CO C a A a b A R A A K M M C C P n 1 e m e a i o t t w m m g M 1ARTICLE IN PRESSG ModelULHER-3372; No. of Pages 8 Journal of Cultural Heritage xxx (2017) xxx–xxx Available online at ScienceDirect www.sciencedirect.com riginal article haracterization of membrane metal threads by proteomics nd analysis of a 14th c. thread from an Italian textile leksandra K. Popowicha,b, Timothy P. Clelandb, Caroline Solazzob,∗ Department of Chemistry, University of Alberta, 11227 Saskatchewan Drive, Edmonton, Alberta T6G 2G2, Canada Smithsonian’s Museum Conservation Institute, 4210 Silver Hill Road, Suitland, MD 20746, USA a r t i c l e i n f o rticle history: eceived 22 January 2018 ccepted 9 March 2018 vailable online xxx eywords: etal threads embrane ollagen yprus gold roteomics anoLC-Orbitrap MS/MS a b s t r a c t Beginning in the 13th century, membrane metal threads – made out of animal skins (leather, parchment, and vellum) or membranous material (e.g., stomach, intestine) coated with metal – were the most popular variety of decorative metal threads used in European textiles. This work provides the proteomics ground- work for the identification of the species and the type of membrane used in the manufacture of a 14th century membrane gilded thread. A protocol for small sample extraction and nanoLC-Orbitrap MS/MS analysis was first tested on standards of pig peritoneum and cow intestine metal-coated with or without the presence of an egg adhesive. The proteomes of each membrane were characterized and compared by qualitative and quantitative bioinformatics; in addition to the predominant collagen proteins in each membrane type, minor tissue-specific proteins (e.g., smooth muscle proteins from intestine standards) were detected. Species-specific collagen peptides (i.e., from collagen I and collagen III) were confidently identified to determine the species of origin, regardless of the application of metal and egg-based adhe- sives. Likewise, the thin layer of egg adhesive was successfully characterized with the detection of egg white (ovalbumin, ovotransferrin, lysozyme) and egg yolk (vitellogenin I, II, III) proteins. When applied to the thread from a 14th century Italian textile, this comprehensive methodology resulted in the iden- tification of seven collagen I and III peptides specific to cow, as well as other proteins suggesting that the ancient thread was made with intestine or stomach membrane without the use of an egg-based adhesive. © 2018 Elsevier Masson SAS. All rights reserved.. Introduction Decorative metal threads have been extensively used for the mbellishment of textiles since ancient times. Many examples of etal threads exist in artifacts of cultural importance, and even arlier references to lavish gold and silver textiles can be found in ncient texts, including a description of an ephod containing gold n the Old Testament of the Bible (Exodus 39:2–3 “They hammered ut thin sheets of gold and cut strands (. . .)”). The popularity of extiles woven or embroidered with metal threads persisted, and heir use is frequently associated with textiles intended to portray ealth or symbolic importance [1]. Metal threads were most often ade of gold, silver, or their alloys, often gilt, with the fabricationPlease cite this article in press as: A.K. Popowich, et al., Characterizat 14th c. thread from an Italian textile, Journal of Cultural Heritage (201 ethod differing by region and changing over time [2]. Five cate- ories describing the use of metals in textiles have been defined: I. etal applied with adhesive to already woven fabrics, II. Metal wire ∗ Corresponding author. E-mail address: SolazzoC@si.edu (C. Solazzo). https://doi.org/10.1016/j.culher.2018.03.007 296-2074/© 2018 Elsevier Masson SAS. All rights reserved.or flattened strips used directly in weaving, III. Metal wire or strips wound around a fiber core, IV. Metallic surface applied to organic wrappings (cellulosic or proteinaceous) wound around a fiber core, and V. Metallic surface applied to organic strips (cellulosic or pro- teinaceous) without a fiber core [2–6]. Protein metal threads (Categories IV and V) were made from membranous tissues (e.g., stomach or intestinal walls of animals), although skin has also been used as a substrate [2,7]. Research at the end of the 19th century suggested that gilt membranes were made using the intestines from slaughtered animals [8]. Cow intestine was similarly used in the manufacture of gold foil, also called gold- beater’s skin, during the same time period. Metal was applied to the membrane with metal leaves or by mottle gilding, using either the natural exudates of the organic membrane or an additional adhe- sive [7]. Reports have described the use of egg white, egg yolk, animal fat, animal and fish glues, gums, and clays as adhesives but to date, no scientific investigation has been carried to substantiate theion of membrane metal threads by proteomics and analysis of a 7), https://doi.org/10.1016/j.culher.2018.03.007 presence and nature of the adhesives [4,7,9]. In category IV threads, also known as gilt membranes or Cyprus gold, the gilded membrane was cut into thin strips and wound around a silk or linen core [2]. No adhesive was used between the gilded organic wrapping and the IN PRESSG ModelC 2 Cultural Heritage xxx (2017) xxx–xxx fi t a b m L m u a a f c g o h r t i m p u fi e p m [ a t i i h t t a t m d a s r a t t 2 a d [ h a c p m a d V i t P e b m Table 1 Reference standards. Reference Membrane Adhesive Fixation Metal leaf type Pig untreated Peritoneum None None None Pig gilt Peritoneum None None Gilt silver Pig gilt heat Peritoneum None Heat Gilt silver Pig gilt egg white Peritoneum Egg white None Gilt silver Pig gilt egg yolk Peritoneum Egg yolk None Gilt silver Cow untreated Gut None None None Cow silver Gut None None Silver Cow silver heat Gut None Heat Silver Cow silver egg white Gut Egg white None Silver Fig. 1. 