ABSTRACT
A 500-year-old unpreserved Mary Rose sample, historically containing an iron bolt, was analyzed using enrichment cultures and 16S sequencing. The novel community of bacteria present demonstrates a biological pathway of Fe and S oxidation and a range of acid-generating metabolisms, with implications for preservation and biogeochemical cycling.
TEXT
In 1545, King Henry VIII's warship the Mary Rose sank off the Southern coast of the United Kingdom during an engagement with a French invasion fleet. Between 1967 and 1968, the starboard hull was rediscovered intact, the high level of in situ preservation due to the ship's burial in clay sediments, limiting wood degradation (30). In 1982, after 437 years, the Mary Rose was raised from the seabed with the aim of restoration, preservation, and long-term conservation of the ship and artifacts. Excavation of the Mary Rose into an oxic environment, however, poses significant challenges for the conservation of this historic maritime treasure (22). The acidification of raised marine archaeological timbers, and the appearance of sulfate salts on wood surfaces, is a degradative phenomenon initially observed in the hull of the Swedish Vasa (35), and later in Mary Rose wooden artifacts (22), despite conservation efforts.
Archaeological wood buried in marine sediments characteristically accumulates reduced iron and sulfur compounds (RISCs) due to the biogeochemical cycling of Fe and S (7). The reductive processes of the sulfur cycle dominate in anoxic marine sediments high in organic matter (23). Microbially generated hydrogen sulfide (H2S) penetrated the Mary Rose and partially oxidized to S0, other oxidative intermediate states, and organosulfur compounds (42). Iron sulfides (FeS and FeS2) form when hydrogen sulfide ions react with available Fe2+ (33, 42), such as soluble Fe(II) ions from corroding iron nails used during the construction of the Mary Rose (4). Approximately 1 mass percent sulfur is present in the Mary Rose hull, of which pyrite (FeS2) accounts for a third of the total reduced sulfur (34). The oxidation of RISCs upon excavation, and the resulting production of sulfuric acid, is of major conservation concern, as acid hydrolysis of cellulose and hemicellulose diminishes wood strength and structure and, with mostly lignin remaining, leads to cellular collapse. This study aimed to determine the community of bacteria present in Mary Rose waterlogged archaeological timber excavated from marine sediments and stored in museum conditions prior to conservation with polyethylene glycol (PEG) and subsequent drying. Furthermore, this study was aimed to determine if biological oxidation of RISCs was contributing to acid production in the Mary Rose, with implications for its preservation.
In 2005, the Mary Rose stempost was raised from the seabed (latitude/longitude, 49°52.2628′N, 006°26.5928′W), an important archaeological excavation providing the central structure of the bow section that was missing from the intact hull and defining the shape of the hull at the bow (Fig. 1). On excavation, a sample of waterlogged wood was stored in a sealed aluminum bag, standard storage protocol at the Mary Rose Trust. In 2010, the sample was analyzed. The stempost sample historically housed an iron bolt, and the wood in proximity to the bolt hole was degraded and covered in soft yellow deposits. Distinct zones of coloration were observed along the length of the stempost moving away from the historic source of iron to black stained wood typical of anoxic conditions (Fig. 2). Samples and pH measurements were taken along the timber; wood covered in yellow deposits taken from the bolt hole was used as inocula for microbiological cultures (Fig. 2, “A”).
Isometric drawings of the Mary Rose hull and stempost. (A) The intact Mary Rose hull recovered from the seabed in 1982 is drawn in relation to the stempost raised in 2005, which provided crucial information about the shape of the ship's bow. The missing areas and proposed silhouette of the hull are represented by dashed lines. (B) Detailed isometric drawing of the stempost and adjoining apron timber with dimensions and locations of iron bolts, raised from the seabed in 2005 at a latitude and longitude of 49°52.2628′N, 006°26.5928′W. C, location of the stempost sample (MRT04A) used in this study. Drawings courtesy of the Mary Rose Trust.
The Mary Rose stempost sample used in this study, with pH values indicated. The areas of wood measuring pH 1.7 are correlated with proximity to the bolt hole that historically contained an iron bolt and yellow sulfur salt deposits on the surface of the wood. Originally, extensive yellow deposits surrounding and internal to the bolt hole existed; however, these were removed and used as initial inocula in enrichment cultures. A, location of wood and sulfur deposits used as inocula for MRT04A Fe and S cycling enrichment culture SM18.
Chemolithotrophic and heterotrophic Fe and S oxidizers were enriched and isolated for characterization using liquid and solid agarose media as described previously (19, 21). Ten wood samples were taken along the length of the stempost and ground in liquid N2 using a sterile pestle and mortar, and DNA extractions were performed as described previously (37) and then pooled to create a combined sample. A 1,080-bp 16S rRNA gene fragment was amplified using nested PCR with primer pairs 27f/1492r and 357f/1392r (27, 39). Cloning and transformation were performed using the TOPO TA cloning kit (Invitrogen). A total of 96 clones were selected from the stempost clone library, plasmid purified, and sequenced (Functional Biosciences, Inc.), but only 58 contigs were assembled due to sequence quality. Contigs were assembled using CAP3 (16) and checked for chimeric sequences using Greengenes (6) and employed as search queries on BLAST (1). Sequences were aligned using the ClustalW alignment algorithm. All phylogenetic analyses were performed in MEGA 5.05 (38), and sequences were submitted to GenBank.
Iron oxide deposits, sulfur oxidation, and extensive growth physically associated with sulfur were observed in culture SM18, concurrent with a decrease in pH (see Fig. S2 in the supplemental material) [media contained 0.4 g/liter MgSO4 · 7H2O, 0.2 g/liter (NH4)2SO4, 0.1 g/liter K2HPO4, 5 g/liter S8, 5 mM K2S4O6, 3.5 mM Fe(II)SO4, 1% NaCl (pH 3.0)]. This culture was selected for a 16S rRNA gene analysis of bacterial diversity. A total of 96 clones were sequenced and 95 contigs assembled. A phylogenetic analysis of 16S rRNA sequences representative of operational taxonomic units (>97% divergence) from the unpreserved stempost and enrichment culture SM18 are shown (Fig. 3 and 4). BLAST search results are given in Table S1 in the supplemental material.
Neighbor-joining phylogenetic tree of 16S rRNA gene sequences from the stempost sample clone library with related taxa. Evolutionary distances were computed using the maximum composite likelihood method with 1,000 bootstraps, and all positions containing gaps and missing data were eliminated. A total of 92 nucleotide sequences, including 18 stempost clone library sequences representing operational taxonomic units (OTUs), were included in the analysis. The rate variation of all sequences was modeled with a gamma distribution. Substitution pattern and rates were estimated under the general time reversible model (+G) shape parameter of 0.2754. The mean evolutionary rates among sites estimated for 5 discrete categories were 0.00, 0.04, 0.23, 0.86, and 3.88 substitutions per site, and there were a total of 881 positions in the final data set. An automatically generated neighbor-joining tree was used as the initial tree for estimating maximum-likelihood values; the maximum log likelihood for this computation was −27160.733. The tree is rooted by Rhodothermus marinus.
Neighbor-joining phylogenetic tree of 16S rRNA gene sequences from the enrichment SM18 coculture clone library with related taxa. Evolutionary distances were computed using the maximum composite likelihood method with 1,000 bootstraps. A total of 95 sequences were included in the analyses. The rate variation of all sequences was modeled with a gamma distribution. Substitution pattern and rates were estimated under the general time reversible model (+G) shape parameter of 0.3903 The mean evolutionary rates among sites estimated for 5 discrete categories were 0.00, 0.04, 0.24, 0.88, and 3.84 substitutions per site, and there were a total of 1,079 positions in the final data set. An automatically generated neighbor-joining tree was used as the initial tree for estimating maximum-likelihood values; the maximum log likelihood for this computation was −8165.931. The tree is rooted by gammaproteobacteria outgroup taxa E. coli and A. ferrooxidans.
The SM18 clone library revealed a two-species coculture of Alicyclobacillus and Acidiphilium spp. (66% and 34% of sequences, respectively) (see Fig. S1 in the supplemental material). BLAST searches against the Alicyclobacillus-like sequences retrieved the highest similarity scores (96.6 to 98.7%) to Alicyclobacillus aeris (a ferrous iron- and sulfur-oxidizing bacterium isolated from the acid mine drainage [AMD] of copper sulfide ores) (9), represented by clones MRT18_01 and MRT18_06. An obligate halophilic, facultative chemolithotrophic iron-oxidizing Alicyclobacillus strain, MRT18, was isolated in pure culture [media containing 1.25 g/liter (NH4)2SO4, 0.5 g/liter MgSO4 · 7H2O, 0.025% tryptone soya broth (wt/vol), 10 mM galactose, 25 mM Fe(II)SO4]. This isolate grew optimally at 37°C, with a pH value of 2.5, required a minimum of 1% NaCl for growth, and oxidized Fe2+ in media containing 4% NaCl (Fig. 5). Endospore staining confirmed the presence of terminal spores, and the identity was confirmed by sequencing (GenBank accession number JX840943). Unlike A. aeris and other iron-oxidizing Alicyclobacillus species, the halo-obligate strain MRT18 did not grow with yeast extract as an organic carbon source, with either ferrous iron or potassium tetrathionate as electron donors.
Iron oxidation by Alicyclobacillus MRT18 grown in a series of NaCl concentrations, 2 to 20% (wt/vol). Ferrous iron concentrations are depicted by solid lines (left-hand vertical axis), and ferric iron concentrations are depicted by dashed lines (right-hand axis). Measurements were taken from the culture supernatant and therefore do not include the extensive Fe3+ deposits on the sides and bottom of the conical flasks. Media contained 1.25 g/liter (NH4)2SO4, 0.5 g/liter MgSO4 · 7H2O, 0.025% tryptone soya broth (wt/vol), 10 mM galactose, and 25 mM Fe(II)SO4. Concentrations of ferrous iron were measured using the ferrozine colorimetric assay (28), and concentrations of ferric iron were measured with a turbidimetric assay (36). Absorbance was measured using an Ultrospec 2000 UV/Visible spectrophotometer (Pharmacia Biotech). NaCl percent trials were performed in triplicate, and each assay measurement was also taken in triplicate. The average values are presented. A control series of each NaCl concentration contained sterile media with ferrous iron but no Alicyclobacillus. Values for all NaCl percent control series were combined to provide a single control series.
The Mary Rose stempost sample contains several species of acidophilic bacteria capable of iron and sulfur cycling. The Acidiphilium sequences from culture SM18 were also detected in the stempost (clones MRT19_10 and MW07E), sharing 97.6% similarity with the acidophilic alphaproteobacterium Acidiphilium sp. JL2 and forming a well-supported clade (91%) (Fig. 4) with the Acidiphilium cryptum cluster. The cooccurrence of Acidiphilium species with acidophilic chemolithotrophic Fe and S oxidizers is common in culture (32) and extreme environments such as acid mine drainage (AMD) (12). All Acidiphilium species characterized so far are capable of dissimilatory ferric iron reduction under microaerobic or anaerobic conditions (20, 29). Chemolithotrophic growth on sulfur is known in some Acidiphilium species, such as Acidiphilium acidophilum (14), and Acidiphilium cryptum can oxidize sulfur in the presence of organic substrates (11, 13). This suggests either a source of organic carbon is produced in culture SM18 to support heterotrophic growth or autotrophic sulfur oxidation is occurring. The Acidiphilium strain represented by clone MRT19_10 enriched in coculture SM18 has yet to be isolated in pure culture to confirm that it is capable of chemolithotrophic growth.
Acidobacteria-like species were present in the stempost (clone MW01C), sharing 97% similarity to the heterotrophic iron-reducing Acidobacterium capsulatum isolated from AMD (24), and thought to be capable of ferrous iron oxidation (40, 41). The Mary Rose stempost clone library also included 2 sequences (MW11C and MW05C) with 97.1% similarity to the gammaproteobacterium Thiobacillus prosperus, a halotolerant chemolithotrophic iron and sulfur oxidizer (17, 31) which formed a monophyletic clade with the type species Thiobacillus prosperus DSM 5130. The T. prosperus-like strain has so far proved unculturable. Growth of many acidophilic Fe oxidizers, such as Acidithiobacillus ferrooxidans and Thiobacillus species, is inhibited by chloride irons (43). The lack of such species, the presence of T. prosperus-like sequences, and the requirement of strain MRT18 for NaCl suggest that an environment high in salt is determining the community structure of acidophilic Fe and S oxidizers in marine archaeological wood. This work is of relevance to bioleaching of metal ores, and the potential use of halotolerant acidophilic iron and sulfur oxidizers (such as Alicyclobacillus MRT18 and the T. prosperus-like clone) in biomining operations using seawater or brackish sources of water as opposed to freshwater.
