A combined physical–chemical and microbiological approach to unveil the fabrication, provenance, and state of conservation of the Kinkarakawa-gami art

Kinkarakawa-gami wallpapers are unique works of art produced in Japan between 1870 and 1905 and exported in European countries, although only few examples are nowadays present in Europe. So far, neither the wallpapers nor the composing materials have been characterised, limiting the effective conservation–restoration of these artefacts accounting also for the potential deteriogen effects of microorganisms populating them. In the present study, four Kinkarakawa-gami wallpapers were analysed combining physical–chemical and microbiological approaches to obtain information regarding the artefacts’ manufacture, composition, dating, and their microbial community. The validity of these methodologies was verified through a fine in blind statistical analysis, which allowed to identify trends and similarities within these important artefacts. The evidence gathered indicated that these wallpapers were generated between 1885 and 1889, during the so-called industrial production period. A wide range of organic (proteinaceous binders, natural waxes, pigments, and vegetable lacquers) and inorganic (tin foil and pigments) substances were used for the artefacts’ manufacture, contributing to their overall complexity, which also reflects on the identification of a heterogeneous microbiota, often found in Eastern environmental matrices. Nevertheless, whether microorganisms inhabiting these wallpapers determined a detrimental or protective effect is not fully elucidated yet, thus constituting an aspect worth to be explored to deepen the knowledge needed for the conservation of Kinkarakawa-gami over time.

. Kinkarakawa-gami wallpapers. In (a) is depicted the verso of one of the wallpapers and the points 1 (written portion) and 2 (non-written portion) that were chosen-for all the artefacts-for the analyses, while the wallpapers' recto and the investigated sampling points (numbers) are represented in (b) INV_11 (3)(4)(5)(6)(7)(8)(9)(10), (c) INV_13 (3)(4)(5)(6)(7)(8)(9), (d) INV_15 (3)(4), and (e) INV_20 (3)(4)(5) www.nature.com/scientificreports/ Traces of Sulphur (S) and manganese (Mn) were observed on the wallpapers' verso, although S was solely recognized in the not-written areas, except for INV_15, while a higher distribution of Ca and iron (Fe) was seen in this side as compared to the recto (Fig. 6a, b, Supplementary Fig. 4a). Since these two elements, along with lead (Pb), were preponderant in the written portions of INV_11 and INV_13 (Fig. 6b, Supplementary  Fig. 4a), they may be components of the ink (e.g., iron gall ink) used for the selvages 12 , which are a trademark of Kinkarakawa-gami 1 . This conclusion cannot be drawn for INV_15_2 and INV_20_2, due to the strong presence of arsenic (As) and mercury (Hg) ( Supplementary Fig. 4a), deriving from the coloured areas of the recto (Fig. 6c,  Supplementary Fig. 5b) and/or a metal diffusion from one overlapping wallpaper to another, which may have hidden Ca and Fe signals.
Gilding of the paper with metals. Tin (Sn) was ubiquitously present in the wallpapers, revealing a greater signal distribution on their recto than their verso (Fig. 6), which is coherent with the gilding technique used by Japanese artisans, who covered the paper surface with a lacquer mordant and a metal foil (i.e., silver, gold or tin) and adhered to the embossed pattern 7 . In this regard, CLSM imaging of the wallpapers' recto showed an amorphous and heterogeneous surface attributable to the artefact's pattern, the inorganic pigments, and/or organic pigments (Fig. 5b). To better validate whether the wallpapers featured a multilayer structure as described in the literature 1,2 , a 3D reconstruction of a portion of INV_15 recto only partially covered by the motif pigmentation was performed (Supplementary Video 2). As a result, two distinct layers were detected: the top layer displayed the amorphous structure characterising the pigmentation, while the bottom layer featured an underlying matrix of fibres resembling that of the verso (Fig. 5c, d, Supplementary Video 2). Since the gilding practice was first carried out by Rottmann, Strome & Co factory in Yokohama (1882) yet abandoned from 1890 onwards 2 , the presence of multilayers containing Sn within the wallpapers suggested their belonging to the second period of Kinkarakawa-gami production.  