CRISPRi-mediated transcriptional silencing in iPSCs for the study of human brain development

Summary This protocol describes the design and use of CRISPRi-mediated transcriptional silencing in human iPSCs, for loss-of-function studies in brain development research. The protocol avoids single cell selection, thereby eliminating side effects of clonal expansion and sites of viral integration. We also describe a neural progenitor differentiation protocol and discuss the challenges of target-specific lentiviral silencing, efficient silencing levels, and off-target effects. For complete details on the use and execution of this protocol, please refer to Johansson et al. (2022).

1. Order and make maxiprep of CRISPRi cloning backbone from Addgene (Addgene_71237). 2. Design 3 guide RNA (gRNA) sequences for the gene of interest according to the CRISPick portal (https://portals.broadinstitute.org/gppx/crispick/public). The software automatically selects guide RNAs that target a region +25-+75 bp downstream the transcription start site (TSS) of the target gene ( Figure 1 and troubleshooting 1). See also ''CRISPick: How It Works''. a. Choose a reference genome (Human GRCh38), mechanism (CRISPRi), and Enzyme (SpyoCas9). b. Type in the gene of interest. c. Tick the "report unpicked sequences" box. d. Wait for the software to generate the reports. e. Download the ''Picking Results'' .txt file. f. Open with Excel, scroll to Column AR and find the software's preferred picks. g. Choose 3 guides for each gene of interest. h. The gRNA sequence is in column S.
CRITICAL: In order to minimize the effects of guide-specific off-targets, at least two working guide RNAs are needed (see more below).
Note: A minimum of 3 guides are always designed for each gene. The success rate of the gRNA varies between different genes. While many genes are efficiently silenced with most tested guides, some genes are more challenging to target. In these instances, many guides may have to be tested in order to achieve sufficient silencing.
Note: Consider what level of silencing would be sufficient for the experiment, it is not always critical nor desired to achieve 99% knock-down efficiency.
Optional: Investigate the potential off-targets of the designed gRNAs. See troubleshooting 5.
Optional: Investigate the position of the guides and the predicted TSS in the USCS genome Browser https://genome.ucsc.edu/cgi-bin/hgGateway. This is shown below using ZNF248 (simplified schematic, Figure 2) as an example.
Select the human GRCh38 assembly (Figure 2, red circle) and type in the name of your gene of interest for a first view (Figure 2, blue circle). Then, type in the predicted TSS from column K in the CRISPick design sheet using this format: chr10:37857610 (green circle). This will show one nucleotide, highlight it and zoom out to see its location in the gene. In the ZNF248 example, the predicted TSS (dark green line) falls at the start of two main isoforms.
Once you highlighted the nucleotide and zoomed out, type in the first gRNA sequence. The browser will show the number of places it finds this sequence (which should only be one). Click on ''browser''(Figure 2, pink circle) and get to the sequence that can now be highlighted. Do the same with the other gRNA target sequences.
In the ZNF248 example, the selected guides are found downstream of the TSS (ZNF248 runs in antisense direction) and can be seen to have some sequence overlap (this is often unavoidable as the target window is quite small) ( Figure 2). However, there is one isoform (Figure 2, red arrow) which might have an alternative TSS. This needs to be kept in mind if one does not know which isotype is expressed. However, in the given example, the RNA-sequencing tracks from the hiPSC-derived forebrain neural progenitor cells (fbNPCs, our cells of interest, Johansson et al., 2022) show no evidence of transcription of the isoform-specific exon (red rectangle), suggesting that the designed CRISPRi guides should be sufficient for silencing all the expressed isoforms. If there is evidence of a different TSS for some isoforms and it is not possible to tell which isoforms are expressed in the cells of interest, then first use the TSS defined by the software. If the transcriptional silencing is not sufficient, then consider manually designing guides for the alternative TSS and combine the two guides (see troubleshooting 4).
3. Order the oligos with specific overhangs for BsmBI cloning. Insert the 20 bp target gRNA sequence between the overhangs. 4. Insert the guides into the CRISPRi lentiviral backbone ( Figure 3) containing both the guide RNA under the U6 promoter and dead-Cas9-KRAB and GFP under the Ubiquitin C promoter (pLV hU6-sgRNA hUbCdCas9-KRAB-T2a-GFP, RRID:Addgene_71237; Thakore et al., 2015). See here for PDF cloning protocol, part A (automatic download). 5. Lentiviral vectors were produced according to (Zufferey et al., 1997) and were in titers of 10 8 -10 9 TU/mL as determined by qPCR of genomic DNA after serial dilution transductions. We recommend using an MOI between 5 and 15.
CRITICAL: To obtain high transduction and high silencing efficiency, it is crucial to obtain a high titer. Software and algorithms n/a n/a n/a n/a n/a n/a ll OPEN ACCESS

