Droplet-based functional CRISPR screening of cell–cell interactions by SPEAC-seq

Abstract

Cell–cell interactions are essential for the function and contextual regulation of biological tissues. We present a platform for high-throughput microfluidics-supported genetic screening of functional regulators of cell–cell interactions. Systematic perturbation of encapsulated associated cells followed by sequencing (SPEAC-seq) combines genome-wide CRISPR libraries, cell coculture in droplets and microfluidic droplet sorting based on functional read-outs determined by fluorescent reporter circuits to enable the unbiased discovery of interaction regulators. This technique overcomes limitations of traditional methods for characterization of cell–cell communication, which require a priori knowledge of cellular interactions, are highly engineered and lack functional read-outs. As an example of this technique, we describe the investigation of neuroinflammatory intercellular communication between microglia and astrocytes, using genome-wide CRISPR–Cas9 inactivation libraries and fluorescent reporters of NF-κB activation. This approach enabled the discovery of thousands of microglial regulators of astrocyte NF-κB activation important for the control of central nervous system inflammation. Importantly, SPEAC-seq can be adapted to different cell types, screening modalities, cell functions and physiological contexts, only limited by the ability to fluorescently report cell functions and by droplet cultivation conditions. Performing genome-wide screening takes less than 2 weeks and requires microfluidics capabilities. Thus, SPEAC-seq enables the large-scale investigation of cell–cell interactions.

Introduction

Cell–cell interactions have a central role in physiology and pathology¹. The immune system, for instance, depends on intercellular communication to mount and regulate immune responses against targets such as pathogens and tumors. Therefore, defining basic physiological mechanisms and candidate targets for therapeutic intervention requires tools that enable the comprehensive and unbiased investigation of cell–cell communication.

Mechanisms of cell–cell communication are usually studied via transcriptomic or proteomic approaches to predict interacting molecules based on cell expression programs^2–4. However, clustered regularly interspaced short palindromic repeat (CRISPR)-based screening approaches enable genome wide genetic perturbation studies to define the molecular mechanisms regulating biological processes of interest⁵. These techniques have advanced the identification of cell-intrinsic regulatory mechanisms but have not yet been extensively used for the study of cell–cell interactions, their biological consequences and the mechanisms involved. To address this challenge, we developed a tool for the systematic perturbation of encapsulated associated cells followed by sequencing (SPEAC-seq)⁶. SPEAC-seq is a platform that uses genome-wide CRISPR libraries, cell coculture in picoliter droplets and microfluidic sorting to enable forward genetic screens of regulators of cell–cell interactions. Here, we provide a detailed experimental workflow and discuss technical aspects of SPEAC-seq.

Comparison with other methods

Several methods have been developed to study regulatory cell–cell interactions. One approach is to combine expression datasets with known ligand–receptor pairs to infer communication between cells^4,7–11. Recently developed spatial technologies that combine microscopy and molecular profiling enable the detection of colocalization and can be combined with bioinformatic prediction of ligand–receptor interactions¹². Although bioinformatic approaches are useful for hypothesis generation, they are limited by the lack of methods to experimentally test predicted cell–cell interactions at scale.

Various methods have been developed to molecularly record cell–cell interactions, for instance, via proximity-dependent enzymatic^13–19 or photocatalytic labeling reactions^20–22. Labeling of nearby cells is also possible via fluorophore transfer^23,24, barcoded viral transfer²⁵ or synthetic receptor circuits^26–28. However, these methods often rely on a priori knowledge of interactions of interest to genetically engineer one or both interaction partners, while also generally lacking the ability to survey functional interaction outcomes. Moreover, the analysis of physically attached cells by flow cytometry²⁹, although versatile for use with ex vivo tissues and free of engineered genetic manipulations, lacks sensitivity for weak membrane interactions or paracrine signaling via secreted factors. Finally, retroviruses and yeast cells can be engineered for pooled screening of large molecule libraries to test well-defined protein–protein interactions, but they disregard the biological context in which those interactions occur^30–32.

CRISPR-based methods enable the genome-wide interrogation of the mechanisms that regulate cellular processes of interest. However, these methods are usually applied to the study of cell-intrinsic regulatory mechanisms; their ability to study cell–cell communication is limited. Hence, there is a need for a platform that enables the use of CRISPR-based approaches to study cell–cell communication through contact-dependent and contact-independent mechanisms.

Droplet microfluidic approaches are widely used in single-cell genomics protocols³³, but can also be applied to study cell–cell interactions³⁴. Thus, we developed a microfluidic-based platform that combines CRISPR-based genetic perturbations with cell coculture in droplets to study mechanisms of cell–cell communication.

Overview of the procedure

In brief, SPEAC-seq utilizes two microfluidic devices for co-encapsulation of cells and for droplet sorting; these are fabricated using soft lithography (Steps 1–14). Lentiviruses encoding genome-wide CRISPR screening libraries are used to transduce cells of interest and disrupt individual genes (Steps 15–28). Cells with individually disrupted genes and fluorescent reporter cells are then co-encapsulated in picoliter droplets via microfluidic pairing, followed by coculture in droplets (Steps 29–38). The droplets containing cocultured cells then undergo fluorescence-activated droplet sorting to enrich droplets inside of which genetic perturbations led to reporter activation (Steps 39–46). Guide RNAs are recovered from droplets enriched for reporter activation, amplified by PCR, and sequenced to compare against droplets without reporter activation (Steps 47–57) (Fig. 1). SPEAC-seq can be completed in less than 2 weeks.

Fig. 1 |. Overview of the SPEAC-seq workflow.

Microfluidic PDMS devices are prepared following photolithography techniques using SU-8 negative photoresist³⁶ (Steps 1–5). PDMS devices are bonded to glass slides using oxygen plasma and the resulting channels are rendered hydrophobic with Aquapel treatment (Steps 6–14). The genome-wide CRISPR knockout screening library is amplified (Steps 15–19), packaged into lentiviral particles (Steps 20–26) and cell populations to be perturbed are stably transduced (Steps 27–28). Reporter cells are prepared as needed (Step 29). In our application of SPEAC-seq, primary microglia were transduced with genome-wide CRISPR knockout libraries, and primary astrocytes from p65^EGFP NF-κB reporter mice were prestimulated at subthreshold cytokine concentrations to prime the reporter circuit⁶. Co-encapsulation of a CRISPR screening cell and a reporter cell in single droplets is performed with a microfluidic device (Steps 29–36). Droplet emulsion containing co-encapsulated cells is collected and cells interact with each other over 24 h (Steps 37–38). Cocultured droplets are reinjected into a second microfluidic device operated in combination with lasers, fluorescence detectors, a field-programmable gate array (FPGA) and a high-voltage amplifier for fluorescence detection and real-time droplet sorting (Steps 39–45). Reporter fluorescence resulting from NF-κB activation (act.) is detected within microfluidic system and droplets are positively sorted via dielectrophoresis (Step 45). Positive and negative droplets are collected, and emulsions are chemically broken with perfluorooctanol (PFO) for cell recovery (Steps 46–48). gRNA sequences are amplified by PCR and libraries are prepared for sequencing (Steps 49–57).

