Our paper on the microbiota of Andean mummies gut contents (Santiago-Rodriguez et al.2016) has recently raised some questions detailed by Eisenhofer, Cooper and Weyrich (2017)., where the authors boldly claim an ‘unacceptable standard for future work’. This is not the first time our work has come under criticism by the same group of colleagues, and although we work in an area that is always open to questions, we can only publish data we are certain are not artifacts brought about by extant contamination. We completely agree with the authors in that it is extremely difficult to work with these precious ancient samples. However, this is where the agreement stops. The same group had raised the very same criticisms to a previous paper (Weyrich, Llamas and Cooper 2014), and our proper rebuttal was published side by side (Cano et al.2014). In spite of this effort to clear up any possible doubts, our colleagues have not resorted to constructive criticism, but rather raised even more doubts in spite of our efforts to clear them up. In a similar manner, the same group has raised doubts as to the veracity of a paper of the presence of Enterococcus spp. in an ancient sample of a woolly mammoth (Goncharov et al.2016). Although we cannot vouch for the data published in the latter paper, it does demonstrate that our esteemed colleagues are on the prowl for publications dealing with ancient DNA. It is good to know that there is a group that is acting as a watchdog group, and we appreciate their efforts to point out possible errors in the ways of others. However, in all our publications we have tried to be as transparent as possible with all the methods used. In fact, another group of colleagues also critiqued a later paper we published (Rivera-Perez et al.2015, Fellows-Yates, Andrades-Valtueña and Herbig 2016) for some of the same reasons, and all their questions were answered in a rebuttal (Rivera-Perez, Cano and Toranzos 2016). The current critique by Eisenhofer, Cooper and Weyrich (2017) questions the use of agarose gels as a means of detecting possible contamination. Perhaps it is not clear in our published paper, but these gels were also run on negative controls after PCR amplification, not as stated by Eisenhofer, Cooper and Weyrich (2017) where they are correct in stating that agarose gels do not have the exquisite sensitivity needed for this type of work. Readers can relax, since these controls were designed precisely as our colleagues state. In fact, all reagents are used only once, store fresh and so there is no possibility of contamination. We also include negative controls in all our PCR reactions; thus, if there was any type of contamination, including in the reagents, this would show up in the gels, once again, after PCR amplification. Our colleagues seem to confuse fecal content of the intestine with coprolites, which may also be a source of their doubts. Coprolites are archaeological artifacts one might step on during an excavation, since they are found externally to any mummy, and as such present other sets of problems with possible contamination that are far greater than those encountered when using fecal contents obtained from within the mummies, like the samples analyzed in the criticized study. Once again, we were under the impression that we answered many questions on coprolites, but our colleagues do not seem to be satisfied. Additionally, we work with well-known and respected pathologists of mummies, who are also co-authors in this paper. Perhaps we were not completely clear in our description of the samples, thinking that those groups working with ancient DNA would be familiar with the procedures used when handling mummies. The samples were obtained under surgical conditions, as would be required for any modern surgery; this is done as a means of working with invaluable historical samples, such as mummies. The possibility of contamination under these conditions is minimal, but all controls are in place during the DNA isolation (which is performed in a separate specialized ancient DNA laboratory, totally DNA-free) and especially during PCR amplification, if this step is to be carried out. If all these controls and care are not enough in the eyes of our colleagues, there is little else we can do; although our goal is not to please this highly critical group, but rather to prevent any type of contamination that will lead to the wrong conclusions. The use of 16S rRNA gene amplification in ancient microbiome analyses has its limitations, and we are painfully aware of these; however, it may not be clearly explained that the length of the V4 region is within the ‘typical length’ expected with ancient DNA samples. Although it has been recently suggested that the V4 variable region may be impractical for ancient DNA analyses because the target size is approximately >200 bp (Ziesemer et al.2015), and that authentic ancient DNA might not be targeted when amplifying the V4 region, experts that have been in this field