14th century Italian textile (#21714-1) from the History Museum in Graz, Austria, ©Universalmuseum Joanneum, Museum für Geschichte, Kulturhistorische Sammlung/Institute of Conservation, University of Applied Arts Vienna/Elisabeth Delvai 2017. The arrow points to where the thread was sampled. In the bottom right,ARTICLEULHER-3372; No. of Pages 8 A.K. Popowich et al. / Journal of ber core; instead, the twist of the metal membrane thread around he core was sufficient to keep the material in place. Migration of dhesive and wrapping material into the core fiber has however een observed [5]. In category V threads the metal was applied to embrane strips and used in textiles without being wound [4]. Membrane metal threads were first used in the 11th century by evantine traders in Cyprus [1]. Beginning in the 13th century, gilt embranes became the most common variety of metal threads sed in European textiles, especially those from Italy and Spain, s well as Western Asia because they were flexible, lightweight, nd inexpensive [7]. Substantial focus has been given to identi- ying the metals used in metal threads, as studies show that the omposition of the metal and gold/silver ratios are suggestive of eographic origin of the metal thread [6,7,10]. Recently, a range f electron microscopy techniques and micro-Raman spectroscopy ave been used to characterize cross-sections of metal threads evealing details of the production technology through study of he 3D texture [11]. Conversely, little attention has been given to dentifying the membrane type or animal species used in organic etal threads, although the elucidation of this information would rovide valuable insight into the origin, production technique, and se of membrane metal threads in the medieval period [1]. Identi- cation of the membrane portion of the substrate poses difficulty, specially when a textile is in a deteriorated state. Using mor- hology on the microscopic scale, differences between leather, embranous material, parchment, and vellum have been made 1,2]; however, the appearance of the substrate can be drastically ltered from its original state because of decomposition, embrit- lement, wear, and treatment damage, making visual identification ndefinite and oftentimes impossible. In contrast to morphological dentification, DNA amplification and molecular biology techniques ave been used to identify the animal species in organic metal hreads found in textiles from the 11–15th centuries [9]. Unfor- unately, this study could only narrow down the identification to few candidate species. Here, we have adopted a bottom-up pro- eomics approach to analyze a 14th c. membrane thread: the entire embrane sample is characterized through a single extraction and igestion of the whole extract, including the membrane proteins nd other binding proteins if present. The complex mixture is eparated by liquid chromatography before analysis by an Orbit- ap Velos mass spectrometer and the data obtained are searched gainst a public database. In the absence of historical informa- ion on the membrane’s animal origin, any species can be targeted hrough a proteomics database search. . Research aims To resolve the identification issue of the organic substrate of metal thread several centuries old, a proteomics approach was evised. Proteomics has been used successfully to identify binders 12] and collagen-based substrates such as parchment [13], but as never been used to characterize membrane metal threads. The im of this project is to show that a complex organic substrate, omposed of a membrane base and protein binders, can be com- rehensively characterized, including the type and species of the embrane and presence of proteinaceous adhesives. To test the pplicability of proteomics to ancient samples, two series of stan- ards (Table 1) were prepared at the University of Applied Arts ienna (Austria) from the abdominal membrane of pig and from ntestine of cow, then subjected to different treatments (heat fixa- ion, egg white, or egg yolk used as adhesives for the metal coating).Please cite this article in press as: A.K. Popowich, et al., Characterizat 14th c. thread from an Italian textile, Journal of Cultural Heritage (201 ig and cow were chosen as species for their commonality and asy availability, and as probable source obtained in the past from utchered animals. The proteome of each membrane was deter- ined by extracting proteins in samples of less than one milligram,the image of the thread was acquired with HIROX KH-8700 3D digital microscope (Hirox-USA, Inc., NJ), courtesy of Thomas Lam (Smithsonian’s Museum Conservation Institute). Software: PowerPoint 2013. thus revealing tissue-specific proteins, while peptides specific to the membrane species and the egg adhesive were characterized in untreated and treated samples. Finally, the developed protocol was applied to a membrane metal thread from a 14th century Ital- ian textile (Fig. 1), thought to be made from an animal’s internal organ and not skin. 3. Materials and methods 3.1. Standards The gilt membrane reference standards were prepared at the Universität für angewandte Kunst (University of Applied Arts Vienna, Austria) based on what is known of traditional protocols [14,15]. Pig peritoneum, which is the membrane sack