Skip Navigation


MBE Advance Access originally published online on January 25, 2007
Molecular Biology and Evolution 2007 24(4):998-1004; doi:10.1093/molbev/msm015
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
24/4/998    most recent
msm015v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by Malmström, H.
Right arrow Articles by Holmlund, G.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Malmström, H.
Right arrow Articles by Holmlund, G.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

© The Author 2007. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution. All rights reserved. For permissions, please e-mail: journals.permissions@oxfordjournals.org

Research Articles

More on Contamination: The Use of Asymmetric Molecular Behavior to Identify Authentic Ancient Human DNA

Helena Malmström*, Emma M. Svensson*, M. Thomas P. Gilbert{dagger}, Eske Willerslev{dagger}, Anders Götherström* and Gunilla Holmlund{ddagger}

* Evolutionary Biology, Uppsala University, Uppsala, Sweden
{dagger} Centre for Ancient Genetics, Niels Bohr Institute and Biological Institute, University of Copenhagen, Copenhagen, Denmark
{ddagger} The National Board of Forensic Medicine, Department of Forensic Genetics and Forensic Toxicology, Linköping, Sweden

E-mail: anders.gotherstrom{at}ebc.uu.se.


   Abstract
TOP
 Abstract
Introduction
 Material and Methods
 Results and Discussion
 Acknowledgements
 References
 
Authentication of ancient human DNA results is an exceedingly difficult challenge due to the presence of modern contaminant DNA sequences. Nevertheless, the field of ancient human genetics generates huge scientific and public interest, and thus researchers are rarely discouraged by problems concerning the authenticity of such data. Although several methods have been developed to the purpose of authenticating ancient DNA (aDNA) results, while they are useful in faunal research, most of the methods have proven complicated to apply to ancient human DNA. Here, we investigate in detail the reliability of one of the proposed criteria, that of appropriate molecular behavior. Using real-time polymerase chain reaction (PCR) and pyrosequencing, we have quantified the relative levels of authentic aDNA and contaminant human DNA sequences recovered from archaeological dog and cattle remains. In doing so, we also produce data that describes the efficiency of bleach incubation of bone powder and its relative detrimental effects on contaminant and authentic ancient DNA. We note that bleach treatment is significantly more detrimental to contaminant than to authentic aDNA in the bleached bone powder. Furthermore, we find that there is a substantial increase in the relative proportions of authentic DNA to contaminant DNA as the PCR target fragment size is decreased. We therefore conclude that the degradation pattern in aDNA provides a quantifiable difference between authentic aDNA and modern contamination. This asymmetrical behavior of authentic and contaminant DNA can be used to identify authentic haplotypes in human aDNA studies.

Key Words: contamination • ancient DNA • authentication


   Introduction
TOP
 Abstract
 Introduction
Material and Methods
 Results and Discussion
 Acknowledgements
 References
 
Contaminating modern human DNA hampers studies on ancient human DNA. Ancient and modern haplotypes and alleles are often identical and there is no conclusive way of identifying contaminating modern DNA. As the contamination mainly derives from exogenous DNA in the material (Richards et al. 1995Go; Hofreiter et al. 2001Go; Malmström et al. 2005Go), replication in an independent laboratory will not eliminate the problem. Contaminant DNA largely behaves like ancient DNA (aDNA) in amplicon cloning and often yields a singular sequence or sequence variation that is the result of degradation of one original contaminant sequence (Malmström et al. 2005Go; Sampietro et al. 2006Go). Thus, authentication procedures designed to avoid and detect contaminating DNA in ancient tissue (Cooper and Poinar 2000Go) are of little help in studies on ancient human remains. The problem was acknowledged as early as a decade ago (Handt et al. 1994Go; Richards et al. 1995Go; Handt et al. 1996Go), and more recently there has been several reports on different aspects of contamination in ancient human DNA studies (Bandelt 2005Go; Gilbert et al. 2005Go; Kemp and Smith 2005Go; Malmström et al. 2005Go; Salamon et al. 2005Go; Bouwman et al. 2006Go; Sampietro et al. 2006Go). Although the problem is widely recognized, there remains a strong desire to work with genetics in ancient humans nevertheless (Dalton 2005Go; Haak et al. 2005Go; Sampietro et al. 2005Go).

