Synthetic uORF construct

So I’ve long wanted finer control of protein abundance, and to date, have had the greatest success messing with the Kozak sequence to alter translation rate, thus modulating the amount of protein steady-state abundance. That said, I’ve wondered if there are other aspects that could be further manipulated to increase the dynamic range of the amount of protein steady-state abundance. This at some point led me to try playing around with upstream open reading frames (uORFs), that can interfere with the translation rate of the downstream protein (in my case, a green fluorescent reporter). We recently made a vector with one such uORF, so I looked at what effect having that uORF had on green fluorescence of the cells.

The actual vector plasmid name is “AttB_2xuORF_mGreenLantern-T2A-shBle-IRES-mCherry-P2A-PuroR”. As you can tell, red fluorescence is behind an IRES, and should be unaffected by the uORF. That’s indeed what we see, with the red distribution being a control construct without the uORF, and the blue distribution being the identical construct with a uORF immediately preceding mGreenLantern (Note: YL2-A is the fluorescence channel for red fluorescent emission).

Now if I gate on that bright red population, and look at the amount of green fluorscence, there is indeed a difference (BL1-A is the channel to look at for green fluorscence), although the effect isn’t huge. Looking at the distribution geometric means, it looks like the uORF construct is roughly 3.23-fold less bright than the control, so roughly a half-log less bright. Now the reason I’m underwhelmed is that I can get a roughly 1.5-fold difference in fluorescence using different Kozak sequences, so the uORF that I created doesn’t exhibit nearly the same magnitude of effect. Eh, that’s designing and testing constructs for you. That said, I suppose if I were to combine both, then I could likely get > 2 logs of dynamic range.

PS. Yes, I know there are simpler ways to modulate protein abundance, like transcriptionally through modulating the amount of expression. One day we’ll do more of that, but for now, translational control it’s been.

Cell surface FP brightness

When doing the Green FPs in HEK 293T cells experiment, we noticed how the same fluorescent protein, EGFP, could have vastly different brightness depending on the construct we were using.

Put another way, cytoplasmic EGFP gave us really high green fluorescence intensity, but a different construct wherein that same EGFP sequence was preceded by a signal sequence (thus causing it to become a secreted / extracellular protein), and succeeded by a transmembrane domain (thus causing the extracellular EGFP molecule to be anchored to the plasma membrane) gave us cells that were roughly 30 to 100-fold less bright. The amount of fluorescence of the transmembrane version was further susceptible to other sequence considerations; for example, addition of a his tag right after the signal sequence (such that the his tag is the most distal sequence on the protein, with the tag flapping around as part of a flexible N-terminus), resulted in a ~ 3-fold reduction in fluorescence as compared to an untagged version. My guess is that this repetitive, pseudo-charged region was interfering with efficient translation into the rough ER, but who knows.

Still, how do I explain this result? Well, I have no evidence-supported answer, but my guess is that it’s about translation process bandwidth and overall real-estate. I’m guessing that cytoplasmic translation and accumulation has pretty high bandwidth, where there are plenty of ribosomes to translate cytoplasmic protein, and there’s plenty of space to accommodate them. In contrast, I’m assuming there are comparatively fewer ribosomes capable of translating transmembrane proteins at the rough ER, and that the overall real-estate on cell-associated membranes (particularly in the vesicular pathway leading up to and including the plasma membrane) is also less (while an imperfect approximation, I’m thinking of it kind of like a difference in surface area or volume of a sphere, type of thing). Although who knows; maybe that’s all incorrect, and it’s more about the signal sequence (possibly from CD8?) and the transmembrane domaine sequence (seemingly from PDGFR-beta) that I used.

Edit 1: Although again, I need to keep reminding myself. Since these are cell surface proteins on adherent cells, some of the reduced signal with the transmembrane protein may be due to some proportion of proteins getting cleaved off the cells during routine trypsinization. I’ve talked to Olivia about trying a side-by-side experiment of resuspending the cells with trypsinization or Versene with gentle agitation. Stay tuned to see if I need to update the above plot or not!

ASC Speck formation

One of the possible projects is understanding how protein sequence variants impact inflammasome formation. There are multiple possible assays for this, but a classic one is observing the formation of ASC specks, where a nucleation of activated sensor (say, MEFV / Pyrin or NLRP3) nucleates the oligomerization of all of the ASC in the cell into a single gigantic filament.

To try to assess this, we have a construct where ASC is directly fused to mCherry. There is clear speck formation by microscopy, as you can see here.

Clearly there is a bunch of really bright puncta when MEFV and ASC-mCherry is present (right), very few such puncta when no MEFV is there (middle), and no puncta at all when there is no mCherry-tagged ASC (left).