We describe the experimental steps required to perform SPEAC-seq, starting from the fabrication of polydimethylsiloxane (PDMS) devices. Techniques required for this workflow but described in other protocols include photolithography³⁵, the setup of a microfluidic droplet sorter³⁶ and CRISPR library handling³⁷.

Development of the protocol

The SPEAC-seq protocol was developed by combining functional genomics and microfluidics to provide a platform for the systematic interrogation of cell–cell interactions. Genome-wide CRISPR libraries have been established as a useful tool for probing the effects of individual genes in an unbiased manner by combining perturbations with functional readouts to establish genotype–phenotype links. Meanwhile, microfluidic technologies have been increasingly applied to study cells at the individual level, for instance, as part of single-cell RNA-sequencing (RNA-seq) protocols where individual cells are co-encapsulated with individual beads to capture RNA³⁸. However, a technique for functional genome-wide CRISPR screening in conjunction with in-droplet coculture was lacking.

Our initial study focused on identifying microglial regulators of astrocyte inflammatory responses using an NF-κB fluorescent reporter⁶. First, we established the stable transfection of a genome-wide CRISPR screening library into microglia, each harboring a Cas9-driven genetic perturbation. The gene-targeting sequences (guide RNA (gRNA)) incorporated into these cells later serve as a read-out for enrichment analyses, as is common in other CRISPR screening methodologies. The scalability of this technology is highlighted by the fact that unbiased genome-wide functional screens are possible with three to four gRNA coverage for each gene; we used libraries of ~80,000 gRNAs. Second, we established the use of NF-κB fluorescent reporter astrocytes as a way to fluorescently report their inflammatory activation, identifying an optimal dose of cytokine stimulation to prime the reporter circuit without inducing fluorescence. Notably, the SPEAC-seq workflow is easily adaptable to study various cellular functions, such as receptor activation, transcription factor activation or gene expression, by exchanging the fluorescent reporter assay, strengthening its potential applications across fields of study. Third, we developed the use of two microfluidic devices in succession to enable cocultivation and sample enrichment. The first microfluidic device enables the co-encapsulation of one cell bearing a CRISPR-driven genetic perturbation with one cell bearing a reporter in a media-in-oil droplet, where cell–cell interactions can take place and lead to activation of the fluorescent reporter. To establish this step, we compared different timepoints and medium compositions to optimize viability and droplet stability, both critical parameters for the effective sorting of droplets based on reporter fluorescence. The second microfluidic device is operated in combination with a fluorescence-activated droplet sorter to enrich for droplets displaying fluorescent reporter activity, enriched for regulators of the circuit of interest. The staining of each set of cells with a different fluorescent dye before co-encapsulation ensures that only droplets with one-to-one pairing are included in the analysis. Finally, sorted droplets are broken to lyse cells, isolate DNA and amplify gRNA sequences by PCR. Next-generation sequencing (NGS) is used to quantify gRNA sequences enriched in each sorted fraction, providing a pooled read-out for the genome-wide CRISPR screening of interaction regulators. SPEAC-seq therefore provides a high-throughput platform combining functional genomic screening with microfluidic cell–cell interactions to study regulators of defined cell responses.

Advantages and applications

SPEAC-seq possesses numerous advantages over other methods described above. The microfluidic generation of a confined coculture space with controlled loading ensures that each droplet analyzed models a one-on-one interaction, with no confounding effects from differences in cell abundance or presence of confounding cell types, as would be the case in in vitro or in vivo functional genomic screens. Therefore, SPEAC-seq provides a well-controlled environment in which defined interaction partner pairing minimizes technical assay heterogeneity.

SPEAC-seq finds application in the study of biological systems where cell–cell interactions control tissue homeostasis or pathology. For example, SPEAC-seq has been successfully implemented for the unbiased discovery of microglia-derived regulators of astrocyte NF-κB activation, providing mechanistical insight into neuroinflammatory functions performed by astrocytes during neuroinflammation⁶. Further potential applications of SPEAC-seq include the unbiased discovery of regulators of T cell activation, antigen presentation or cellular cytotoxicity. Of note, the adaptation of SPEAC-seq to study cell types other than those discussed here is possible with few modifications: namely, the optimization of transfection conditions, coculturing conditions and fluorescent reporter operation. The adaptation potential of SPEAC-seq is amplified by the availability of several fluorescent reporter assays and functional genomics platforms building on CRISPR or other functional genomics technologies⁵. Thus, SPEAC-seq could be used to study the regulation of multiple biological processes, such as epigenetic regulation, promoter activation, gene expression, RNA splicing, protein-level activation events, receptor signaling and metabolic function, among others. Not least, the high scalability and throughput of SPEAC-seq allows for the study of thousands of perturbations per experimental run. We expect this platform to be a valuable tool to study regulators of cell–cell communication in multiple biological contexts.

Limitations

SPEAC-seq has important limitations as an in vitro technique, primarily owing to its limited recapitulation of physiological cell–cell interactions. Although primary tissues derived from genetically engineered mouse strains may be used as reporter cells, the application of CRISPR screening, however, requires the introduction of endonucleases and targeting oligonucleotides before encapsulation and interaction screening. Techniques for rapid perturbation of human-derived cells via transient electroporation of ribonucleoprotein complexes may expand the usage of SPEAC-seq beyond cells in culture³⁹. As SPEAC-seq studies interactions that result in the activation of fluorescent reporters, it is limited by the availability of such reporters and their regulation by cell–cell interactions in droplets. In addition, the culture of individual cells in picoliter droplets is linked to limited nutrient availability, suboptimal gas exchange and a lack of growth factor support or cell adhesion/anchoring⁴⁰. These factors limit the length of coculture of interaction partners and may alter tissue-specific or context-specific functions that exist in vivo. Overall, we highlight the value of SPEAC-seq as a discovery platform to guide functional validation studies of screening hits. Although scale-up of SPEAC-seq for the genome-wide screening of functional interaction regulators is possible, the number of cells needed as input is large and may be prohibitive for certain cell types. We expect that improvements in microfluidic encapsulation techniques will improve the capture rate and reduce cell loss⁴¹. Moreover, the technical expertise needed to implement this protocol spans molecular biology and engineering of costly optical and electronic systems, potentially limiting widespread use. Commercial droplet sorters are becoming available, and additionally, double encapsulation of droplets may be sufficient to adapt this protocol for sorting with traditional flow cytometers⁴².

