), redox sensing through interaction with the transcription factor Nrf2 (8), promotion of angiogenesis (9) and kidney fibrosis (10, 11). Homozygous deletion of peroxidasin causes eye defects (12) and high expression has been identified as a determinant of melanoma cell invasion (13).
Mammalian peroxidasin is a typical heme peroxidase that catalyzes the two electron oxidation of thiocyanate and iodide as well as bromide (14). It also oxidizes typical peroxidase substrates including tyrosine, urate and nitrite (15). Furthermore, HOBr is a strong oxidant that reacts with numerous biological molecules. Its favored reactions are the oxidation of cysteine and methionine residues and the conversion of amine groups to bromamines (16,17). HOBr also converts tyrosine to 3-bromotyrosine (3-Br-Tyr) and 3,5dibromotyrosine at an appreciable rate (16,18). These brominated products are stable and unique making them ideal biomarkers for detecting the formation of HOBr and its reactions with proteins.
A key question about peroxidasin is whether the HOBr that it produces forms the collagen IV cross link exclusively or undergoes other reactions. Our earlier work showed that peroxidasin releases free HOBr as detected by the conversion of added NADH to its bromohydrin (19). Here, by monitoring the formation of 3-Br-Tyr in proteins, we examine whether the activity of peroxidasin during cell growth and in isolated decellularized ECM leads to HOBr-mediated protein modification. We used mouse epithelial-like PFHR9 cells, which are differentiated embryonal carcinoma cell that secrete appreciable amounts of basement membrane proteins. This cell line has been used in many studies to characterize the crosslinking of collagen IV (4,20). When PFHR9 cells produce ECM, they secrete peroxidasin which is embedded in the matrix and able to catalyze the formation of the cross link.

Peroxidasin catalyzes the formation of 3bromotyrosine during cell growth
As previously reported, PFHR9 cells express peroxidasin which catalyzes the formation of the sulfilimine cross link in the collagen IV that they release during the deposition of the ECM in culture (4,19,21). This unusual cross link was characterized by Vanacore et al (21) as a linkage between the sulfur of Met 93 and the amine nitrogen of hydroxyLys 211 in the non-collagenous domain (NCD; Figure 1 A). We first confirmed the formation of dimer in the NCD of collagen IV in our system by isolating the ECM after 6-8 days' culture and analysing the NCD by SDS-PAGE and densitometry quantification. The majority (>70 %) of the isolated NCD of collagen IV from WT cells was cross-linked (Figure 1 B, C). The NCD can be joined by one or two sulfilimine bonds, resulting in two dimers with different electrophoretic mobility: dimer 1 (upper band) and dimer 2 (lower band) (Figure 1 B). Phloroglucinol (PHG; 1,3,5trihydroxybenzene) is an effective, reversible inhibitor of peroxidasin (22). The cross-linking decreased to approximately 15 % in ECM isolated from WT cells grown in the presence of PHG, or from peroxidasin-knockout cells.
We then measured 3-Br-Tyr levels in the ECM from PFHR9 cells cultured under the same conditions. 3-Br-Tyr and Tyr were detected using LC/MS/MS with isotopically labelled internal standards (Figure 2 A). As shown in Figure 2 B, Tyr and both bromine isotopes of 3-Br-Tyr were detected in ECM from WT cells (left panel) whereas the 3-Br-Tyr peaks in ECM isolated from peroxidasin-knockout cells were scarcely detectable (right panel). Using the standard calibration curve, the level of 3-Br-Tyr in ECM isolated from the WT cells was calculated to be 1.1 mmol/mol tyrosine and decreased significantly to 0.14 mmol/mol when WT cells were grown in the presence of PHG (Figure 2 C). 3-Br-Tyr formation mirrored the profile for dimer formation seen in Figure 1 C.
To test substrate requirement for the formation of 3-Br-Tyr during growth of cells, PFHR9 cells were cultured in media supplemented with 50 or 100 µM bromide. It should be noted that a trace amount of bromide (~6 μM in 100 mM NaCl) is present in culture medium as contamination in sodium chloride (5). The level of 3-Br-Tyr increased with increasing bromide concentration, and appeared to plateau above 50 μM bromide (Figure 3 A). Analysis of dimer formation in the same samples showed that cross-linking was near maximal with the trace amounts of bromide present in the culture medium (Figure 3 B).