MATERIALS AND EQUIPMENT
None of the following media has been filtered when prepared.

Figure 3. Workflow for lentiviral production
After designing of the desired guide RNAs, the resulting oligonucleotides (yellow) are annealed and cloned into the backbone of the CRISPRi plasmid (green), which is then used, along with standard 3 rd generation lentiviral plasmids (blue), to produce the gene specific CRISPRi lentiviral vector.

Component Storage Dilution
iPS-Brew XF, Basal Medium À20 C 1:1 iPS-Brew XF, 503 Supplement À20 C 1:50 Penicillin/streptomycin (stock: 10,000 U/mL) À20 C 0.5% *This is the only iPSC media that we used, although we can only speculate, we do not expect the use of other media to influence CRISPRi efficiency.

STEP-BY-STEP METHOD DETAILS
Significance of not using clonal expansion When using lentiviral delivery, the lentiviral units will integrate into the genome at different sites, some of which can lead to a phenotypic alteration in the cell. Additionally, clonal expansion of isogenic clones can lead to alterations in the transcriptome (Carcamo-Orive et al., 2017). If one uses cells of mixed clonal origins, the effect of viral integration and clonal expansion will be reduced and the signal to noise ratio of transgene expression will be increased.

Establishing the CRISPRi cells
Timing: 2 weeks In this step, human iPSCs are transduced with the previously designed CRISPRi lentiviral vector. This results in the creation of cells with a constitutive downregulation of the target gene.
1. Creation of CRISPRi cells ( Figure 4). a. Culture hiPSCs on Laminin521 (Lam521)-coated wells and feed everyday with IPSBrew. Use 900 mL PBS +/+ and 25 mL of Lam521 (0.5 mg/cm 2 ) and incubate in the fridge for 12-16 h (or at 37 C for 2 h). Split every 3-4 days with Accutase (incubation for 10 min at 37 C), seeding 100,000-200,000 cells per well in a 12-wells plate with 900 mL of iPSBrew. (See Grassi et al., 2019;Johansson et al., 2022). b. On the day of transduction, split the cells with Accutase (incubation for 10 min at 37 C). Count the cells. c. Seed 200,000 cells per well in a 12-wells plate. Use one well per condition. Use double amount of media (1.8 mL) with 10 mM of Rock inhibitor (RI). d. Plate some untransduced cells as negative control. e. Directly after seeding the cells, dilute each virus 1:10 in iPSBrew and add the equivalent of MOI10 to each well (the first time, use MOI5 and 15 as well to select the best MOI for the virus and cell line combination). Add one virus per well. f. Change the media every day and add 10 mM Rock Inhibitor until the day after the first split ($70% confluency). The first split ideally takes place on day 3 post-transduction. g. On the 3 rd day post-transduction, split cells with Accutase (incubation for 10 min at 37 C). They should now behave quite normally. h. On day 7-10 split again with Accutase (incubation for 10 min at 37 C), collect cells, and take samples for RNA and cryopreservation.