Experimental design

CRISPR screening (Steps 15–28)

SPEAC-seq makes use of genome-wide pooled CRISPR construct libraries to investigate the effects of individual genes on a defined cell–cell interaction, thus screening for functional regulators. In developing SPEAC-seq, we used the well-established lentiCRISPRv2 Brie library for genome-wide gene knockout via Cas9 (ref. 43). At present, various CRISPR-enabled modalities have been developed, including transcriptional activation, translational repression, epigenetic modification, base editing and knock-in⁵. Therefore, SPEAC-seq is amenable for adaptation to a variety of applications, cellular interactions, biological contexts and molecular functions. Importantly, experimental scale-up is largely a function of the screening library size and the transduction efficiency of target cells (Step 27), which must be optimized for each application to ensure sufficient gRNA representation³⁷. In addition, important quality steps should be considered if establishing a new CRISPR library, as recombination events during library amplification may affect the equal representation of all constructs, and differences in gRNA targeting efficiency may bias enrichment results obtained with SPEAC-seq⁵ (Step 15).

Reporter assay selection (Step 29)

SPEAC-seq uses a fluorescent reporter assay associated with a specific biological outcome to trigger droplet sorting. Thus, the system depends on detectable fluorescence to enrich for cells carrying specific gRNA perturbations that elicit phenotypes of interest (Step 45). The flexibility afforded by this design is advantageous, as several fluorescent reporter assays exist for various functional read-outs, and molecular methods allow for the customized generation of new genetic reporter circuits, molecular fusions and molecule-specific tags. Our previous application of SPEAC-seq investigated activation of the transcription factor NF-κB following cell–cell interaction⁶. However, it is conceivable to apply SPEAC-seq in the study of any signal transduction cascade as a response to a cell–cell interaction, with the condition that a fluorescent signal can be engineered as a measure of that specific process.

Controls (Steps 15–46)

It is critical to optimize the fluorescent reporter assay for its use in droplet cocultures with each specific cell type pair. Including negative and positive controls of known effects on the reporter system is essential to titrate stimulation (Step 29), coculture (Step 38) and sorting conditions (Step 45). The use of negative controls, such as by encapsulation of reporter cells with no CRISPR cell, is needed to quantify baseline reporter activity and establish the fluorescence threshold for reporter activation detectable with the microfluidic system. Furthermore, positive controls known to activate the reporter (for example, TNF and IL-1β for NF-κB activation) are also needed to optimize reporter detection within the microfluidic setting. In some scenarios, it might be informative to prestimulate reporter cells to prime the reporter system or to simulate a certain physiological or pathological condition in vitro. It is also important to consider the signal-to-noise ratio when designing the assay. For this reason, we performed activation of the CRISPR cell type with lipopolysaccharide (LPS) to stimulate communication between microglia and astrocytes⁶. Thus, one additional control consists of the non-activated CRISPR cell type encapsulated with a positively activated reporter cell type. Sequencing of gRNAs enriched in this condition can help to reduce noise. Finally, the correct detection of droplet loading and gating (Step 45) should be validated by inclusion of appropriate controls, including each cell type encapsulated on its own with or without its corresponding cell dye.

Cell handling, microfluidic encapsulation and coculture (Steps 15–38)

The SPEAC-seq workflow involves cocultivation in droplets as a defined reaction space to functionally study cell–cell interactions. Therefore, ensuring cells are viable and healthy is central to the undistorted modeling of cell–cell interactions. Cell lysis before encapsulation may also affect neighboring cells in suspension. For this protocol, we prepare two cell types separately. One fraction undergoes spinfection with a CRISPR screening library followed by antibiotic selection³⁷ (Steps 15–28). Here, we note that a low multiplicity of infection of typically 0.3–0.5 virus particles per target cell ensures that few cells receive multiple gRNAs. The other cell type encodes the fluorescent reporter system and is prepared as needed (Step 29). In our application, we prestimulate cells with low-dose cytokines. For both fractions, viability should be measured before encapsulation. In particular, antibiotic selection should be performed with the lowest possible lethal dose and thorough washing is needed to remove selection agents before encapsulation with interaction partner cells.

Our co-encapsulation device works with a simple flow focusing junction, which allows droplet generation at high speed (Fig. 2a) (Step 36). With this setup, it is possible to encapsulate 835,000 cells per hour (assuming a concentration of 1.4 × 10⁶ cells/mL), minimizing the time that cells spend outside of culture conditions. Use of a nontoxic and inert density gradient medium, such as iodixanol (OptiPrep) is essential when producing cell suspensions before encapsulation because it ensures that cells will not settle to the bottom of the syringe during injection into the microfluidic system and facilitates control of droplet loading as a function of cell density (Box 1). During operation of the co-encapsulator device, it is essential to continuously check the integrity and uniformity of the formed droplets. Lastly, optimization is needed for in-droplet cocultivation conditions to maximize viability within the context of the studied interaction and reporter system (Step 38). We observed high levels of viability after a 24-h incubation period (and reasonable levels up to 72 h), with the limiting factors probably being nutrient depletion and suboptimal gas exchange.

Fig. 2 |. Operation of microfluidic devices for co-encapsulation and sorting.

a, Microfluidic design of co-encapsulator device. Droplet generation oil is used to encapsulate two aqueous cell suspensions flowing at equal rates. b,c, Microscopy images during co-encapsulation (b) and of the resulting co-encapsulated emulsion (c). d, Exemplary probability distribution for Poisson loading of droplets. e, Microfluidic design of sorting device. Multiple oil inlets control the spacing of droplets and regulate the flow around the concentric electrode. The electrode channel is filled with NaCl for electrical conductivity. A grounded moat is filled with NaCl to isolate channels upstream of the electrode from the electric fields generated by the sorting pulse. Fluorescence detection within microfluidic sorting system occurs via three aligned lasers and three PMT detectors in combination with dichroic mirrors and bandpass filters. Sorted droplets are further spaced by pressurized air. f, A microscopic image of droplet passing laser beam (left) and being sorted at electrode into the proximal channel (right). g, A schematic of fluorescence detection from droplets showing three detection channels. Droplet size can be detected based on background staining of the aqueous medium. Loading of droplets with each cell type can be detected by use of cell population-specific cell-permeable dyes. Cell viability is established by leakage of cell dyes. Reporter activation is detected based on EGFP fluorescence. h, A schematic of gating for sorting of droplets. Firstly, intact droplets are gated based on droplet size to exclude fragmented or coalesced droplets. Secondly, EGFP reporter fluorescence is used to gate droplets containing an activated reporter cell. Finally, gating on the cell dye of CRISPR-perturbed cells is used to sort droplets with the correct pairing. Image f adapted with permission from ref. 6, AAAS.

Key points.

• SPEAC-seq uses CRISPR screening, cell coculture in droplets and microfluidic sorting to enable forward genetic screens of regulators of cell–cell communication.

• The procedure provides a detailed experimental workflow and discusses technical aspects. SPEAC-seq enables the defined pairing of interaction partners, is adaptable to different CRISPR screening modalities, is scalable for unbiased genome-wide discovery of functional regulators and can be adapted with few modifications to study a variety of cell types.