containing the abdominal organs, was obtained from the Veterinary Univer- sity of Vienna (Vienna, Austria). Cattle gut was obtained from a local butcher. The fat was scraped off the membrane and the membrane washed and rinsed, then stretched and pinned. Gilt silver leaves or silver leaves were immediately applied on the damp membrane using egg yolk, egg white, or no adhesive (Fig. S1). The standardsion of membrane metal threads by proteomics and analysis of a 7), https://doi.org/10.1016/j.culher.2018.03.007 were allowed to dry overnight. Select standards were subjected to heat fixation with a heated spatula at approximately 80 ◦C. Refer- ence standards are listed in Table 1. ING ModelC Cultur 3 H t p b p b [ t 3 h ( ( f ( c I Q A 3 a p E b t a f m r T 4 b U o t a w 2 d c t s e w a a 3 w l ( h s 3 BARTICLEULHER-3372; No. of Pages 8 A.K. Popowich et al. / Journal of .2. Historical sample The original sample (#21714-1) comes from the collection of the istory Museum at the Universalmuseum Joanneum (Graz, Aus- ria). The textile, a silk fabric with metal fibers woven into it, was roduced in Italy in the 14th century, depicts dogs or cats alongside irds. The sample was taken from a large piece of fabric that is well reserved. Scanning electron microscopy showed that the mem- rane was coated with gilt silver, and wound around bast fibers 16], although most of the metal has worn away from the sampled hread, as seen in Fig. 1. .3. Reagents Iodoacetamide (IAM) and tris(hydroxymethyl)aminomethane ydrochloride (TRIS HCl) were obtained from Sigma–Aldrich St. Louis, MO); guanidine hydrochloride (GuHCl), methanol MeOH), acetonitrile (ACN), and Optima grade formic acid rom Fisher Scientific (Fair Lawn, NJ); ammonium bicarbonate ABC) from VWR International LLC (West Chester, PA); tris(2- arboxyethyl)phosphine (TCEP) from Thermo Scientific (Rockford, L). Sequence grade modified trypsin from Promega (Madison, WI). uantitation of protein was determined with the Pierce BCA Protein ssay Kit from Thermo Scientific (Rockford, IL). .4. Sample preparation Standards were cut with a razor in squares of less than 1 mg nd were solubilized in 600 L of 4 M GuHCl and 50 mM Tris HCl at H 8, using five cycles of 30 s homogenization with a Bead Ruptor lite (Omni International) at 7 m/s, with 5 min dwell time at −20 ◦C etween each cycle and following the final cycle. Protein extrac- ion was allowed to continue by shaking the samples overnight t room temperature. All samples were centrifuged at 10,000 rpm or 20 min and the protein concentration of the supernatant was easured by the Pierce BCA Protein Assay. The supernatant was educed with 500 mM TCEP for a final concentration of 50 mM CEP and pH adjusted to 8.0. A 250 L aliquot was alkylated with 0 mM IAM and vortexed for 45 min in the dark. This was followed y buffer exchange with 100 mM ABC using 3000 MWCO Amicon ltra-0.5 mL centrifugal filters. Samples were digested with 0.5 g f trypsin (trypsin:protein ratio 1:100), overnight at 37 ◦C. For pep- ide cleanup, the samples were loaded on a C18 stage tip made from n Empore SPE Extraction Disk (3 M) [17]. The tips were washed ith 0.1% FA solution. Peptide mixtures were then eluted in 2× 0 L of 80:20 (v/v) acetonitrile:0.1% FA. All samples were then ried down to a final volume of 10 L. Soluble proteins were extracted from 100 g of an ancient 14th entury thread sample, with the fiber core already removed, using he above described protocol, with the entire 600 L extraction olution being reduced and alkylated. In addition to the soluble xtracts, the insoluble fraction that was pelleted by centrifugation as retained and diluted in 100 mM ABC, after which both soluble nd insoluble fractions underwent the sample preparation steps as bove. .5. Protein analysis by nanoLC-Orbitrap MS/MS The desalted samples were diluted and 0.5–2 g of total protein as injected and analyzed by LC-MS/MS. The peptides were first oaded onto an in-house packed Thermo BioBasic C18 precolumn 30 mm × 75 m i.d.) after which they were separated on an in-Please cite this article in press as: A.K. Popowich, et al., Characterizat 14th c. thread from an Italian textile, Journal of Cultural Heritage (201 ouse packed analytical column (210 mm × 75 m i.d.) made of the ame stationary phase, using a Thermo Scientific Dionex Ultimate 000 UHPLC system with the following gradient: 2% B 0–8 min, 55% 98 min, 90% B 100–103 min, 2% B 104–120 min, where buffer A PRESS al Heritage xxx (2017) xxx–xxx 3 is 0.1% FA in H2O and buffer B is 0.1% FA in ACN. The UHPLC was directly coupled to a Thermo Scientific LTQ Velos Dual Pressure Linear Ion Trap mass spectrometer which analyzed the peptides in positive mode using the following parameters: MS1 60,000 resolu- tion, 100 ms acquisition time, 1 × 106 automatic gain control (AGC), MS2 15,000 resolution, 250 ms acquisition time, 5 × 105 AGC, top 8, 30 normalized collision energy (NCE) higher-energy collisional dissociation (HCD). 