That said, any high-throughput compatible experiment can’t simply rely on microscopy (without a bunch of extra infrastructure), and it’s easier to make it a FACS compatible assay. Thus, the big question was whether we could see a difference in flow. Here are the results:

As you can tell, the diffuse cells (left) have a streak of points below the slope = 1 diagonal,  the ASC only cells (diffuse with a handful of puncta) have a similar streak of points but with a small shadow of points along the slope = 1 diagonal (probably the completely spontaneous ASC speck formation cells; middle), and the MEFV and ASC-mCherry overexpressing cells are almost exclusively puncta forming and also almost exclusively have points running down that slope = 1 diagonal. Thus, it does seem we can distinguish ASC speck formation using flow cytometry.

To make things even easier, one can turn it into a ratiometic density plot, where I’ve divided the red fluorescence height by the red fluorescence area. The differences are relatively subtle, so you have to make sure your axes are zoomed into the right region, but once you do, you can definitely see that there is a different distribution for the ASC speck cells. Cool!

Green FPs in HEK293T cells

At one point, I was a doe-eyed postdoc reading about new fluorescent proteins (FPs) with improved brightness and thinking it was potentially important to incorporate new FPs into my constructs for cell culture work. I have since come to realize that the intrinsic gains to fluorescence published in those papers do not necessarily translate to brightness in my experiments. The reason are probably multifactorial, including:
1. Commonly accessible equipment is generally optimized for EGFP (or similar FPs), so newer FPs with slightly different excitation and emission spectra may not be captured well with existing microscopes or flow cytometers.
2. The FP brightnesses are typically assess in-vitro, and there may be other factors in eukaryotic cell cytoplasms that may affect the FP brightness (eg. FP half-life).

Well, I’ve still ordered a handful of different green fluorescent proteins anyway, and figured it was worth doing a side-by-side comparison in the transgenic system we use in the lab. This was all done with the HEK293T G542Ac3 (LLP-Tet-Bxb1attP_Int-BFP_IRES-iCasp9-BlastR) cells, which were stably recombined with a single copy of each fluorescent protein at a common genomic locus. The construct organization was: Bxb1attB_[Green FP]_IRES-mCherry-2A-PuroR. Olivia did these recombinations, selected the cells, and ran the cells on the ThermoFisher Attune Flow cytometer, with Sarah’s help. This is what the results look like:

Some interpretations:
1. Rather minimal (~3-fold) difference between mGreenLantern and UnaG. I suppose if we were ever in a situation where we needed every unit of green brightness possible, that we would go with mGreenLantern. That said, UnaG has some benefits; namely, it’s 58% the size of EGFP/mGreenLantern/mNeonGreen, and it lacks the VERY ANNOYING identical sequences at the N- and C-termini (MVSKG … DELYK) which makes molecular cloning a potential pain.
2. Conceptually, I like the idea of fuGFP, but that 20-fold diminished green fluorescence compared to EGFP is potentially problematic. Who knows, maybe I’ll turn this into a target of directed evolution at some point…?

Cell surface localization assay

About the same time I got inspired to try making the vesicular protease assay, I figured I’d also build an assay to try to look at cell-surface localization of proteins.

Going back to that post about dL5 and malachite green (MG), I believe that MG doesn’t cross the the plasma membrane particularly well (Although now that I look at it, there are both ways to increase and decrease it’s cell permeability). Thus, I figured I’d test the dL5 fluorogen activating peptide in two forms; one where it’s sent out to the cell surface using a signal peptide (but anchored to the plasma membrane with a transmembrane domain), and another where its expressed as an intracellular, cytoplasmic protein.

Well, we made the constructs, and Sarah recombined the cells and tested them, and here are the results.

Well, ignore the blue distribution for now (since this is a construct testing a different hypothesis), and only look at the red and orange distributions for now. As you can tell, in the absence of MG, the signal is pretty low (I should probably through in some unrecombined landing pad cells on that plot to show what the background level is). On the other hand, MG addition causes cells encoding the extracellular dL5 to exhibit ~ 3e4 near infrared MFI, while the intracellular dL5 cells had about 1e4 MFI. So while that’s only a 3-fold difference, the standard deviations of those distributions were pretty tight, such that there was rather small overlap between those two distributions. So while this is a single, one-off experiment, it looks like this assay format may work.

Probably some additional knobs that can be turned to try to improve signal over background. First, is maybe this effect is somewhat MG concentration dependent, and reducing the amount of MG that is added may add some more dynamic range. Also, there are those less cell permeant MG derivations, which will likely improve the range (albeit, these are likely far harder to get than OG MG)

Vesicular protease cleavage assay

Obviously we make a lot of recombinant DNA constructs and create a bunch of different assays to try to understand biology. Well, at some point I became curious if I could come up with a protease cleavage assay for measuring how various peptide sequences could be substrates for vesicular proteases inside cells. I figured we’d start to test this by looking at furin cleavage, since 1) furin is quite ubiquitous, 2) furin cleavage is relevant for multiple disease-relevant membrane proteins, like many viral entry glycoproteins, and 3) including SARS-CoV-2 spike, where the furin cleavage site in that protein is thought to contribute to its dynamics of spread and pathogenicity during infection.