for decades agree that an amplicon of <300 bp is still within what is considered reasonable in ancient DNA analyses (Drancourt et al.1998, Roberts and Ingham 2008). In addition, the V4 region has been better characterized and is known to have a better phylogenetic resolution (Yang, Wang and Qian 2016); thus, the use of one region over the other may not be very high in the importance scale. We would, however, also emphasize that the use of one specific region allows for a better comparison between samples in the literature. For instance, Yang et al. also demonstrated that the V3 region does not give a satisfactory phylogenetic resolution and analysis, and that V4, V5 and V6 perform the best. These recent data may suggest that the V3 region, although in the range of 50–160 bp, may not be the best for phylogenetic analysis. Unfortunately, results based on the V3 region were selected over the V6 region in the study of the dental calculus microbiome of subjects from the Neolithic and Industrial revolutions (Adler et al.2013). It may also be surprising to the dogmatic, but it is entirely possible that complete intracellular genomes may remain undegraded over millenia, if these ancient microorganisms were rapidly dehydrated, as is the case of the process of natural mummification. We also explained in a previous rebuttal (Rivera-Perez, Cano and Toranzos 2016) that intracellular DNA (such as the nucleic acids that remain undegraded) is more likely to be in large pieces rather than the ‘typical ranges between 50–160 bp’, as stated by our colleagues. To infer that only small pieces will be found may be more in the realm of an opinion than a proven fact. We agree that physicochemical variables that are part of the taphonomic conditions will also greatly influence the nucleic acids, but we also argue that intracellular nucleic acids are more protected against these conditions. It is noteworthy that recent published data indicate that other macromolecules (Cappellini et al.2011) can also survive the taphonomic conditions over millennia, going against the previously accepted norm. Indeed, we are aware that MapDamage does not work properly on microbiomes unless there is a high genome coverage that could enable the identification of nucleotide misincorporations (Ginolhac et al.2011). MapDamage analyses were performed following the tutorial and parameters described in https://ginolhac.github.io/mapDamage/. Interestingly, MapDamage has given differing results when working with ancient microbial DNA. It is possible that the absence of mismatches directly at the 5΄ end may be a technical artifact resulting from the mapping tool utilized, which could have been one of the potential reasons of not observing the expected nucleotide misincorporation in our study; however, Maixner et al. MapDamage output from modern DNA, Otzi's DNA and Treponema denticola clearly demonstrates that our MapDamage output does not resemble the MapDamage output of modern DNA (Maixner et al.2014). This, in turn, indicates that it is highly unlikely that we were working with modern contaminants. Maixner et al. also showed an increased DNA damage in the T. denticola specific reads, but the damage patterns occurred at an order of magnitude higher compared to Otzi's DNA. Importantly, they showed that the C to T misincorporation pattern in T. denticola is not restricted to the 5΄ end. These and our data clearly indicate that MapDamage may give very differing results depending on the ancient DNA sequenced (i.e. human vs. microbial), the type of microorganism and bioinformatic tools utilized (i.e. mapping) (Maixner et al.2014). Finally, our colleagues boldly state that the publications by one of our coauthors in the 1990s have been ‘widely discredited’. We find this comment to be acceptable only as an ‘alternative fact’, which may be fine within the political arena, but is completely out of place in a scientific journal. We are not aware of any scientific literature that has ‘widely discredited’ these data. In fact, in our previous rebuttal to our colleagues we clearly explained our findings and all the controls used, in great detail, and all the extreme care taken to avoid contamination. Any of us would surely resort to a withdrawal of a paper that has been discredited by a proper study. We can rest assured that there is no ‘wasted money and valuable time’ (as our colleagues state) in the line of work we do. In fact, we appreciate the money and valuable time spent by our colleagues on critiquing a paper, and we can all be sure that none of us is spending money and valuable time on an endeavor that may damage the credibility of the field. The credibility of the field may be damaged, however, by dogmatic approaches in any field of science. The best we can do is to present the best data we can obtain under the proper conditions, and trust the reviewers and the readers to draw their own conclusions based on those data; this is, after all, part of the peer-review process.