Droplet sorting and library preparation (Steps 39–57)

Droplets displaying activation of the fluorescence reporter are sorted using a microfluidics station³⁶ (Fig. 2e). It is important to handle droplets carefully to avoid loss in between steps, although some droplet coalescence is expected. For example, we suggest collecting droplets in a syringe so that no pipetting is needed for reinjection into the sorter (Step 37). Within the droplet sorting device, the reinjection channel narrows to arrange droplets in single file, followed by spacing the droplets apart with a spacing oil inlet (Step 45). Our microfluidic sorter uses a concentric electrode design⁴⁴, whereby a curved channel flows around a rounded electrode to maximize the time droplets spend in the dielectrophoretic field (when activated). In addition, this design minimizes false-positive and false-negative sorting that results from polydisperse emulsions by flowing bias oil at a high rate to reliably focus droplets of different sizes to the waste channel. When droplets are sorted, they enter a separate output channel, which includes an air inlet to produce flow and to separate sorted droplets from each other.

Operation of this microfluidic sorter is not much different from a fluorescence-activated cell sorting (FACS) machine: fluorescence detection is tuned by setting the gain voltages to the photomultiplier tubes (PMTs), gates are set to threshold for uniform droplet sizes and exclude coalesced doublets, and sorting gates are set based on fluorescence detection from the reporter cell type. Prior optimization of the sorting parameters using positive and negative controls for each fluorescent reporter assay, cell type pair and experimental setup are necessary. The dielectrophoretic pulse parameters also need to be optimized for each system, and a high-speed camera can be used to verify sorting of individual droplets.

Expertise needed to implement the protocol

SPEAC-seq combines the use of genetic and molecular methods of cell biology with the use of microfluidic systems and NGS to enable high-throughput functional interaction screening. Knowledge of the biological system of interest is needed to design a valuable SPEAC-seq screening experiment, including the selection of a CRISPR library, reporter assay and optimization of experimental conditions. Photolithography requires specialized equipment and expertise to optimize microfabrication. Engineering knowledge is required to set up the droplet sorter, including handling lasers, optical system alignment and integration of electronic hardware with the instrument control software LabView. This protocol assumes that researchers have a working droplet sorter before starting any functional interaction screening experiments. Standard biological laboratory skills and equipment are needed for cell culture, DNA cloning, transfection, in vitro stimulation and PCR amplification. Library preparation and sequencing require special equipment such as a Bioanalyzer, a sequencing instrument and a computing environment for processing sequencing data. Analysis of SPEAC-seq results requires knowledge of a statistical programming language such as R.

Materials

Reagents

Microfluidic mold fabrication and PDMS chip production

• SU-8 photoresist 2025–2100 (Kayaku Advanced Materials)

  CAUTION Wear appropriate protective equipment when handling SU-8, as it is an irritant on skin or eye contact.

• SU-8 developer (Kayaku Advanced Materials)

  CAUTION Wear appropriate protective equipment when handling SU-8, as it is an irritant on skin or eye contact.

• Silicon wafers (3 inch diameter, Type-P, single-side polished; University Wafer, cat. no. 447)

• Isopropanol molecular grade (Fisher BioReagents, cat. no. BP26181)

  CAUTION Avoid skin contact with isopropanol. Handle with care as isopropanol is highly flammable.

• Acetone (Sigma-Aldrich, cat. no. 179124)

  CAUTION Avoid skin contact and inhaling fumes when handling acetone.

• Aquapel Glass Treatment (Aquapel, cat. no. 47100)

  CAUTION Aquapel is toxic and moisture sensitive. Use only in a fume hood with appropriate protective equipment.

• SYLGARD 184 Silicone Elastomer Kit (Dow, cat. no. 4019862)

  CAUTION Avoid eye contact. Wear appropriate protective equipment.

• Microscope glass slides (75 × 50 × 1 mm; Fisher Scientific, cat. no. 12–550C)

• Fluorinert FC-40 (Sigma-Aldrich, cat. no. F9755)

  CAUTION Do not inhale thermal decomposition products. Wear appropriate protective equipment.

Oligonucleotides

Primers (synthetized at Integrated DNA Technologies):

P5 staggered primer for Illumina sequencing
0 nt stagger	5′-AT GATACGGCGACCACCGAGAT CTACACT CTTT CCCTACACGACGCT CTT CCGAT CT GAT GT CCACGAGGT CT CT-3′
1nt stagger	5′-AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATCTCGATGTCCACGAGGTCTCT-3′
2 nt stagger	5′-AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATCTGCGATGTCCACGAGGTCTCT-3′
3 nt stagger	5′-AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATCTAGCGATGTCCACGAGGTCTCT-3′
4 nt stagger	5′-AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATCTCAACGATGTCCACGAGGTCTCT-3′
5 nt stagger	5′-AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATCTTGCACCGATGTCCACGAGGTCTCT-3′
6 nt stagger	5′-AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATCTACGCAACGATGTCCACGAGGTCTCT-3′
7 nt stagger	5′-AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATCTGAAGACCCGATGTCCACGAGGTCTCT-3′
8 nt stagger	5′-AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATCTGAAGACCCTTGTGGAAAGGACGAAACACCG-3′
P7 index primer for lentiCRISPRv2
5′-CAAGCAGAAGACGGCATACGAGATNNNNNNNNGTGACTGGAGTTCAGACGTGTGCTCTTCCGATCTCCAATTCCCACTCCTTTCAAGACCT-3′

Plasmids and libraries

• LentiCRISPRv2 pooled gRNA library (Addgene, cat. no. #73632, RRID: Addgene_73632)

• Packaging plasmid psPAX2 (Addgene, cat. no. #12260, RRID: Addgene_12260)

• Envelope plasmid pCMV-VSV-G, Addgene, cat. no. #8454, RRID: Addgene_8454)

Microfluidic reagents

• 3M Novec 7500 Engineered Fluid (HFE-7500) (Gallade Chemical, cat. no. 98021229285)

  CAUTION Avoid direct contact as it may cause irritation of skin, eyes and airways. Wear appropriate protective equipment.

• Qx200 Droplet Generation Oil for EvaGreen (Bio-Rad Laboratories, cat. no. 1864006)

• 1H,1H,2H,2H-Perfluoro-1-octanol (PFO) (Sigma-Aldrich, cat. no. 370533)

  CAUTION Avoid contact with skin. Use only in a fume hood with appropriate protective equipment.

• Sterile syringes (3 mL, 10 mL; BD, cat. nos. 309657, 309604)

• PE/2 tubing (Scientific Commodities, cat. no. BB31695-PE/2)

• Luer Stub 27ga 0.5 in (Instech Laboratories, cat. no. LS27K)

• NaCl (Sigma-Aldrich, cat. no. S9888)

• Parafilm M (Sigma-Aldrich, cat. no. P7668)

• Falcon bacteriological Petri dish with lid (150 × 15 mm; Corning, cat. no. 351058)

• Precision knife (0.5 mm thick blade; Grainger, cat. no. 3ZH07)

  CAUTION Handle sharp objects with care.

• Scotch tape (3M, cat. no. S-12187)

• Drierite silica gel desiccant (Drierite, cat. no. 21005)

Cell lines

HEK293FT (Thermo Fisher Scientific, R70007; RRID: CVCL_6911)

CAUTION The cell lines used in your research should be regularly checked to ensure they are authentic and are not infected with mycoplasma.

