At work, samples excavated from the ocean floor are brought back to the lab and spread out on a Petri dish containing some 'media': Nutrients for growth that are combined with agar to form a solid surface! Bacteria that were present in the samples will then grow on the dish, forming a 'colony': Generally a small circle of billions of cells that all started from a single cell/spore in the sample! Each individual colony is then transferred to its own plate and grown so that you have a pure isolate: A pure strain of one bacterial type! Once we have those, they are transferred to a liquid media (also contains nutrients, but no agar, so it stays liquid) and grown in an incubator (keeps bacteria warm and shakes them to aerate the media).
This is called liquid fermentation! Whatever chemical compounds the bacteria might produce are often secreted from the inside of the cells into the surrounding liquid media. Afterwards we extract these compounds from the liquid media using a resin that broadly absorbs most compounds that would be interesting! This resin is then washed with organic solvents which transfers all these interesting compounds from the resin into the organic solvents. Organic solvents are things like ethanol, or acetone (in nail polish remover), dichloromethane (in paint thinner), etc.! These solvent washes, which now contain the compounds produced by the bacteria during fermentation, can then be concentrated and this produces a "crude extract": It contains a whole lot of compounds that were either produced by the bacteria or were present in the liquid media to begin with!
This crude extract (we generate thousands of extracts each year) is then tested in a screen to see if it has any interesting bioactivity! For example, somebody might develop a screen to see if a compound kills cancer cells. A simplified method might involve something like putting cancer cells into a dish, adding this crude extract, and then observing whether or not the cancer cells survive. In a real setting this would be very high throughput...each "dish" would be a tiny 'well' (about the volume of a few drops of water) arrayed in a 'plate'...a rectangular piece of plastic that contains a grid of these wells (usually plates contain 96, 384, or 1536 wells). So then each well of each plate would contain a few drops of a suspension of cancer cells. We add a different extract to each well and see if cancer cells live or die! This way you can screen thousands of compounds very quickly.
You could also do this to look for new antibiotics! Instead of putting cancer cells into each well, you'd put some sort of pathogenic bacteria that you want a new antibiotic against. In reality these screens often end up being a lot more complicated because people don't just want to kill cancer or bacteria cells outright (often these compounds would be toxic to regular human cells too), so they develop a screen that targets something really specific, like a particular protein involved in cancer pathways or something!
If one of these crude extracts turns up as a 'hit' on a screen, we then go about finding out what the specific compound is in the extract that's responsible for this interesting bioactivity! Remember that the crude extract contained everything the bacteria made, which can be hundreds to thousands of compounds. To do this, we perform a fractionation of the crude extract, in which these hundreds of compounds in a single extract are separated based on their chemical properties.
This generates a series of fractions, so let’s say I had an extract with a thousand compounds that I separated into 100 fractions. Now each fraction contains maybe 10 compounds. Each of these fractions is then rescreened, and hopefully we find that the bioactivity is in a particular fraction. Now we know the active compound is one of 10! We then repeat this fractionation process on these 10 compounds, rescreen them and find the active one, and then identify what that compound is. If it's a new compound, somebody would go about studying whether or not it could be a useful drug. So for example, using the screen, we might know that the compound kills pathogenic bacteria, but it wouldn't be a useful drug if it's too toxic to people, or gets broken down in the body too quickly, things like that! This might involve a “clinical trial”: Trying the drug on real people after it's been through plenty of tests and development!
After work, I go home and dine with my wife Akemi! Then I go to bed!