It has been previously suggested that a comparison of the degree of degradation between different sources of DNA in an ancient sample (i.e., between the authentic source DNA and the contaminant sources of DNA) might be used as a tool to authenticate ancient DNA. The argument has been referred to as "appropriate molecular behavior" (Cooper and Poinar 2000Go). The logic behind this argument is as follows. Postmortem, DNA molecules degrade as a loose function of temperature and time (cf. Smith et al. 2001Go). Thus for any given source of DNA, over time there will be the generation of an increased number of short fragments and a decrease in the number of longer fragments. As contaminant sources of DNA are younger in age than the true endogenous DNA sequences, it is to be expected that the relative levels of short to long DNA fragments derived from the contaminant should be lower than for the endogenous DNA. Furthermore, the average fragment size of modern contaminant DNA should be higher in comparison to ancient degraded DNA (Noonan et al. 2005Go).

In this paper, we quantify DNA fragments of different sizes derived from contaminant human and authentic ancient dog and cow DNA. Prior to DNA extraction, the powdered samples were pretreated with bleach, a method which has been proposed as an effective means to minimize the carryover of contaminant DNA sequences from the bone in to the final DNA extract (Kemp and Smith 2005Go; Salamon et al. 2005Go). As part of this study, we investigate the efficiency of the decontamination method, through comparison of the data generated here with data previously generated from extractions on the same specimens, performed in the absence of the bleach pretreatment. We use the extracted DNA to measure the quantitative relation between long and short contaminant fragments and authentic DNA. We predict that a decrease in targeted fragment length will result in a significantly higher proportion of authentic DNA.


   Material and Methods
TOP
 Abstract
 Introduction
 Material and Methods
Results and Discussion
 Acknowledgements
 References
 
We extracted, amplified, and sequenced DNA from 23 prehistoric dog bones and teeth that had yielded DNA in a previous study (Malmström et al. 2005Go) and a bone and a tooth from 2 additional dogs (table 1) together with 9 extraction controls amplified in duplicates for all fragments and 4 polymerase chain reaction (PCR) controls, also for all fragments according to previously published protocols (Yang et al. 1998Go; Malmström et al. 2005Go). The majority of this specific material was excavated more than 10 years ago, and the previous study showed that the DNA was of good quality (Malmström et al. 2005Go). However, as the material was believed to be severely contaminated with modern human DNA, 2 modifications were added to the extraction protocols. After thorough sandpaper polishing, bones and teeth were incubated in 0.1 M HCl for 5 min, washed 3 times in ddH2O (DNA free ELGA grade) and once in 95% EtOH. Following powdering, the samples were soaked in 0.5% bleach for 15 min and washed 3 times in LiChrosolv water (Merck, Darmstadt, Germany) prior to DNA extraction. The HCl and bleach treatment had not been used in the previous study. Two fragment sizes (148 and 112 bp, denoted H148 and H112, respectively) were targeted for amplification with primers (table 2) specific for the human mitochondrial D-loop, and similar respectively) were targeted with primers (table 2) specific for the dog mitochondrial D-loop. A third, shorter fragment size (93 bp, denoted D93) in the dog D-loop was included to monitor the degradation pattern previously discovered in ancient material (Poinar et al. 2006Go). The human fragment was identical to the one targeted in the previous study, whereas the dog fragment was not (Malmström et al. 2005Go). DNA was quantified with real-time PCR following Malmström et al. (2005)Go. Eleven samples were replicated in an independent laboratory (table 1). The long fragments (H148 and D152) were amplified and quantified, and the dog fragment was sequenced.