OPEN ACCESS
k. Select the two guides that produces the most efficient silencing for further studies. Ideally, the silencing efficiency is R 90%.
Optional: FACS GFP+ cells prior to investigating silencing efficiency and freezing. This is preferable if the transduction efficiency is low. If feasible, FACS at the endpoint of the experiment is still recommended in case of low levels of silencing or selection during differentiation. If low downregulation levels are reached with all of the designed guides, see also troubleshooting 4.
Note: Selection via FACS is needed at either the start or end of experiment. Selection at the start gives a better indication of actual silencing efficiency and it is often better for long-term experiments to start with a purer population. Selection at the end is not always possible due to complex differentiation protocols (for instance, with cerebral organoids the association into single cells needed for selection via FACS destroys some cell populations). In case of fbNPC (described here), we sorted only at the end of the experiment as these cells are well suited for single cell dissociation. For cerebral organoid differentiated described in Johansson et al., (2022), we sorted the cells prior to start of differentiation and then used singlenuclei dissociation for analysis. If selection via FACS is not possible, puromycin selection can be an alternative method to establish the CRISPRi cells (see troubleshooting 2).
Note: We recommend freezing serval vials of cells early in procedure to be able to use similar passage numbers for repeat or follow-up experiments as well as to avoid a decrease in efficiency (due to silencing and/or uneven growth patterns of transduced and un-transduced cells) with increased passage number and repeated freeze-thaw cycles. In our hands we have not seen much effect of these processes, but this can also depend on the gene of interest.
Note: We have not tested other culture media or coating matrices but anticipate that other media or matrices will not significantly affect the procedure.

Differentiate hiPSCs into fbNPCs
Timing: 3 weeks Here, the steps to differentiate the hiPSCs into forebrain NPCs (fbNPCs) are described ( Figure 5, Grassi et al., 2019 andJohansson et al., 2022). The CRISPRi cells have also been differentiated into cerebral organoids (see Johansson et al., 2022) with efficient silencing shown to persist after four months, see also expected outcomes.
2. Thaw the two CRISPRi cell batches with the most efficient transcriptional silencing (e.g., guide1 and guide2), the LacZ CRISPRi control (experimental control), and include one untransduced control (FACS control). 3. Split the cells once before the start of differentiation. 4. Coat one well in a 6-well plate (2.5 mL PBS + 150 mL Lam111) per condition. Leave 12-16 h at 4 C. 5. Day 0.
i. Prepare FACS media and collection media (B27 media + Rock Inhibitor (10 mM).
ii. Collect the cells using Accutase (10 min incubation at 37 C) to ensure single cell suspension. iii. Resuspend the cells in FACS media and filter prior to FACS. iv. Prepare collection tubes with collection media (B27 media + Rock Inhibitor (10 mM). CRITICAL: Bring untransduced cells to the FACS to set the gate for GFP and Draq7 (cell death marker).
Note: As the cell density increases the protocol calls for increased volume of media.
Note: One sample is needed for RNA isolation (RNeasy Mini kit (50) (QIAGEN, cat. # 74104)) to re-assess the transcriptional silencing efficiency at the end of the experiment. Collect additional samples depending on the scientific question and assays. We typically collect 3 3 200,000 cells per condition.
Optional: If the gene of interest encodes for a protein the silencing can also be validated using WB.
Silencing can also be validated using RNA-sequencing, via histone methylation marks-mediated CU-T&RUN or ChIP, where reduction of H3K4me3 and/or enrichment of H3K9me3 markers confirm the silencing.

Timing: 4 days
This part of the protocol involves combining or verifying the effect of the CRISPRi using two different guides. This varies depending on the read-out. RNA-sequencing is used here as readout, and the combination of guides will be described below (Figure 7).
Most guides will have off-targets. However, two guides are very unlikely to have the same off-targets. Thus, combining the readouts of two guides will minimize the off-target effect of each guide. Here, we combine the two CRISPRi conditions into one, and analysis is performed with control vs CRISPRi cells (in our case n=4 and 8, respectively).

EXPECTED OUTCOMES
Using this protocol, we generally get very high transduction efficiency (60-95%) and very high silencing efficiency ( Figure 8A). Importantly for this protocol and the use of CRISPRi for long-term transcriptional silencing, the construct we have chosen is stably expressed, and after 2 weeks of differentiation we detected no reduction in the percentage of positive cells ( Figure 8B). In cerebral organoids, we have stable transcriptional silencing even in 4-month-old organoids (Johansson et al., 2022).
Note: A change in the percentage of positive cells may occur with increased time and/or differentiation due to phenotypic alterations (e.g., survival or proliferation) rather than transgene silencing.

TROUBLESHOOTING Problem 1
There is another gene close to my gene of interest running in the opposite direction (See design, cloning and production of CRISPRi lentiviruses).

Potential solution
This might not be a problem, CRISPRi works best when the Cas9 is guided to the DNA downstream of the transcription start site. However, there can still be interference. The solution is then to design a few more guides targeting around +75-+90 bp downstream of the transcription start site.
The CRISPRick software first chooses guides within the +25-+75 bp downstream of the transcription start site. However, one can manually pick guides outside of this range. Choose a guide that has a high pick order and comments are only ''outside of range''. Design qRT-PCR guides for the nearby gene as well and test the effect at the same time as the gene of interest.

Problem 2
We cannot use GFP selection (See establishing the CRISPRi cells).

Potential solutions
Use the sister construct Addgene Plasmid #71236), that has selection via puromycin. We have not used this plasmid, but we anticipate it is equally efficient. We have however used puromycin in these cultures for selection of other vectors with a final concentration of 0.5 mg/mL. However, we have found the NPCs to be sensitive to puromycin (i.e., even with a puromycin resistance gene they thrive less) and we therefore recommend the user to do a robust selection during the iPSC stage and carefully optimize the puromycin selection during the differentiation procedure.
Modify the plasmid to express RFP instead of GFP (also available from the lead contact).

Problem 3
There are not enough green cells after transduction (See establishing the CRISPRi cells).

Potential solutions
Use a range of MOIs to test for survival, transduction, and transcriptional silencing efficiency (Figure 9). If the titre is low, try to modify production to achieve a higher titre. Expand the cells more before the sort and take cells from 2 or 3 wells to the FACS. Ensure the cells are not over-confluent when collecting for transduction. Use cells that have been split only once or twice after thaw.

Problem 4
Not sufficient transcriptional silencing with either of the designed guides (See Establishing the CRISPRi cells).
We have found that the efficiency of CRISPRi is not only guide and position dependent, but also locus dependent. For example, in the case of ZNF558, CUT&RUN showed that the H3K4Me3 activation mark over the ZNF558 promoter was absent after ZNF558 CRISPRi (Johansson et al., 2022), indicating complete silencing. In other experiments, using the same methodology but targeting different genes, all the designed guides were much less efficient in reducing mRNA levels. However, even though complete silencing was not achieved, a very significant reduction in mRNA levels was found (>85%), and this can be sufficient for many experiments.
In addition, where genes that are crucial for cell survival are involved, cells do not survive with complete silencing, but this method might give cells with the highest level of transcriptional silencing they can tolerate. This allows for functional studies not possible using CRISPR-KO.

Potential solutions
In the first instance, using a higher MOI is recommended. There is a definite correlation between CRISPRi efficiency and MOI (see Figure 9). Design new guides that may achieve more efficient silencing. Investigate if there is a possibility of an alternative TSS. If so, one can design guides manually, using the CRISPR tracks (selecting green guides) around where the alternative site might be.

Problem 5
The guides are not regulating the same genes (See design, cloning and production of CRISPRi lentiviruses).
This would suggest that the guides are not specific and bind too many off-targets. In this work, the analysis of the guides was combined as described in step 4, in order to filter out genes that are regulated by only one of the guides and therefore not true targets (see also Johansson et al., 2022).

Potential solutions
Combine the analysis of both guides (control vs CRISPRi). Include a third guide in the analysis. Design new guides. Use in combination with another approach such as siRNA.
Look for the expected off-targets for each guide ( Figure 10).
• Go to UCSC Genome Browser.
• Paste in the guide sequence of choice.
• In the BLAST search window, confirm that the guide sequence only appears once in genome and press the browser link (see Figure 2). • Highlight as described above and zoom out 1.53.
• Click on the guide corresponding to the sequence (red arrow). The PAM sequence is seen as a thinner line in the browser track. • All information about that guide, including predicted off-targets, can be found here.
Of the top 10 hits we recommend no overlap with potential off-targets. However, these are only potential off-targets and might not be affected and in addition other targets might be. We recommend handling off-targets with the use of combinatorial analysis, meaning only look at phenotypes seen with two guides.

RESOURCE AVAILABILITY
Lead contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Dr. Pia Johansson (pia.johansson@med.lu.se).

Materials availability
Plasmids generated in this study are available upon request.

Data and code availability
This study did not generate/analyze datasets or code.