3.6. Bioinformatics analysis PEAKS 8.0 (Bioinformatics Solutions Inc.) was used to search the RAW data for matches against the UniProt database (www.uniprot.org) of publicly available sequences. The mam- malian database was imported on June 15th, 2017, while a customized database was created containing the complete Uniprot databases of Sus scrofa and Bos taurus to which were added Gallus gallus egg proteins (Timestamp 14th July 2017). Searches were car- ried out using trypsin as enzyme, three missed cleavages, peptide mass tolerance (PMS) of 10 ppm, fragment mass error tolerance (MS/MS) of 0.02 Da, carbamidomethylation as a fixed modification, and the following variable modifications as configured in PEAKS: deamidation (NQ), hydroxylation/oxidation (RYFPNKD), and oxi- dation (M). PEAKS PTM and SPIDER were both enabled to identify unspecific PTMs. Results were filtered using an FDR of less than or equal to 1% for peptide spectrum matches, a protein score of −10lgP ≥ 20, and 1 unique peptide. Quantitative bioinformatics analysis was performed using Pro- teome Discoverer 2.2 (Thermo Fisher). Searches were carried out by MS Amanda 2.0, using the same search parameters used in PEAKS. Protein quantification was performed in Proteome Discoverer 2.2, based on summed abundance of unique plus razor peptides, with data normalized to collagen I alpha 1 (COL1A1). Collagen peptides from pig and cow standards identified as “unique” to a protein group by PEAKS 8.0 software, when searching against the Mammalia protein database, were fur- ther investigated using the Basic Local Alignment Search Tool (BLAST) available online from both Uniprot (http://www.uniprot. org/blast searched against UniProtKB protein database) and the National Center for Biotechnology Information NCBI (https://blast.ncbi.nlm.nih.gov/Blast.cgi searched against NCBI Mammalia database). 4. Results and discussion 4.1. Membrane characterization The proteomes of peritoneum and intestine were examined to evaluate their differential composition, and were further analyzed using quantitative bioinformatics to search for protein markers that could potentially specify the membrane type. The peritoneum is a continuous membrane that lines the abdominal wall (parietal peri- toneum) and covers the abdominal organs (visceral peritoneum); it is composed of a thin layer of mesothelium, which is made of a monolayer of squamous epithelial cells on a basement mem- brane, supported by a layer of connective tissue [18]. The basement membrane is composed of a specialized layer of extracellular matrix protein, with type IV collagen being the most abundant protein [19–22]. Other predominant protein constituents in base- ment membrane include laminin, nidogen (NID), and perlecan (HSPG) [18,23,24]. Although intestinal tissue also contains a base-ion of membrane metal threads by proteomics and analysis of a 7), https://doi.org/10.1016/j.culher.2018.03.007 ment membrane, in contrast to peritoneum, intestine has a layer of muscle tissue, specifically smooth muscle tissue. Smooth mus- cle proteins include calponin-1 (CNN1) and myosin-11 (MYH11) [19,25–27]. ING ModelC 4 Cultur s c p f m c m p l ( l t w d H r w m d r g 1 f O F b C 2 s 6 g EARTICLEULHER-3372; No. of Pages 8 A.K. Popowich et al. / Journal of Analysis of the peritoneum (Table S1) and intestine (Table S2) tandards showed that there was minimal difference between the ollagen proteins, other extracellular proteins, blood and serum roteins, and cellular protein content of these two membranes (see ull results in SI files S1 and S2). As expected, multiple smooth uscle proteins were found in the intestine samples only, which orrespond to the muscular layer present in intestine. Basement embrane specific proteins were, interestingly, not detected in eritoneum samples, but very low levels of NID1, HSPG, and col- agen IV alpha 1 (COL4A1) were observed in the intestine samples Table S2). These proteins are difficult to extract because of their ow solubilities [22,28]. The collagen IV network for instance con- ains disulfide bonds making it highly insoluble and extraction ith reducing agents, such as TCEP used here, is required to break isulfide bonds and can somewhat improve extractability [29–31]. owever, here the top layer of the membrane was scraped to emove fat during preparation and most likely the mesothelial layer as removed alongside the fat. The basement membrane is also uch thinner than the connective or muscle layers, so lower abun- ance of these proteins is expected. The abundances of collagen and muscle proteins, normalized elative to collagen I alpha 1 (COL1A1), are shown in Fig. 2 using the ene names to allow for interspecies comparison. Collagen III alphaPlease cite this article in press as: A.K. Popowich, et al., Characterizat 14th c. thread from an Italian textile, Journal of Cultural Heritage (201 (COL3A1), and to a lesser extent collagen I alpha 2 (COL1A2), were ound in higher abundance in the intestine than the peritoneum. f all the muscle proteins identified in the intestine, actin, desmin, ig. 2. Normalized relative abundance of collagen and muscle proteins (indicated y their gene codes) in peritoneum, intestine, and a 14th century membrane thread. orresponding protein names: COL1A1, collagen I alpha 1; COL1A2, collagen I alpha ; COL3A1, collagen III alpha 1; COL4A1, collagen IV alpha 1; ACTA2, actin, aortic mooth muscle; CNN1, calponin-1; DES, desmin; MYL6, myosin light polypeptide ; MYL9, myosin regulatory light polypeptide 9; MYH11, myosin 11; TAGLN, trans- elin; TPM1, tropomyosin alpha 1 chain; TPM2, tropomyosin beta chain. Software: xcel2013 and PowerPoint 2013. PRESS al Heritage xxx (2017) xxx–xxx and myosin-11 were found with the highest abundances in all cow standards and as such represent useful markers for membrane dif- ferentiation. 