Inorganic pigments. XRF spectroscopy revealed the recurrent presence in the artefacts of Si, S, Fe, Mn, Zn, As, whereas Hg, copper (Cu), chromium (Cr), and Pb were detected in specific coloured portions (Fig. 6), indicating the use of several inorganic pigments for the decoration. Although most of these pigments have strong IR contributions below 600 cm −113 , some absorption bands attributable to the yellow pigment orpiment (As 2 S 3 , Sekio; 1035, 913, and 820 cm −1 ) and its degradation (-AsOFe stretching mode 830-840 cm −1 ) were frequently found within the µATR-FTIR spectra ( Table 2, Supplementary Table 1-4), agreeing with the detection of As and S through XRF spectroscopy (Fig. 6). Since up to 1885 the wallpaper production did not involve As-compounds 2,6 , these results suggested artefacts' manufacture between 1886 and 1889, which is a rather tight timeline unlikely

Organic binders
Unidentified binders

Mordants
Green vitriol ✓ ✓ Figure 3. Statistical analysis performed on µATR-FTIR spectra for the 3770-2750 cm −1 interval. The hierarchical relationship between the spectra is represented by the dendrogram depicted in (a), while (b, c) show the estimation of the number of clusters using the average silhouette width and the obtained silhouette of clusters, respectively. PCA results, µATR-FTIR spectra, and the most representative (i.e., deepest) curve for each identified cluster (highlighted with different colours) are illustrated in (d-f). For clarity, the most external spectra of the clusters are underlined in (d) with the grey colour. www.nature.com/scientificreports/ to be achieved through other techniques (e.g., Accelerator Mass Spectroscopy (AMS)-radiocarbon dating featuring ± 20 years of experimental error) 14 .
Fe ubiquity on the recto's background ( Supplementary Fig. 4b) may instead indicate the application of natural iron oxide pigments, such as yellow-Odo-, brown-Taisha-and red-Kincha-ochres 15 , to obtain gold-bronze shades, which is another distinctive feature of this Japanese art 7 . Moreover, these pigments contain silicates and  www.nature.com/scientificreports/ manganese dioxide (MnO 2 ) as either components or impurities 16 , supporting the frequent detection of Si and Mn in traces (Fig. 6). An additional trace element observed in INV_11_7 and INV_13_3 was S ( Supplementary  Fig. 4b), whose contribution might rely on the use of tin sulphide (SnS 2 ), which was known in Asia since 300 BC 17 , as a bronze-gold pigment for these backgrounds.  Fig. 6c). In this regard, the simultaneous presence of S and Hg (INV_15_3), or S and As (INV_20_5; Supplementary Fig. 4b) suggested that red mercury-(HgS; cinnabar or Shu) and yellow/red arsenic-sulphides (Sekio or red realgar-Yuo, AsS or As 4 S 4 ) were the main pigments of INV_15 and INV_20 recto's backgrounds 15 . As and Zn traces in INV_15_3 could be also linked to Shu, which generally contains these elements as impurities 17 . Furthermore, Zn-based paint driers may have been applied to favour the pigments' deposition and solidification on the tin foil 18 , while Chinese white (zinc oxide; ZnO) and zinc sulphide (ZnS) could be used as white pigments 19 . To improve the brightness of the latter, artisans even mixed them with lead white [basic lead carbonate; 2PbCO 3 ·Pb(OH) 2 ] 20 , justifying the Pb contribution observed in INV_11_5 ( Supplementary Fig. 4b).