Well, the first construct that may have worked is a pretty simple one, where I have EGFP targeted to extracellular space using an N-terminal signal peptide, but retained on the cell surface by having it C-terminally fused to a transmembrane domain. Anh has actually been using this construct as part of his undergraduate research project. Well, to turn this construct into one that can study Furin cleavage, I modified it to encode a R-R-A-R peptide between EGFP and the transmembrane domain. Thus, if Furin cleaves this construct, the EGFP protein is now no longer tethered to the cell, and presumably escapes into the media once it reaches the cell surface.

So this is an N=1 experiment so far, so it’s not the most conclusive. That said, there does seem to be decent separation between the construct with and without the furin cleavage site, where cells with the construct with the EGFP that potentially gets released had ~ 10-fold less fluorescence than the construct that could get released. I suppose if this reproduces, it could be worth trying to turn this into a library-based experiment for studying protease cleavage within intracellular compartments.

… Though now that I’m thinking about it further, I’ll definitely need to ask how these cells were prepped. If they were prepped with trypsin (which is likely), this was likely more of a trypsin cutting assay rather than a furin one.

Edit 10/27/22:
Well, so to avoid the whole trypsin complicated, recombined the above constructs in suspension HEK cells that are floating and don’t ever need to be detached off the plate for flow. Here’s what that plot looks like.

So ya, we’re still seeing the ~ 10-fold effect there, so it looks like that’s the real dynamic range of the assay for measuring furin protease cleavage.

Fluorogen activating peptides

So I had collected this data over 10 months ago, when I made an initial foray into trying the fluorogen activating peptides dL5 in the presence of Malachite Green, a dirt cheap small molecule (You can use it to try to kill fungus eating at your fish in your aquarium). A recent email reminded me about fluorogen activating peptides again, so I went back and looked at this data again, and it really isn’t bad.

In summary. 1) Works well, giving ~ 100-fold increase to fluorescence from background when highly expressed. 2) Probably want to do at least 500 ng/mL or perhaps even 1ug/mL to improve signal over background. 3) Still get specific signal after washout, but you may as well leave it on and get max signal.

Split mCherry

I had a product that could have benefitted from using split mCherry to serve an AND function. Put the split mCherry in my usual mCherry spot in the recombination vector (2A’d with Puromycin, also), and I couldn’t see any visible fluorescence when both the small and big fragments were in the same cell. After some time, I saw a paper using split super-folder mCherry fused with the Spycatcher system, so I used that and that seemed to allow us to now see a shift in fluorescence off the background. I found another paper using an improved split super-folder mCherry (sfmCherry3C), and tested that, which seemed to work slightly better.

So while not perfect, in that there isn’t *complete* separation from the background distribution, it’s shifted away enough that it should serve my purposes (for now). Though, well, I’ll probably keep playing around with this (perhaps adding something like a leucine zipper?) and seeing if that helps increase fluorescence.

Dark culture media with RUBY

I like playing around with various recombinant DNA tools to see how they work and figure out if they’ll do something useful for me. Sometimes it works out amazingly, like iCasp9, which is a fantastic negative selection transgene. Other times, it’s not so clearly a success…

I recently ordered RUBY from Addgene (originally used by the depositing lab to darken plant roots) to see if I could turn my cultured cells visibly darker. I had actually messed around a little with this concept previously using tyrosinase, where it worked in making the cells darker, albeit one could only see the darker color either as a centrifuged cell pellet, or perhaps as a large overgrowing “colony” on an originally sparse culture plate. Well, I shuttled RUBY into my recombination vector and selected for cells expressing it. I don’t even think I saw dark cells this time (though I didn’t look very closely, since this is just a fun side-experiment), but I did notice that these cells had much darker media than their recombined siblings, like the cells expressing fluorescent proteins in the two wells to the right of it. So I’m not exactly sure what’s happening, but the chromogenic small molecule is definitely making it out into the culture media.

I guess I’ll just file this observation for now and see if it ever comes in useful at some point the future! hahaha.

Update: FYI, RUBY is *BIG*. Like ~ 4kb kind of big, since it encodes multiple enzymes within a chemical pathway. So def not a small chromoprotein kind of thing.

HEK293Ts with melanin

I think synthetic biology is really cool, and I like playing around with recombinant DNA elements so I can see how well they work in my own hands. If they work OK, then I just let that knowledge stew in the back of my brain until I can eventually figure out a use for it. Reading this paper by Martin Fussenegger made me realize just how easy it is to make cultured cells express melanin. Here was my first foray in creating melanin in HEK cells by overexpressing tyrosinase/

Cells pelleted in the tubes on the left are expressing tyrosinase. The cells pelleted in the tubes on the right are not.

Doesn’t quite work well enough to use as a general reporter (it’s really hard to tell in a cell monolayer, and only becomes noticeable as colonies of cells or in a pellet, like above), but still kind of fun to see. Let’s see if I find an eventual use for this in some future work.