Cell handling

• Fetal bovine serum (FBS) (Corning, cat. no. 35–070-CV)

• Dulbecco’s modified Eagle medium (DMEM)/F12, GlutaMAX (Thermo Fisher Scientific, cat. no. 10565042)

• DMEM/F12, no phenol red (Thermo Fisher Scientific, cat. no. 21041025)

• Penicillin–streptomycin (Thermo Fisher Scientific, cat. no. 15140122)

• Hank’s balanced salt solution (Thermo Fisher Scientific, cat. no. 14025134)

• OptiPrep density gradient medium (Sigma-Aldrich, cat. no. D1556)

• Pluronic F-68 (Thermo Fisher Scientific, cat. no. 24040032)

• CellTrace Far Red Cell Proliferation kit (Thermo Fisher Scientific, cat. no. C34564)

• CellTrace Calcein Red-Orange (Thermo Fisher Scientific, cat. no. C34851)

• Cy5-alkyne (Sigma-Aldrich, cat. no. 777358)

• LT1 transfection reagent (Mirus, cat. no. MIR2305)

• Opti-MEM reduced serum medium with GlutaMAX (Thermo Fisher Scientific, cat. no. 51985034)

• STBL4 electrocompetent cells (Thermo Fisher Scientific, cat. no. 11635018)

• Stable outgrowth medium with catabolite repression (SOC) (Thermo Fisher Scientific, cat. no. 15544034)

• LB agar plates with 100 μg/mL ampicillin (Sigma-Aldrich, cat. no. L5667)

• Miller’s LB Broth (Research Products International, cat. no. L24400)

• Lenti-X Concentrator (Takara Bio Inc., cat. no. 631232)

• Polybrene infection/transfection reagent (EMD Millipore, cat. no. TR-1003-G)

• Puromycin HCl (Sigma-Aldrich, cat. no. P9620–10ML)

• Trypan blue solution (Corning, cat. no. 25–900-CI)

• LPS from Escherichia coli 0111:B4 (Invivogen, cat. no. tlrl-eblps)

• Recombinant IL-1β (R&D Systems, cat. no. 401-ML-005)

• Recombinant TNF-alpha (R&D Systems, cat. no. 410-MT-010)

Buffers and molecular biology

• Ex Taq DNA Polymerase, reaction buffer and dNTP mix (TaKaRa Bio Inc., cat. no. RR001B)

• KAPA library quantification kit (Roche, cat. no. 07960140001)

• AMPure XP beads (Beckman Coulter, cat. no. A63880)

• Agilent high sensitivity DNA kit (Agilent Technologies, cat. no. 5067–4626)

• Nuclease-free water (Thermo Fisher Scientific, cat. no. 10977015)

• Qiagen Blood and Tissue DNA isolation kit (Qiagen, cat. no. 69504)

• Qiagen EndoFree Plasmid Maxi kit (Qiagen, cat. no. 12362)

Equipment

CAUTION This protocol involves the use of specialized equipment found inside a microfabrication clean room, for which appropriate protective equipment and training is required. Follow safety procedures established by microfabrication facilities.

• Spin Coater (Headway Research, model no. PWM32-PS-R790)

• Contact mask aligner (Karl Süss, model no. MJB3)

  CAUTION Do not look directly into sources of ultraviolet light.

• Ceramic hot plates (VWR International, model no. 97042)

  CAUTION Handle hot surfaces with care.

• Ultrasonic Bath (Bransonic)

  CAUTION Wear appropriate protective hearing equipment.

• Profilm3D profilometer (Filmetrics)

• Rapid Core biopsy punch (ID 0.75 mm, OD 1.07 mm; Ted Pella, model no. 15115–2)

  CAUTION Handle sharp biopsy punch with care.

• Cutting mat (Ted Pella, model no. 15087–1)

• Graefe forceps (Fine Science Tools, model no. 11051–10)

• Kimwipes (Kimberly-Clark Professional, model no. 34120)

• Oxygen plasma etcher (Technics, model no. 500-II)

• Bright-field inverted microscope (Motic, model no. AE31)

  CAUTION Never look directly through the eye piece while operating the lasers. Always use the high-speed camera to visualize microfluidic circuits.

• Nexus breadboard (Thorlabs, cat. no. B3030F)

• Syringe pumps (Harvard Apparatus, cat. no. 70–2226)

• CompactRIO field programmable gate array (FPGA) module (National Instruments, cat. no. NI-9154)

• Analog output module (National Instruments, cat. no. NI-9263)

• Analog input module (National Instruments, cat. no. NI-9223)

• Trek high voltage amplifier (Advanced Energy, cat. no. 609E-6)

  CAUTION Handle high voltages with extreme care and never monitor output directly. Instead, use the voltage monitor that is scaled down to 1/1,000th of actual output voltage.

• Lasers (OptoEngine, cat. nos. MLL-FN-473, MLL-III-532, MRL-III-640)

  CAUTION Laser safety should be taken seriously, as the lasers involved in this protocol can severely and permanently damage vision. Use appropriate protective eye wear.

• Dichroic mirrors (Semrock, cat. nos. FF552-Di02–25×36, Di03-R488-t1–25×36, Di01-R405/488/532/635, FF593-Di03–25×36)

• Brightfield filter (Semrock, cat. no. FF740-Di01–25×36)

• Emission bandpass filters (Semrock, cat. nos. FF01–446/510/581/703–25, FF01–673/11–25, FF01–572/28–25, FF01–517/20–25)

• Laser safety goggles (Laserglow Technologies, cat. no. AGF5565XX)

  CAUTION Laser safety should be taken seriously, as the lasers involved in this protocol can severely and permanently damage vision. Use appropriate protective eye wear.

• Photomultiplier tubes (Thorlabs, cat. nos. PMM01 and PMM02)

• Pressurized air or nitrogen (Airgas)

  CAUTION Pressurized gas tanks should be stored and handled with care. Only use appropriate pressure regulators when opening.

• Gas regulator (Harris, cat. no. KH1013)

• Benchtop centrifuge (Eppendorf, model no. 2231000768)

• Vacuum desiccator (Bel-Art, model no. F42027–0000)

• CO₂ Incubator (Panasonic Healthcare)

• Imperial III Incubator (Lab-Line)

• Veriti Thermal cycler (Applied Biosystems)

• ViiA 7 Real-Time PCR System (Applied Biosystems)

• Gene Pulser Xcell Electroporation System with Electroporation Cuvettes (Bio-Rad, model no. 1652660)

• 2100 Bioanalyzer Instrument (Agilent Technologies)

• Hemocytometer (Incyto, model no. DHC-N01–5)

• Razor blades (0.22 mm thick; cat. no. VWR 55411–050)

  CAUTION Handle sharp objects with care.