4.2. Species identification The long-term survival of collagen proteins makes them excellent markers for species identification in ancient artifacts. Previously, collagen peptide markers have been identified by pep- tide mass fingerprinting for species identification of bones [32] and skin substrates [13,33]. In the analysis of archeological animal glue, often these proteins are the only proteins detected in suffi- cient abundance to allow species identification [34–36]. In artworks where animal glue was commonly used as binder in paintings and frescoes, collagen peptides specific to bovine have previously been identified in an 18th century gilt sample [37]. As shown in Fig. 2, COL1A1 and COL1A2 were the most abundant collagen chains found in the intestine and abdominal membranes, followed by COL3A1. Several more collagen proteins (types IV, V, VI, and XIV) were identified from the LC–MS/MS analysis of the untreated (adhesive and metal free) pig and cow standards (Tables S1 and S2), but with much lower coverages. Having further estab- lished the abundance of collagen I and III chains in the membrane standards, species identification was focused on these proteins: the species markers identified in the standard membranes were also used as a proxy to determine whether the collagen chains, and by consequence species identification, would be affected by the manufacture processes. Distinctive markers for pig and cow were identified by searching against the Uniprot Mammalia protein database in the untreated pig (SI file S1) and cow (SI file S2) membrane samples, and by select- ing peptides uniquely found in the proteins and species of interest (COL1A1, COL1A2, and COL3A1 of Sus scrofa and Bos taurus). The selected peptides were further identified using BLAST (Uniprot and NCBI) as being specific to the Sus scrofa species (Table 2 for pig sam- ples) or the bovid clade (Table 3 for cow samples) that includes the Bos, Bubalus, and Bison genera. There are indeed no peptide spe- cific to the Bos taurus species in the targeted proteins. All collagen peptides are shown in SI File S3. No COL1A1 specific peptides were found in either pig or cow: unsurprisingly, cow COL1A1 shares 97% of its sequence with that of pig, for instance. COL1A2 and COL3A1 have greater degrees of sequence variation between animal species. Six COL1A2 and nine COL3A1 peptides were identified in the untreated pig sample (corresponding to respectively five and six different sequence seg- ments, Table 2), and six COL1A2 and three COL3A1 peptides were identified in the untreated cow sample (corresponding to respec- tively four and one different sequence segments, Table 3), after manual inspection of the data in Peaks. With 15 and 9 peptides identified in the collagen chains for pig and cow respectively, species markers were successfully identified in both membrane types. The addition of the gilt or silver layer did not significantly affect the detection of the marker peptides and no degradation on the protein chains was observed because of the metal. Metal adducts on peptides were searched for but none were found; the metal seems to wear off easily (a phenomenon also observed on ancient textiles) and was completely filtered after sol- ubilization of the proteins. Because of the presence of the egg yolk binder, some of the collagen peptides in the pig were not detected (five markers were not identified at all in the pig sample with egg yolk). A similar result was observed for the egg white in both theion of membrane metal threads by proteomics and analysis of a 7), https://doi.org/10.1016/j.culher.2018.03.007 pig and cow, but to a lesser extent. Overall, there was no dramatic loss of collagen peptides after treatment, suggesting that species identification should be possible in ancient samples using collagen peptides only. ARTICLE IN PRESSG ModelCULHER-3372; No. of Pages 8 A.K. Popowich et al. / Journal of Cultural Heritage xxx (2017) xxx–xxx 5 Table 2 Specific collagen peptides identified for the Sus scrofa (pig) standards, with peaks scores given in −10lgP. Protein Peptide sequence Position Standards Untreated Gilt Gilt heat Gilt egg white Gilt egg yolk GNDGSVGPVGPAGPIGSAGPPGFPGAPGPK 235–264 +2 Hpro 53.33 29.74 62.93 34.58 + 3 Hpro 65.91 45.91 68.18 26.13 GPTGPAGVR 424–432 32.71 31.52 33.19 31.13 33.97 COL1A2: A0A1S7J1Y9 PIG GPTGDPGKNGEKGHAGLAGAR 499–519 + 1 Hpro + 1 deam 52.20 52.83 + 2 Hpro + 1 deam 54.85 44.77 58.02 29.46 48.44 IGPPGPSGISGPPGPPGPAGK 795–815 Sus scrofa domesticus + 3 Hpro 55.28 61.94 57.17 60.78 58.03 + 4 Hpro 32.49 34.73 33.84 29.58 28.41 + 5 Hpro 29.41 32.28 21.26 GenBank: BAX02569.1 GYPGNPGPAGAAGAPGPQGAVGPAGK 949–974 + 2 Hpro 63.31 74.20 65.44 70.27 76.47 + 3 Hpro 65.50 76.37 67.61 82.88 75.54 PGPAGAAGAPGPQGAVGPAGK 954–974 + 1 Hpro 57.89 48.23 51.82 59.98 + 2 Hpro 61.39 67.05 60.37 63.22 61.06 GEVGPAGSPGPSGSPGQR 352–369 + 2 Hpro 51.76 57.04 54.05 GVAGEPGRDGVPGGPGLR 520–537 + 2 Hpro 40.98 42.81 39.54 39.81 40.43 + 3 Hpro 40.82 42.65 43.86 40.22 43.83 GDSGAPGERGPPGAVGPSGPR 676–696 + 2 Hpro 42.37 51.10 50.62 41.45 47.46 GPPGAVGPSGPR 685–696 + 1 Hpro 39.00 39.76 37.97 41.53 + 2 Hpro 25.83 41.45 COL3A1: F1RYI8 PIG GAPGEKGEGGPPGIAGQPGGTGPPGPPGPQGVK 829–861 + 4 Hpro 32.24 20.28 40.95 Sus scrofa + 5 Hpro 17.36 32.64 18.25 + 6 Hpro 27.57 19.91 31.68 RefSeq: NP 001230226.1 GEGGPPGIAGQPGGTGPPGPPGPQGVKGER 835–864 + 4 Hpro 35.04 33.69 + 5 Hpro 25.78 GEGGPPGIAGQPGGTGPPGPPGPQGVK 835–861 + 3 Hpro 62.51 64.77 64.68 + 4 Hpro 34.83 33.35 39.03 26.88 GSPGPQGPPGAPGPGGISGITGAR 934–957 + 2 Hpro 65.66 63.52 56.79 68.32 + 3 Hpro 58.40 65.56 60.57 39.16 56.83 NGDRGETGPAGPAGAPGPAGSR 1062–1083 4 t p I b ( y s u r e a t t e ( r a d w t+ 1 Hpro + 1 deam Total # peptides .3. Characterization of the egg adhesive Egg white is composed of the proteins ovalbumin (50%), ovo- ranferrin (12–13%) and lysozyme (3%) [38,39], while egg yolk’s rimary protein components are the proteins vitellogenins (I, II, and II), with vitellogenin II being the most abundant [40]. The results of oth pig and cow standards containing egg white or yolk adhesive Table S3) identified the corresponding egg white proteins and egg olk proteins. Overall, proteomic analysis of the standards demon- trated that the presence of vitellogenins in a sample indicates the se of egg yolk, whereas the presence of ovalbumin, ovotransfer- in, and lysozyme, without any vitellogenins, indicates the use of gg white alone. Ovomucoid (11% of total egg white protein) was lso detected with 61% coverage and 16 peptides in the cow intes- ine egg white standard, but with only 3% coverage and 1 peptide in he pig peritoneum egg white standard. In addition, standards with gg yolk adhesive contained small quantities of egg white proteins SI Files S1 and S2 and Table S3 for quantitative data), most likely esulting from incomplete separation during the preparation of thePlease cite this article in press as: A.K. Popowich, et al., Characterizat 14th c. thread from an Italian textile, Journal of Cultural Heritage (201 dhesive. The capabilities of quantitative bioinformatics analysis to istinguish between yolk and whole egg based on the levels of egg hite proteins has yet to be examined. In addition to detecting hese characteristic egg proteins, the egg containing standards did46.17 29.06 55.25 15.67 56.69 25 22 25 21 16 not experience any reduction in collagen or tissue protein signal (Fig. 2). 4.4. Analysis of a 14th century membrane metal thread The proteomics methodology was applied to the analysis of a membrane metal thread from a 14th century Italian textile and the results searched against the entire Mammalia protein database (SI File S4). Several proteins specific to smooth muscle tissue were detected in the ancient sample, including MYH11 and desmin (Tables S4 and S5). The normalized relative abundances of colla- gen and muscle proteins are shown in Fig. 2. Overall, the presence of multiple proteins found only in muscle tissue suggest that the membrane may be from intestine or stomach, both of which have a muscular layer and are known to have been commonly used in medieval gilt membranes [2,7]. COL1A1, COL1A2, and COLIIIA1 were detected with the high- est number of peptides from Bos taurus (45, 42, and 38 peptides, respectively). In spite of an average loss in protein coverage ofion of membrane metal threads by proteomics and analysis of a 7), https://doi.org/10.1016/j.culher.2018.03.007 16% on the main three collagen chains, most bovine markers identified in the reference samples were observed in the ancient sample: seven out of nine of these peptides (e.g., COL1A2 pep- tide GAP*GAIGAP*GPANGDR [Fig. 3]) were detected in the soluble ARTICLE IN PRESSG ModelCULHER-3372; No. of Pages 8 6 A.K. Popowich et al. / Journal of Cultural Heritage xxx (2017) xxx–xxx Table 3 Specific collagen peptides identified for the Bos Taurus (cow) standards, with peaks scores given in −10lgP. Protein Peptide sequence Position Standards 14th c. thread Sol. 14th c. thread Ins. Untreated Silver Silver heat Silver egg white GAPGAIGAPGPAGANGDRa,b,c 674–691 + 2 Hpro + 1 deam 54.48 58.18 55.47 53.06 44.55 16.37 GAPGAIGAPGPAGANGDRGEAGPAGPAGPAGPRa,b,c 674–706 + 2 Hpro + 1 deam 63.42 62.63 58.48 68.55 33.75 72.00 COL1A2: CO1A2 BOVIN SGETGASGPPGFVGEKa,b,c 829–844 + 1 Hpro 48.91 46.16 45.48 Bos Taurus GYPGNAGPVGAAGAPGPQGPVGPVGKa,b,c 947–972 + 2 Hpro 78.21 78.88 72.41 80.92 59.44 47.41 RefSeq: NP 776945.1 + 2 Hpro + 1 deam 51.76 AGPVGAAGAPGPQGPVGPVGKa,b,d 952–972 + 1 Hpro 61.47 43.40 48.78 IGQPGAVGPAGIRa,b,c ,d 1066–1078 39.13 + 1 deam 36.14 33.59 + 1 Hpro 42.92 42.94 38.72 41.03 40.43 37.86 + 1 Hpro + 1 deam 32.48 34.05 + 2 Hpro 32.15 COL3A1: Q08E14 BOVIN GAPGEKGEGGPPGAAGPAGGSGPAGPPGPQGVKa,b,c ,d ,e 828–860 + 3 Hpro 80.88 75.98 45.82 68.08 61.29 + 4 Hpro 38.30 36.80 Bos Taurus GEGGPPGAAGPAGGSGPAGPPGPQGVKa,b,c ,d ,e 834–860 + 2 Hpro 86.72 84.30 59.22 48.45 39.79 RefSeq: NP 001070299.1 GEGGPPGAAGPAGGSGPAGPPGPQGVKGERa,b,c ,d ,e 834–863 + 3 Hpro 79.56 81.63 46.83 64.53 39.64 Total # peptides 10 8 11 8 10 7 Sol.: soluble fraction; Ins.: insoluble fraction. a Bos taurus. b Bos mutus. f t m 1 o r w w w 5 fi m m 1 2c Bos indicus. d Bubalus bubalis. e Bison bison bison. raction and five in the insoluble fraction, as shown in Table 3. Pep- ide IGQPGAVGPAGIR, frequently used to identify bovine by peptide ass fingerprinting because of its two characteristic peaks at m/z 192 (no hydroxyproline) and m/z 1208 (one hydroxyproline), was bserved in the historic sample with deamidation on the glutamine esidue Q. This modification, commonly found in ancient proteins, as not observed in the standard samples. Finally, no egg proteins ere detected, suggesting that the membrane was likely prepared ithout the use of an egg-based adhesive. . Conclusions The analysis of a 14th century thread demonstrates for the rst time the capabilities of proteomics for the analysis of ancient embrane metal threads, despite small sample sizes and complex atrices. In addition, the study was successful in: - Finding protein markers to differentiate the type of membrane made from an internal organ. The possibility that animal skins (leather, parchment) were used for metal threads will be fur- ther explored, and in particular the susceptibility of each type of membrane to degradation. - Identifying collagen species markers independently of the pres- ence of binders and metal. While bovine was clearly identified in the ancient sample, there is, to our knowledge, little infor-Please cite this article in press as: A.K. Popowich, et al., Characterizat 14th c. thread from an Italian textile, Journal of Cultural Heritage (201 mation regarding the species used in membrane threads. The methods of fabrication and species used are likely to vary by geographic origin or for other reasons such as stylistic, qual- ity of materials, cost, etc. Knowledge of the species could haveimportant implications for determining the provenance of the threads. 3- Identifying egg-based adhesives alongside the membrane pro- teins. The complete absence of egg peptides in the ancient sample indicates that such a binder was likely not used. A collagen-based glue, whose proteins would be undistinguish- able from the membrane collagens, would require careful separation during sample preparation. Although beyond the scope of this study, the present work will need to be broadened to the analysis of different combinations of membrane + protein binders to best characterize membrane metal threads in ancient textiles. The proteomics study of membrane metal threads represents a new application to the field of cultural heritage. The expertise and care necessary to make textiles with these threads, as well as the high cost of the materials used (e.g., gold, silk) make them some of the most valuable items of ancient textiles. The implementation of proteomics to determine the choice of membrane type, animal species and use of protein binder, in conjunction with the analysis of the metal layer, will result in a more thorough understanding of the origin and technology of these metal threads. Fundingion of membrane metal threads by proteomics and analysis of a 7), https://doi.org/10.1016/j.culher.2018.03.007 Aleksandra K. Popowich was supported by an Alberta Smith- sonian Internship Program Scholarship from the Government of Alberta. ARTICLE IN PRESSG ModelCULHER-3372; No. of Pages 8 A.K. Popowich et al. / Journal of Cultural Heritage xxx (2017) xxx–xxx 7 F m the d D A d t G D o C f o M A f j R [ [ [ [ [ [ [ [ [ [ [ [ [ [ [ [ [ig. 3. Spectrum of the unique cow COL1A2 peptide (GAP*GAIGAP*GPANGDR) fro eamidated asparagine is underlined. Software: PowerPoint 2013. isclosure of interest The authors declare that they have no competing interest. cknowledgments We particularly thank Elisabeth Delvai who provided samples uring completion of her master degree at the Institute of Conserva- ion, University of Applied Arts Vienna, as well as the Museum für eschichte, Universalmuseum Joanneum. The authors also thank r. Bernhard Pichler, head of Archaeometry Department, University f Applied Arts Vienna, Dr. Robert Koestler, director of the Museum onservation Institute and Mary Ballard, textile conservator at MCI or their help in coordinating the project and valuable discussions n metal threads. Finally Dr. Thomas Lam and Asher G. Newsome, CI, are thanked for their technical support. ppendix A. Supplementary data Supplementary material related to this article can be ound, in the online version, at http://dx.doi.org/10.1016/ .culher.2018.03.007. eferences [1] M. Járó, Gold embroidery and fabrics in Europe: XI–XIV centuries, Gold Bull. 23 (1990) 40–57. [2] N. Indictor, The Use of Metal in Historic Textiles (Personal Notes), New York, 1987. [3] N. Indictor, R.J. Koestler, The identification and characterization of metal wrappings in historic textiles using microscopy and energy dispersive X-ray spectrometry: problems associated with identification and characterization, Scan. Electron Microsc. 2 (1986) 491–497. [4] N. Indictor, M. Ballard, The effects of aging on textiles that contain metal: impli- cations for analyses, in: International Restorer Seminar Veszprem, Hungary, 1989, pp. 67–75. [5] N. Indictor, R.J. Koestler, M. Wypyski, A.E. Wardwell, Metal threads made of proteinaceous substrates examined by scanning electron microscopy: energy dispersive X-ray spectrometry, Stud. Conserv. 34 (1989) 171–182. [6] N. Indictor, R.J. Koestler, C. Blair, A.E. Wardwell, The evaluation of metal wrappings from Medieval textiles using scanning electron microscopy–energy dispersive X-ray spectrometry, Text. Hist. 19 (1988) 3–22. [7] M. Braun-Ronsdorf, Gold and silver fabrics from Medieval to modern times, CIBA Rev. (1961) 2–16.Please cite this article in press as: A.K. Popowich, et al., Characterizat 14th c. thread from an Italian textile, Journal of Cultural Heritage (201 [8] F. Bock, Goldstickereien und Webereien in alter und neuer Zeit und das dazu verwandte Goldgespinnst, G.P.J. Bieling, Nürnberg, 1884. [9] D. De Reyer, A.Y. Jeantet, S. Pilbout, A. Anglo, M. Monnerot, Les lamelles des fils métalliques organiques dans les textiles médiévaux: approche méthodologique de leur origine biologique, Stud. Conserv. 47 (2002) 122–133. [ [ 14th century membrane metal thread. Hydroxylated proline is denoted by * and 10] E. Hoke, I. Petrascheck-Heim, Microprobe analysis of gilded silver threads from Mediaeval textiles, Stud. Conserv. 22 (1977) 49–62. 11] T.G. Weiszburg, K. Gherdán, K. Ratter, N. Zajzon, Z. Bendo˝, G. Radnóczi, Á. Takács, T. Váczi, G. Varga, G. Szakmány, Medieval gilding technology of historical metal threads revealed by electron optical and micro-raman spectroscopic study of focused ion beam-milled cross sections, Anal. Chem. (2017). 12] C. Tokarski, E. Martin, C. Rolando, C. Cile Cren-Olivé, Identification of proteins in renaissance paintings by proteomics, Anal. Chem. 78 (2006) 1494–1502. 13] S. Fiddyment, B. Holsinger, C. Ruzzier, A. Devine, A. Binois, U. Albarella, R. Fischer, E. Nichols, A. Curtis, E. Cheese, M.D. Teasdale, C. Checkley-Scott, S.J. Milner, K.M. Rudy, E.J. Johnson, J. Vnoucˇek, M. Garrison, S. McGrory, D.G. Bradley, M.J. Collins, Animal origin of 13th-century uterine vellum revealed using noninvasive peptide fingerprinting, Proc. Natl. Acad. Sci. U.S.A. 112 (2015) 15066–15071. 