Levels of 3-bromotyrosine are higher in ECM than in intracellular proteins
To establish whether 3-Br-Tyr is formed on intracellular as well as extracellular proteins, lysates from PFHR9 cells were separated into intracellular extract (ICE), ECM and NCD. The three fractions were hydrolyzed and analyzed by LC/MS/MS. The level of 3-Br-Tyr detected in the intracellular fraction, (0.13 mmol/mol Tyr) was approximately 10% of that in the total ECM whereas the level in the NCD was 1.4-fold higher (Figure 4). 3-Br-Tyr levels in each fraction increased 1.2-1.4-fold when cells were grown in the presence of added bromide, and substantially decreased when PHG was present or in ECM from peroxidasin-knockout cells ( Table 1).

Formation of 3-bromotyrosine in the isolated decellularized ECM
We also investigated whether peroxidasin embedded in the decellularized ECM promotes 3-Br-Tyr formation. Uncross-linked ECM was isolated from WT cells grown in the presence of PHG and treated with hydrogen peroxide and bromide. The level of 3-Br-Tyr measured at one hour increased significantly in the presence of both bromide and hydrogen peroxide, but not when either reagent was added separately (Figure 5 A). A similar trend was observed for collagen IV crosslinking (Figure 5 B). The levels of 3-Br-Tyr increased progressively over time (Figure 6 A), and with increasing concentrations of bromide and hydrogen peroxide (Figure 6

Thiocyanate modulates the formation of 3bromotyrosine
Thiocyanate (SCN -) is a pseudohalide that competes with bromide as a substrate for peroxidasin (14). Previously we showed that physiological concentrations of thiocyanate interfere with the cross-linking of collagen IV (19). Here we investigated whether thiocyanate modulates the formation of 3-Br-Tyr. PFHR9 cells were grown in culture medium supplemented with thiocyanate (0 -400 μM), either without or with added bromide (50 and 100 μM) to test for substrate competition. The levels of 3-Br-Tyr in ECM isolated from the cells decreased progressively from 1.2±0.3 to 0.27±0.03 mmol/mol tyrosine with increasing concentration of thiocyanate in the medium (Figure 7 A). Supplementing the medium with bromide increased the starting level of 3-Br-Tyr and the IC50 for inhibition by thiocyanate. Dimer formation in the NCD of cells treated with thiocyanate in the same way was also progressively inhibited but to a lesser extent than tyrosine bromination (Figure 7 B).
We also studied the effect of thiocyanate in the isolated ECM system. Uncross-linked ECM from PFHR9 cells grown in the presence of PHG was decellularized and treated with increasing concentrations of thiocyanate (0-400 µM), in addition to bromide (50 or 100 µM) and 50 µM hydrogen peroxide. Similar to the results in the cell culture system, bromination of tyrosine residues in decellularized isolated ECM decreased with increasing concentration of thiocyanate, with near complete inhibition at 400 µM (Figure 7 C). Increasing the bromide concentration also led to a higher IC50. Thiocyanate also caused a decline in dimer formation but to a much lesser extent than the inhibition of 3-Br-Tyr formation (Figure 7 D).
In earlier work, we showed that urate modulates the formation of the sulfilimine cross link in collagen IV (19). Here we tested whether it also affects the formation of 3-Br-Tyr. When added to culture medium without added bromide, up to 400 µM urate caused a gradual decline in 3-Br-Tyr in the ECM, from 1.2±0.2 to 0.68±0.2 mmol/mol Tyr. When 100 µM bromide was also added, the level of 3-Br-Tyr decreased from 1.7±0.3 to 1.3±0.3 mmol/mol Tyr. In contrast to the inhibition of dimer formation in the decellularized ECM system, urate scarcely affected dimer formation in cell culture, with no observable inhibition in the presence of 100 µM bromide (data not shown).

