Friday 15 August 2014

A look at Nature's profits

Who in their right mind pays thirty-two bucks for a pay-per-view article?
I assume nobody. If you can understand a jargon filled peer-reviewed article on muonic neutrinos, the chances are you know someone whose academic library has access. But if that is the case why bother?

I would guess it is make the library subscription prices a bargain, after all one issue of Nature bought from the shop costs ten pounds (15 USD) whereas a annual personal subscription comes to 4 USD per issue. Namely the usual bulk sale. Apart from the small details that (a) in both Nature and Science the per issue cost printed on the cover (10 units of some currency) has not changed for decades and (b) newsagents do not sell New Scientist half the time, so will never sell Science or Nature.

As it turns out, a private university in Texas (Trinity University) with 2.5k students did a trial where the library  cancelled its subscriptions and used the saved money as a fund to buy all the 30$ articles students and faculty needed. Suprisingly, it was a success according to the article I found —wikipedia does not mention if they still do it, but it does say the library has an acquisition budget of 1.8 million dollars per annum.
Annoyingly, few numbers are given in that article, except for the number of articles bought in December 2007, which was a meager 220, meaning that, assuming the students were illiterate, each faculty read only one article per month. I think I can safely say that random articles of the internet are not to be trusted.

Nature Publish Group had a quibble with the University of California over its extortionate library subscription of over one million dollars per annum. Assuming that the pricing depends on the student population —I do not know if that is what they do, I am just guessing— the university pays seven dollars for each of the 160,000 students per year to access NPG articles. Which would mean that a library subscription is cheaper than pay-per-view if on average a student reads more than one NPG article for his/her whole degree. So my guess could be right.

Six dollars per year per student sounds like a bargain: it is one 33rd of the price of a yearly subscription, which sounded like a bargain to start with. So, they are loosing money and not making it, right? Actually, they still manage a 15-19% profit margin.

In the Select Committee on Science and Technology, there is an interesting (phantasmagorical) deposition by NPG (link) where they show some very weird figures:

  1. Annual income of Nature: 30M GBP
  2. Articles per year in Nature: one thousand (1/10 of those submitted)
  3. The "front half" (the intelligible bit) takes up 2/3 of the cost
  4. Editorial board: 43% cost
  5. conferences and overseas travel: 1%
  6. printing and distribution: 31%
Now, I fully see why it is called Nature: it is as badly organised as a biological system.
So they pay the editorial board 13 million per year to review one thousand articles. The worse case scenario is that they have 650 editors paid equally 20 k GBP pa. to review twenty articles annually. My guess is that there are many fewer editors than 100 (@ 100, 130 k GBP pa for ten articles per month), who get paid serious money (unlike the peer-reviewers) and go on loads of conferences.
That two thirds of the costs go to the nice summary articles that only people with the printed versions read baffles me entirely.
Basically, I thought these chaps were evil, but it turns out they just like spending money badly like the rest of us.

Update: my google search victim (NPG) is not the worst offender. It actually is Elsevier, which makes 34% profit margin, which "makes Murdoch look like a socialist" according to the Guardian...

Tuesday 29 July 2014

Metabolic engineering Breaking Bad

Disclaimer: Obviously, this is a speculative what-if out of intellectual curiosity and by no means condones narcotics and their creation.
The logo of Breaking Bad had it been bio.
Half the issues in Breaking Bad could have been solved if they had been using biocatalysis and metabolic engineering. The catch is that nobody has made a production strain and it is not a simple task.

Biocatalysis: greener and safer

Some time soon, biocatalysis and metabolic engineering will replace many heterocatalytic processes as it is more efficient, cheaper, safer and greener.
Worldwide, there is a big problem of exploding methamphetamine labs. This problem could be fixed by switching from toxic and dangerous heterocatalysis processes to green and safe biocatalysis ones.


So if Walt and Jessie wanted to win the green chemistry award, what would they need to do?
Their major problem is the starting material and the production steps. So the whole lot. Therefore, they need to do metabolic engineering.
The major issue is that methamphetamine is not a natural compound, so extensive engineering would be needed to produce the final steps. However, once they laboriously made a production strain, they would need to set up a large-scale bioreaction, i.e. a brewing tank, extract the product by phase-separation, remove the solvent and purify by crystallisation as they normally do. Then the only worry then is that they may be as contaminant-prone as Hank is at brewing.

Starting point

Methamphetamine is a compound that looks like phenyalanine, but on the chiral α-carbon there is methyl group instead of a carboxyl one and the amine group is methylated.
The natural molecules most similar to methaphetamines are pseudoephedrine and ephedrine. These two diastereomers differ from the former in having a hydroxyl group (in different chiral orientations) on the carbon adjacent to the benzene ring. The biosynthetic pathway is known (PMID 22502775), but requires eleven steps, which have not been assembled exogenously in an orderly way.
Additionally, to convert ephedrine to methamphetamine new enzymes need to be engineered for the hydroxyl reduction, which is problematic. Consequently, ephedrine biosynthesis might not be the best route and instead something more radical may be in order.

Ephedrine biosynthesis

Ephedrine gets its name from the genus Ephedra, whose members produce it. Unfortunately the selective pressures for plant secondary metabolism are rather unusual (cool) and as a result the metabolic routes get a tad tortuous.
If one were to forget how plants like to do things, one would guess a simple pathway with a similar logic to threonine biosynthesis be present. Namely, the carboxyl is twice reduced and the remaining hydroxyl is isomerised to the right place.