View this table:
[in this window]
[in a new window]

  		Table 1 Number of Human and Dog mtDNA Templates in Each Dog Specimen

 

View this table:
[in this window]
[in a new window]

  		Table 2 Primers and Probes for Human and Dog mtDNA and for Cattle/Human 16S

 
Additionally, DNA from 34 historic and medieval cattle remains was included in the study and typed with an alternative method to support any results from the dog material. The cattle material was from urban contexts in western Sweden, one sample set was from an 18th century harbor (Marstrand), and yet one from a 13th century town (Skara, table 3). Thus, the cattle material was 10 ± 5% of the age of the dog material and yet from areas with about the same climate and average year temperature as the dog material. The cattle material was extracted in duplicates (serving as independent observation as there is a large variation in contamination content in duplicate extractions [Malmström et al. 2005Go]) together with 21 extraction and 10 PCR controls. As we suspected that the material could be well preserved and thus not ideal for contamination studies, we wanted to assure that the proportion of contaminant DNA was sufficiently large for further analyses. Therefore, DNA was extracted according to previous protocols (Yang et al. 1998Go; Malmström et al. 2005Go) but without the additional exposure to HCl and bleach. We did, however, perform a test where we exposed the bone powder from a subset of the samples to 2 different bleach concentrations (0.5 and 3%, table 3). Conserved primers (table 2) were designed to amplify both authentic ancient cattle mtDNA and contaminating human mtDNA for fragments of 4 different sizes (70 bp, 124 bp, 178 bp, and 180 bp) in the 16S rRNA gene. We targeted a substitution (A/T) in nucleotide position 2750 according to accession number V00654 (cattle) and at nt 2952 according to AB055387 (human), where A is specific for cattle DNA and T for human DNA. The substitution was identified and proportionally quantified with pyrosequencing (Ronaghi et al. 1998Go) using a previously reported protocol (Götherström et al. 2005Go), with the addition of proportional allele quantification as implemented in the pyrosequencing software (Gruber et al. 2002Go; Neve et al. 2002Go).


View this table:
[in this window]
[in a new window]

  		Table 3 Proportion of Cattle Fragments (A status) in Each Cattle Sample without Bleach Treatment and with Treatment of different Bleach Concentrations

 
Deviations between the variance of several of the sample groups of the dog samples rendered the use of parametric statistics problematic. Therefore, our statistical analyses involved the nonparametric Mann–Whitney U test. The test provides reasonable power without being dependent upon the shape of the variance. We compared the amount of contamination in the dog samples with previously published data (Malmström et al. 2005Go). We also compared the contamination yield in the dog bones with the contamination yield in the negative extraction controls in a similar way. For authentication, we compared the proportion of dog DNA compared with human DNA in the replicated samples that yielded data on both species (n = 7) with the original data set. To monitor changes, we restricted further statistical testing to samples that yielded quantifiable DNA for both short and long fragments (23 and 14 samples amplified with human or dog-specific primers, respectively, GenBank accession number DQ860843–DQ860864 and AY673648–AY673672 for the previously published data set). We calculated to what extent short fragments exceeded long fragments in the DNA extracts (amount of H148 or D152 fragments/amount of H112 or D111 fragments). The cattle samples were quantified in a different manner to the dog samples and thus yielded a different type of data. We used {chi}2 to calculate whether we had identified an excess of samples with an increase in the proportion of authentic aDNA (cattle DNA) compared with contaminating human DNA as the fragment size decreased. This was done for the shortest fragment (70 bp) compared with the increasingly longer fragments (124 bp, 178 bp, and 180 bp). The samples that had yielded results for all 4 fragments were used for a simple regression with size as the independent variable and proportion of aDNA as the dependent variable. As we were interested in the trend, we normalized the data set by dividing all observations in each sample with the one observed for 70 bp prior to calculations. All calculations were performed on STATISTICA 7.


