Beyond the Lab

My final cells with the edited gene, stored for future use. CRISPR and Recombineering.

(Left to Right)

Don't trust anyone, or at least, bacterial colonies. I'll tell you why. The classic way to knock out a gene is to replace it with an antibiotic resistance cassette, in my case one conferring Zeocin resistance, and then look for survivors. The reason we replace it with an antibiotic resistant marker rather than a random series of amino acids is to check for proper insertion. Theoretically, if a colony grows on a plate treated with the Zeocin antibiotic, the cassette went in; if it doesn't, the original gene is still there. However, I soon found out that bacterial cells can trick you, and this wasn’t a very nice realization. I’ll get into that later.

To build the cassette, I had to make two "armplicons," short DNA sequences flanking my gene and attach them through PCR to either end of the cassette. These armplicons let the cassette know which gene to replace. The way to get the Zeocin resistant cassette into the cells is by electroporation, which is as violent as it sounds. My M. smegmatis and Jucho cells were then zapped at 2.5 kV to blow temporary holes in the membrane and to allow the cassette to go inside.

My first two electroporations gave me colonies growing happily on Zeocin, which should have meant success. It didn't. When I checked it with another PCR, it showed the original gene still sitting there untouched. It was then that I learned from my mentor that my cells grow slowly enough to the point where they can develop their own antibiotic resistance (even if they never took the cassette). So all I had on my plate were a bunch of sorry disguised excuses for edited cells. 

By electroporation number three, I'd learned not to trust growth as the finish line. I ran four confirmation PCRs instead of one, and this time the gene was truly gone: three confirmed Jucho colonies and four mc²155 colonies carrying a clean knockout of MSMEG_2508.

Approach two: CRISPR-Cas9. Recombineering worked, but it's slow, and my summer internship didn’t have room for three more rounds. The backbone that we inserted into our cells carried CRISPR gene editing with personalized guide RNA’s (that tell the machinery where to cut), something called mCherry to make the edited cells pink, and another antibiotic resistance marker. Like the other strategy, I electroporated this into mc²155 and Jucho, and then CRISPR did what it does: cut.

M. smegmatis  and Jucho aren't great at cleanly resealing double-strand breaks, so repair is sloppy and usually leaves a small deletion that shifts the reading frame and kills the gene. Turns out DNA is not a very good surgeon, more like an ER doctor with 30 patients on their hands. As a control, I targeted an essential gene, MKD8_4346. Those cells failed to survive, exactly as predicted, so I could trust the CRISPR tool.

After this, I streaked out colonies on plates and checked for deletions by running a gel. A gel can tell you something looks different. It can't tell you what happened at the base-pair level, so the cleanest candidates went off for Sanger sequencing for a more accurate read. This whole process was done multiple times until, ultimately, six of eight mc²155 colonies carried a single deletion in MSMEG_2508. Not a perfect eight-for-eight, but I'll take it. 

With knockouts from both routes, I could finally run the experiment I'd been building toward all summer: a DCT mating assay. Donor and recipient strains, each with a different resistance marker, are combined on solid medium overnight, then put on a plate with both antibiotic markers. Only true transconjugants survive. I set up two matings: Δ2508 against wild-type mc²155, and Δ2508 against Δ69KO, a gene already established as a player in kin recognition. If anything came back strange with the edited MSMEG_2508 cells, it would make us think that that gene does indeed play a role in kin recognition. Unfortunately, my last day came before the DCT assays had finished. My mentor emailed me a week or so later saying that they hypothesized that MSMEG_2508 did inhibit DCT given that there were less transconjugants on the plates than expected. 

Wadsworth taught me how to be persistent and continue even when an experiment seemed to refuse to work. Even though I was only able to successfully edit out one gene, I produced many edited cells. I’ll have saved the next scientist who decides to pick up my project months of trying to edit this gene out and give them more time to run DCT assays, pin down exactly how this gene contributes to kin recognition, or find a condition or regulator that shuts down ESX-1 secretion entirely. As I mentioned in my last blog post, the genes governing cell contact and conjugation in a harmless lab strain like M. smegmatis are conserved in pathogens like M. tuberculosis, where the same ESX secretion systems help the bacterium dodge the immune system. 

My time here also brought me the closest I’ve been to doing scientific research with real-world health outcomes, which is why I came into the lab every day so adamant on making progress. I constantly envisioned, in the back of my mind, my work here being a part of a larger endeavor to create non-antibiotic therapies for TB that could be brought to New Yorkers and other communities around the world.