Proper authentication of ancient DNA is essential, yes; but so are undogmatic approaches

FORNACIARI, GINO
2017-01-01

Abstract

Our paper on the microbiota of Andean mummies gut contents (Santiago-Rodriguez et al.2016) has recently raised some questions detailed by Eisenhofer, Cooper and Weyrich (2017)., where the authors boldly claim an ‘unacceptable standard for future work’. This is not the first time our work has come under criticism by the same group of colleagues, and although we work in an area that is always open to questions, we can only publish data we are certain are not artifacts brought about by extant contamination. We completely agree with the authors in that it is extremely difficult to work with these precious ancient samples. However, this is where the agreement stops. The same group had raised the very same criticisms to a previous paper (Weyrich, Llamas and Cooper 2014), and our proper rebuttal was published side by side (Cano et al.2014). In spite of this effort to clear up any possible doubts, our colleagues have not resorted to constructive criticism, but rather raised even more doubts in spite of our efforts to clear them up. In a similar manner, the same group has raised doubts as to the veracity of a paper of the presence of Enterococcus spp. in an ancient sample of a woolly mammoth (Goncharov et al.2016). Although we cannot vouch for the data published in the latter paper, it does demonstrate that our esteemed colleagues are on the prowl for publications dealing with ancient DNA. It is good to know that there is a group that is acting as a watchdog group, and we appreciate their efforts to point out possible errors in the ways of others. However, in all our publications we have tried to be as transparent as possible with all the methods used. In fact, another group of colleagues also critiqued a later paper we published (Rivera-Perez et al.2015, Fellows-Yates, Andrades-Valtueña and Herbig 2016) for some of the same reasons, and all their questions were answered in a rebuttal (Rivera-Perez, Cano and Toranzos 2016). The current critique by Eisenhofer, Cooper and Weyrich (2017) questions the use of agarose gels as a means of detecting possible contamination. Perhaps it is not clear in our published paper, but these gels were also run on negative controls after PCR amplification, not as stated by Eisenhofer, Cooper and Weyrich (2017) where they are correct in stating that agarose gels do not have the exquisite sensitivity needed for this type of work. Readers can relax, since these controls were designed precisely as our colleagues state. In fact, all reagents are used only once, store fresh and so there is no possibility of contamination. We also include negative controls in all our PCR reactions; thus, if there was any type of contamination, including in the reagents, this would show up in the gels, once again, after PCR amplification. Our colleagues seem to confuse fecal content of the intestine with coprolites, which may also be a source of their doubts. Coprolites are archaeological artifacts one might step on during an excavation, since they are found externally to any mummy, and as such present other sets of problems with possible contamination that are far greater than those encountered when using fecal contents obtained from within the mummies, like the samples analyzed in the criticized study. Once again, we were under the impression that we answered many questions on coprolites, but our colleagues do not seem to be satisfied. Additionally, we work with well-known and respected pathologists of mummies, who are also co-authors in this paper. Perhaps we were not completely clear in our description of the samples, thinking that those groups working with ancient DNA would be familiar with the procedures used when handling mummies. The samples were obtained under surgical conditions, as would be required for any modern surgery; this is done as a means of working with invaluable historical samples, such as mummies. The possibility of contamination under these conditions is minimal, but all controls are in place during the DNA isolation (which is performed in a separate specialized ancient DNA laboratory, totally DNA-free) and especially during PCR amplification, if this step is to be carried out. If all these controls and care are not enough in the eyes of our colleagues, there is little else we can do; although our goal is not to please this highly critical group, but rather to prevent any type of contamination that will lead to the wrong conclusions. The use of 16S rRNA gene amplification in ancient microbiome analyses has its limitations, and we are painfully aware of these; however, it may not be clearly explained that the length of the V4 region is within the ‘typical length’ expected with ancient DNA samples. Although it has been recently suggested that the V4 variable region may be impractical for ancient DNA analyses because the target size is approximately >200 bp (Ziesemer et al.2015), and that authentic ancient DNA might not be targeted when amplifying the V4 region, experts that have been in this field for decades agree that an amplicon of <300 bp is still within