Blog / Making the Cut / tips-and-tricks

Improved knockout with Cas9

Jacob Corn

Cas9 is usually pretty good at gene knockout. Except when it isn’t. Most people who have gotten their feet wet with gene editing have had an experience like that in the following gel, in which so...


Cas9 is usually pretty good at gene knockout. Except when it isn’t. Most people who have gotten their feet wet with gene editing have had an experience like that in the following gel, in which some guides work very well but others are absolute dogs.

That’s a problem if you have targeting restrictions (e.g. when going after a functional domain instead of just making a randomly placed cut). So what can one do about it?

TL;DR Adding non-homologous single stranded DNA when using Cas9 RNP greatly boosts gene knockout.


The problem

There have been a few very nice papers showing that Cas9 prefers certain guides. I refer to these as the One True Guide hypothesis, with the idea being that Cas9 has somehow evolved to like some protospacers and dislike others. The data doesn’t lie, and there is indeed truth to this - Cas9 likes a G near the PAM and hates to use C. But guides that are highly active in one cell line are poor in others, and comparing very preference experiments in mouse cells vs worms gives very different answers. That’s not what you’d expect if the problem lies solely in Cas9’s ability to use a guide RNA to make a cut.
But of course, Cas9 is only making cuts. Everything else comes down to DNA repair by the host cell.

Our solution

In a new paper from my lab, just out in Nature Communications, we found that using a simple trick to mess with DNA repair can rescue totally inactive guides and make it easy to isolate knockout clones, even in challenging (e.g. polyploid) contexts. We call this approach “NOE”, for Non-homologous Oligonucleotide Enhancement.
(The acronym is actually a bit of a private joke for me, since I used to work with NOEs in a very different context, and Noe Valley is a nice little neighborhood in San Francisco)
How does one perform NOE? It’s actually super simple. When using Cas9 RNPs for editing, just add non-homologous single stranded DNA to your electroporation reaction. That’s it. This increases indel frequencies several fold in a wide variety of cell lines and makes it easy to find homozygous knockouts even when using guides that normally perform poorly.
The key to NOE is having extra DNA ends. Single stranded DNA works the best, and even homologous ssDNAs one might use for HDR work. We tend to use ssDNAs that are not homologous to the human genome (e.g. a bit of sequence from BFP) because they make editing outcomes much simpler (NHEJ only instead of NHEJ + HDR). But double stranded DNAs also work, and even sheared salmon sperm DNA does the trick! Plasmids are no good, since there are no free ends.
We know that NOE is doing something to DNA repair, because while this works in many cell lines, the molecular outcomes differ between cells! In many cells (5/7 that we’ve tested), NOE causes the appearance of very large deletions (much larger than you would normally see when using Cas9). But in 2/7 cells tested, NOE instead caused the cells to start scavenging little pieces of double stranded DNA and dropping them into the Cas9 break! The junctions of these pieces of DNA look like microhomologies, but we haven’t yet done the genetic experiments to say that this is caused by a process such as microhomology mediated end joining.

What's going on here?

How can making alterations in DNA repair so drastically impact the apparent efficacy of a given guide? We think that our data, together with data from other labs, implies that Cas9 cuts are frequently perfectly repaired. But this introduces a futile cycle, in which Cas9 re-binds and re-cuts that same site. The only way we observe editing is when this cycle is exited through imperfect repair, resulting in an indel. Perfect repair makes a lot of sense for normal DNA processing, since we accumulate DNA damage all the time in our normal lives. We'd be in a sorry state indeed if this damage frequently resulted in indels. It seems that NOE either inhibits perfect repair (e.g. titrating out Ku?) or enhances imperfect repair (e.g. stimulating an ATM response?), though we are still lacking direct data on mechanism at the moment.

What is it good for?

The ability to stimulate incorporation of double stranded DNA into a break might be useful, since non-homologous or microhomology-mediated integration of double stranded cassettes has recently been used for gene tagging. But we haven’t explicitly tried this. We have also found NOE to be very useful for arrayed screening, in which efficiency of the edit is key to phenotypic penetrance and subsequent hit calling.
Importantly, NOE seems to work in primary cells, including hematopoietic stem cells and T cells. We’ve been using it when doing pooled edits in unculturable primary human cells, and find that far higher fractions of cells have gene disruptions when using NOE. We’ve so far only worked in human cells with RNP, and I’m very interested to hear peoples’ experience using NOE in other organisms. We haven’t had much luck trying it with plasmid-based expression of Cas9, but other groups have told me that they can get it to work in that context as well. 

