TL;DR: Statistics is everywhere, and simulating bottlenecks that happen during routine lab procedures such as dilutions of cells can potentially help you increase reproducibility, and at the least, help you better conceptualize what is happening with each step of an experiment.

I’m still working on getting a cell counter for the lab. In the meantime, we’ve been using an old school hemacytometer to count cells before an experiment. Sarah had used a hemacytometer more recently than me, and knew to dilute the cells from a T75 flask 10-fold to get them into a countable range for the hemacytometer. She said she had performed the dilution by putting 10 ul cells in 90 ul media (and then putting 10 ul of the dilution into the hemacytometer). But as she said this, she asked whether it was OK to perform the dilution as described; a grad student in her previous lab had taught her to do it that way, but a postdoc there said it was a bad idea. My immediate response was that if the cells are sufficiently mixed, then it should be fine. And while that was my gut reaction, I realized that it was something I could simulate and answer myself using available data. Would the accuracy of the count be increased if we diluted 100 ul of cells into 900ul of media, or 1ml of cells into 9ml of media?

Here are the methods (skip if you don’t want to dive in and want to save yourself a paragraph of reading): To me, it would seem the answer to whether the dilution matters depends on how the cells are dispersed in the media / how variable the count is when the same volume is sampled numerous times. Sarah’s standard practice is to count four squares of the hemacytometer, so I had four replicate counts for each volume pipetted. She had repeated this process three times by the time I had performed the analysis, giving me a reasonable dataset I could roll with. I got the mean and standard deviations for each of the three instances, all corresponding to a volume of 0.1 ul. They were all quite similar, so I created a hypothetical normal distribution from the average mean and standard deviation. Next was seeing how different ways of performing the same dilution impacted the accuracy of individual readings. I recreated the 10 ul cells by sampling from this distribution 100 times, 100 ul cells by sampling 1,000 times, and 1 ml by sampling 10,000 times, and taking the mean. I repeated this process 5,000 times for each condition, and looked at how wide each distribution was.

I then turned the counts into concentration (cells / ml):

Instead of stopping there, I thought about the number of cells I was actually trying to plate, which was 250,000. The number the distributions were converging to was ~ 27.3 (black line), so I used that as the “truth”, and saw how many “true” cells would be plated if I had determined the volume needed to be plated based on each of the repeat concentrations calculated by each of the conditions of dilutions. The resulting plot looked like this:

So as you can tell based on the plot, there are slight differences in cells plated depending on imprecision propagated by the manner in which the same 10-fold dilution as performed: while all distributions are centered around 250k, the 10 ul dilution distribution was quite wide, while the 1 ml in 9 ml dilution resulted in cell counts very close to 250k each time. To phrase it another way, ~15% of the time, a 10 ul in 90 ul dilution would cause the “wrong” number of cells to be plated (less than 24k, or more than 26k). In contrast, due to the increased precision, a 100 ul in 900 ul dilution would never result in the “wrong” number of cells being plated. So speaking solely about the dilution, the way the dilution was being performed could have some light impacts on the accuracy of how many cells would be actually plated.

I was going to call this exercise complete, but I ran this analysis by Anna, and she mentioned that I wasn’t REALLY recreating the entire process; sure I had recreated the dilution step, but we would have also counted cells from the dilution in the hemacytometer to actually get the cell counts in real life. Thus, I modified the code such that each dilution step was followed by a random sampling of four counts (using the coefficient of variation determined from the initial hemacytometer readings), and taking the mean of those counts; this represented how we would have ACTUALLY followed up each dilution in real life. The results were VERY different:

In effect, the imprecision imparted by the hemacytometer counts seemed to almost completely drown out the imprecision caused by the suboptimal dilution step. This was pretty mind-blowing for me; especially considering that I would have totally missed this effect had I not run this post by Anna. Now fully modeling the counting process, a 10 ul in 90 ul dilution would cause the “wrong” number of cells (less than 24k, or more than 26k) to be plated ~ 42.5% of the time, and a 100 ul in 900 ul dilution would still cause a “wrong” cell number to be plated ~ 42.2 % of the time; almost identical! Thus, while a 100 ul in 900 ul dilution does impart some slightly increased accuracy, it’s quite minor / negligible over a 10 ul in 90 ul dilution. So while in a sense this wasn’t the initial question asked, it’s still effectively the real answer.

At the end of the day, I think the more impactful aspect of this exercise is the idea that even routine aspects of wet-lab work are deeply rooted in stats (in this case, propagation of errors caused by poor sampling), and that the power of modern computational simulations can be used to optimize these procedures. There’s something truly empowering to having a new tool / capability that gives you new perspectives on procedures you’ve done a bunch of times, and allows you to fully rationalize it rather than relying on advice given to you by others.

Here’s the code if you want to try running it yourself.
Acknowledgements: Thanks to Sarah for bringing this question to my attention. Also, BIG THANKS to Anna for pointing out where I was being myopic in my analysis, which got me to a qualitatively different (and more real-life relevant) answer. It really is worth having smart people look over your work before you finalize it!

I recently bought myself a new computer for the office, which meant that I had to download and install all of the key software I used for work. I decided to write down what these were so future me (or future employees) could have it as a reference. All of the following links / commands are for use with a mac.

Everyone should install these:

Google Chrome

Google Drive & Sync or Google Drive File Stream

Microsoft Office

Anaconda

R (requisite for R Studio)

R Studio 
packages worth installing: tidyverse, shiny, ggrepel

XQuartz (requisite for Inkscape on older macs)

Inkscape

Pymol

iTerm2

Sublime Text 3

FlowJo (v10)

Optional (but useful):

Box Sync Installer

Dropbox desktop app

MacDown

A plasmid Editor (ApE)

Snapgene Viewer

Flux

Sim Daltonism

ImageOptim

The Unarchiver

4 Peaks

Skim

GitHub Desktop

Zoom Desktop

Skype Desktop

OBS (Open Broadcast Software)

Spyder (Pymol IDE)

Forticlient VPN

Cyberduck

Useful installations from the command line

Samtools

First go to the Samtools directory

$ ./configure

$ make

$ make install

$ export PATH=bin:$PATH

Homebrew

$ xcode-select –install

$ ruby -e “$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/install)”

$ brew doctor

Gifski

$ brew install gifski

FFmpeg

$ brew install ffmpeg

PyPDF2

$ conda install -c conda-forge pypdf2

A note about editing your .bash_profile

Packages to install in RStudio

You can install these as you need them, but some of these packages are so useful you may as well do it up front when setting up your computer.

install.packages(“tidyverse”)
install.packages(“ggrepel”)
install.packages(“patchwork”)
install.packages(“googlesheets4”)
install.packages(“MASS”)
install.packages(“ggbeeswarm”)

Adding printers in WRB 5th floor

Note: You will not be able to set up the printer when on the “CaseGuest” wireless network. The printer can be setup while on the “CaseWireless” wireless network, though it may be hard to access it from the lab-space. In that case, a direct link to the ethernet is probably the best way to go. When connected to the land-line, you may first encounter an error saying the identity of any websites you go to cannot be confirmed. This may be because you are on a new computer, in which case you have to first register your computer with CWRU IT by typing in “setup.case.edu” into a browser.

The most convenient printer is the black-and-white one near my office. To add this printer on a mac: 1) Go to Systems Preferences (such as through the apple icon op) 2) Go to Printers and Scanners 3) Press the “+” sign to add a printer 4) Type in the IP address “129.22.253.166”. 5) The protocol should be “Line Printer Daemon – LPD” 5) You can rename it to something like “Nearby Black and White” to make it easier to remember. 6) It’s fine to leave Use as “Generic PostScript Printer” 7) Click on Duplex Printing (Since it’s useful). Voila!

5/31/22 edit: On my newest Mac, the process was somewhat different. Here, after going to “System Preferences” > “Printers & Scanners”, go to the second “IP” tab with the globe on it, add in the “129.22.253.166” to the address field, set the protocol to “Line Printer Daemon – LPD” with all other settings kept the default except for the location (I write something like “WRB 5-east mailroom”). Note: I click on the option of duplex printing, which allows you to print on the front and back sides.

6/13/23 edit: If, for some reason, the printer on our side of the floor is down, you can always use the printer on the other side of the floor. Same instructions as above, except the IP address is “129.22.253.139”.

VAMP-seq is a valuable tool for identifying loss-of-function variants for a protein when you don’t have a specific assay for characterizing that protein’s activity. This is because almost all proteins can be “broken” by mutations that cause the protein to mis-fold, mis-traffick, or any other perturbations that may cause the protein to disappear from its normal cellular compartment. In the case of PTEN, we identified 1,138 variants that were lower than WT abundance. Notably, ~ 60 % of variants characterized as pathogenic in people were loss-of-abundance variants, confirming the importance of this property.

VAMP-seq uses a genetic fusion between your protein of interest and a fluorescent moeity (eg. EGFP) to assay its steady-state abundance. Fusions with unstable variants of your protein results in cells expressing unstable fusion proteins that don’t really fluoresce. In contrast, fusions with stable variants (such the WT protein) may fluoresce brightly. Unlike western blots which are ensemble measurements, this assay can be performed at a single-cell level. Correspondingly, this single-cell fluorescence readout makes it possible to test a large library of variants in parallel, in a multiplexed format.

Still, when thinking about using VAMP-seq, you must first consider its limitations. While strong loss-of-abundance variants will be non-functional, variants of intermediate abundance may still be sufficiently functional in many contexts. Furthermore, while low abundance correlates with inactivity (and pathogenicity, in many clinical genetics contexts), WT-like abundance in no way indicates activity (or benignity when observed in people). For example, active site mutants often destroy protein function while having little to no effect on protein folding and abundance.

There are also proteins that are inherently incompatible with VAMP-seq. Secreted proteins won’t work because you lose the single-cell, genotype-phenotype link needed for the single-cell assay to work. Marginally stable or intrinsically disordered proteins likely won’t work due to a lack of destabilizing effect. Obligate heterodimers won’t work, though you may be able to get around it by overexpressing the protein partner, such as what I did with MLH1 for assessing PMS2 (See Fig 6). Proteins that cannot be tagged are problematic; this likely includes proteins that normally exist in crowded complexes, or that have key trafficking motifs on their termini. Proteins that are toxic to cells when overexpressed also poses problems, though one of my new landing pad platforms may help with that.

If your protein of interest passes those criteria, the rest is empirically confirming that there is enough signal over background to run the assay. A good literature search is a great place to start. You should look for 1) evidence that an N- or C-terminal tag works well (both good expression and normal protein activity), and 2) known destabilized variants that could serve as controls when performing the preliminary experiments. Ideally, there will be a clear difference in fluorescence distributions that are separable by thresholds used in FACS sorting, and that these differences are physiologically meaningful (as far as we know). Comparing /correlating the results of western blotting with the fluorescence distributions of EGFP-tagged protein is helpful (see below for PTEN). If the initial EGFP distributions between WT and the destabilized variants don’t seem super crisp, see what it looks like when you take the EGFP:mCherry ratio. As you can tell in the below figure, the EGFP:mCherry ratio is quite handy for increasing the precision of each distribution, as it divides out much of the heterogeneity in transcription / translation between cells.

Regardless, I recommend to most people that they clone both N- and C- terminal fusions, and minimally look at the MFI values of the cells expressing each fusion, as low MFI will likely mean low dynamic range of the assay (from too little signal over background). Ideally, both WT and controls will be tested in both contexts. If the signal is relatively high but there’s concern that the large GFP fusion is causing problems, you could try 1-10/11 split EGFP. This only requires fusion with the ~ 15aa beta-strand 11 of EGFP (and separately co-expressing the larger fragment in the cells), though it’s not completely free of steric hindrance as it requires the spontaneous complex formation of the two subunits for fluorescence. This format also worked with PTEN (See SFig 1d), though I noticed a ~ 10-fold hit in overall fluorescence using the splitGFP format.

Once it passes all those tests, then follow the steps in the Nature Genetics paper. Good luck!

Cas9 is a dream biotechnology, since you can use it to quite easily and specifically disrupt almost any DNA sequence you’re interested in modifying. Its been particularly helpful in work with cultured mammalian cells. In this document, I will describe how I use Cas9 to make gene knockouts in HEK 293T cells.

So although I’ve heard good things about transfecting guide-RNA loaded Cas9 ribonucleotide protein complexes into cells, I haven’t done this yet. Instead, I’ll describe a likely less optimal, but super cheap and easy method of co-transfecting Cas9 and guide-RNA expressing plasmids into cells to create knockouts. Specifically, I’ll describe my method of knocking out a gene by removing a big chunk of coding sequence, either ablating its start codon, or getting rid of an important exon to only create truncated or frameshifted non-functional proteins.

Equipment & Materials:

1) Tissue culture hood & incubator
2) Transfection reagent
3) Fluorescence Activated Cell sorter (FACS)
4) PCR machine, polymerase, Gibson master mix.
5) Agarose gel electrophoresis apparatus and some form of GelDoc.
6) Sanger sequencer (or sequencing service)

Step1: Make custom guide-RNAs for your gene of interest

As mentioned above (and shown in the figure), to create genetic knockouts, I like to delete out entire exons. The main reason is that it’s really easy to genotype with PCR. It’s also nice in that if the change is made, even if the repair is an in-frame indel, the deletion will be so large that the protein should still be loss-of-function. Lastly, exon deletion can often be achieved by targeting adjacent introns; this means that even if you stably express the Cas9 / gRNA combo, you can re-express a cDNA version and have it not be susceptible to disruption.

I first find guide RNAs that will target the DNA I want to disrupt. Addgene seems to have a pretty handy list of guide-RNA sequence design applications here: (https://www.addgene.org/crispr/reference/). By default, I still tend to go to this classic website: (http://crispr.mit.edu/). Have it come up with a pair of seemingly specific guides, per the strategy above. Note: When using a U6 Pol3 promoter, I’ve read that if your guide RNA doesn’t start with a G, you should plan to include an extra G in front of the guide to increase transcription.

Addgene lists a bunch of protocols from the original labs here: (https://www.addgene.org/crispr/reference/#protocols). Depending on how you like to do your molecular cloning, you may want to follow one of those. I’ll describe making a custom guide-RNA using Gibson Assembly, which is absolutely amazing.

Using a really basic U6-promoter guide-RNA expressing plasmid (like so: XXXX), design a pair of long primers that overlap in their 5′ ends with the guide-RNA sequence, but hybridize to sequences adjacent to the guide-RNA sequence but point in the opposite directions. The ~20nt guide-RNA sequence makes for a great Gibson overhang.

The steps here are:
1) Amplify plasmid, appending the guide-RNA sequence to the termini of the amplicon. I use Kapa Hifi, with 40ng template plasmid (30ul total), and ~7/8 cycles.
2) DPNI digest. I pipet 1ul of DPNI directly into the 30ul reaction tube, and incubate at 37*C for 2 hours (I suspect you don’t need all two hours).
3) “Purify” your intended plasmid. At first, I used to run my amplifications out on an agarose gel and gel extract (using a Qiagen gel extract kit). Nowdays, I just use a Zymo Clean and Concentrate kit (first setting aside some sample to run on a gel afterwards, for diagnostic purposes only). I elute in a small volume (6 ul for the Zymo kit).
4) I then make equal volume mixtures (eg. 1ul “insert”, 1ul “vector”), and mix in the corresponding amount of 2x Gibson Master Mix (So 2ul for the above example). I incubate this at 50*C for 0.5 to 1 hr. Standard protocol here.
5) I mix the sample (usually 4ul) into 50ul of chemically competent E.coli, and transform (eg. “Heatshock” transformation).
6) In the case of the Church lab plasmids, they’re Kan resistant, so don’t forget to recover before plating onto Kan plates.

Normally, I end up seeing tens to hundreds of colonies the next day. I then usually pick 2 colonies to grow up and miniprep. When I run Sanger, usually at least one colony contains plasmid DNA with the intended guide-RNA encoded. If neither colony has the intended plasmid, I normally pick two more and screen those. I don’t remember the last time this protocol failed (possibly hasn’t in my dozen attempts), but I think if it does it’s likely going to be from there being too much template, so I would then re-do the process really making sure DPNI is doing its job (and maybe gel extracting if I have to).

Step2: Transfecting the guide-RNA

I’ve only ever done this in HEK 293T cells, since that’s where I do most of my tissue culture work (since they’re so easy to handle). Next step is to transfect cells with the guide-RNA and Cas9 encoding plasmids. I tend to do this in 6-wells, using Fugene 6.

For this transfection, I either mix the guide-RNA plasmids (half plasmid for guide 1, the other half plasmid for guide 2) with PX458 (which encodes Cas9 2A-linked with EGFP), or a plasmid expressing untagged Cas9 along with a plasmid encoding a fluorescent protein. Since transfection usually results in many many copies (hundreds? thousands?) of plasmid getting in per cell, cells expressing the fluorescent protein (generally) should have been transfected with guide RNAs and Cas9 also (and thus are the cells most likely to have been edited).

Then, approximately two days after transfection, I sort for transfected cells using the fluorescent marker. I usually do this in a 96-well plate, sorting single cells into individual wells (to create clonal lines), while taking one of the corner wells and sorting 100+ cells into this one. This gives me an improved bulk population as backup in case the single clone growouts fail, and gives me a well that makes focusing onto the right focal plane a lot easier when using the light microscope (since hard to correctly focus on wells where it’s not clear if there are even any cells there).

If in a rush, I’ll check on the wells in approximately a week. By then, you can definitely see colonies of cells growing by brightfield microscope. I would then trypsinize and replate to keep them growing happily. If I’m not in a rush, I’ll sometimes let the cells grow for a couple weeks before checking; at this point, the media color starts becoming yellowish in wells containing growing cells, and clonal lines that grew out are large enough that if I transfer them to a new well, I’ll be able to do an expt with those cells in a few days.

Step3: PCR Genotyping

Once I have enough cells (say, at least half a 24-well plate well’s worth), I’ll extract genomic DNA using a Qiagen DNA easy kit. I’ll then use the primer combinations shown above in the figure. Notably, primer combo BC should only amplify if there is a wild-type copy present (DNA from unmanipulated cells should serve as a positive control). This should be a binary “amplifies or doesn’t” readout. The other primer combo also gives a slightly orthogonal perspective on what might be present at the intended locus within the genome. If unmodified, combo AC should give a large band (or no band, if these are really large introns). If modified, combo AC should give a much smaller product (I aim for a few hundred bases for easy visualization by agarose gel electrophoresis). To verify that this small band is what I think it is, I gel extract the band and sequence using Sanger. Conveniently, you can even use either primer A or primer B as the Sanger sequencing primer if you didn’t order a dedicated internal primer. In the handful of times I’ve done this, I think the predominant modified product that I’ve seen is literally a blunt-ended Non-Homologous End Joining of the two DNA fragments without any resection / removal of terminal nucleotides. Thus, if this is indeed the major modified product, I’m able to design my guides such that NHEJ ligation results in a frameshift.

Using the above PCR, I go through and screen the colonies to find one that looks to be a complete knockout (no amplification with primers BC, only small band with primers AC). Of course, actual protein knockout should be confirmed by western blot.

Voila! Knockout cells to now use in your experiments!

Some references:

I think this project / paper from super grad student Molly Gasperini in Jay Shendure’s lab is what originally made me think about using paired gRNA deletions to make knockouts.

I’ve tried to link to specific protocols in the above descriptions. For any details that are ambiguous, the methods described in my 2018 NAR paper should largely describe the basic steps / reagents used.

PS: Ed Anderson, a Postdoc at UNC Chapel Hill, made the excellent point that the molecular cloning in this post (ie. Gibson assembly steps) could be performed with In-Vitro Assembly (IVA). We certainly had some success with this method in my postdoc lab (albeit we used it for parallelized NNK library generation), though could certainly be used here for simpler and most cost-effective routine molecular biology.