Detection of bone glue treatment as a major source of contamination in ancient DNA analyses.код для вставкиСкачать
AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 118:117–120 (2002) Detection of Bone Glue Treatment as a Major Source of Contamination in Ancient DNA Analyses Graeme J. Nicholson,1 Jürgen Tomiuk,2 Alfred Czarnetzki,2 Lutz Bachmann,3* and Carsten M. Pusch2 1 Institute of Organic Chemistry, University of Tübingen, D72076 Tübingen, Germany Department of Human Genetics, University of Tübingen, D72074 Tübingen, Germany 3 Natural History Museums and Botanical garden, Zoological Museum, University of Oslo, N-0562 Oslo, Norway 2 KEY WORDS amino acid; paleogenetics; racemization; skeletal remains ABSTRACT Paleogenetic investigations of ancient DNA extracted from fossil material is for many reasons susceptible to falsification by the presence of more recent contamination from several sources. Gelatine-based bone glue that has been used extensively for nearly two centuries by curators to preserve hard tissues contributes nonauthentic DNA to paleontological material. This fact has been frequently neglected and is barely mentioned in the The vast majority of ancient DNA analyses employ the polymerase chain reaction (PCR). While the enormous sensitivity of the method allows the amplification of minute amounts of authentic ancient DNA, even traces of contaminating nonauthentic DNA will inevitably lead to artifacts. Consequently, numerous control experiments are required to verify PCR-based results in ancient DNA analyses (Handt et al., 1994; Cooper and Poinar, 2000). Contaminating nucleic acids may stem from various sources. These include human DNA, derived from the persons performing the genetic experiments or from people who previously handled the specimen, and from edaphon DNA, primarily derived from bacterial or fungal growth. Another source of contamination is seen in the numerous substances used for hard-tissue conservation (Rixon, 1976; Brommelle et al., 1984; Horie, 1987; Collins, 1995). From the early 19th century up to the present, gelatine-based glue has been widely used, because it is cheap, easily available, and very effective (Lepper and Lewis, 1941; Shelton and Johnson, 1995). Although some authors have expressed suspicions that such conservation techniques may cause severe complications if not hindrance for the analysis of old biomolecules (Horie, 1987; Hall et al., 1993; Cooper, 1994; Shelton and Johnson, 1995), this source of experimental pitfalls is underestimated. Bone glue is produced from bones, hides, sinews, ligaments, and gristle which are degreased, demineralized with acid, and swollen by a prolonged treatment (up to 20 weeks) with aqueous alkali (usually Ca(OH)2) at ambient temperature, before extraction and partial hydrolysis of the remaining collagen © 2002 WILEY-LISS, INC. literature. Now paleogeneticists, curators, and conservators are faced with the problem that treatment of samples with adhesives and consolidants for conservatory purposes has seldom been recorded. Here, we show that racemization of amino acids, and in particular serine, is an excellent indicator for the treatment of paleontological samples with glue. Am J Phys Anthropol 118:117–120, 2002. © 2002 Wiley-Liss, Inc. with hot water. The resulting viscous and colloidal solution penetrates the material to be conserved and hardens therein. DNA, however, is coextracted in this procedure. Thus bone glue is a rich source of nonauthentic DNA for paleontological samples. Moreover, due to the processing of collagen for glue production (i.e., extreme pH), such DNA is expected to be severely degraded as is the authentic fossil DNA. Thus, it is unlikely that both sorts of DNA can be distinguished by physico-chemical properties. Glues that were produced decades ago most probably contain even more impurities than today’s products. Therefore, it is of importance for paleogeneticists to know whether or not a find has been treated with natural hardeners (Cooper, 1994). In addition, curators and conservators would also appreciate a