• Pipettes (0.5–10 μL, 2–20 μL, 20–200 μL and 100–1,000 μL, Eppendorf Research plus)

• Pipette tips (10 μL (Celltreat, cat. no. 229016), 20 μL (USA Scientific, cat. no. 1120–1810), 200 μL (USA Scientific, cat. no. 1111–1210), 1,000 μL (Thermo Fisher Scientific, cat. no. 02-707-404))

• Serological pipettes (5, 10 and 25 mL; Celltreat, cat. nos. 229005B, 229010B-4 and 667225B)

• Sterile microcentrifuge tubes (1.5 mL; USA Scientific, cat. no. 1615–5500)

• PCR 8-tube strips (0.2 mL; USA Scientific, cat. no. 1402–4700)

• DNA LoBind tubes (Fisher Scientific, cat. no. 13–698-790)

• Falcon conical tubes (15 mL and 50 mL; Corning, cat. nos. 352196 and 352070)

• Cell strainer (70 μm and 100 μm; Celltreat, cat. nos. 229484 and 229486)

• Falcon 5 mL round-bottom tube with 35 μm strainer (Corning, cat. no. 352235)

• Cell Scraper (Celltreat, cat. no. 229310)

• Falcon T175 flask (Corning, cat. no. 353112)

Equipment setup

Microfluidic equipment setup

The microfluidic equipment setup required for this study has been previously described³⁶. Setup and optimization of this system is essential for droplet sorting in SPEAC-seq. In brief, a steel breadboard is used to mount the following core pieces of hardware in fixed position relative to each other: optical setup with lasers, microscope with high-speed camera and stage for microfluidic device, and fluorescence detector setup. Three lasers of red, green and blue wavelengths, mirrors and dichroic mirrors are setup to align all lasers to one laser line, which is guided through the back of the microscope onto the microfluidic device staged on the microscope. A camera mounted on the microscope helps the user to safely position the laser in the desired position on the microfluidic circuit to excite fluorescence in the passing droplets; the high-speed camera additionally makes high-framerate recordings to verify droplet sorting. Fluorescence coming from droplets in the microfluidic circuit is guided back into the objective of the microscope and out into a set of PMT detectors, using dichroic mirrors and bandpass filters to specifically reflect emission wavelengths to each dedicated PMT. These detectors are connected to a FPGA with analog output and input modules that control the gain voltage required to tune PMT sensitivity and that receive the PMT detection voltage. LabView is used to run the FPGA, which integrates the fluorescence signals to capture data and then controls the sorting pulse via an analog output module connected to a high-voltage amplifier. The amplified electric pulse is transmitted to a saltwater-filled syringe that fills the electrode channel of the microfluidic device. Microfluidic device preparation is described below. Further, four programmable syringe pumps are used to control the flow rates of each channel during co-encapsulation and sorting. A pressurized air tank is connected to a regulator and connected to the microfluidic device via PE/2 tubing. Droplet emulsions and sorted outputs are collected either directly in syringes or in conical tubes as described below.

Reagent setup

Virus production medium

20% (vol/vol) FBS in DMEM/F12. This medium can be stored at 4 °C protected from light for up to 6 months as long as no contamination is present.

Coculture medium

10% (vol/vol) FBS, 18–20% (vol/vol) OptiPrep, 1% (vol/vol) Pluronic F-68, 1% penicillin–streptomycin in DMEM/F12 without phenol red.This medium can be stored at 4 °C protected from light for up to 6 months as long as no contamination is present.

20% (v/v) PFO solution

Combine one part PFO with four parts HFE-7500 in a fume hood. This solution should be prepared fresh on the day of droplet breakage.

Aquapel

Filter Aquapel through a syringe-attachable 0.22 μm filter to sterilize before application. Sterile-filtered single-use aliquots can be stored in a jar of Drierite desiccant up to a year.

LB medium

Following manufacturer’s instructions, weigh 25 g of LB granulate and dissolve in 1 L of deionized water, followed by standard autoclaving.

NaCl solution

To prepare a 2 M NaCl stock solution dissolve 11.6 g in 100 mL of deionized water.

Software

• LabView with FPGA module (National Instruments, 776670–35, 778694–35)

• R (v3.6.0 or higher; The R Foundation; https://www.r-project.org/)

• Python (v3.7 or higher, Python Software Foundation; https://www.python.org/)

• Cutadapt⁴⁵ (https://cutadapt.readthedocs.io/en/stable/)

• MAGeCK⁴⁶ (https://github.com/liulab-dfci/MAGeCK)

Procedure

Preparation of microfluidic devices

● TIMING 4 d

• 1
  Fabricate silicon wafer molds for the two microfluidic devices inside a microfabrication clean room by following standard soft photolithography techniques described elsewhere³⁶. In brief, order photolithography masks with microfluidic device designs (Supplementary Data 1 and 2) printed at 20,000 d.p.i. or higher resolution. Fabricate each layer to the channel heights described below; choose SU-8 negative photoresist with the appropriate viscosity following manufacturer’s instructions. Place the silicon wafer into the spin coater, dispense ~3–5 mL SU-8 onto the wafer and perform spin coating at the corresponding speed according to SU-8 manufacturer’s instructions. Perform a soft bake of coated silicon wafer on a hot plate following SU-8 manufacturer’s recommendations for temperature and duration depending on thickness of layer. Align photolithography mask with wafer and perform UV exposure following SU-8 manufacturer’s instructions for energy dosage depending on layer thickness. Perform a post-exposure bake following SU-8 manufacturer’s instructions. Develop in a glass dish containing SU-8 developer by gently agitating and finally wash off excess photoresist with isopropanol. Hard bake following SU-8 manufacturer’s instructions. Use a three-dimensional optical profilometer to characterize feature height and optimize fabrication protocol if height is incorrect.

  	Co-encapsulator	Sorter	SU-8 suggestion
  Layer 1	70 μm	60 μm	SU-8 2025
  Layer 2	200 μm	100 μm	SU-8 2075

  CRITICAL Silicon master microfabrication protocols must be empirically optimized and will vary depending on the specific equipment used and SU-8 age. Consult with photoresist manufacturer datasheets (e.g., https://kayakuam.com/wp-content/uploads/2019/09/SU-82000DataSheet2025thru2075Ver4-3.pdf) and with microfabrication facility staff for equipment details and optimization recommendations.

• 2To prepare a PDMS device from the silicon master, mix ~30 g of PDMS elastomer and 3 g of curing agent (1:10 ratio) in a plastic cup and mix well. Bubbles are normal. Place in a vacuum chamber for 20–30 min or until air bubbles are no longer visible.
• 3Tape the silicon wafer to a Petri dish and pour in the PDMS mix slowly. Cover the Petri dish with its lid to minimize dust contamination. Degas in a vacuum chamber for at least 30 min or until all bubbles are gone. Use a clean pipette tip to pop any remaining bubbles or push them outside of the margin of the mold.

  CRITICAL STEP If any bubbles remain trapped, they may interfere with properly making the channels.