How do I try it?

So if you’re interested, give it a shot. The details are all in our recent Nature Communications paper, but feel free to reach out if you have any more questions. This work was done by Chris Richardson (the postdoc who brought you flap-annealing HDR donors), Jordan Ray (an outstanding undergrad who is now a grad student at MIT), and Nick Bray (a postdoc bioinformatics guru).

X Close

Living protocols for genome editing

Jacob Corn

The field of genome editing is moving at breakneck speed and protocols are rapidly evolving. We've already made a f...


The field of genome editing is moving at breakneck speed and protocols are rapidly evolving. We've already made a few different posts on tips, tricks, and protocols for genome editing and regulation. But effectively sharing protocols and making sure that they're up to date is a daunting task. Much better to have a community-driven effort, where a starter protocol can be tweaked and updated as new developments come along.

That's why I'm happy to share that we've recently started putting our methods on This is an open repository for protocols, which the great feature of "forking". This means you can start from a protocol that you like, tweak it as desired, make a record of the tweaks, and re-publish your changes. Everything is also linkable to a DOI, which means you can potentially reference online protocols from within papers.

IGI protocols for T7E1 assays, in vitro transcription of guide RNAs, Cas9 RNP nucleofection, and more are available at

Here's an explanatory video, from the team.


X Close

Adventures in CRISPR library preparation

Benjamin Gowen

For the last couple of months, a few of us at the IGI have been generating new sgRNA libraries for CRISPRi and CRISPRa. After scraping colonies off of nearly one hundred extra-large LB-Agar plate...


For the last couple of months, a few of us at the IGI have been generating new sgRNA libraries for CRISPRi and CRISPRa. After scraping colonies off of nearly one hundred extra-large LB-Agar plates, it was time to fill the lab with the sweet smell of lysed bacteria and DNA prep buffers. We were working with 21 separate sublibraries, totaling around 250,000 sgRNAs. Plasmid prep on this scale is a completely different beast from anything I had done before, so we decided to share some thoughts on what works (and what doesn't!) for efficiently prepping sgRNA libraries.

Prepping the work station

We were worried about other plasmids sneaking into our preps--especially individual sgRNA plasmids that get used frequently in our lab. We doused and scrubbed our benches and vacuum manifold with 70 % ethanol and RNase-Away before starting, and a few times throughout the day. This should hopefully destroy or denature any stray plasmids hanging around. It's also worth cleaning out your vacuum trap and putting fresh filters in the vacuum line, since old dirty filters can really weaken vacuum power.

Do all the DNA prep at once

For me, it's much more efficient to spend a couple of days solely devoted to high-throughput DNA prep than to spread the work out over several days, a few columns at a time. 


The initial lysis and neutralization steps in most plasmid preps are time-sensitive, so there's a limit on how many samples one person can process at once. We found that a team of 3 people (each processing 8 samples at once) maximized our throughput without us bumping into each other too much. After eluting DNA off the columns, once person can manage the DNA precipitation while others start on the next round of samples.

Starting material

Scraping the colonies off of a 23x23 cm LB-Agar plate gave us an average bacterial pellet mass of 1.1 g (range 0.5-1.6 g). This meant that each plate of bugs got its own maxiprep column (see below for kit recommendations). If you're working with bugs from liquid culture or other plate sizes, you can pool or aliquot the samples to get a similar pellet mass per column.

Plasmid prep kits

We wound up trying several different plasmid prep kits, and the clear winner in our hands was the Sigma GenElute HP Plasmid Maxiprep Kit. The columns are compatible with the QIAGEN 24-port vacuum manifold we already had in the lab, the protocol was amenable to doing 24 preps in a batch, and the house vacuum system in our building was strong enough to pull liquid through all 24 columns at once. Importantly, all of the columns ran consistently and reasonably quickly. One slow or plugged column is an annoying but solvable problem when doing 4 or 5 preps, but it can really back up the pipeline when doing multiple batches of 24. Our  average yield from this kit was 1.4 mg per prep.

Kits to avoid:

  • Sigma GenElute HP Plasmid Megaprep: Sigma advertises 4 times the yield from a megaprep column compared to their maxipreps. Some of our samples could be pooled, so we thought pooling 4 samples into one megaprep would be faster than running them as 4 individual maxipreps. Boy were we wrong! The megapreps had to be processed one or two at a time, and thus didn't scale well at all. Worst of all, the megaprep columns were NOT compatible with the QIAGEN vacuum manifold. We managed to fix this with tubing and adapters, but the house vacuum system was only strong enough to pull on one or two of the larger megaprep columns at a time. For us, mega preps took far more time and gave about half the yield we would have expected from just grinding through 4 times as many maxipreps.
  • QIAGEN Plasmid Plus Maxiprep Kit: 1 out of the 8 columns we used  stalled while running the cleared lysate. That column had to be left on the vacuum overnight. Our yields were also lower than the Sigma maxipreps. 
  • QIAGEN HiSpeed Plasmid Maxiprep Kit: These don't scale well at all. The columns aren't compatible with a vacuum manifold, and the QIAprecipitator syringe filters require a lot of manipulations to each individual sample. After the first 4 samples, I ditched the QIAprecipitator step altogether. Precipitating the DNA with a 45 minute spin was much faster when dealing with 10 or 20 preps at once.

We're always interested in ways to make the next sgRNA library prep easier than the last. If you have your own favorite plasmid prep kit or other tricks for efficient library preparation, feel free to leave a comment. 

Special thanks to the other members of Team DNA Prep--Gemma Curie, Amos Liang, and Emily Lingeman. I'd still be running maxipreps if it weren't for them!

X Close

Choices choices

Jacob Corn

 If you're just getting into CRISPR/Cas9 editing, you might be a bit overwhelmed by all of the options available to you. Especially if you're working in human cells. I mean, just take a look at at...


 If you're just getting into CRISPR/Cas9 editing, you might be a bit overwhelmed by all of the options available to you. Especially if you're working in human cells. I mean, just take a look at at the Addgene mammalian CRISPR page! Which one to choose!? There are various tags, numbers of NLSes, tag/NLS locations, promoters, sgRNA+Cas9 in one vector, and the list goes on.

A postdoc in the lab, Pei-Chun Lin, wanted to prepare for her own knockout projects by testing a few of the Cas9 systems in an apples-to-apples comparison. This is the most vanilla use of Cas9: make a single cut and watch a gene product go away. Pei-Chun obtained a HEK293 cell line stably expressing YFP and tested several sgRNAs to find a few that worked with varying efficiencies, then used those two guides with five different setups, including both all-in-one and separate sgRNA and Cas9 vectors. She transiently transfected each Cas9/sgRNA for 72 hours, then read out cutting efficiency at various times with both YFP fluorescence (to monitor disappearance of the protein) and the T7 endonuclease assay (to monitor genomic indels caused by Cas9 cutting).

As you might expect, all systems work when given a great sgRNA (phew!), with varying efficiency (compare guide YFP#6 between Cas9s). The Cas9 system with the lowest apparent activity (Addgene #50661) has the big benefit of inducibility, so there's still a great reason to use it. But the different efficiencies mean that sgRNAs that look decent with one Cas9 vector can appear inactive when paired with another (e.g. compare guide YFP#4 when used with Cas9 #41815 vs #43861).



Pei-Chun's data also highlight the temporal difference between mutation and protein abundance (as might be caused by highly stable proteins). Genomic edits were easily detectable after 3 days by T7E1, but it took 6 days to even begin detect a change in protein abundance. Not unexpected, given the known stability of XFPs, but keep this in mind next time you plan a Western or other protein-based assay to read out Cas9 activity!




Here's all the data as a PDF

X Close

Looking at genome-wide guide libraries

Jacob Corn

Genome-wide sgRNA libraries for knockout screening are all the rage, and researchers have very generously made them available on Addgene. But which one shou...


Genome-wide sgRNA libraries for knockout screening are all the rage, and researchers have very generously made them available on Addgene. But which one should you choose for your project? Data is always better than guesswork, and I've started looking at what's out there in a relatively systematic way. Note that this is a work in progress, so results will probably change a bit, but the general gist will probably be similar. Also, details in scoring of guides is a controversial topic, so for now let's discuss stick to big picture themes everyone can probably agree upon.

Below is a smoothed distribution of scores (lower is better, scale is arbitrary), based on several metrics for three different guide libraries related to two papers (score on the x-axis and frequency of that score on the y-axis). Why three libraries for two papers? Turns out one Addgene-deposited library isn't the same as what's tested in the paper. Thankfully the authors were very transparent and deposited both sets of sequences into Addgene. Kudos to them!

The first thing to notice is that most guides score below 50. That's good! It means those guides have only been penalized for relatively trivial things. Hence, the libraries are pretty good and most guides should be functional (a no-brainer, given the awesome papers on these libraries). The second thing to notice is that some libraries have a peak around a score of 50. That's bad, because 50 is the penalty added to guides containing Pol III terminator sequences. These are likely to be useless guides, since they won't even be fully transcribed, but at least they should be silent in the library. The third thing to notice are the jagged peaks  above a score of ~100. These guides are relatively rare in the libraries, but are potentially scary. Guides can only get a score this high if they have likely off-target sequences that occur within coding regions. In fact, guides get +100 for each off-target coding region. So the saw-tooth pattern >200 represents guides in these libraries that could potentially knockout more than one gene other than the one target. Hence, when using these libraries it's very important that your phenotype of interest occur with more than one guide (as stressed in the papers). Trusting just one guide could lead you very far astray due to off-target effects.

The above isn't mean to dissuade anyone from using these libraries. They're an incredible and unprecedented resource and the originators have done the community a huge service by making them available for a nominal fee on Addgene. But I think all seasoned scientists know that it's dangerous to treat things as a black box, and that looking under the hood never hurts. I'm sure library improvements will be a hot topic for some time to come and that can only be a good thing for people who want to use these for new biology.


X Close

How to make a guide RNA for a Cas9 knockout

Jacob Corn

Guide RNA lore is split across multiple papers, people, and places, and I'm frequently asked about the "best" way to make a guide RNA for Cas9. The following is the state of the art as I understan...


Guide RNA lore is split across multiple papers, people, and places, and I'm frequently asked about the "best" way to make a guide RNA for Cas9. The following is the state of the art as I understand it, as of today (8/11/14), split into several steps. The steps below assume you want to use Streptococcus pyogenes Cas9 to cut a gene to introduce an insertion/deletion ("indel") to make a knockout (the simplest use case) using a double-strand cut (wild type Cas9). The process may differ if you want to (for example) use CRISPRi to inhibit transcription. I've used * to mark steps that would be at least somewhat altered for other applications or if you're using less common parts (e.g. Cas9 from another species, different guide RNA promoter, etc).   Before you start

  1. Decide what kind of targeting you want to do. Here we're considering double-stranded cutting to make a knockout via introduction of an indel.
  2. Decide which Cas9 you'll use. Here, we'll assume you're using Streptococcus pyogenes (aka "Spy"). This choice would affect the protospacer adjacent motif ("PAM") you'll look for.*
  3. Get the genomic sequence you want to target from NCBI Gene or elsewhere (e.g. if you're targeting an intergenic region).*
  4. For knockouts, you generally want to introduce an indel as close to the 5' end of the coding region as possible. This will have the highest likelihood of creating a protein-destroying frameshift.*

Finding Guides

  1. Use one of many servers to find guide RNAs in the region you'd like to cut. For example, CRISPR-MIT, E-CRISP, or CHOPCHOP. Which tool you choose is mostly personal preference, and each has their own model for scoring guide RNAs.
  2. Each target site will either be ~23 bases ending in "GG" (guide binds Crick strand) or ~23 bases starting in "CC" (guide binds Watson strand). The protospacer adjacent motif ("PAM") refers to those last or first three bases and is present in the DNA you're targeting but should not be used in the guide RNA. Hence, the guide RNA itself will be ~20 bases and lack a PAM. SpyCas9 can also use "AG" (Watson "CT") as a PAM, but not as efficiently.
  3. The exact length of the guide doesn't seem to matter very much; anything from 17-27 bases (remember, guides don't include the PAM) seems basically OK (with some qualifiers).

Choosing a guide Now you have a (possibly very long) list of potential guides. Each one has an associated score. How do you choose which one to use? Here's a semi-ordered list of factors to consider, from most to least important. Consider these qualitative, rather than a quantitative score.

  1. For a knockout, it doesn't matter which strand the guide RNA binds, but CRISPRi guides should be complementary to the non-coding strand.*
  2. Guides should be perfectly complementary to the region you want to target in the 8-12 bases closest to the PAM.
  3. Never choose a guide that has any significant off-target sites (perfect match for the 8-12 bases closest to the PAM) in a coding of the genome.
  4. Never use a guide with >=3 U's in a row, since these sequences act as Pol III terminators. This is obviously not applicable if you are using a Pol II vector instead of the common U6 promoter vectors.*
  5. Prefer guides with a PAM of NGG instead of NAG.*
  6. Avoid sequences with significant secondary structure (The Vienna web server is a great place to check this). You should also avoid guides that disrupt the secondary structure of the 3' constant region (the most common constant region sequence is "GUUUUAGAGCUAGAAAUAGCAAGUUAAAAUAAGGCUAGUCCGUUAUCAACUUGAAAAAGUGGCACCGAGUCGGUGCUUUUUU").
  7. GC content should be between ~ 30-80%, the higher the better (but not too high!).
  8. Avoid additional G's after the PAM. For example, a genomic sequence of "|AGG|CCAT" is probably OK (where "AGG" is the PAM). But "|AGG|GCAT" is probably not a good idea, and "|AGG|GGGG" is a definite no-no.*
  9. Some groups have shown that U's are disfavored in the -1, -2, and -4 position (counting back from the first base of the constant region). Other groups haven't seen this. Your mileage may vary.
  10. Prefer guides in DNAse hypersensitive regions (as annotated by ENCODE on the UCSC genome browser). This isn't a necessity, but probably won't hurt.
  11. It's recently been shown that microhomology at the site of cutting can substantially increase the chance of getting an out-of-frame indel. This doesn't affect cutting itself, but could help you get the knockout.

Construction of the final guide

  1. Take the guide sequence you chose above and append the constant sequence "GUUUUAGAGCUAGAAAUAGCAAGUUAAAAUAAGGCUAGUCCGUUAUCAACUUGAAAAAGUGGCACCGAGUCGGUGCUUUUUU" to the 3' end.
  2. If your guide does not begin with a 5' G, just add one. This increases efficiency of transcription from the U6 promoter and does not need to be homologous to the region you're targeting.*
  3. Add cloning sites appropriate for the expression vector you've chosen. For example, if using the Zhang's lab pX330 vector, append "CACC" to the 5' end of the Watson strand and "AAAC" to the 5' end of the Crick strand.*
  4. Order oligos, anneal, and ligate.

The above might seem like a lot, but it's really not all that bad. You'll quickly get a feel for what makes good vs bad guides. Since it's so easy to test multiple guides, I always recommend making at least two guides per knockout you'd like to make. That way if one is a dud, you aren't caught flat-footed. Obviously, there are many *s in the list above, denoting steps that might be a bit different if your application or parts differ from SpyCas9 making a double-strand break for the purposes of a knockout. The toolbox is always expanding, so options abound! But hopefully the above provides a general idea of how to get started. Do you have another neat trick to share? Did I miss something important? Want to expound on the best way to make a CRISPRi guide (a whole other ball of wax)? Feel free to leave a comment!

X Close



March 12, 2020 0 Comments

Welcome to Lena

Welcome to Lena Kobel, who joins the lab as a Cell Line Engineer. Lena has a long history in genome engineering, with previous experience in Martin Jinek’s...

October 16, 2018 2 Comments

Bootstrapping a lab

Today I’m going to talk about setting up a lab from a 10,000 foot view. I got thinking about this because my social media feed was recently filled with people announcing...

June 12, 2017 1 Comment

Shapers and Mechanists

There’s a series of cyberpunk short stories and a book written in the 1980s by Bruce Sterling called The Schismatrix. It centers around two major offshoots...

June 1, 2017 1 Comment

Backpacking season

It’s important to spend time outside the lab. And before you ask, that’s not why the blog has been dormant. I was teaching this last semester (a general biochemistry...

November 9, 2016 0 Comments

Sequence replacement to cure sickle cell disease

My lab recently published a paper, together with outstanding co-corresponding authors David Martin (CHORI) and Dana Carroll (University of Utah), in which...

September 12, 2016 1 Comment

Improved knockout with Cas9

Cas9 is usually pretty good at gene knockout. Except when it isn’t. Most people who have gotten their feet wet with gene editing have had an experience like that...

August 29, 2016 0 Comments

Safety for CRISPR

This post is all about establishing safety for CRISPR gene editing cures for human disease. Note that I did not say this post is about gene editing off-targets....

July 5, 2016 0 Comments

CAR-Ts and first-in-human CRISPR

(This post has been sitting in my outbox for a bit thanks to some exciting developments in the lab, so excuse any “dated” references that are off...

May 25, 2016 0 Comments

CRISPR Challenges – Imaging

This post is the first in a new, ongoing series: what are big challenges for CRISPR-based technologies, what progress have we made so far, and what might we look...

May 17, 2016 0 Comments

Ideas for better pre-prints

A few weeks ago, Jacob wrote a blog post about his recent experience with posting pre-prints to bioRxiv. His verdict? “…preprints are still an experiment rather...

Contact Us

Questions and/or comments about Corn Lab and its activities may be addressed to: