Part 2: Tryptophan discovery: How do we know what we know?

[mike_malaska@deepspace.social]

Part 2: Tryptophan discovery: How do we know what we know?

As promised, we gonna circle back on that tryptophan discovery.
(For Part 1 in this series see here: https://deepspace.social/@mike_malaska/115653001232082594)

And this gets to an even bigger question in #astrobiology (and life in general):
“How do you know what you know?”

We gonna dive deep into the analytical chemistry of that reported tryptophan detection, digging into the supplementary information analysis.

[thread]

[mike_malaska@deepspace.social]

This will get to the root of a big problem in #astrobiology. How do we actually know what molecules we detect in an alien environment when we have no clue what to expect?

I remember a pamphlet at a chemistry exhibit at a local science museum when I was a kid. It said “Chemistry is taking thing apart and putting them back together.”

Synthetic chemistry puts stuff together.

Analytical chemistry takes a thing apart, then tries to figure how the pieces fit together to make the original thing.

[mike_malaska@deepspace.social]

Another way to think about analytical chemistry (and #astrobiology) is like the game “20 questions”. You're trying to guess the structure of a molecule and you only get to ask questions. After each question you narrow down the possibilities. It is hard to get the exact structure. Unless you ask the right questions, you only get degenerate (multiple possible) answers.

Think of it as having a bin of possibilities, and you are trying to throw things out of the bin that don’t fit the data.

[mike_malaska@deepspace.social]

Example game of 20 questions, you ask:

1) is it bigger than a breadbox? (yes)
2) is it made of metal? (yes)
3) does it have an engine? (yes)
4) can a human fit inside it? (yes)
5) is it used for transportation? (yes)
6) does it have wheels? (yes)

After 6 questions, you’ve got lots of information. But possibilities still include trains, sports cars, trucks, buses, small aircraft, commercial jets, and paddlewheel steamers.

The data is “consistent with” a sports car, but other things. too.

[mike_malaska@deepspace.social]

A few key phrases, but they are very important when reading papers like this:

“consistent with” – it means it could be, but we could be fooled by something similar. (“=Mmmmmmmaybe”)

“diagnostic for” – this is an ambiguous term. I take it to imply “indicators for..” (=”probably”)

“conclusive for”, “positively identified as” – beyond a shadow of a doubt you’ve got it! (=“Absolutely!!!)

[mike_malaska@deepspace.social]

More phrases:

"targeted analysis" – means you know what you are looking for, and you expect it.
(=“Is X in here?”)

"untargeted analysis" – you have no clue what is in there. Pure exploration. (="What are the characteristics of things in here?")

“De novo structure elucidation” – from absolute scratch (no prior knowledge) you figure out the exact molecular structure. This is the most rigorous. (="starting with no preconceived biases, we found this.")

[mike_malaska@deepspace.social]

How do you absolutely definitively prove a chemical structure?

Let’s look at the legal definition in patent literature (think pharma, where REALLY big money is at stake).

The US Patent office (USPTO) says: “one must define a compound by ‘whatever characteristics sufficiently distinguish it’.”

Link US PTO (patent office) legal definition for compound definition.
Direct link: https://www.uspto.gov/web/offices/pac/mpep/s2163.html

[mike_malaska@deepspace.social]

That's a beautifully circular definition. It means if I can come up with another molecule that fits your observations, then you don't have a unique ID.

(thus invalidating your discovery.)

It puts the onus on the inventor (or discoverer) to have a solid defensible characterization from alternate possibilities.

(Key question: "You claim your identification uniquely fits the data, how did you verify that? Didja check ALL possible molecules????")

[mike_malaska@deepspace.social]

Any actual astrobiological detection is going to get A LOT of scrutiny since “extraordinary claims require extraordinary evidence”.

So yeah, go hardcore, satisfy the skeptics. If the underlying molecular detections are not solid, you are building your evidence on sand.

(And.....fun fact – MS-only analysis has NEVER been able to identify a complex organic chemical structure de novo.)

[mike_malaska@deepspace.social]

So how do natural product chemists* perform de novo structure elucidation?

Firstly, is you start with something you know is pure. You check with GC, HPLC under a few different column conditions to make sure that there is only one peak and not something else hiding under it. You should only see one retention time peak and no other bumps or shoulders.

Then you know that you are only dealing with one thing and you don’t have to worry if Signal A is coming from Compound B.

[mike_malaska@deepspace.social]

*(in my past career, I was not a natural product chemist, but was second in the drug discovery chain.

You know those stories where they identify a deep sea sponge alkaloid compound that kills cancer cells? Yeah, my old job was to figure how to modify that natural product to make it a better drug candidate: increase activity, remove toxicity, improve solubility, etc.

"Here's 20 mg of a funky paspalinine derivative, chemically modify it to make it work better. Go."
((it was actually very fun.))

[mike_malaska@deepspace.social]

For de novo structural elucidation, the workhorse is NMR (nuclear magnetic resonance). With NMR you can run a bunch of different techniques. My favorite minimal set for an unknown is 1H NMR, 13C NMR, DEPT-135, HSQC (1H-13C correlation), HMBC (multiple bond coupling experiments) and maybe ROSY (coupling) or NOSY (through-space interaction) experiments if you need it.

[mike_malaska@deepspace.social]

The beauty about NMR is that it is completely non-destructive and (unlike MS) does not lie. Unless you have a weird relaxation delay (you can compensate for that) you will see all the protons (and usually the carbons) under standard instrument settings.

There are a lot of other weird funky pulse sequences you can run too if you need to try to figure stuff out.

And, the best part, you can recover all the sample when you are done.

[mike_malaska@deepspace.social]

The two downsides of NMR is it can fool you with symmetry. But a simple MS run can help sort that out. The other bummer is that you can’t extract really hardcore purity information from it.

NMR is only quantitative to about 95%. So you need another technique to check for high purity. (such as: GC, LC, elemental analysis)

[mike_malaska@deepspace.social]

The absolute best structural confirmation is single-crystal high resolution X-ray analysis. That gets you the positions and orientations of all the molecules. If it is a really highly resolved structure, you can see (or infer) the hydrogen atoms.

But X-ray crystallography is really hard (and expensive) to do right off the bat. NMR will always be the tool of choice.

[mike_malaska@deepspace.social]

(and in a funny twist, protein structure is usually figured out by X-ray crystallography. But proteins don’t normally exist in crystal form, they are fluidy-loosey-goosey flexible. Crystallization can introduce new conformational artifacts.

To get it right and determine the protein bonding interactions in native solution you use…..NMR!)

[mike_malaska@deepspace.social]

For MS, a HRMS (high resolution mass spectrometry) experiment is my first choice. That gives you the molecular ion to lots of decimal places and thus gives you the exact mass and thus empirical formula so you know what atoms and how many of them you are dealing with. It doesn’t give you structure, but it nails the empirical formula.

[mike_malaska@deepspace.social]

MS also has lower resolution options. That can give you a rough idea of the weight, maybe of the parent molecule, and maybe some pieces and parts as it breaks into chunks. You might combine it with a front-end separation (like GC or HPLC) so you can analyze mixtures of molecules by separating them first then sending little tiny subsamples of them into the MS chamber as they come off a column.