• 4Incubate at 60 °C overnight.
• 5Using a precision knife, carefully cut around the margins of the design. Be aware not to puncture or break the silicon wafer. Lift the PDMS slab and place it on a cutting mat, cutting off any rough edges that might compromise bonding. With features facing up, use a 0.75 mm biopsy punch to make holes for all fluid inlets and outlets, including electrode channels. Seal the patterned side of the slab with scotch tape to keep it dust free.

  CRITICAL STEP Do not use too much force when cutting the PDMS slab as it may shatter the silicon wafer. Additionally, ensure that the biopsy punch is sharp when punching inlet/outlet holes, otherwise it might generate ports that are leaky when connecting PE/2 tubing.

• 6Prepare punched PDMS slab and a glass slide for plasma bonding. Clean dust off of the PDMS device using scotch tape repeatedly. Clean a glass slide in an ultrasonic bath three times using a slide holder filled with acetone first, then isopropanol and finally deionized water. Dry the glass slide with pressurized air.

  CRITICAL STEP Clean surfaces are essential for good plasma bonding in Step 7.

• 7Place clean glass slide from Step 6 and PDMS slab inside a plasma etcher with the features facing up. Treat with oxygen plasma for 30 s at 150 W and 300 mTorr. Then, quickly repressurize the chamber with nitrogen, retrieve both pieces and press the upward-facing sides together gently. Place on a flat surface and press firmly for >30 s. Bake on a hot plate at 150 °C for 10 min to aid bonding. Alternatively, bake overnight at 60 °C.

  CRITICAL STEP For optimal bonding, the pieces should be pressed together as soon as possible after exposure to oxygen plasma.

  PAUSE POINT Devices can be stored at room temperature (21 °C) indefinitely at this point by sealing the ports with scotch tape.

• 8Treat the channels with Aquapel to make them hydrophobic. To do this, fill a syringe with sterile-filtered Aquapel, attach a Luer stub and connect PE/2 tubing. Then, using tweezers, connect the other end of the PE-2 tubing to the outlet and/or inlet holes and slowly push sufficient Aquapel out manually until all channels are covered and air is displaced. Placing a light source horizontally to the device can aid in visualizing the fluid inside the channels. Use Kimwipes to clean excess fluid off the top of the device afterward.
• 9Allow exactly 5 min of contact time with Aquapel.

  CRITICAL STEP This is a time-sensitive step; Aquapel precipitates upon reaction with air, which may plug channels if treatment exceeds 5 min.

• 10Fill a syringe with air and attach a Luer stub and PE-2 tubing and attach to outlet and/or inlet holes to manually flush out Aquapel from all channels of Aquapel.
• 11Using similar technique as in Step 8, connect a syringe loaded with FC-40 to the inlet/outlet channels and flush through the device.
• 12Flush out all FC-40 oil by manually pushing air through, making sure to very thoroughly dry all channels. Check that every module is clear of liquid to avoid residue build-up that might plug the device.
• 13Bake PDMS devices at 60 °C overnight.
• 14Cover PDMS devices with scotch tape to prevent dust contamination until use. Tape can be cut to the size of one module; so that other modules stay clean during operation of the device.

  PAUSE POINT Devices can be stored at this point by sealing the ports with scotch tape.

  Aquapel may degrade, therefore it is advised to use it within 6 months.

Preparation of CRISPR screening

● TIMING 2 weeks

• 15Mix 400 ng pooled CRISPR library DNA with 100 μL STBL4 electrocompetent cells. This calculation is sufficient for 20,000 gRNAs and should be scaled up appropriately for each library.
• 16Electroporate 35 μL of this mixture at 1.8 kV (Ec1 setting) and immediately add 500 μL stable outgrowth medium, shake for 1 h at 30 °C and spread out onto an LB agar plate containing ampicillin.
• 17Grow for 16–18 h at 30 °C.
• 18Mechanically scrape cells with cold LB, transfer to a microtube and centrifuge at 5,000g for 5 min.
• 19Purify DNA using the endotoxin-free Plasmid Maxi kit (Qiagen) following the manufacturer’s protocol, except for the following modifications. Directly lyse the bacterial pellets by resuspending in P1, P2 and P3 before adding to the column. Use prewarmed elution buffer at 50 °C for elution.

  PAUSE POINT DNA libraries can be stored long-term at −20 °C.

• 20Seed 15 million HEK293FT cells per T175 flask in antibiotic-free virus production medium and culture overnight at 37 °C.
• 21After the overnight incubation, replace the culture medium with Opti-MEM reduced serum medium.
• 22Combine 37 μg of purified library DNA from Step 19, 46 μg DNA of packaging plasmid psPAX2 and 4.6 μg DNA of envelope plasmid VSV-G with 500 μL Opti-MEM reduced serum medium at room temperature and mix by pipetting.

  In a separate tube, combine 500 μL Opti-MEM reduced serum medium with 262.8 μL LT1 transfection reagent and mix by pipetting. A 1:3 ratio of DNA (in μg) to LT1 reagent (in μL) is recommended as a starting point.

  CRITICAL STEP Transfection conditions may need optimization if using other libraries.

• 23Combine the DNA and transfection reagent mixes and incubate for 20 min at room temperature.
• 24Add 1 mL transfection mix to the cell flask and shake to distribute. Incubate at 37 °C, 5% CO₂ for 6 h.
• 25After 6 h, replace transfection medium with virus production medium. Incubate for another 48 h.
• 26Collect lentivirus-containing media and concentrate using Lenti-X concentrator per the manufacturer’s protocol. Resuspend pellets in 1:100 of the original volume in PBS and store aliquots at −80 °C. Use an aliquot to titrate the lentivirus concentration following the qPCR Lentivirus Titer kit manufacturer’s protocol.

  PAUSE POINT Lentivirus can be stored at −80 °C short term. Avoid freeze-thaw cycles.

• 27Spinfect cells by seeding at the predetermined cell density in addition with 8 μg/mL polybrene and lentiviral supernatant then centrifuging at 30 °C.

  CRITICAL STEP For high-efficiency transduction of genetically screened cells, spinfection in the presence of polybrene can be used. Transduction parameters such as cell density, polybrene concentration, multiplicity of infection and centrifugation conditions need to be optimized for each individual application and library³⁷.

• 28Finally, select stably transduced cells using 2 μg/mL puromycin-containing media for 6 d.

  CRITICAL STEP Antibiotic selection conditions need to be optimized for each cell type.

Cell encapsulation and coculture

● TIMING 2 h + 1–3 d incubation time

• 29Collect cells for co-encapsulation. Find recommendations regarding required cell numbers for SPEAC-seq in Box 1. Preconditioning of CRISPR and reporter cells will be specific to each setting. In the original study, p65^EGFP astrocytes were stimulated with 10⁻⁷ μg/mL TNF and IL-1β 24 h before co-encapsulation⁶. Microglia transfected with CRISPR libraries were stimulated with 100 ng/mL of LPS 24 h before co-encapsulation⁶.
• 30Wash cells with Hank’s balanced salt solution (HBSS) and proceed to stain with CellTrace dyes for 30 min at 37 °C. One cell type can be traced with 2 μM CellTrace Far-Red and the other with 1 μM CellTrace Calcein Red-Orange, AM. Additionally, add 200 nM Cy5-alkyne to the Far-Red-stained population as a background stain for droplet size detection.
• 31Wash cells with coculture medium and create a single-cell suspension by passing the cell suspension through a 35-μm strainer.
• 32Establish viable cell number using Trypan blue and a hemocytometer or an automated cell counter.