   Results and Discussion
TOP
 Abstract
 Introduction
 Material and Methods
 Results and Discussion
Acknowledgements
 References
 
Bleach has previously been used on limited amounts of powdered bone and tooth material as a means to decontaminate samples (Kemp and Smith 2005Go; Salamon et al. 2005Go). In our data, we found that the level of contamination was significantly lower (P < 0.001, Z = 7.7924) among the dog extractions where we treated the powder with bleach than when powdered bone and tooth from the same samples were not treated with bleach (the previously published data set of Malmström et al. [2005]Go). The amount of authentic aDNA also decreased in these specimens, but to a lesser degree (77% of the authentic aDNA was lost, whereas >99% of the contamination was lost, fig. 2A). The cattle material shows a similar effect when exposed to bleach. Furthermore, we noted that with the cattle material, an increase in the bleach concentration from 0.5% to 3% did not appear to enhance the result or provide a higher concentration of authentic aDNA (fig. 3 and table 3). There was significantly more contamination left in the ancient dog extracts than in the extraction and PCR controls (P = 0.0005, Z = –3.47517; the controls contained a maximum of 55 molecules and an average of 17 molecules per sample; none contained dog DNA). For the dog-specific primers, the yield ranged from 0 to 14821 (the highest amount was observed in D93) molecules, where the average of all 25 samples was 1491 molecules for D93, 879 molecules for D111, and 188 molecules for D152 (table 4). The contaminating human molecules ranged from 0 to 248 molecules (the highest amount was observed in H148), where the average of all 25 samples was 55 molecules for H112 and 29 molecules for H148 (table 4). The yield appeared to be somewhat higher in the replication (the highest amounts were for D152, which yielded a maximum of 2,034 starting molecules and H148, which yielded a maximum of 1,268 starting molecules). As it was a general trend for contamination as well as for authentic ancient DNA, it could simply be that the experiment was somewhat more efficient in the replication. However, the proportion between human contamination and ancient dog DNA in the replication did not deviate significantly from the original sample set (P = 0.28, Z = 1.0702). The increase of DNA yield with decreased fragment size was significantly higher for authentic aDNA than for contaminating human DNA when samples that yielded DNA for long as well as for short fragments were considered (P = 0.0011, Z = –3.2569, figs. 1 and 2A). This trend was further confirmed by quantification in the cattle material (fig. 2B, table 4). For the cattle material, the average percent of cattle DNA in the negative controls was 2.3 ± 7.8 (standard deviation [SD]) for 70 bp, 0.3 ± 0.6 (SD) for 124 bp, 0.1 ± 0.6 (SD) for 178 bp, and 0 ± 0 (SD) for 180 bp. Only in 1 case out of 84 did we register a sufficiently high proportion of cattle DNA (35% in one of the 70 bp amplifications from a negative extraction control) to conclude with certainty that there actually was cattle DNA in the negative control (Gruber et al. 2002Go; Neve et al. 2002Go). A significant difference for more authentic aDNA in the shorter fragments than in the longer fragments was obtained in 2 cases out of 3, when longer fragments were compared with the 70 bp fragment. This pattern was evident when the size difference increased (70 bp/124 bp: P = 0.25, {chi}2 = 1.35, n = 54; 70 bp/178 bp: P = 0.037, {chi}2 = 4.36, n = 48; 70 bp/180 bp: P = 0.024, {chi}2 = 5.06, n = 59; and 1 degree of freedom in all cases). A simple regression also indicated a significant correlation for fragment size and proportion of contamination (P = 0.0035, F = 8.736). However, this difference is not evident when fragments are visualized on agarose gels after conventional PCR. All of the 25 dog samples showed presence of human-specific amplicons for the H112 fragment and 23 of them did so also for the H148 fragment, whereas 17 showed dog-specific amplicons for the D111 fragment and 15 did so for the D152 fragment.


Figure 2
View larger version (17K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		FIG. 2.— Illustration of contamination in ancient material. (A) Frequency of dog samples with specific amounts of contaminating (H112) and authentic ancient molecules (D111), n = 25. Although the amount of contaminating molecules rarely exceeds 100 and never 300 molecules in the samples exposed to HCl and bleach, the amount of authentic ancient molecules may reach several thousands. (B) Frequency (in %) of extractions from cattle samples yielding specific proportion of authentic ancient DNA, n = 117 for 70 and 124 bp, and n = 113 for 178 and 180 bp. More samples provide short authentic aDNA fragments compared with contaminating DNA fragments, whereas less samples provide long authentic aDNA fragments.

 

Figure 3
View larger version (10K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		FIG. 3.— Cattle remains were treated with bleach in different concentrations (0%, n = 64; 0.5%, n = 77; and 3%, n = 68) prior to pyrosequencing. The proportion of contamination appears to decrease with bleach treatment, but whether the treatment is with 0.5% or 3% bleach appears to have little effect on the final result.

 

View this table:
[in this window]
[in a new window]

  		Table 4 Average and Maximum Values for Numbers of Dog and Human Mitochondrial Fragments in Dog Samples and Proportion of Human Fragments in Cattle Samples

 

Figure 1
View larger version (12K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		FIG. 1.— Size-related increase of template frequencies. The first box illustrates the relation between the H112 human fragment and the H148 human fragment, where there are 3.4 times more of the shorter than of the longer fragment. The second and third box illustrates the same relation between the D111 and the D152 dog fragment where there is 5.9 times more of the shorter than the longer fragment, and for the D93 and the D152 dog fragment where there is 9.6 times more of the shorter than the longer fragment. The dog observations are based on 14 samples, and the human observations are based on 23 samples.

 
Aggressive pretreatment of the dog material, in our case with HCl and bleach, eliminated a large proportion of the contaminant DNA (fig. 2A and 3). We also conclude that authentic aDNA will show a more rapid increase in yield with decreased fragment size than contaminating DNA, even when using material excavated a century ago, which has been well handled since and thus likely contains a large number of old contaminant DNA molecules. We could detect this pattern with 2 different quantification methods and in 2 different types of data sets. This asymmetrical behavior is the only known detectable and quantifiable difference between contaminating modern human and ancient human DNA, and we therefore suggest that it can be used to support claims for authentic ancient human DNA. Typically, a human aDNA extract yields several different haplotypes, both authentic and contaminant. However, when an internal shorter type-specific fragment is targeted, the number of ancient haplotype copies should increase disproportionately compared with the contaminant haplotypes. Quantification of haplotypes or alleles in amplicons of different fragment lengths should thus allow researchers to single out authentic human DNA.


   Acknowledgements
TOP
 Abstract
 Introduction
 Material and Methods
 Results and Discussion
 Acknowledgements
References
 
We thank Love Dalén, Rolf Quam, Jan Storå, Cecilia Anderung, and Maria Vretemark for assistance and the Swedish Research Council for financial support. We also thank 2 anonymous reviewers for substantial comments that helped to improve the study.


   Footnotes
 
Lisa Matisoo-Smith, Associate Editor


   References
TOP
 Abstract
 Introduction
 Material and Methods
 Results and Discussion
 Acknowledgements
 References
 

    Bandelt HJ. (2005) Mosaics of ancient mitochondrial DNA: positive indicators of nonauthenticity. Eur J Hum Genet 13:1106–1112.[CrossRef][ISI][Medline]

    Bouwman AS, Chilvers ER, Brown KA, Brown TA. (2006) Brief communication: identification of the authentic ancient DNA sequence in a human bone contaminated with modern DNA. Am J Phys Anthropol 131:428–431.[CrossRef][ISI][Medline]

    Cooper A and Poinar HN. (2000) Ancient DNA: do it right or not at all. Science 289:1139.

    Dalton R. (2005) Caveman DNA hints at map of migration. Nature 436:162.[Medline]

    Gilbert MTP, Rudbeck L, Willerslev E, Hansen AJ, Smith C, Penkman KEH, Prangenberg K, Nielsen-Marsh CM, Jans ME, Arthur P. (2005) Biochemical and physical correlates of DNA contamination in archaeological human bones and teeth excavated at Matera, Italy. J Archaeol Sci 32:785–793.

    Götherström A, Anderung C, Hellborg L, Elburg R, Smith C, Bradley DG, Ellegren H. (2005) Cattle domestication in the Near East was followed by hybridization with aurochs bulls in Europe. Proc Biol Sci 272:2345–2350.

    Gruber JD, Colligan PB, Wolford JK. (2002) Estimation of single nucleotide polymorphism allele frequency in DNA pools by using pyrosequencing. Hum Genet 110:395–401.[CrossRef][ISI][Medline]

    Haak W, Forster P, Bramanti B, Matsumura S, Brandt G, Tanzer M, Villems R, Renfrew C, Gronenborn D, Alt KW, Burger J. (2005) Ancient DNA from the first European farmers in 7500-year-old Neolithic sites. Science 310:1016–1018.[Abstract/Free Full Text]

    Handt O, Hoss M, Krings M, Pääbo S. (1994) Ancient DNA: methodological challenges. Experientia 50:524–529.[CrossRef][ISI][Medline]

    Handt O, Krings M, Ward RH, Pääbo S. (1996) The retrieval of ancient human DNA sequences. Am J Hum Genet 59:368–376.[ISI][Medline]

    Hofreiter M, Serre D, Poinar HN, Kuch M, Paabo S. (2001) Ancient DNA. Nat Rev Genet 2:353–359.[CrossRef][ISI][Medline]

    Kemp BM and Smith DG. (2005) Use of bleach to eliminate contaminating DNA from the surface of bones and teeth. Forensic Sci Int 154:53–61.[CrossRef][ISI][Medline]

    Malmström H, Storå J, Dalén L, Holmlund G, Götherström A. (2005) Extensive human DNA contamination in extracts from ancient dog bones and teeth. Mol Biol Evol 22:2040–2047.[Abstract/Free Full Text]

    Neve B, Froguel P, Corset L, Vaillant E, Vatin V, Boutin P. (2002) Rapid SNP allele frequency determination in genomic DNA pools by pyrosequencing. Biotechniques 32:1138–1142.[ISI][Medline]

    Noonan JP, Hofreiter M, Smith D, Priest JR, Rohland N, Rabeder G, Krause J, Detter JC, Pääbo S, Rubin EM. (2005) Genomic sequencing of Pleistocene cave bears. Science 309:597–599.[Abstract/Free Full Text]

    Poinar HN, Schwarz C, Qi J, et al. (2006) Metagenomics to paleogenomics: large-scale sequencing of mammoth DNA. Science 311:392–394 (13 co-authors).[Abstract/Free Full Text]

    Richards MB, Sykes BC, Hedges REM. (1995) Authenticating DNA extracted from ancient skeletal remains. J Archaeol Sci 22:291–299.

    Ronaghi M, Uhlén M, Nyrén P. (1998) A sequencing method based on real-time pyrophosphate. Science 281:363–365.[Free Full Text]

    Salamon M, Tuross N, Arensburg B, Weiner S. (2005) Relatively well preserved DNA is present in the crystal aggregates of fossil bones. Proc Natl Acad Sci USA 102:13783–13788.[Abstract/Free Full Text]

    Sampietro ML, Caramelli D, Lao O, Calafell F, Comas D, Lari M, Agusti B, Bertranpetit J, Lalueza-Fox C. (2005) The genetics of the pre-Roman Iberian Peninsula: a mtDNA study of ancient Iberians. Ann Hum Genet 69:535–548.[CrossRef][ISI][Medline]

    Sampietro ML, Gilbert MTP, Lao O, Caramelli D, Lari M, Bertranpetit J, Lalueza-Fox C. (2006) Tracking down human contamination in ancient human teeth. Mol Biol Evol 23:1801–1807.[Abstract/Free Full Text]

    Smith CI, Chamberlain AT, Riley MS, Cooper A, Stringer CB, Collins MJ. (2001) Neanderthal DNA. Not just old but old and cold? Nature 410:771–772.

    Yang DY, Eng B, Waye JS, Dudar JC, Saunders SR. (1998) Technical note: improved DNA extraction from ancient bones using silica-based spin columns. Am J Phys Anthropol 105:539–543.[CrossRef][ISI][Medline]

Accepted for publication January 22, 2007.


Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?



This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
24/4/998    most recent
msm015v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by Malmström, H.
Right arrow Articles by Holmlund, G.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Malmström, H.
Right arrow Articles by Holmlund, G.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?