what is considered reasonable in ancient DNA analyses (Drancourt et al.1998, Roberts and Ingham 2008). In addition, the V4 region has been better characterized and is known to have a better phylogenetic resolution (Yang, Wang and Qian 2016); thus, the use of one region over the other may not be very high in the importance scale. We would, however, also emphasize that the use of one specific region allows for a better comparison between samples in the literature. For instance, Yang et al. also demonstrated that the V3 region does not give a satisfactory phylogenetic resolution and analysis, and that V4, V5 and V6 perform the best. These recent data may suggest that the V3 region, although in the range of 50–160 bp, may not be the best for phylogenetic analysis. Unfortunately, results based on the V3 region were selected over the V6 region in the study of the dental calculus microbiome of subjects from the Neolithic and Industrial revolutions (Adler et al.2013). It may also be surprising to the dogmatic, but it is entirely possible that complete intracellular genomes may remain undegraded over millenia, if these ancient microorganisms were rapidly dehydrated, as is the case of the process of natural mummification. We also explained in a previous rebuttal (Rivera-Perez, Cano and Toranzos 2016) that intracellular DNA (such as the nucleic acids that remain undegraded) is more likely to be in large pieces rather than the ‘typical ranges between 50–160 bp’, as stated by our colleagues. To infer that only small pieces will be found may be more in the realm of an opinion than a proven fact. We agree that physicochemical variables that are part of the taphonomic conditions will also greatly influence the nucleic acids, but we also argue that intracellular nucleic acids are more protected against these conditions. It is noteworthy that recent published data indicate that other macromolecules (Cappellini et al.2011) can also survive the taphonomic conditions over millennia, going against the previously accepted norm. Indeed, we are aware that MapDamage does not work properly on microbiomes unless there is a high genome coverage that could enable the identification of nucleotide misincorporations (Ginolhac et al.2011). MapDamage analyses were performed following the tutorial and parameters described in https://ginolhac.github.io/mapDamage/. Interestingly, MapDamage has given differing results when working with ancient microbial DNA. It is possible that the absence of mismatches directly at the 5΄ end may be a technical artifact resulting from the mapping tool utilized, which could have been one of the potential reasons of not observing the expected nucleotide misincorporation in our study; however, Maixner et al. MapDamage output from modern DNA, Otzi's DNA and Treponema denticola clearly demonstrates that our MapDamage output does not resemble the MapDamage output of modern DNA (Maixner et al.2014). This, in turn, indicates that it is highly unlikely that we were working with modern contaminants. Maixner et al. also showed an increased DNA damage in the T. denticola specific reads, but the damage patterns occurred at an order of magnitude higher compared to Otzi's DNA. Importantly, they showed that the C to T misincorporation pattern in T. denticola is not restricted to the 5΄ end. These and our data clearly indicate that MapDamage may give very differing results depending on the ancient DNA sequenced (i.e. human vs. microbial), the type of microorganism and bioinformatic tools utilized (i.e. mapping) (Maixner et al.2014). Finally, our colleagues boldly state that the publications by one of our coauthors in the 1990s have been ‘widely discredited’. We find this comment to be acceptable only as an ‘alternative fact’, which may be fine within the political arena, but is completely out of place in a scientific journal. We are not aware of any scientific literature that has ‘widely discredited’ these data. In fact, in our previous rebuttal to our colleagues we clearly explained our findings and all the controls used, in great detail, and all the extreme care taken to avoid contamination. Any of us would surely resort to a withdrawal of a paper that has been discredited by a proper study. We can rest assured that there is no ‘wasted money and valuable time’ (as our colleagues state) in the line of work we do. In fact, we appreciate the money and valuable time spent by our colleagues on critiquing a paper, and we can all be sure that none of us is spending money and valuable time on an endeavor that may damage the credibility of the field. The credibility of the field may be damaged, however, by dogmatic approaches in any field of science. The best we can do is to present the best data we can obtain under the proper conditions, and trust the reviewers and the readers to draw their own conclusions based on those data; this is, after all, part of the peer-review process.
2017
Toranzos, Gary A; Santiago Rodriguez, Tasha M; Cano, Raul J; Fornaciari, Gino
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11568/863409
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? 1
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact