Lecture #6 - CRISPR Part #2 Flashcards
Goal of Screen or Selection
Overall - associate a gene with a phenotype
Goal of a screen or selection – identify a process of interest + predict the likely phenotype of a mutant unable to carry out that process + to devise a method of identifying mutants with that phenotype
Selection
Overall - A growth condition that allows for the propagation of genetically altered cells
IF the phenotype is based on survival THEN you can take the mutant library and subject it to stress and select for mutants that survive
- Only gives mutants of interest (not looking at every mutant individually)
Example – looking at localization of a protein –> use a cell sorter that gives cells where the protein is localized in a certain place
Screen
Cells are plated on a condition where both mutant and WT can grow and be distiguished phenotypically
- Screens = discovery based + unbiased
Example - KO gene 1 by 1 –> THEN you spatially separate each mutant in a well and look at each well/mutant individually to see if there is a particular phenotype
- Need to screen every mutant
Foward Genetics
Generate Mutants (Usually at random) and screen for interesting phenotype
- Start with a phenotype and want to know the gene
Example – Look for cells that arrest at a particular cell stage
Reverse genetics
RNAi every gene in the genome and look for genes that when knockdown down cause a defect in process (Ex. Cause a effect in cell division)
- Start with a gene and want to know what happens when you change that gene
- Finding phenotype associated with the gene mutations
Issue - there are 100s of potential phenotypes –> How do you know you will find a phenotype with the KO
Example – kinase gene and see the phenotype
CRISPR and reverse Genetics
Can easily do reverse genetics using a CRIPSR screen – Can KO every gene and use a high throughput way of looking at phenotypes
Genome wide CRISPR/cas9 Screens
Overall – Using CRISPR to do genome wide KO Screens
CRISPR/cas Lentivirual Plasmid
Lentiviral plasmid encoding for CRISPR/cas system contain:
1. Sequence needed to generate lentivirus
2. U6 promoter
3. EFS promoter
- Fond upstream of cas9
4. 2 – Epitope-tagged Spcas9 expressed from EF1a
5. Selectably Marker
- Example - Purmycin N-actyle traferase
- Used for insertion of lentivrius into the genome
Right image - The SpCas9 can be stably into cells first and evaluated for high expression BEFORE introducing a guide separately
U6 promoter in lentivirus plasmud
Location – Upstream of cloning site (MCS) to insert the protospacer species
- After U6 have the remaining sgRNA scaffold sequence
- ALSO Downstream of U6 promoter = have second promoter (EFS)
MCS = used to insert a sgRNA sequence that targets the gene of interest
What do you target when KO a gene
Target upstream 5’ exon with sgRNA to cause a frameshift and early stop codon
Why:
1. Because the more upstream the codon is the more likely it is to trigger nonsense mediated decay
2. Because if create a frameshift in an early exon then it is more likley that we will make a non-functional protein product even if the mRNA is translated
How do you validate a KO
Can validate the KO with sequencing or western blot or immunofloresnce mircroscopy
How can we decide what guide will most efficiently cut our upstream exon?
Overall - have online tools that we can use to design our sgRNA
Example tool = CRISPR website
CRSIPR website
CRSIPR website – looks at the target sequence + genome we are using + cas9 ortholog being used to cut DNA
Predicted results from CRISPr website returns a number of parameters:
1. Predicted cut sites in the genome
2. Predicted specificity (Ex. off target effects)
3. Predicted cutting efficinecey
4. Number and location of possible off-taregts
Goal of designing guide
Goal - high affinity cutting with avoiding off-target cuts
IF have off target cuts - Aim for high mismatch + intergenic (intron) off-targets
Clonal lines from CRISPR KO
Since every cell may generate a different edit researchers often derive cell lines from single clones to be used experimentally
- Example of different edits - every cell could generate a different set of indels (some may be in frame)
Methods for generating clonal cell lines :
1. FACs sorting
2. Limiting dilution
What may be present in clonal lines
Clonal lines may carry any passenger mutations –> SO you must characterize of phenotypes
Often several clonal lines with good KO can be evaluated together to reduce the risk thats an unrelated mutation generating the phenotype
- Use several lines with good KO to see if a phenotype that we are seeing is because of the gene we are actually intending to KO
Example CRISPR KO expeirment timeline
Day 1 – Traduce cels with cas9 + sgRNA expressing lentivirus
Day 2 – Apply selection to kill non-tranduced cells
- Control = no transduction
- Select for sgRNA
Day 5-10 – After selection –> evaluate cells to see if they have the KO if interest by western blot or Immunoflorusnece or Sanger seqeucnes
- OPTIONAL – create a clonal line before evaluating the KO
Answer – All of the above
Whole genome CRSIPR KO screens
Overall – a multiplexed version of a single KO experiment
- take a single KO experiment and scale it to be genome wide to discover unknown functions of genes
Screen requires a change in fitness or phenotype as the result of the KO
- The change in phenotype/fitness is required to generate a robust hit list
Includes - Synthetic lethality screen OR growth in a particular genetic background OR growth in the presence of a drug
What is missed in CRIPSR screens
Often miss essential genes due to ubiqitous cell death when those sgRNA are introduced
Two ways to run CRISPR screens in lab
- Arrayed
- Pooled
Pooled Screens
All cells are grown in a single vessel
Treatment must cause change cell fitness due to gene KO to affect sgRNA abudnece
Pooled = less expensive and less time consuming/labor intensive
Pooled = used for selection
- Have a flask or plate of cells with all of the mutants together –> Subject to a selective pressure
Phenotypes used for a pooled screen
Issue – Pooled approaches are limited to growth phenotypes
Limited to cell proliferation or survival OR to cell-autonomous phenotypes that are selectable by sorting using flouresence or cell surface markers
Example of Pooled CRISPR Screen workflow
Process:
1. Generate cas9 expressing cels –> ensures that Cas9 is highly expressed and leads to efficient gene KO
2. Transduce with library of sgRNAs across the whole genome and select to make sure that the sgRNA is expressing in cells
3. Treat cels with drug or negative control for 42 days
4. Compare gene essentiality between samples by counting sgRNA represnetation using NextGen Illumina Sequencing
Array Screen
Overall - Each sgRNA is isolated in a single known cell
- Array = different mutants are in each well and you look at the behavior of cell in each well
Phenotype assessment is used to rank hits (Ex. Floruenscnce detection)
- Can screen for phenotypes and then assign sgRNA to those after
Array = more expensive and time consuming because each reagent is seperatley prepared
- May use special facilities that use automation for the handling of plates
How do we figure out which sgRNA is in our cells
Overall - Sequencing sgRNAs of surviving cells
Process:
1. Do large scale genome extraction
2. PCR from constant region within sgRNA expression locus
- Include barcodes in this step if multiplex sequencing (Barcodes allow us to sequence using NGS)
3. Next Generation Illumina Sequenicng
4. Compare gene essentiality between samples by counting sgRNA representation relative to control
- Create the hit list by comparing essnetiality between samples by counting sgRNA representation relative to the control
Example sgRNA that improves fitness
Example – If a KO improves fitness then that guide will be OVER represented in the treated samples relative to DMSO
IF the KO decreases fitness the guide will be UNDER represented
Improving Robustness of a Hits list
Goal in a whole genome CRISPR KO experiment it is important to know whether a phenotype is due to the traget KO or an off target effect
Methods to improve robustness of hit list:
1. Multiple sgRNAs per gene (4-6) –> IF all of the sgRNAs give a similar phenotype then it is unlikley to be the result of off-target effect (know due to target and NOT due to off-target effect)
2. Secondary screen with a new set of separately transduced cells
3. Deeper sequencing depth
- Used if looking for depletion of sgRNA in a synthetic lethality screen
Other CRIPSR screens
KO screens due to INDELS caused by cas9 cleavage is1 mechnaism BUT you can also pair dCas9 with a transcription activator or repressor to over express or knockdown genes in a genome wide screen
CRISPRi - KRAB repressor protein localized to transcriptional start sites
CRISPRa - VP64 transcriptional activator localized to TSS
Genrate a genome scale human CRISPR/cas9 KO library - What sequences would you target?
Target Exon1/2 of protein coding genes
Generate a genome scale human CRISPR/cas9 KO library - How many total sgRNAs?
20,000 Protein coding genes and need 4-6 per need to validate –> 80,000 total
Generate a genome scale human CRISPR/cas9 KO library - How to generate a library of sgRNAs?
