Yesterday evening, we held the first talk organised by us on Second Nature. It was given by Dr Phil Holliger from the Medical Research Council and was all about Ancient DNA and ways to repair it more accurately.

_Picture by Troy McLuhan" _

About 40 or so people came to the talk, and there were lots of questions for Phil afterwards. His slides, talk and the full discussion are all in the full post below.

All in all, I thought it was really successful – an excellent turnout and loads of really positive feedback. Couple of technical glitches to iron out, and I think we’ve learnt something about what does and doesn’t work on slides in Second Life.

Most interestingly, I now feel much more confident of doing these talks, and I want to try audio next: we will still do text, because so many people still can’t use voice, but I want to try doing audio and text simultaneously, see how that works out. Maybe it will be really distracting for those who do have audio (a bit like watching subtitles on the TV) but I still think it’s worth a try. Maybe next week, maybe the week after – stay tuned for news.

Anyway, I want to say a huge thank you to Phil for coming and being our very first guinea pig and thank you to everyone for coming – see you at the next one!

Below are the slides and text of the talk. The text has been slightly cleaned up to remove unneccesary paragraph breaks and typos, but no actual content changes. The discussion has been entirely cleaned up to remove noise and to make it more readable. I’ve also moved some of the comments about – i.e. put the answers underneath the questions to which they actually relate. I did that because I thought it would be much better for people who couldn’t attend to follow. If anyone thinks this is disastrous or tantamount to censorship, let me know in the comments. I do still have the full transcript, of course, and anyone who wants a copy is very welcome.

Anyway, on with the talk: I welcomed Dr Phil Holliger from the Medical Research Council in Cambridge, and he took it away…

——————————————————-

In 79 A.D., Mount Vesuvius erupted and buried two towns. One of these was Pompeii; the other was Herculaneum, a seaside resort for wealthy Romans.

Among the villas uncovered by excavation was the summer residence of Julius Caesar’s father-in-law, Lucius Calpurnius Piso. The picture on the top right shows a reconstruction of the villa forming part of the Getty museum in LA.

Inside the villa a large number of what appeared to be sticks of charcoal were discovered, some of them bundled together. These turned out to be ancient papyrus scrolls carbonized by the volcanic heat. The “Villa dei Papyrii” as it became known, contained the only known library of antiquity.

While some part of the text remained legible, large parts of the scrolls were initially uninterpretable. However, the information encoded in this ancient library was not lost but had to wait for better technology. Today, multi-spectral imaging promises a breakthrough in deciphering the fragile scrolls.
———————————————————

Just like the scrolls of Herculaneum, many specimens of paleontological, archaeological, or forensic interest contain a wealth of information. But this information is written in their DNA.

Ancient DNA sequences have been isolated from a wide variety of sources and have provided information about human migration, animal and crop domestication and the genetic relationship between modern Homo sapiens and its closest extinct relative H. neanderthalensis.

Only minuscule amounts of DNA survive in the ancient samples. DNA is therefore amplified from ancient samples by an iterative process called the polymerase chain reaction or PCR, shown on the right panel.

DNA molecules can be mass-produced, amplified from incredibly small amounts by PCR. Even a single molecule of DNA can be copied and amplified exponentially until millions of copies. The copying process is carried out by thermostable polymerases like Taq polymerase, derived from Thermus aquaticus, a bacterium that lives in the boiling hot springs of Yellowstone National Park.
———————————————————

Ancient DNA can provide a window to the past but damage to the DNA makes the encoded information hard to read. Even under the best burial conditions, DNA continues to deteriorate, as the many DNA repair pathways, which maintain the integrity of the genome in living organisms no longer work. This damage either limits the length of continuous sequence that can be recovered or renders even well preserved specimens unuseable despite the demonstrable presence of DNA (by hybridization).

We reasoned that genetic information encoded in such samples might not be lost but simply inaccessible due to the fact that the DNA polymerases cannot read across sites of damage. Think of a zipper. The zip stops at a broken tooth, even though there are many other teeth past the site of damage. We need a tool that can get past these sites of damage. These are Polymerases capable of replicating across DNA damage, and they should therefore be able to allow the deciphering of previously unreadable ancient DNA sequences just like modern imaging is helping decipher the burnt scrolls of Herculaneum.

However, no such polymerases suitable for PCR exist in nature. We therefore decided to generate such polymerases in the laboratory using one of nature’s tricks: evolution.
———————————————————

Darwinian evolution can be applied not just to organisms but to molecules too. Thus, molecular properties can be improved by iterative cycles of mutation, selection and amplification.
———————————————————

We devised a strategy for the selection of polymerase function, which we call “compartmentalized self-replication” or CSR.

CSR is based on a simple feedback loop, whereby a polymerase replicates its own encoding gene. Compartmentalization serves to isolate individual self-replication reactions from each other.
Compartmentalization is a crucial aspect of life. All living organisms are made from cells, which encase the genome and the proteins it encodes within a lipid membrane.

We use a different approach to nature. In our case, polymerase genes and the polymerase they encode are encapsulated in water droplets dispersed in oil, i.e. a water-in-oil emulsion. These “artificial cells” ensure genotype-phenotype linkage, i.e. they ensure that a polymerase only replicates only their own encoding gene and no other.

In such a system adaptive gains by a polymerase directly (and proportionally) translate into more “offspring”. In other words, if a polymerase (purple spheres) is well adapted to the selection conditions, it can replicate its own encoding gene and produce many copies of itself, i.e. “offspring”. However, if a polymerase (yellow hexagon) is poorly adapted, it cannot self-replicate and its gene will disappear from the gene pool.
———————————————————

When random changes in the polymerase genes proved unproductive we turned to a method called molecular breeding. This is like breeding of animals for distinctive traits, but instead of animals we are using our different Polymerasees, and by choosing ones with properties we like we get a Selection, or a library, of different types with variations on the properties that we are interested in.

Molecular breeding samples only functional diversity and therefore a large number of our new types are active.

As shown in the left panel, we recombined three thermophilic polymerase (DNA pol I) genes from the genus Thermus: Taq polymerase from (Thermus aquaticus), the standard polymerase used for PCR amplification of ancient DNA, Tth (T. thermophilus) and Tfl (T. flavus)) to create a polymerase library for selection.

The right panel shows a three-dimensional model of Taq polymerase. Residues deriving from Tth or Tfl that we find in our selected polymerases are shown in different shades of blue. The darker the blue the more often they occur. This just to illustrate how the offspring of an evolution experiment can comprise a patchwork of elements of the parental genes.
———————————————————

Evolving polymerases, which combine the processivity and selectivity required for PCR amplification with a high tolerance for template damage is challenging. Furthermore, damage tolerance should be generic as detailed information about the forms of DNA damage in ancient samples is lacking (and damage may vary depending on burial conditions).

In the top panel: Many lesions (red X) abrogate base pairing and yield distorted, non-cognate 3’ structures, similar to mismatches.

While natural polymerases readily extend of matched primer terminus (ending in a cognate GC base-pair), they stall at mismatches or sites of damage (G.X mispair). Extension is significantly slowed down not just at the 3’ end but also up to four bases downstream (highlighted in red).

In order to maximize tolerance to such distorted primer-template structures, we decided to select for polymerases capable of extending a primer 3’ terminus preceded by up to four mismatched bases.

We now look at some of the results we generated.

The bottom left panel shows the selection scheme. Two independent aqueous compartments of the water-in-oil emulsion are shown. Polymerases (such as Pol1 (left compartment)) that are capable of utilizing quadruple mismatch primers (AGGG·AGGG, GGTG·GGTG) to replicate their one encoding gene (pol1) produce “offspring” and increase their copy number in the post-selection population, while polymerases like Pol2 (right compartment) that are unable to utilize quadruple mismatch primers disappear from the gene pool

After three rounds of CSR selection, we recovered a diverse set of polymerases with novel properties including the generic ability to utilize single, double and quadruple mismatches (as seen in the bottom right picture for the polymerases called 3D1). They could also process unusual DNA structures and bypass template lesions such as abasic sites or hydantoins, as will be shown in the following two slides.
———————————————————

We examined primer extension reactions using a radiolabelled primer and looked at products on a polyacrylamide gel (top left). We examined extension of three different quadruple mismatches: (M1: GGTG·GGTG, M2: AGGG·AGGG)(used for selection), and the unrelated (M3: TTTT·TTTT) and compared it to extension of matched primer (M0). While Taq was unable to extend any of the mismatches, the selected polymerases 3A10 and 3D1 yield extension products with M1-3 but extension products are predominately shorter than M0.

Possible primer template configurations and expected main product lengths (N+1) are illustrated in the top right and bottom pictures. Matched primer-template sequences (M0) at primer 3’ end are shown in blue, mismatched and misaligned structures are shown in red. This illustrates that the selected polymerases can deal efficiently with non-cognate DNA structures.
———————————————————

Again we measured polymerase activity by examining the ability to extend a radiolabelled primer.

Top left: On an undamaged template all three polymerases Taq and the selected polymerases 3A10, 3D1 display approximately the same activity. To the left, the chemical structure of the undamaged base T is shown.

The top right template contains an abasic site at the + 1 position (marked by a red AP). To the right, the chemical structure of an abasic site is shown. Abasic sites are among the most frequent forms of DNA damage and are generated by spontaneous depurination or depyrimidination and as the end product of a number of oxidation-induced DNA damage pathways.

As can be seen, while Taq polymerase can insert a nucleotide opposite the abasic site (see arrow), it cannot bypass it and remains stuck in the +1 position. In contrast, the selected polymerases 3A10 (and to a lesser extent 3D1) can efficiently bypass the site of damage, inserting mostly an A opposite the abasic site.

The bottom left and right templates contain hydantoins at the + 1 position (marked by a red 5H, 5M). High levels of hydantoins have been found in some ancient samples and associated with PCR failure. Their chemical structures are shown to either side of the gel pictures. ( Hydantoins are oxidative degradation products of the pyrimidine bases. 5-hydroxy-hydantoin derives from C and 5-methyl-5-hydroxy-hydantoin derives from T.)

These lesions show the same general picture. Taq polymerase can insert a nucleotide opposite the hydantoins (see arrow), but bypass them poorly. The selected polymerases 3A10 also have problems in bypassing the lesions, but 3A10 is significantly better than Taq on either lesion, inserting mostly an A in both cases.

———————————————————

The ability of 3A10 to bypass template damage is reflected in PCR amplification. This shows a gel picture of the PCR amplification of DNA containing 2 abasic sites. As can be seen while all natural polymerases such as Taq but also Tth and Tfl fail to yield a PCR amplification product, 3A10 yields a strong amplification band. The other weaker bands are from other selected polymerases.

———————————————————

Left panel: The ability of selected polymerases to efficiently bypass template lesions in PCR encouraged us test their activity for the recovery of ancient DNA. We performed subsequent experiments using a blend of Taq with the most promising selected polymerases (3A10, 3D1 and others) (rather than testing individual combinations) in order to minimize wastage of precious ancient samples and maximize the chances of success. We first performed 56 PCR amplifications at limiting dilutions of ancient DNA derived from a 47,000 year-old cave bear (Ursus spelaeus) bone and scored successful amplifications for blend and Taq alone. We found that the blend yielded amplification products at between 2 – 5-fold lower concentrations of ancient DNA than Taq and indeed did yield amplification products at DNA concentrations, where Taq no longer generated any

Right panel: Normalizing PCR activity on a dilution series of “modern” DNA showed that this was not due to higher PCR efficiency of the blend. On the contrary, Taq appeared to be more than an order of magnitude more active at low template concentrations (of “modern” DNA), suggesting that the blend requires more template than Taq to produce an equivalent PCR signal.

The increased template DNA requirement of the blend suggests that the increased ability of the blend to amplify ancient DNA represents an underestimate of the blend’s potential. Moreover, it implies that the blend can tap into a pool of DNA molecules that are inaccessible to Taq, presumably because they are damaged.
———————————————————

To make sure that sample heterogeneity and random variation are not the source of the above effect, we performed a further 608 independent PCR amplifications from two different samples of cave bear bone (~47,000 and ~60,000 years-old respectively), and scored the number of PCR amplicons at limiting dilution. The blend yielded a larger number of amplicons (8-150%) than Taq in all but one experiment, confirming previous results.

In conclusion, molecular breeding and directed evolution by CSR have allowed the isolation of polymerases, which enhance the recovery of genetic material from Pleistocene specimens, presumably due to their ability to amplify damaged DNA.

Ice age genomics is upon us. Largely, thanks to novel sequencing methods, such as the Roche /454 sequencer, which also utilizes emulsion PCR,

Polymerases such as those described here should benefit the recovery of ancient DNA and may speed up sequencing as they are pre-adapted to emulsion PCR.

Polymerases capable of amplifying damaged DNA may also reduce bias towards modern DNA contamination and enable novel applications in palaeobiology, molecular archaeology and historic and forensic medicine.

The novel polymerases described here are really just a step in a direction, but they show that we can use evolution to improve our molecular tools. Further improvements should be within reach and hopefully will render ever more ancient sequences readable to us.
———————————————————

————————————————————-

There then followed a discussion: Questions and answers below.

Troy McLuhan: So how does it fill in the blanks at the “broken teeth of the zipper”?

PhilH Akina: Troy, the polymerase inserts essentially a random letter and a consensus sequence can be reconstructed by repeating the experiment several times and sequencing the results. As damage occurs at random, the “broken zip” is in a different place each time

BobbyDaz Carling: Did you use the 454 durring your studies?

PhilH Akina: BobbyDaz, no we didn’t use 454 during the studies

Duriel Akula: Do you think it is possible to rationalize from structural analysis which polymerase structures are more amenable to directed evolution based on the the impact of mutations on the stability and the capacity to generate diversity ?

PhilH Akina: Duriel, unfortunately our ability to do that is currently not ery good but you can sometimes make an educated guess

BobbyDaz Carling: how did you assemble the genome in the correct order if there were sections missing?

PhilH Akina: We didn’t actually do any sequencing, but all currently sequenced ice age genomes are assembled by relying on homology with the genomes of extant mammals

Monika Sonic: how different are they?

PhilH Akina: They’re obviously quite close but it’s hard to tell how close until the complete sequence is there. The hope is that at some point a contiguous sequence can be asembled from ancient sequences alone when the reads start to overlap

Zen Zeddmore: if such mismatches as quoted are allowed, how strong is the significance of the concensus

PhilH Akina: Depending on the number of errors which occur sporadically in your sequences, you can derive a statistical measure of the confidence of the consensus

Duriel Akula: a fallow up from my question would then be … do you believe that having structures for some start -end points of the directed evolution would likely help you learn what is an evolvable polymerase?

PhilH Akina: Duriel, do you mean high res 3D structures?

Duriel Akula: quite hard to get for these a structure from the start point and modelling for the mutations

PhilH Akina: yes, you’re right, the structures are hard to obtain and the effects of the mutations tend to be subtle so only very high resolution will do

Max Chatnoir: Can you infer from the kind of error tolerance you see what sorts of changes you might be seeing in the active site?

PhilH Akina: we don’t have a high resolution structure at the moment but the behaviour of the polymerases with various bulky substrates suggests that we have increased the space available in the active site, thereby making the polymerase less fussy

Morrhys Graysmark: PhilH, how close are we to resurrecting extinct species, particularly those that disappeared more recently, like the passenger pigeon?

PhilH Akina: It’s one thing to sequence the genome of an extinct species to reasonable accuracy, and this is almost certainly going to happen. But it is quite another to try to resurrect that species – you’d need to synthesise their genome and mammalian genomes are huge, and then somehow implant that genome into an unfertilised egg and find a suitable host animal for gestation and so on. I don’t think this is going to happen any time soon

Max Chatnoir: But if you get enough continuous sequence you could look at genomic organization. That would be pretty interesting.

PhilH Akina: Max, yes, it would

Habermas Aya: Is there a limit to how ancient of DNA can be so reconstructed, in theory?

PhilH Akina: For specimens buried in caves, DNA can be found up to maybe a few hundred thousand years maximum.

Habermas Aya: No Jurasic Park anytime soon

PhilH Akina: but there have been recent results from ice cores in Greenland, I think, which suggest that deep frozen ancient DNA might persist for a million years years or more

Virtualandy Capra: what about specimens found in peat bogs?

PhilH Akina: Peat bogs are bad because they’re acidic which speeds up DNA degradation. Also, they’re not really cold enough

Zen Zeddmore: so hurry and sign up at alcor for the life of your future :)

Hiro Sheridan: Phil has there ever been any neanderthal DnA recovered?

PhilH Akina: Yes, in fact, the Neanderthal genome project is underway. Our collaborator Svante Paabo has already sequenced more than a million base pairs

Zen Zeddmore: I am very curious also about this CSR. This seems like a very powerfull methodology. Do you have any links to more info?

PhilH Akina: Zen, yep, if you IM me your email address, I can send you some links

Editor’s note: Links to more resources will be added to the bottom of this transcript later today or next week

Troy McLuhan: Do you know of other teams evolving molecules to solve other problems?

PhilH Akina: Troy, evolutionary methods are well establshed in molecular biology. For example, at the MRC Laboratory of molecular biology where I work, Greg Winter has developed antibody display methods which allow the generation of human antibodies staright from their genes by evolution. One of these is called humira, and has recently been licenced for treatment of rheumatoid arthritis

Max Chatnoir: That’ll make some old bones happy.

Duriel Akula: Comment: ancient genomes can also be predicted from phylogenetic reconstruction

Hiro Sheridan: Has anybody recovered DNA from animals trapped in amber – the idea behind Jurassic park?

PhilH Akina: Some people have tried, Hiro, to my knowledge unsuccessfully

———————————————————-

And that was the end of the talk. We stuck around for a bit and Phil and I chatted to people afterwards. I hope you all enjoyed it: any feedback very welcome, in the comments or to j.scott@nature.com. Otherwise, see you next time!