The painted areas of the wallpapers' recto were deeply analysed focusing on the most representative colorations (i.e., red, green, brown, and black). Similar chemical compositions were observed for the red colour in  17 . The presence of Pb may be instead ascribed to (1) lead white for the outline of the leaves 15 , (2) lead-tin yellow 21 , or (3) yellow lead oxide (PbO) 22 to achieve diverse green shades. Regardless, all the green areas exhibited S and As, whose distribution varied depending on the different colour shades, as noted in INV_11_3, INV_11_9, INV_13_4, and INV_13_8 ( Supplementary Fig. 5b). Thus, Sekio 15 seemed to be constantly present in the green areas 23 , suggesting (1) a combined use of Verdigris, Scheele green, or emerald green with sulphide compounds in INV_11_3 to improve both brightness and durability of the pigments 24 , as well as (2) the addition of Sekio to an organic dark (blue/purple) pigment to obtain the green colour 15,25,26 . Except for INV_20, the intense Fe signal characterising the recto's green-painted sections (Fig. 6e) could imply the combination of green/yellow inorganic pigments with natural ochres 15 , as also supported by the detection of Si, Mn, and Zn in traces (Fig. 6e, Supplementary Fig. 5b). Nevertheless, µATR-FTIR spectra of INV_20_3 exhibited -SO 4 asymmetric (1120-1040 cm −1 and 827 cm −1 ) and symmetric (983 cm −1 ) stretching, -FeOSOFe vibration (955 cm −1 ), as well as -FeO stretching (830 cm −1 ) characteristic of the Fe-based mordant green vitriol (FeSO 4 ·xH 2 O) ( Table 2, Supplementary Table 4), which was likely used in this wallpaper. Several brown drawings were part of INV_11 and INV_13 (Table 1), where Fe detection (Fig. 6f, Supplementary Fig. 5c) might infer to the utilization of Taisha 7 , also supported by the low content of Si, Mn, and Zn, which confer diverse yellow-brown and red-brown shades to the pigment 16 . Additionally, lead white and/or lead oxide 15  www.nature.com/scientificreports/ and INV_13_9 showed a relatively high distribution of both As and S (Fig. 6f) attributable to the addition of Sekio or Yuo to get unique brown shades 23,26 . Finally, the black colour was only present in INV_11_10, whose elemental analysis revealed traces of S and the three ubiquitous elements Ca, Fe, and Sn (Fig. 6g), implying the use of an organic black pigment for this colour 15 .
Organic pigments. Several IR absorption bands typical of the natural purple-pigment Indigo were detected in the green areas of INV_11_3 and INV_13_8 (Fig. 2a, (Fig. 4d, Supplementary Fig. 3b) Fig. 3b). Moreover, INV_15_4 partitioning within cluster 2 for the 1840-719 cm −1 interval (Fig. 4c-e) may be due to the presence of other organic substances having absorption bands in common with Indigo. Indeed, any of typical IR contributions of this organic pigment were observed in INV_15_4 μATR-FTIR spectrum (Supplementary Table 3), which also featured a low depth value, as well as a high PC2 score, with respect to the other spectra clustering together (Fig. 4d, Supplementary Fig. 5b).
Organic binders. The identification of inorganic pigments highlighted the need for Japanese artisans to use binders (e.g., proteinaceous binders, vegetable oils, waxes, and lacquers) for their painting 15 . Although the type of these substances is generally determined through FTIR, the wallpapers' complexity impaired their univocal identification, since they had similar IR absorption bands and their specific vibrational modes often overlapped with each other (Supplementary Tables 1-4) 30,31 . In this regard, the clustering analysis in the 1840-719 cm −1 interval revealed that spectra featuring Indigo's IR contributions grouped in cluster 2, while cluster 1 featured the verso spectra and those where the organic pigment was absent or less present, in which overlapping vibrational modes of cellulose, lignin, silicates, carbonates, and organic binders were ubiquitous (Fig. 4c, Table 2). Similarly, although four clusters were identified for the µATR-FTIR spectra in the 3770-2750 cm −1 interval (Fig. 3c-e), several absorption bands attributable to these substances were detected (Supplementary Tables 1-4). Nevertheless, the type of organic substances was presumed, when possible, by comparing their IR absorption bands with those characterising the organic binders (Supplementary Tables 1-4 Tables 3-4), which is in agreement with the Japanese traditional use of animal glue (nikawa), gelatine cubes, and beeswax as binders 15,25 . Since IR absorption bands distinctive of proteinaceous binders or waxes are more present in the 3770-2750 cm −1 interval 30,31 , the partitioning of µATR-FTIR spectra of INV_15 recto in cluster 3 (average s i = 0.66), INV_13_9 and INV_20_3 in cluster 1 (Fig. 3c, d) statistically supported these observations. Particularly, the MBD-based analysis of cluster 1 revealed a high degree of similarity between these µATR-FTIR spectra, representing both the deepest curve (INV_13_9) and the core (70%) of the cluster itself, while INV_11_4.1 was the most external spectrum (Fig. 3f, Supplementary Fig. 2a (Fig. 4d). Several µATR-FTIR spectra displayed also vibrational modes distinctive of organic vegetable lacquers ( Table 2,  Supplementary Tables 1-4), which grouped in cluster 3 in the 3770-2750 cm −1 interval (Fig. 3c-e). In this regard, the so-called lacquering procedure for the wallpapers' production involved the mixing of plant-extracted lacquers-likely the Urushi one, based on the Japanese tradition 32 -with Odo, Taisha, and Kincha to confer the yellow-gold background characteristic of the Kinkarakawa-gami's recto 7 . Indeed, the Urushi lacquer features a strong IR contribution around 3400 cm −1 attributable to -OH stretching 32 , which, alongside the vibrational modes of proteinaceous binders, was a distinctive trait of the µATR-FTIR spectra belonging to cluster 3 (first www.nature.com/scientificreports/ interval; Fig. 3c-e). In line with this, INV_15_4 resulted the deepest curve identified for this cluster (Fig. 3f), while INV_13_5, where the -OH stretching of the Urushi lacquer was absent (Supplementary Table 2), was the most external one (Fig. 3d, Supplementary Fig. 2a).

Isolation and identification of microorganisms populating the wallpapers. A microbiological
evaluation of specific areas of these artefacts was carried out to assess the role of microorganisms in either the degradation or damage prevention of the wallpapers, as well as the hindrance in identifying IR contribution of organic substances due to the interference derived from macromolecules' vibrational modes 33,34 . A wide array of microorganisms populates different works of art, as their organic and inorganic substances constitute carbon and essential element sources for bacteria and fungi 34 . The microbial presence on artistic items is also favoured by environmental factors (e.g., low ventilation, high humidity) and the poor state of conservation of the artefacts, which allows the microbial growth under oligotrophic conditions 34 . Particularly, microorganisms tend to irreversibly attach to the artefact surface, forming communities defined as a biofilm, where its complex hydrogel matrix-made of proteins, lipids, and polysaccharides-confers to bacteria and fungi protection from external factors 35 . Additionally, less than 10% of microorganisms populating different niches can be cultured through standard procedures 34 , making it almost impossible to entirely identify the microbial community associated with artistic items. Several bacteria and fungi (  (Table 3). These results are in line with the studies reporting the isolation of cultivable microorganisms from wall or easel paintings 34,36 .
The microbial isolates C 4 (11)3, C 4 (13)1, C 7 (11)1, and C 4 (11)2 closely related to B. coreaensis, B. pocheonensis, M. chokoriensis, and Cladosporium cf. ramotenellum respectively (Table 3), and autochthonous of Kinkarakawagami, are commonly found in several environmental matrices of Korea, China, Thailand, and Japan 37-40 , confirming the eastern provenance of these wallpapers. Besides, these microorganisms hold enzymatic assets (i.e., xylanases and laccases) that enable them to degrade and depolymerize (1) lignin and cellulose compounds 37,40-44 , (2) lacquers [45][46][47] , and (3) organic pigments 46,48 (Table 3). Laccases are instead Cu-polyphenol oxidases firstly identified in the exudates of the Japanese Rhus verniczfera plant 50 from which the Urushi lacquer is extracted 32 , yet also recognized as enzymatic catalyst of most fungal strains 47 . Since these enzymes are responsible for the degradation of phenol substrates 45,47 , the partial identification of vibrational modes typical of urushiol polymerization or Indigo (Supplementary Tables 1-4) could be ascribed to the biotic deterioration of both Urushi lacquer and organic pigment, as Japanese lacquer IR contributions were found within sampling points where fungal strains were isolated (Tables 2, 3). A similar conclusion can be made for C 4 (11)3 and C 4 (11)2 isolates phylogenetically related to B. coreaensis and C. ramotenellum respectively, whose production and secretion of xylanases and laccases is one of their distinctive metabolic traits 37,40,44 . Further, Bacillus [C 4 (13)2], Staphylococci [C 5 (13)2 and C 4 (15)1)], and Micrococci [C 2 (11)1 and C 5 (13)1] species are among the most persistent strains capable of metabolizing honey and beeswax 51 , while M. chokoriensis [C 7 (11)1] and C. ramotenellum [C 4 (11)2] are proficient in degrading gelatine and starch, which were two of the most used binders in Japanese tradition 15,27 . Table 3. Identification of bacterial and fungal strains isolated from the four Kinkarakawa-gami analysed. C n (X)m: C n indicates the chosen sampling point, as depicted in Fig. 1; X the catalogue number of the wallpaper from which the bacteria and/or fungi were isolated; m represents the number of bacterial or fungal isolates investigated. www.nature.com/scientificreports/ The presence of microorganisms can also be linked to their tolerance and/or resistance towards a broad spectrum of metal and metalloid compounds 52 , which are the main components of inorganic pigments. For instance, bacteria belonging to the Firmicutes and Actinobacteria phyla can detoxify their surrounding environment from toxic metals or metalloids, even using them as a terminal electron acceptor to produce energy 52,53 . Bacillus, Staphylococcus, and Micrococcus spp. are overall present in paintings where carbonates (chalk-CaCO 3 ) and silicates (quartz-SiO 2 ) are abundant 54,55 , due to their ability to overcome the challenge deriving from these minerals. Bacillus spp. are also highly tolerant towards Pb- 56 and Fe-containing compounds 57 , being able to transform lead acetate [Pb(CH 3 CO 2 ) 2 ] 56 , hematite (Fe 2 O 3 ) and its hydrated forms 58 , as well as to oxidize Fe(II) to Fe(III), as in the case of B. pocheonensis 59 . Analogously, S. epidermidis, M. chokoriensis, and K. rizophila strains were studied for their resistance against Pb-containing compounds [60][61][62] , while the fungal strain Hirsutella spp. showed tolerance to Ca-and Fe-sulphates 63 . Thus, the elemental composition of INV_11 and INV_13 (Fig. 6) justified the presence of microorganisms ( Table 3) that hold metabolic traits allowing them to survive on metal-rich wallpapers. Although INV_15_4 displayed a high amount of Shu (HgS) (Supplementary Fig. 5a), one bacterial [C 4 (15)1] and two fungal [C 4 (15)2 and C 4 (15)3] strains related to S. epidermidis, Penicillium georgiense, and Aspergillus costaricaensis were isolated (Table 3), highlighting their great tolerance towards Hg-containing compounds 64,65 . Indeed, these fungal strains were responsible, among others, for the darkening of cinnabar (Shu), as reported elsewhere 66 . The isolation of S. epidermidis and Hirsutella strains could also derive from anthropogenic/environmental contamination, as they are human and nematode pathogens respectively 63,64 , reflecting the artefacts' poor state of conservation.
Besides, the local heterotrophic microflora (i.e., Bacillus, Micromonospora, and Micrococcus species) can produce secondary metabolites with antimicrobial properties as a defence mechanism under stress conditions 34,67-69 . Thus, Firmicutes and Actinobacteria could act as biocontrol agents for the preservation of cultural heritage 34 . These observations, along with the advanced state of deterioration that was macroscopically visible for INV_20, may indicate the key role played by bacteria belonging to Bacillus, Micromonospora, and Micrococcus genera in controlling and preventing the further degradation of the colonized wallpapers; indeed, INV_20 was the only artefact from which any cultivable microorganism was not isolated.

conclusion
This multidisciplinary study allowed to unveil physical-chemical features regarding the composition, manufacture, and dating of the collectively important cultural heritage represented by Kinkarakawa-gami works of art. The experimental evidence gathered was corroborated through the uncommon yet resourceful and innovative in blind functional data statistical analysis. Indeed, given the complexity of the studied wallpapers in terms of IR vibrational modes ascribed to (1) substances used for their fabrication, (2 physical-chemical degradation products, and (3) the presence of a microbiota, this statistical approach has proved greatly helpful to identify and confirm trends, differences, and similarities observed among the four Kinkarakawa-gami artefacts. Microbiological investigations supported the Eastern provenance of these wallpapers, however, whether the cultivable microbes act as deteriogen or biocontrol agents is yet to be defined; hence, DNA sequencing-based technology (e.g., study of the microbiome) represents the new frontier to unveil the identity of uncultivable microorganisms and, alongside physical-chemical characterisation, will improve the development of innovative, promising, and eco-friendly restoration strategies for the conservation of cultural heritage.

experimental section
Materials. The four wallpapers here studied belong to the V. Ragusa-O'T. Kiyohara collection of Palermo (Italy). Given the complexity and richness of these artefacts in terms of details and depicted colours, an extensive sampling of the wallpapers (Table 1) was performed to thoroughly analyse them.
Tryptic soy, malt extract, and agar technical were purchased from Sigma-Aldrich® (Milan, Italy), while QIAquick PCR purification kit was obtained from QIAGEN (Milan, Italy).

X-Ray fluorescence (XRF) spectroscopy. A Tracer III sd Bruker AXS (Bruker, UK) equipped with
Rhode anode and working at 40 kV and 11 µA was exploited for XRF analyses, whose acquisition time was 30 s. Element identification and XRF spectra analysis was performed by using ARTAX® software, which was provided with the instrument, while R 3.6.1 and OriginPro® 2016 software were used for the representation. Elemental compositions of wallpaper sampling points are reported in terms of Net Area (× 10 4 a.u.), which represents the integral intensity of X-ray emission characteristic of each element obtained after performing the Bayes deconvolution and deduction of the background intensity 70 . These results are to be considered as a qualitative estimation regarding the presence and abundance of diverse elements within the chosen sampling points, as any appropriate standard was not used to determine the concentration of each element.
Attenuated total reflectance-Fourier transform infrared (ATR-FTIR) spectroscopy. ATR-FTIR spectra were recorder by using a µFTIR Lumos (Bruker, UK) equipped with a Platinum ATR and an IR microscope featuring 0.1 µm as lateral resolution. The spectra were collected in the 4000-600 cm −1 range, with a resolution of 2 cm −1 and 60 scans per each sampling point, and they were subsequently analysed through the software OPUS(7.5)®, which was provided with the instrument, as well as OriginPro® 2016.
Statistical analyses of µATR-FTIR spectra. Based on the structure of µATR-FTIR spectra, they can be assimilated to complex sets of data (i.e., functions, where each spectrum corresponds to a distinct function) varying over a continuum (i.e., wavenumber range) that, taken together, can be considered as single curves. Hence, in blind functional data analysis (FDA) was applied as the statistical methodology on these spectra 71 www.nature.com/scientificreports/ that were split, based on the wavenumber range acquisition, in 3770-2750 cm −1 and 1840-719 cm −1 intervals, which were singularly analysed by performing a hierarchical clustering of µATR-FTIR curves with the respect to their shape. The quantification of each curve's cohesion to its own cluster as compared to the separation from the other clusters was obtained through a silhouette (s i ) measurement, while the optimal number of clusters was determined by maximizing the average s i to the configuration obtained in the hierarchical clustering. Functional Principal Component Analysis (FPCA) was then performed to derive the PCs inside the final clusters 71 and find an optimal orthogonal linear projection of the curves on a d-dimensional subspace (R d when d = 2 or d = 3) to minimize the expected value of the squared error due to the projection. Lastly, a data depth algorithm based on the Modified Band Depth (MBD) was constructed focusing on centrality and separation of the spectra 72 , allowing to identify both the most representative and the most external curves of each cluster. All the statistical analyses were performed by using the R 3.6.1 software; a more extensive review of the performed statistical analyses and the R-packages used is reported in the Supplementary Methods section.
fluorescence microscopy. Fragments of the wallpapers were imaged both on recto and verso sides by a Leica TCS SP5 fluorescence confocal laser scanning microscope (CLSM), using a 40X-1.25 NA objective (Leica Microsystems, Germany). Images were acquired under two-photon excitation at 830 nm in the 450-700 nm emission range. The samples were soaked in glycerol during the imaging procedure. The same setup was applied for a 3D reconstruction of the wallpaper multilayers. The data were analysed by ImageJ software.
Microbiological analyses. The sample areas of interest of Kinkarakawa-gami were gently swiped with sterile cotton swabs, which were then suspended in the physiological solution (sodium chloride 0.9% w/v) for 30 min. Afterward, the suspensions were serially diluted, being aliquots (100 μL) spread onto both tryptic soy and malt extract agar plates to isolate either bacterial or fungal strains respectively, whose biomass growth was carried out at 30 °C for 5 days under static conditions. Polymerase chain reaction (PCR) was performed (Supplementary Methods), following thermocycler conditions described elsewhere 73 , on the extracted and purified genomic DNA to obtain the 16S rRNA gene product and the internal transcribed spacer (ITS) region in the ribosomal RNA operon. To identify-at genus level-the isolates retrieved from Kinkarakawa-gami's art, PCR products were purified through QIAquick PCR purification kit, according to the manufacture's protocol, sequenced (BMR Genomics, Padova, Italy), and searched for nucleotide homology with other microorganisms (Supplementary Methods).