Company makes oligios on array (give you 80,000 pieces of DNA) –> Take oligio library of guides and clone that into a plasmid backbone
- Each gRNA they make have unique spacer BUT the DNA flanking the spacer is the same (ends of each guide are the same) –> use the end constant region to PCR amplify the library and clone them into the same plasmid backbone which will recognize the similar ends
For cloning process you don’t need to know anything about the guide seq to turn the complex oligio library into a complex plasmid library –> once have complex plasmid library you transform them into packaging cells
Packaging cells
Cells that are engineered to be able to produce lentiviral vector
NEED to use a lentiviral vector for genome scale CRIPSR KO (need an integrating virus)
Generate a genome scale human CRISPR/cas9 KO library - What method would you use to identify the mutated gene?
Western blot for protein or Next Gen sequencing
- Because the sequence surrounding the guides is the same = do 1 PCR to amplify all of the guides and submit for deep sequencing to see which guides are there
Once you pool the cells –> generate a hit list -> do deep sequecning
- Need to do deep sequencing before and after selection
Exercise #2 – Positive Selection –> Idetofying a Norovirus receptor - SET UP
Looking for genes that codes for the norovirus receptpr
- Norovirus kills the cells that it infects
- Lentivirus will NOT kill cells (just delivers CRIPSR to cells)
Have – CRISPR/cas9 Mouse KO lentiviral library + mouse Norovirus + Mouse microglial cells that die after infceted with norovirus + plates and media
Exercise #2 – Idetofying a Norovirus receptor - How would you perform the selection?
Process – transduce cells –> Need to select for intiatial cas9 insertion using Puromycin –> THEN add gRNA library –> select for cells that take up the gRNA through blastcodyine –> give cells norovirus –> sequence the survivors
-BEFROE need to sequence the starting populations
End – you get a list of hits and you want to rank the list
- Looking at genes that are not essential for the cell BUT are essential for the viruses
Start with 20,000 genes KO –> after selection with norovirus find the 1 gene that you want (given answer for receptor)
Why do you seqeunce survors in norovirus example
Cells where the norovirus receptor has been KO will survive (Receptor KO = no Recpetor = cells won’t get infected = cells survive)
Screening for survivors –Survivors are no longer killed by norovirus because the norovirus receptor was KO by the guide
Exercise #2 – Idetofying a Norovirus receptor - How would you rank and identify hits?
End – you get a list of hits and you want to rank the list
To rank list – you look at sgRNA enrichment/depletion in surviving cells and compare that to the sgRNA you sequenced at the start
- Look at the ratio of each guide for how abundent it is in the starting library compared to the surviving/selecting library
Combining guides when ranking hit list
For each gene across the 4-6 targets you combine the information to get a per gene numver (get read per million per gene) –> IF you sequences 1 million reads of uninfected and a million reads of infected norovius samples (have 1 million reads in each) –> you want to know how many reads per million were there for gene A and how many reads per million were there for gene B
- Reads per million – tells you relative to all other genes how abundent is gene A
- Looks relative to other genes how abundant a specific gene is
Can do a ratio (before infection and after infection)
Answer – D – done so that you know that what you look at is not random targets
- Hope all gRNA will behave in the same way –> Show that the hit is real
B = how you get the hit list
Lentiviruses
Lentiviruses = LOOK like a lentivirus BUT they have a different genome that tricks them into carrying the CRISPR plasmid
NOW when the lentivirus infects a cell IT does NOT do normal viral reproduction BUT instead your DNA from plasmid is integrated into the target cells and the target cell will NOT make anymore lentiviruses
- IF used WT lentivirus then the target cell would start making lentirviuses
- Virus just brings the CRIPSR cargo and that CRIPSR cargo interagtes and then gets expressed
Enabled by the specilized packaging ecells
Packaging cells (Depth)
Packaging cells – have expression constructs that allow them to produce virus BUT the cells do NOT produce the viral genome
Because do not make viral genome –> the virus that is produced needs a genome to package –> it will will package the plasmid that was introduced
- Can package plasmid because you put sequences on the plasmid that allow it to be recognized and packaged by the vrius
Lentivirus Transduction
Once you have the virus THEN you do Transduction so you have CRISPR cas9 expressiing cells –> THEN do selection with Puromycin –> then do CRIPSR selection or screen
- In selection - Select for cells that got the Cargo = make puromycin resistance gene
GOAL - Want 1 cell to be infect with 1 virus that is carrying 1 gRNA
- Each cell should have a different guide
What groups do you compare in a screen
Need to compare treated to untreated starting time 0 and untreated that have had guides for same amount of time
Compare to untreated with gudies for the same amount of time so that you know the change in growth is due to KO the specific gene (ex. Gene for heat restsice) and not because it is a guide that targets a different gene that also leads to the same growth phenotypes
Amplifying sgRNA
PCR for time 0 samples + your treated + untreated library
gRNA integration into the genome is different in each cell BUT the plasmid containing the gRNA in each cell have a U6 promoter and the scafoled RNA –> MEANS you can do PCR using primers that binds to the U6 promoter and the scafold –> Allows you to amplify the guides that are different in every cell in 1 PCR
- Do 1 PCR for all of the guides in the population because have the same seuqnec (U6 promoter and scafold) flanking the variable sgRNA
Use of the barcodes
Allows you to take 9 experiments in the different tubes and add 9 different adapters that will each have a different barcode (sepweate well wehn making library) –> taggs the amplicon with a barcode that labels it as a specific experiment (EX. Labels as untreated rep 1) –> can take the 9 samples and put in 1 tube to give to core
- By reading the bacrcode you know which expeirment the sequence read came from
- NOW can pool in a flow cell and deconvolute once you have the sequences
What do the end PCR amplicons look like
End – have the same adapter on the guides BUT a different barcode and the actual gRNA will be different
Counting the sgRNAs
Data Analysis for screen
Overall - look for enriched or depleted guides
Image – shows what a hit list would look like –> have some number of reads for a guide in each group and look at the reaction to see if the guide is enriched or depleted
- Top image – Enriched gRNA
- Bottom image – Depleted gRNA
Example – Screen hits
Image shows genes that have hits (lines = 4 guids for gene)
In ALL images – at time 0 the guides are found as frequently as any other gene (enrichement of any guide is low in starting library because all of the genes are there)
Top left image – shows a good hit –> all 4 guides are very abundent
- After selectove pressure all of the guides are enriched
Top right – Intermediate hit –> 1 guide worked well (is very enriched) ; 1 guide is kind of enriched ; 2 guides are low and not enriched
- Hit may be driven by 1 guide
Bottom left – 3 out of 4 of the guides work = good hit (better than top right)
Bottom right = non-hit because only 1 guide works well
Example MOIs
MOI of 10 –> most cells get 10 viruses BUT some will get 9 viruses or 8 or 12 etc.
- Shows why you want 100 cells with the same gRNA
MOI of 1 –> 33% of the cells are unedited + 33% of cells get 1 virus + 33% of cells get multiple guides
- Means 1/3 of the library has 2+ guides
END - WANT MOI of 0.1 (1 virus for every 10 cells)
MOI of 0.1
MOI of 0.1:
- Most cells are uninfected BUT that is ok because we will select for infected cells using puromycin
- 10% of the cells get a single virus
- almost no cells get 2+ guides
Having a MOI of 0.1 = Might get many uninfected clones BUT make this sacrifice because you don’t want to get 2+ viruses per cell
- DO still get some 2+ = why you need to verify hits after
END - How many cells do you use
MOI of 0.1 = tranduce 80 X 10^6 with 8 X 10^6 viral particles and select for transduacble cells
If want 1 virus every 10 cells = transduce 80 million cells to get 8 million transduction events (8 million virus with 80 million cells to get MOI of 0.1)
Answer – C
Viruses integrate randomly into the genome – could integrate into regions that are repressed = get variability = need to do multiple times
Answer – C
- Compare to the starting population and to transduced cells grown for the same amount of time
Answer A could be a good control to have
KRAS mutations
KRAS mutations drive cancer initiation and progression
- KRAS mutations = one of the most undruggable targets in cancer research
Question we ask – What is a cell with an active KRAS mutation vulnerable to that the WT is not vulnerable to
- Are there genes that when they are KO then a cell that also has a KRAS mutation dies (looking for a synthetic lethal intreaction with KRAS mutations)
Answer – D
Activity #5 – Deisgn a sgRNA librrary GOF screens - Question #1 – How would use CRISPR/cas9 to increase gene expression
Overall - Recruit a transcriptional activator
First - Need to KO the nuclease activity of Cas9 = have dCas9
Attatch dcas9 to protein domains that can activate or repress chromatin in that area and theerfore activate or repress transcrtion in that area OR add domains that recruit RNA polymerase
- Attatch things to cas9 that activate or repress Gende expression
- Can have multiple epitopes on cas9 = can add more copies of activating domain = get more expression
dcas9 and transcriptional modulation with cas9