The final step (not pictured), the N-methylation, would be simply accomplished by a SAM-dependent methyltransferase and is the only part that is correct.
In the isomeration step, one might anticipate that an enamine-ketimine tautomerism followed by an attack by water might occur ruining the effort. However, the isomerisation of homserine to threonine is done via a PLP enzyme which holds the amine, so this isn't the problem.

The problem is plants like to make second metabolism in an OCD way, starting from specific compounds (e.g. geranyl-PP, farnesyl-PP, cinnamoyl-CoA, coumaroyl-CoA, malonyl-CoA and acetyl-CoA), preferably using decarboxylative condensation.
The ephedrine pathway is no exception and shares the beginning of the pathway with many other compounds Phenylanine > cinnamic acid > cinnaomyl-CoA>>benzoyl-CoA. Then the unique part is that the benzoyl-CoA is condensed with pyruvate, reduced and transaminated.
In reality, whereas the full pathway is known, the genes themselves are not, simply because nobody has sequenced E. sinica. Although a group in 2009 has gone to the effort of making yeast take up the DNA of E. glauca via ion implantation —no kidding around there!— and make ephedrine, but they did not sequence a few hybrids or similar, but instead did a lot of tedious work with primers (PMID: 19280123). Consequently, their experiment would need to be repeated, but with sequencing.
Once the various genes are cloned into E. coli, preferable into a strain that overproduces phenylalanine (eg. from PMID: 17880710), the pathway would be optimised, giving an ephedrine-producing strain.

The last step

The last step is the trickiest.
After those few years of work are done, the hydroxyl needs to be removed. Biochemically, hydroxyl groups are normally removed in two steps, the hydroxyl group is removed without adding an electron pair to the molecule by a hydroxylase, therefore leaving a double-bonded carbon, which is then reduced by a reductase. In some rare cases, the hydroxyl is reduced away. The most famous example is ribonucleotide reductase. The mechanism is rather mental and ugly.

The dehydration route is a problematic option however. A modified 3-hydroxyacyl-ACP dehydrase and the enoyl-ACP reductase from the fatty acid biosynthetic pathway seem like good candidates. However, the dehydrated enolamine would spontaneously tautomerise and hydrolyse as mentioned above.
This might not be that catastrophic as the product would be phenylpropanone, which being similar to phenylpropane-dione, the product of the pyruvate-benzoyl-CoA condensation. It might be promiscuously transaminated again by the cathionine transaminase or by phenylamine transaminase.
Nevertheless it is an odd way of doing things.
The N-methylation must be done last as the product is slippery being so hydrophobic. Normally, biochemistry likes to put a handle to hold stuff like that, such as phosphates, CoA and glycosides. In this case, a N-glycosilation would be a good option. The best bet, however, would be to move the N-methylation step after the hydroxyl reduction, the methyltransferase cannot discern between the precursor for ephedrine or pseudoephedrine, so it is probably fairly accommodating towards amphetamine.

Crazy way

The carboxyl group needs to be replaced with a methyl group. This is not an option from a biochemists' perspective as C-C bonds cannot be made that easily, unless by condensation or transmethylation on aromatic structures. In a typical methyltransferase the methyl donor is SAM, while the acceptor is a nucleophile (Lewis base), such as an amine or a hydroxyl that has been deprotonated by a catalytic acid. From a technical point of view as far as I can tell there should not be anything forbidding a PLP and SAM dependent decarboxylative C-methylation. After a decarboxylation the negative charge is partially absorbed by the PLP (electron sink), leaving an nucleophilic secondary aldimine. The enzymatic reside that favoured the departure of the carboxyl group (say, catalytic lysine) might compete with the SAM though.

In the literature there is no sign of such a reaction: there is a decarboxylative O-methyltransferase (PMID: 22247507) and the various cases of SAM and PLP dependent enzymes, e.g. aminomutases, rely on SAM as a radical donor. There are some enzymes that point towards the possibility of such a C-methyltransferase, such as on an enol-ketone tautomerism (PMID: 5490210 and PMID: 17784761). Nevertheless, such an enzyme, if possible, would require a lot of work and luck to pull off. So the safer option might sound a lot longer, but has a higher chance of working...


A side question is how to select for better variants. To evolve a strain to make methamphetamine an way to select for one is obviously needed.
The traditional way would be to assay for those compounds by HPLC, but this would mean that the variants would be screened laboriously (especially in light of the optimisation required).
An option for a high-throughput approach is to make a transcription factor that responds only to the substance needed, so that it activates a fluorescent reporter which can be selected by a FACS (eg. PMID: 22276138). Unfortunately, the targets of amphetamines and those of catecholamines are membrane receptors. So the phenylalanine binding TF, tyrR-econded, would be a good candidate for engineering.


The oral murine LD50 of methamphetamine is slightly higher than capsaicin (55 vs. 46 mg/kg), so it is probably non-toxic to bacteria and due to its hydrophobicity it can be phase-separated easily from the aqueous environment. So a least one bit would be straight-forward.


Given the recombination and sequencing for gene identification, the many rounds of engineering and so forth, it would be five years if they are lucky. So metabolic engineering breaking bad would not start even at its fifth season...
And it would be expensive to do and there is no guarantee that their strain would remain safe — copying Walter's formula was an issue, here it would require only a stolen tube.
In brief, it would actually be a real pain to do and take years to cobble together, so unfortunately, metabolic engineering cannot make Walter and Jessie more green and free of precursor woes...