The Mary Rose stempost contains a phylogenetically diverse range of sequences related to methanogenic, acetogenic, and sulfidogenic anaerobic bacteria with multiple metabolic pathways of acid production. The Cytophaga-Flavobacteria-Bacteroides (CFB) group dominated the clone library, accounting for 72% of sequences. The majority of stempost sequences were phylogenetically related to Meniscus glaucopis (93.5%, clone MW12D), known to ferment a range of sugars, including cellobiose (18). The Meniscus glaucopis-like species is a candidate taxon for the previously described “erosion bacteria” associated with microbial degradation of archaeological wood (2, 26). Other sequences placed phylogenetically in the CFB group had 95 to 98% similarity to Prolixibacter bellariivorans, a sugar-fermenting, facultative anaerobe producing acetate, propionate, and succinate (15), a phenol-degrading sulfidogenic bacterium isolate (25), and Bacteroidetes bacteria isolated from methane-rich sediments (5).
Strictly anaerobic bacteria from the Clostridium-Bacillus subphylum are well represented in the stempost clone library. Sequences MW03C and MW04C shared 97.3% and 89.9% sequence similarity to acetogenic bacteria Dehalobacter sp. strain MS (8) and Dehalobacterium formicoaceticum, respectively. Dehalogenating bacteria are often found associated with sulfate-reducing, iron-reducing, methanogenic, or acetogenic microorganisms (10), of which several candidate taxa are present in the stempost clone library. The remaining clones are of 93.0 to 95.4% sequence similarity to a range of obligately anaerobic sugar-fermenting Clostridia species. Sequence MW11D is placed closest to Alkalibaculum strains in a clade also containing Acetobacterium woodii and Eubacterium limosum. Two sequences from the Ruminococcaceae family occur within the Clostridium leptum cluster; sequence MW01E is well supported in a cluster of nonruminal and ruminal polysaccharide-fermenting anaerobes capable of metabolizing the polymers inulin, cellulose, and cellobiose. The clone sequence MW09A is placed phylogenetically closest to the sugar-fermenting acetic acid bacterium Acetanaerobacterium elongatum (3). The detection of previously uncultured Clostridium sequences belonging to phylogenetic clades dominated by organisms capable of cellulose degradation suggests the wood degradation is facilitated by cellulolytic bacteria in addition to acid production from the oxidation of RISCs.
The stempost supports a unique community of bacteria involved in the biogeochemical cycling of iron, sulfur, halogenated compounds, and cellulosic and polysaccharolytic degradation. A range of metabolisms are detected, from strictly anaerobic to obligately aerobic, that may be contributing to acid pathways in the waterlogged archaeological wood. This reflects the gradient from pH 4.7 to pH 1.7 along the length of the stempost, correlated with a graduation from black anoxic wood to the bolt hole exhibiting yellow sulfur salt deposits typical of iron sulfide-rich wood exposed to oxygenated conditions. The isolation of a novel, spore-forming mixotrophic Alicyclobacillus strain not previously isolated from this environment indicates that spore formation may be part of microbial ecological succession in marine archaeological wood containing reduced forms of Fe and S.
The novel community described provides the first insight into the microbial ecology of marine archaeological waterlogged wood stored in museum conditions before preservation with traditional conservation methods. Until now, the conservation protocols used to limit sulfur deposits have been based on abiotic oxidation of Fe and S compounds. Knowledge of the microbial community present is the first step in designing remediation strategies, including bioremediation, to limit the biological oxidation of RISCs in areas of wood exhibiting low surface pH, where chemical oxidation is rate limited by low pH and Fe3+ generation.
Previously, the biological contribution to the “sulfur problem” in archaeological wood had not been confirmed. This work establishes that species of acidophilic bacteria capable of oxidizing iron and sulfur are present in the Mary Rose warship and is the first demonstration that oxidization of reduced Fe and S compounds is not solely an abiotic phenomenon in archaeological wood.
ACKNOWLEDGMENTS
This work was funded by the Mary Rose Trust and the University of Portsmouth.
Many thanks to Barrie Johnson, Kevin Hallberg, and Barry Grail for their generous scientific support and advice and provision of Acidiphilium strains for use in overlay media. Thanks to Siobhan Watkins for her microbiological advice, Simon Cragg and Christine Hughes for assistance with SEM, and Lai Pang and Julian Mitchell for preliminary findings that informed this research.
FOOTNOTES
- Received 30 July 2012.
- Accepted 22 September 2012.
- Accepted manuscript posted online 28 September 2012.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/AEM.02387-12.
- Copyright © 2012, American Society for Microbiology. All Rights Reserved.