simple method for determining previous glue treatment, since many substances used for conservation of hard tissues are not compatible and will destroy the material rather than conserve it (Howie, 1984; Collins, 1995; Stoneking, 1995). We propose the extent of racemization of serine as the principal parameter for detection of previous treatment with bone glue. *Correspondence to: Lutz Bachmann, Natural History Museums and Botanical Garden, Zoological Museum, University of Oslo, Sars Gate 1, N-0562 Oslo, Norway. E-mail: firstname.lastname@example.org Received 21 September 2000; accepted 3 December 2001. DOI 10.1002/ajpa.10061 Published online in Wiley InterScience (www.interscience.wiley. com). 118 G.J. NICHOLSON ET AL. METHODS AND RESULTS The condition used for the swelling and hydrolysis of the collagen affects the degree of racemization of the constituent amino acids, in particular those susceptible to racemization under basic conditions, i.e., those with electron-withdrawing groups (e.g., ⫺OH, ⫺COOH) close to the center of chirality (Neuberger, 1948; Bada and Schröder, 1975), leading in particular to elevated levels of D-aspartic acid and D-serine. Phenylalanine is also affected, although to a much lesser degree. We determined the [D]/[L] ratio of serine, alanine, leucine, aspartic acid, glutamic acid, and phenylalanine in 64 bone samples, 14 of which are known to have been treated at least once, and 50 of which had not been treated with glue. The age of samples ranged from 2 years (modern control) to ⬃500 000 years B.P., and they originated from a variety of species including pig, cow, horse, bear, rhinoceros, mammoth, reindeer, Neandertals, and anatomically modern humans. Additionally, three different glue samples were examined: two were obtained from a local drugstore, and a third, of vintage origin, has already been used for several years in an osteological collection. The DNA that can be extracted from commercially available glue can easily be detected by a simple experiment. Granulate pills of glue were inoculated into large-size slots of an ethidium bromide-stained 1.5% (w/v) agarose gel (NuSieve/SeaKem, 3:1) and electrophoresed in 1 ⫻ TBE at 8 V/cm. Subsequently, the DNA can be extracted from the gel slice by standard methods (e.g., Pusch, 1997). Rough dotblot hybridization experiments of radioactively labeled DNA extracted from an old bone glue sample to various genomic DNAs spotted onto a nylon membrane reveal in this example a mixture of nucleic acids originating from, e.g., cow, pig, and even human (Fig. 1). This result sufficiently answers the question of whether or not DNA can survive the glue manufacturing process. In particular, there might be a surprising quantity of human DNA in the vintage glue sample used for the dot-blot hybridization experiment. However, it is impossible to identify the source of human DNA detected in the vintage glue sample, because the origin of the glue is unclear. Nevertheless, in the context of ancient DNA analysis, it is important to note that bone glue might contribute nonauthentic DNA to paleontological material. Glues that were produced decades ago most probably contain even more impurities than today’s products. Approximately 1 mg of pulverized bone sample was hydrolyzed in 200 l 6 N DCl in D2O (24 hr/ 110°C), esterified with 200 l 1.5 N DCl in CH3OD (15 min/110° C), and trifluoroacetylated with 100 l trifluoroacetic anhydride (TFAA) (10 min/110°C). Insoluble inorganic salts from the bone were largely removed by decanting the TFAA solution containing the dissolved amino-acid derivatives into a fresh vial Fig. 1. Dot-blot hybridization experiment of radioactively labeled DNA extracted from an old bone glue sample to various genomic DNAs spotted onto a nylon membrane. Spotted DNA samples are as follows (positive hybridization is indicated in bold). A: 1, soil (sample 1); 2, soil (sample 2); 3, Phage X174; 4, pig; 5, Lumbricus; 6, Drosophila; 7, cyprinid fish; 8, dog; A9, mouse. B: 1, soil (sample 3); 2, soil (sample 4); 3, E. coli; 4, Diphyllobotrium; 5, Musca; 6, starfish; 