14] M. Jaro, A. Toth, E. Gondar, Determination of the manufacturing technique of a 10th century metal thread, in: K. Grimstad (Ed.), ICOM Committee for Con- servation, 9th Triennial Meeting, ICOM Committee for Conservation, Dresden, German Democratic Republic, 1990, pp. 299–301. 15] M. Járó, A. Tóth, Scientific identification of European metal thread manufactur- ing techniques of the 17–19th centuries, Endeavour 15 (1991) 175–184. 16] E. Delvai, Medieval fabric fragments from the collection of the Universalmu- seum Joanneum in Graz, Austria, Technological Analysis and Conservation, Institute of Conservation, University of Applied Arts Vienna, Vienna, 2017. 17] J. Rappsilber, M. Mann, Y. Ishihama, Protocol for micro-purification, enrich- ment, pre-fractionation and storage of peptides for proteomics using StageTips, Nat. Protoc. 2 (2007) 1896–1906. 18] V.S. LeBleu, B. MacDonald, R. Kalluri, Structure and function of basement mem- branes, Exp. Biol. Med. 232 (2007) 1121–1129. 19] M.J. Randles, M.J. Humphries, R. Lennon, Proteomic definitions of basement membrane composition in health and disease, Matrix Biol. 57–58 (2017) 12–28. 20] E. Poschl, U. Schlotzer-Schrehardt, B. Brachvogel, K. Saito, Y. Ninomiya, U. Mayer, Collagen IV is essential for basement membrane stability but dispens- able for initiation of its assembly during early development, Development 131 (2004) 1619–1628. 21] R. Timpl, H. Wiedemann, V. Van Delden, H. Furthmayr, K. Kuhn, A network model for the organization of type IV collagen molecules in basement mem- branes, Eur. J. Biochem. 120 (1981) 203–211. 22] K. Kuhn, Basement membrane (Type IV) collagen, Matrix Biol. 14 (1994) 439–445. 23] J.W. Fox, U. Mayer, R. Nischt, M. Aumailley, D. Reinhardt, H. Wiedemann, K. Mann, R. Timpl, T. Krieg, J. Engel, M.-L. Chu, Recombinant nidogen consists of three globular domains and mediates binding of laminin to collagen type IV, EMBO J. 10 (1991) 3137–3146. 24] H. Sakai, S. Takeoka, H. Yokohama, Y. Seino, H. Nishide, E. Tsuchida, Purification of concentrated hemoglobin using organic solvent and heat treatment, Protein Expr. Purif. 4 (1993) 563–569. 25] S.J. Winder, M.P. Walsh, Calponin: thin filament-linked regulation of smooth muscle contraction, Cell. Signal. 5 (1993) 677–686. 26] G.J. Babu, D.M. Warshaw, M. Periasamy, Smooth muscle myosin heavy chain isoforms and their role in muscle physiology, Microsc. Res. Technol. 50 (2000) 532–540.ion of membrane metal threads by proteomics and analysis of a 7), https://doi.org/10.1016/j.culher.2018.03.007 27] S.J. Winder, B.G. Allen, O. Clement-Chomienne, M.P. Walsh, Regulation of smooth muscle actin–myosin interaction and force by calponin, Acta. Physiol. Scand. 164 (1998) 415–426. 28] R. Timpl, J.C. Brown, Supramolecular assembly of basement membranes, Bioes- says 18 (1996) 123–132. ING ModelC 8 Cultur [ [ [ [ [ [ [ [ [ [ARTICLEULHER-3372; No. of Pages 8 A.K. Popowich et al. / Journal of 29] R. Timpl, G.R. Martin, P. Bruckner, G. Wick, H. Wiedemann, Nature of the col- lagenous protein in a tumor basemement membrane, Eur. J. Biochcm. 84 (1978) 43–52. 30] H.K. Kleinman, M.L. McGarvey, L.A. Liotta, P. Gehron Robey, K. Tryggvason, G.R. Martin, Isolation and characterization of type IV procollagen, laminin, and heparan sulfate proteoglycan from the EHS sarcoma, Biochemistry 21 (1982) 6188–6193. 31] M.W. Karakashian, P. Dehm, T.S. Gramling, E.C. Leroy, Precursor-size compo- nents are the basic collagenous subunits of murine tumor basement membrane, Coll. Relat. Res. 2 (1982) 3–17. 32] M. Buckley, M. Collins, J. Thomas-Oates, J.C. Wilson, Species identification by analysis of bone collagen using matrix-assisted laser desorption/ionisation time-of-flight mass spectrometry, Rapid Commun. Mass Spectrom. 23 (2009)Please cite this article in press as: A.K. Popowich, et al., Characterizat 14th c. thread from an Italian textile, Journal of Cultural Heritage (201 3843–3854. 33] D.P. Kirby, M. Buckley, E. Promise, S.A. Trauger, T.R. Holdcraft, Identification of collagen-based materials in cultural heritage, Analyst 138 (2013) 4849–4858. 34] N. Bleicher, C. Kelstrup, J.V. Olsen, E. Cappellini, Molecular evidence of use of hide glue in 4th millennium BC Europe, J. Archaeol. Sci. 63 (2015) 65–71. [ [ PRESS al Heritage xxx (2017) xxx–xxx 35] H. Rao, Y. Yang, I. Abuduresule, W. Li, X. Hu, C. Wang, Proteomic identification of adhesive on a bone sculpture-inlaid wooden artifact from the Xiaohe Cemetery, Xinjiang, China, J. Archaeol. Sci. 53 (2015) 148–155. 36] C. Solazzo, B. Courel, J. Connan, B.E. van Dongen, H. Barden, K. Penkman, S. Taylor, B. Demarchi, P. Adam, P. Schaeffer, A. Nissenbaum, O. Bar- Yosef, M. Buckley, Identification of the Earliest Collagen- and Plant-Based Coatings from Neolithic Artefacts, vol. 6, Nahal Hemar Cave, Israel, 2016, pp. 31053. 37] S. Dallongeville, M. Koperska, N. Garnier, G. Reille-Taillefert, C. Rolando, C. Tokarski, Identification of animal glue species in artworks using pro- teomics: application to a 18th century gilt sample, Anal. Chem. 83 (2011) 9431–9437. 38] H. Sugino, T. Nitoda, L.R. Juneja, Hen Eggs: Basic and Applied Science, CRC Press,ion of membrane metal threads by proteomics and analysis of a 7), https://doi.org/10.1016/j.culher.2018.03.007 Boca Raton, FL, USA, 1997. 39] J. Wu, A. Acero-Lopez, Ovotransferrin: structure, bioactivities, and preparation, Food Res. Int. 46 (2012) 480–487. 40] K. Mann, M. Mann, The chicken egg yolk plasma and granule proteomes, Pro- teomics 8 (2008) 178–191.