Discussion
We have demonstrated that HOBr produced by peroxidasin leads to the formation of 3-Br-Tyr residues in extracellular proteins as well as promoting sulfilimine cross links in collagen IV. We show that bromination occurs during growth of epithelial-like cells in culture, as well as in isolated decellularized ECM. That peroxidasin is responsible for the formation of 3-Br-Tyr is evident from the very low level measured in ECM isolated from peroxidasin-knockout cells and the strong inhibitory effect of PHG. Therefore, peroxidasin activity is not restricted to forming the sulfilimine cross link in collagen IV but also causes other protein modifications.
In cell culture, the level of 3-Br-Tyr increased with increasing bromide concentration, but with no added bromide the extent of bromination was already about half maximal. As observed previously (19), sulfilimine formation was maximal without the addition of bromide. This can be accounted for by the trace amounts of bromide present in the culture medium (5). Our findings indicate that the HOBr produced by peroxidasin shows a degree of selectivity for the cross-linking of collagen IV but even at low concentrations of bromide, there is substantial 3-Br-Tyr formation. At higher concentrations, but still within the physiological range of 20-100 µM, production of HOBr and its reactions with other targets are increased.
3-Br-Tyr residues were also produced in isolated decellularized ECM, when supplied with the peroxidasin substrates, bromide and hydrogen peroxide. Levels of 3-Br-Tyr were in the same range as those formed during cell growth in culture, but added bromide was required and the concentration dependence was more pronounced. The isolated ECM also required a higher bromide concentration for maximal dimer formation. This difference might be due to the longer time in cell culture allowing accumulation of 3-Br-Tyr and NCD dimer. Another possibility is that the cellular system binds or concentrates bromide to a higher effective concentration than that of the medium and this compartmentalisation is lost during isolation of the ECM. It is also possible that steric specificity conferred by peroxidasin binding partners is lost during processing. To determine the location of brominated proteins following cell culture, ICE, which would contain membrane and organelle as well as cytoplasmic proteins, total ECM and the NCD of collagen IV were examined. On a 3-Br-Tyr per Tyr basis, the ECM proteins showed much higher bromination. This is consistent with peroxidasin catalysing bromination once secreted to the ECM, where it has been shown to catalyze the cross-linking of collagen IV (4). It is possible that some activity could occur before secretion. A low but significant amount of 3-Br-Tyr was detected in the intracellular extract. It is unlikely to be due to contamination with ECM proteins, as approximately six times more protein was extracted in the ICE than the ECM and to achieve the observed level of 3-Br-Tyr in the ICE fraction would require half of the total ECM protein to have been solubilized into the ICE. Peroxidasin has been detected in the endoplasmic reticulum where it is synthesized, processed and secreted via the Golgi (1,23), then cleaved by proprotein convertase A at the cell membrane (24). These organelles are possible sites for intracellular 3-Br-Tyr formation but further fractionation is required to establish whether this is the case.
The level of 3-Br-Tyr was significantly higher in the NCD of collagen IV than in the total ECM. To assess whether all the 3-Br-Tyr could reside in the NCD, we considered that the NCD constitutes approximately 7 % of the total ECM protein (~14% of the amino acids in collagen IV which accounts for ~50 % of the structural proteins in basement membranes) (25)(26)(27). As the 3-Br-Tyr:Tyr ratio was only 1.4-fold higher in the NCD, the small fraction of NCD cannot account for the bromination of tyrosine in the ECM. Therefore, while the sulfilimine cross-linking by the HOBr produced by peroxidasin occurs specifically in the NCD of collagen IV, other reactions of HOBr are not restricted to this region. During submission of this manuscript, He et al. (28) reported MS evidence for a form of bromine being present in kidney basement membranes and corroborated our findings that this included 3-Br-Tyr produced via peroxidasin activity. They detected bromination of Tyr residues in the collagen IV NCD, but as shown here it is not restricted to this region.
Thiocyanate is a better peroxidasin substrate than bromide, with rate constants of 1.83x10 7 and 5.6x10 6 M -1 s -1 respectively having been measured for reaction with compound I of a truncated recombinant form that contains the peroxidase and Ig domains (14). Based on direct competition, the product of thiocyanate oxidation, HOSCN, does not cross-link collagen but thiocyanate does inhibit cross-linking (5,19) as recapitulated here. The extent of inhibition is less than might be expected from the above rate constants, suggesting that peroxidasin may have an affinity for bromide, possibly due to location and association with ECM proteins. In the present study thiocyanate inhibited 3-Br-Tyr formation more effectively than collagen IV cross-linking, either when added to cells in culture, or to isolated ECM. The observed IC50 values and their increase with increasing bromide concentration imply that thiocyanate acts as a competitive substrate. However, thiocyanate reacts rapidly with HOBr (a rate constant of 2.3x10 9 compared with 2.6x10 5 M -1 s -1 for tyrosine (17,29) and could also inhibit bromination by direct scavenging. Either way, our results indicate that the reactions of peroxidasin-derived HOBr could be modulated by physiological levels of thiocyanate, which are typically 20-120 µM. As HOSCN is less reactive than HOBr (30,31), high thiocyanate concentrations could be advantageous in protecting from the harmful (presumably collateral) effects of HOBr without compromising the sulfilimine cross link.
Peroxidasin reacts with a range of peroxidase substrates, including urate (15). Urate inhibited bromination activity by the recombinant enzyme (15) and partially inhibited cross-linking in isolated ECM (19). With cultured cells we observed little inhibition of cross-linking or 3-Br-Tyr formation by urate at high physiological concentrations, suggesting the relevance of such inhibition is likely to be minor.
The level of bromination that we detected in ECM proteins was around one in a thousand tyrosines. While this may seem low, similar levels have been observed in experimental and clinical studies of eosinophil peroxidase (18,32). Furthermore, although 3-Br-Tyr is a good biomarker for HOBr, cysteine, methionine, tryptophan, histidine and amino groups are more favored targets than tyrosine (16,17). Peroxidasin can also catalyze dityrosine formation (3). Therefore, the detection of 3-Br-Tyr is an indication that more extensive modifications to protein, and potentially other biomolecules, are likely to occur. The nature of such modifications and their implications need to be established.
Our finding that peroxidasin activity yields 3-Br-Tyr has implications for clinical studies where 3-Br-Tyr has been measured in body fluids and tissues and used as a biomarker of oxidative damage in pathological conditions including asthma, sinonasal inflammation, cystic fibrosis, eosinophilic esophagitis and diabetes (32)(33)(34)(35)(36)(37). In such studies, the formation of 3-Br-Tyr has been attributed to other heme peroxidases, namely myeloperoxidase and eosinophil peroxidase (18,32,38). Our findings raise the possibility that it arises from peroxidasin activity. Peroxidasin is widely expressed in human tissues, including the vascular system (39), with high expression in some cancer cells (13,40). It could act as a resident contributor to oxidative damage of proteins in the ECM, a tissue with poor antioxidant defence. Prooxidant activity of peroxidasin has been proposed to play a role in hypertension and ischemiareperfusion injury (41)(42)(43)(44)(45), and evidence that halogenation of ECM proteins impairs renal function in a mouse model suggests that it might be involved in diabetic glomerular injury (42). Furthermore, tyrosine halogenation decreases the pKa of the phenol group (46), which can affect tyrosine phosphorylation and consequently cell signalling cascades. Therefore, tyrosine bromination by peroxidasin might have a wide impact, especially given the role of the ECM in cell signalling (47).
In summary, we have used 3-Br-Tyr as a specific biomarker to show that the reactions of HOBr produced by peroxidasin are not limited to the cross-linking of collagen IV. Further research is needed to characterize the structural and functional impact of HOBr in peroxidasin-producing cells. Our findings also raise questions about the origin of 3-Br-Tyr in clinical material and open the door to investigating the role of peroxidasin in catalysing oxidative modifications in pathological conditions.

Experimental procedures
Materials DMEM and fetal bovine serum were obtained from Gibco Life Technologies, hydrogen peroxide from LabServ, sodium hypochlorite (Janola) from Pental Products, bacterial collagenase from Worthington, and phloroglucinol (1,3,5-trihydroxybenzene) from Sigma Aldrich. Other chemicals were obtained from Sigma Aldrich, BDH Chemicals, or JT Baker.
Mouse epithelial-like cells PFHR9 (ATCC CLR-2423) were obtained from the American Type Culture Collection. Peroxidasin-knockout PFHR9 cells were generated using CRISPR-Cas9 gene editing, as described by Colon and Bhave (24) and generously donated by Dr Gautam Bhave from Vanderbilt University, Nashville, Tennessee.

Cell culture and isolation of ECM
We followed the method described by Bhave et al. (4). Briefly, PFHR9 cells were seeded at high density (~9x10 6 cells in a 15 cm culture plate) and grown in DMEM (+ 10% fetal bovine serum) for 6-8 days post-confluency to deposit the ECM. The medium was changed daily with supplementation of 50 μM PHG, potassium bromide or sodium thiocyanate as indicated. The cells were lysed with ice cold lysis buffer (10 mM Tris pH 8, 1 mM EDTA, 1 % sodium deoxycholate, 1x Complete® Protease Inhibitor), scraped, sonicated until homogeneous and centrifuged at 20,000 g to separate the supernatant, designated intracellular extract (ICE), from the ECM pellet. The pellet was washed twice in Wash Buffer (1M NaCl, 10 mM Tris pH 7.5) with centrifugation and resuspended in PBS with brief sonication. Protein concentration was measured using the Direct Detect® Infrared Spectrometer. The yield of ECM proteins from a 15 cm culture plate after 6-8 days was approximately 5 mg. Established procedures were used to separate the non-collagenous domain. Isolated ECM was resuspended in digestion buffer (2 mg/ml bacterial collagenase in 50 mM sodium phosphate pH 7.4) and left overnight at 37°C (48). The NCD was purified by anion-exchange chromatography using DEAE resin (49). The soluble digest was mixed with an equal volume of DEAE resin (equilibrated in 50 mM Tris pH 7.4) with shaking for 1 hour at 4°C. The slurry was centrifuged and the soluble unbound NCD in the supernatant was used for analysis. The purified NCD was resolved by SDS-PAGE and stained with Coomassie Brilliant Blue R-250.

Treatment of the isolated decellularized ECM:
ECM isolated from cells grown in the presence of 50 μM PHG was resuspended in PBS and sonicated. Potassium bromide and hydrogen peroxide (concentrations indicated in each figure) were added to 350 μg of ECM protein in a total volume of 300 μl. The reaction was carried out at 37°C for 60 minutes (unless stated otherwise) and stopped by adding 1 mM azide or 5 mM methionine. Other substrates and inhibitors were added at the concentrations stated in each figure.

Detection of 3-bromotyrosine
A stable isotope dilution LC/MS/MS method was used for the detection and quantification of 3-Br-Tyr levels in proteins as previously described (18,33,50,51). Isotopically-labelled standards ( 13 C6 Tyr and 13 C9 3-Br-Tyr) were used to control for experimental variations, such as recovery, matrix effect and ionization. Artefactual bromination was monitored by measuring any conversion of 13 C6 Tyr to 13 C6 3-Br-Tyr and was minimal. The two bromotyrosine isotopes (m/z 260 for the 79 Brcontaining isotope and m/z 262 for the 81 Brcontaining isotope) occurred at 1:1 ratio and their peak areas were summed for quantification. A typical chromatogram is shown in Figure 2 A. Standard calibration curves using the ratios 12 C Tyr: 13 C6 Tyr and 12 C Br-Tyr: 13 C9Br-Tyr were used for quantification.
Protein samples (50 μg) were hydrolyzed with internal standards (6 nmol 13 C6 Tyr and 6 pmol 13 C9 3-Br-Tyr) in 4 M methane sulfonic acid, supplemented with 1 % (w/v) phenol, under nitrogen at 110°C for 18 hours. The hydrolysate was processed by solid phase extraction to remove the acid and eluted with 80 % (v/v) methanol. Eluents were dried and reconstituted in 0.1 % (v/v) formic acid for analysis. Reverse-phase HPLC-MS/MS was performed using a Kinetex® 2.6 μm C18 100Å column (150x2.1 mm), and an Agilent 1290 Binary Pump at a flow rate of 200 μl/minute. The gradient started with 2 % acetonitrile with 0.1 % (v/v) formic acid for four minutes, increased to 80 % acetonitrile over three minutes, maintained this over three minutes then equilibrated with the starting eluent. The analytes were delivered into a Qtrap® 6500 mass spectrometer (Sciex) and detected in multiple reaction monitoring mode using positive ion mode. The ion spray was set to 5.5kV and the temperature was set to 600°C. The collision gas was nitrogen and the collision energy was 25%. The areas of peaks were calculated using Analyst Software v 1.6.2 (Sciex). A linear response was observed over the range 0.1-10 nmol for tyrosine and 0.1-10 pmol for 3-bromotyrosine.

Data availability statement:
All data are available within the manuscript.  Tables   Table 1     ECM from cells grown in the presence of PHG was isolated and treated with 100 µM Br⁻ and/or 50 µM H2O2 for 1 hour at 37°C. Samples were analyzed as described in Figure 1 and Figure 2. Values are the means of ≥ 3 independent biological repeats ± SD. *, p <0.05, ****, p <0.0001 by one-way ANOVA analysis followed by Tukey's multiple comparisons.  3-Br-Tyr and dimer analyses were performed as described in Figure 1 and Figure 2. Values represent the means of 3 independent biological repeats ± SD. The grey dashed lines represent inhibition by 50 µM PHG. IC50 values were calculated from exponential decay curves in each figure. Error bars are shown in one direction for clarity and are not visible when they were within the symbol size. Curves represent non-linear regression lines selected to give best fit to the experimental data points.