7, frog; 8, cow; 9, Tropaeolum. C: 1, soil (sample 5); 2, baltic amber; 3, Lawrist 4 cosmid vector; 4, Sepia; C5, plasmid pUC19; 6, clupeid fish; 7, cat; 8, rat; 9, human (blood). before evaporating off excess TFAA. The amino-acid derivatives, dissolved in approximately 10 l toluene, were separated by enantioselective gas chromatography on a Chirasil-Val capillary and detected by mass spectrometric selective ion monitoring, using the following ions: m/z 138 (serine), m/z 140 (alanine), m/z 182 (leucine), m/z 156 (aspartic acid), m/z 214 (glutamic acid), and m/z 162 (phenylalanine). The [D]/[L] ratio of each amino acid was calculated directly from the respective peak areas. This general technique for racemization control was described in more detail elsewhere (Gerhardt and Nicholson, 1994). The results are summarized in Figure 2. The [D]/[L]-serine values of the glue samples are in the range of 169 –201 ⫻ 10⫺3 and are similar to those of bone samples treated with glue that range from 106 –202 ⫻ 10⫺3 (mean, 151.3 ⫻ 10⫺3 ⫾ 29.7 ⫻ 10⫺3; median, 155.5 ⫻ 10⫺3; 0.1 quantile, 110.2 ⫻ 10⫺3; 0.9 quantile, 184.3 ⫻ 10⫺3). The [D]/[L]-serine values of untreated bone fragments, on the other hand, are significantly lower, ranging from 0 –98 ⫻ 10⫺3 (mean, 28.8 ⫻ 10⫺3 ⫾ 23.9 ⫻ 10⫺3; median, 22.7 ⫻ 10⫺3; 0.1 quantile, 2.3 ⫻ 10⫺3; 0.9 quantile, 72.3 ⫻ 10⫺3). The [D]/[L]-serine values of glue-treated and untreated bone samples are neither normally nor Poisson-distributed. Thus, the values of the two groups cannot be simply summarized by the mean and standard deviation. This is not unexpected, since each sample could have experienced widely different environmental conditions such as time, temperature, pH, or moisture content, all of which will affect the [D]/[L]-serine values. In accordance with the guidelines of Poinar et al. (1996), the bone material used for amino-acid race- DETECTION OF BONE GLUE TREATMENT Fig. 2. Distribution of subsurface [D]/[L]-serine values of 64 bone samples. Bar charts illustrate number of cases of [D]/[L]serine that fall into the indicated 10 ⫻ 10⫺3 intervals. Values are not normally distributed, and thus the mean (⽧), median (䊐), minimum-maximum (whiskers), and 0.1 and 0.9 quantiles (boxed) are depicted. A: Fifty bone samples that were never treated with bone glue. Additional measurements of surface samples derived from 13 untreated bones (indicated by open bars) are summarized in the small bar chart. B: Fourteen bone samples that were treated at least once with bone glue. Arrows indicate [D]/[L]serine values of three glue samples. *Glue sample used for hybridization experiment shown in Figure 1. mization was originally taken from the middle of the compacta in order to avoid any influence of surface contaminants on the values measured. However, the point in question in this study (presence or not of contamination with glue) differs from that of Poinar et al. (1996). Since subsurface sampling is invasive and causes visible damage to the specimen, the question arises as to whether surface sampling, which can be carried out with almost no visible damage, is sufficient in order to identify previous glue treatment of the fossil. Samples taken at different depths (0 –3 mm) from a glue-treated bone specimen showed similar values for [D]/[L]-serine, with a slight tendency to lower values with increasing depth. Since these measurements were made on a well-preserved bone specimen with optically intact compacta, this is also an indication of the high degree of penetration of bone glue into the compacta. Analogous measurements on an untreated bone specimen displayed the opposite 119 tendency. Here, the values of [D]/[L]-serine from the surface were significantly lower than those from subsurface samples. Therefore, additional measurements were performed on surface samples from 13 of the untreated bones previously measured with subsurface sampling. In all cases, the [D]/[L]-serine values of these surface samples were lower than corresponding values from the middle of the compacta, and ranged from 0 –32.6 ⫻ 10⫺3 (mean, 19.2 ⫻ 10⫺3 ⫾ 5.8 ⫻ 10⫺3; median, 18.1 ⫻ 10⫺3; 0.1 quantile, 14.2 ⫻ 10⫺3; 0.9 quantile, 24.2 ⫻ 10⫺3). The data indicate that reliable differentiation between glue-treated and untreated bone samples is possible. The [D]/[L]-serine values from untreated samples and samples treated with bone glue fall within two nonoverlapping intervals. The highest value of an untreated sample with subsurface sampling (98 ⫻ 10⫺3) is still lower than the smallest (106 ⫻ 10⫺3) of a sample treated with bone glue. Differentiation is even more reliable with surface sampling: here, the maximum value determined for an untreated bone specimen was 32 ⫻ 10⫺3. Thus, a [D]/[L]-serine value of 100 ⫻ 10⫺3 can be used as a threshold. [D]/[L]-serine values exceeding this threshold indicate that the material has almost certainly been treated at least once with bone glue. This holds true, regardless of the age of the bones or from which species the material originated. As stated earlier, elevated levels of D-phenylalanine can also be expected, although to a much lower degree. [D]/[L]-phenylalanine values of the 48 untreated samples taken from the compacta range from 1.3–109 ⫻ 10⫺3 (mean, 15.1 ⫻ 10⫺3 ⫾ 24.9 ⫻ 10⫺3; median, 5.7 ⫻ 10⫺3; 0.1 quantile, 3.5 ⫻ 10⫺3; 0.9 quantile, 30.2 ⫻ 10⫺3), and those of 11 untreated ones taken from the surface range from 1.2–30.4 ⫻ 10⫺3 (mean, 8.8 ⫻ 10⫺3 ⫾ 8.9 ⫻ 10⫺3; median, 4 ⫻ 10⫺3; 0.1 quantile, 2.7 ⫻ 10⫺3; 0.9 quantile, 22 ⫻ 10⫺3). The [D]/[L]-phenylalanine values of 14 glue treated samples taken from the compacta range from 13.8 – 20.8 ⫻ 10⫺3 (mean, 18. ⫻ 10⫺3 ⫾ 2.2 ⫻ 10⫺3; median, 18.9 ⫻ 10⫺3; 0.1 quantile, 14.8 ⫻ 10⫺3; 0.9 quantile, 20.3 ⫻ 10⫺3). The three glue samples have [D]/[L]-phenylalanine values of 27.5, 29, and 41 ⫻ 10⫺3, respectively. It is evident that the [D]/[L]phenylalanine ratio is less powerful for the identification of glue treatment than the [D]/[L]-serine ratio, since the intervals of [D]/[L]-phenylalanine values from glue-treated and untreated samples overlap. CONCLUSIONS Amino-acid racemization offers a valuable means of assessing the risk of glue-based DNA contamination of fossil bones. The approach has two major advantages. 1) Amino-acid racemization is a standard procedure in ancient DNA research; most scientists interested in studying fossil DNA extracted from bone material apply the technique prior to their genetic experiments in order to estimate DNA survival. Our procedure is a quick extension of this. 120 G.J. NICHOLSON ET AL. 2) The amount of sample required is small (⬃1 mg). It is sufficient and in fact preferable to sample from the bone surface, thus reducing the destruction of valuable material to a minimum. Samples showing [D]/[L]-serine values higher than 100 ⫻ 10⫺3 should not be used for analyses of ancient DNA, at least when corresponding untreated material is available. If it is inevitable to analyze DNA from glue-treated bones, one should be aware of artifacts caused by glue-based DNA contaminations. Control experiments performed to prove the authenticity of the obtained data must take into account that glue may contaminate the sample with DNA from a variety of species, including humans. Therefore, the use of species-specific primers alone in order to obtain sequence data that make phylogenetic sense is not sufficient to avoid artifacts. The degree of racemization of amino acids has already been proposed as an indicator for the presence of amplifiable authentic DNA in ancient samples (Poinar et al., 1996; Cooper, 1997; Krings et al., 1997). Racemization of aspartate is the principal criterion considered, and a value smaller than 0.08 for [D]/[L]-aspartic acid was proposed for the presence of DNA endogeneous to a find (Poinar et al., 1996). In the case of glue-treated samples, we face the problem that although values measured are always beyond this threshold, they may nevertheless contain authentic DNA (i.e., the proportion of extraneous to endogenous DNA cannot be reliably estimated). Accordingly, contamination from more recent times can allegedly be recognized by [D]/[L] ratios of alanine and leucine being higher than those of aspartic acid (Poinar et al., 1996; Cooper, 1997; Krings et al., 1997). However, this was never observed in any of the samples that had been treated with glue. Therefore, we suggest extending the list of indicators for the presence of amplifiable authentic DNA in ancient samples to include the degree of racemization of serine, in order to take into account glue treatment of the sample. However, bone glue is only one (albeit frequently used) agent to preserve bone material (Howie, 1984). A variety of substances, including shellac, casein, dammar, bee’s wax derivatives, vegetable-based tannins, and fungicidal and insecticidal agents, may have been used in the conservation of a specific find in question (Cooper, 1994). Many of these may complicate ancient DNA analyses, either as a source of contamination with proteins or nucleic acids, or by inhibiting enzyme-based DNA modifications such as the polymerase chain reaction (Horie, 1987; Hall et al., 1993; Shelton and Johnson, 1995). The poor re- cording of treatment of samples with adhesives and consolidants for conservatory purposes makes it necessary to develop methods to identify such treatments, since the investigation of curated and conserved samples will become more and more important in ancient DNA research. LITERATURE CITED Bada JL, Schröder RA. 1975. Amino acid racemization reactions and their geochemical implications. Naturwissenschaften 62: 71–79. Brommelle N, Pye EM, Smith P, Thompson G. 1984. Adhesives and consolidants: preprints of the contributions to the Paris Congress, 2– 8 September 1984. London: International Institute for Conservation of Historic and Artistic Works. Collins C. 1995. The care and conservation of palaeontological material. Oxford: Butterworth-Heinemann, Ltd. Cooper A. 1994. DNA from museum specimen. In: Herrmann B, Hummel S, editors. Ancient DNA. New York: Springer Verlag. p 149 –165. Cooper A. 1997. Neandertal genetics. Science 277:1021–1024. Cooper A, Poinar HN. 2000. Ancient DNA: do it right or not at all. Science 289:1139. Gerhardt J, Nicholson GJ. 1994. Unambiguous determination of the optical purity of peptides via GC-MS. In: Hodges RS, Smith JA, editors. Proceedings of the Thirteenth American Peptide Symposium: Peptides: Chemistry, Structure and Biology. Leiden: Escom. p 241. Hall LM, Ashworth C, Bartsiokas A, Jones DS. 1993. Experiments on inhibition problems in old tissues. Ancient DNA Newslett 1:9 –10. Handt O, Höss M, Krings M, Pääbo S. 1994. Ancient DNA: methodological challenges. Experientia 50:524 –529. Horie CV. 1987. Materials for conservation: organic consolidants, adhesives and coatings. London: Butterworths. Howie FMP. 1984. Materials used for conserving fossil specimens since 1930: a review. In: Brommelle NS, Pye EM, Smith P, Thompson G, editors. Adhesives and consolidants: preprints of the contributions to the Paris Congress, 2– 8 September 1984. London: International Institute for Conservation of Historic and Artistic Works. p 92–97. Krings M, Stone A, Schmitz RW, Krainitzki H, Stoneking M, Pääbo S. 1997. Neandertal DNA sequences and the origin of modern humans. Cell 90:19 –30. Lepper HA, Lewis GE. 1941. Materials for preparation of vertebrate fossils: an analysis of their effectiveness. Am J Sci 239: 17–24. Neuberger A. 1948. Stereochemistry of amino acids. Adv Protein Chem 4:298. Poinar HN, Höss M, Bada JL, Pääbo S. 1996. Amino acid racemization and the preservation of ancient DNA. Science 272:864 – 866. Pusch CM. 1997. A simple and fast procedure for high quality DNA isolation from gels using laundry detergent and inverted columns. Electrophoresis 18:1103–1104. Rixon AE. 1976. Fossil animals remains: their preparation and conservation. London: Athlone Press. Shelton SY, Johnson JS. 1995. Conservation of sub-fossil bone. In: Collins C, editor. The care and conservation of palaeontological material. Oxford: Butterworth-Heinemann, Ltd. p 59 – 72. Stoneking M. 1995. Ancient DNA: how do you know when you have it and what can you do with it? Am J Hum Genet 57:1259 – 1262.