  CRITICAL STEP Accurate cell counting is essential to ensure intended droplet loading.

  Increase accuracy of cell counting by repeating counts or using automated cell counting.

• 33Centrifuge cells for 5 min at 500g and resuspend in coculture medium to the needed cell density. We used CRISPR cells at 0.1 cells per drop (1.39 × 10⁶ cells/mL) and reporter cells at 0.5 cells per drop (6.95 × 10⁶ cells/mL)

  CRITICAL STEP Refer to Box 1 for details on the effects of cell density on droplet loading.

• 34Draw up each cell suspension into a micropipette first, then carefully transfer to a 3-mL syringe and attach Luer stub and PE/2 tubing. To avoid cell lysis, handle the syringe loading carefully. Avoid rapid pulling of the plunger. For easier connection to the ports of the PDMS device, it is useful to cut the PE/2 tubing at a slight angle using a razor blade.
• 35Fill a 10-mL syringe with EvaGreen droplet generation oil and attach it to the syringe pump. Manually push out the liquid until it reaches the end of the PE/2 tubing. Connect the PE/2 tubing to the oil inlet of the co-encapsulator device (Fig. 2a) and prime the device by starting the pump at a rate of 10,000 μL/h.

  CRITICAL STEP It is useful to flow oil through the whole device to flush out any debris and prime the hydrophobic surface.

• 36
  Attach cell suspension syringes to their respective pumps and gently push out the liquid until it reaches the end of the PE/2 tubing. Connect PE/2 tubing to the cell inlet ports and start co-encapsulation at the following flow rates. The flow may take a minute to stabilize.

  Syringe type	Channel	Flow rate
  3 mL	Cell suspensions	600 μL/h
  10 mL	Droplet generation oil	3,000 μL/h

  CRITICAL STEP Remember to generate the necessary control emulsions described above, including (1) reporter cells alone (i.e., not co-encapsulated with any CRISPR cell) as a baseline/negative control; (2) reporter cells alone that are positively activated using, for example, cytokines that induce NF-κB activation as a positive control of fluorescence detection; and (3) subthreshold stimulated reporter cells co-encapsulated with unstimulated microglia (experimental control to exclude spontaneous interactions not part of inflammatory setting).

  ◆ TROUBLESHOOTING

• 37Use a microscope-mounted camera to check the flow stability at the flow focusing junction and the integrity of resulting droplets (Fig. 2b,c). Once continuous droplet generation can be observed, connect a piece of PE/2 tubing to the outlet channel. For droplet collection, insert the PE/2 tubing into a 3-mL syringe loaded with at least 500 μL HFE-7500 as backing. Mount the syringe vertically and leave air on top. The emulsion exiting the device is then collected in the HFE-7500 phase and displaces the air as it enters the syringe. Seal the syringe with Parafilm.

  ◆ TROUBLESHOOTING

• 38Incubate the emulsion in a cell culture incubator at 37 °C for 24 h.

  CRITICAL STEP We encourage all users to perform time course analysis of viability and reporter activation to determine optimal cocultivation conditions⁶ (Fig. 3c).

Fig. 3 |. SPEAC-seq uncovers microglial regulators of astrocyte inflammation.

a, Microglia are known to regulate astrocyte inflammatory functions, although only a handful of factors have been identified. b, GFP activation level in p65^EGFP astrocytes following stimulation to determine subthreshold dose of cytokines used to prime cells without activating reporter circuit⁶. c, Viability of astrocytes cultured in droplets over 3–72 h (ref. 6) (n = 2 biological replicates). Mean and s.d. are shown. d, Droplet stability characterization by image analysis of emulsions made with culture medium with or without phenol red over 24 h (n = 1,045–2,241 droplets per condition). Mean and individual measurements are shown. Polydispersity index calculated as (standard deviation of droplet diameters)²/(mean droplet diameter)⁴⁹. e, Principal component analysis of RNA-seq data from LPS-treated or vehicle-treated microglia cultured in plates or in droplets, showing overlap of transcriptional profiles across culture formats (n = 3–4 biological replicates per condition)⁶. f, Analysis of gRNA enrichment in droplets sorted by fluorescence versus negative droplets⁶. g, Filtering of gRNA candidate hits by RNA-seq dataset of LPS-treated microglia⁶. P value adjusted for multiple hypothesis testing (P adj) below 0.05 considered significant. These data are published as part of our original implementation of SPEAC-seq⁶. Images b,c,e–g adapted with permission from ref. 6, AAAS.

Droplet sorting

● TIMING 2–8 h

• 39Prepare the droplet sorting station. For technical details, refer to a previously published droplet sorting protocol³⁶.
• 40To fill electrode and moat channels, fill two 3-mL syringes with 2 M NaCl solution and connect to the respective inlets using a Luer stub and PE/2 tubing (Fig. 2e). The electrode channel can be filled by pushing air out into the PDMS using a syringe screw clamp. Clamp the electrode syringe tip to the high-voltage amplifier and the moat syringe tip to ground.

  ◆ TROUBLESHOOTING

• 41Use HFE-7500 oil to establish flow around the concentric electrode (Fig. 2e). Load two 30-mL syringes and one 3-mL syringe with HFE-7500 and connect to oil inlets.
• 42Gently invert the syringe holding the droplet emulsion a few times to avoid coalescence from tight packing. Place the droplet-containing syringe vertically on a pump and run the pump until the emulsion reaches the end of the PE/2 tubing. After priming the sorting device with HFE-7500 oil and attaching all other inlets, connect the droplet suspension to the sorting device with PE/2 tubing.
• 43Connect a pressurized air tank with a pressure regulator to the sorting device air inlet using PE/2 tubing and set air pressure to 2–8 psi or until air can be seen entering the device in that channel.

  CRITICAL STEP This is needed to maintain the intended resistance in the sorting output channel and for spacing apart of droplets to further avoid coalescence.

• 44
  Set up the following flow rates:

  Syringe type	Channel	Flow rate
  3 mL	Droplet emulsion	100 μL/h
  3 mL	Spacing oil	500 μL/h
  30 mL	Concentric flow oil	2000 μL/h
  30 mL	Concentric flow bias oil	3000 μL/h
  -	Pressured air	2–8 psi

• 45Ensure that lasers are aligned and focused on the droplet channel upstream of the concentric electrode (Fig. 2f). During sorting, set the PMT gain levels (between 0.4 and 1) and monitor the fluorescence detection to set gating thresholds on the desired droplets (Fig. 2g). Set the peak detection minimum level to exclude baseline noise and droplets containing lysed cells from being processed as a peak (Fig. 2h). Gate droplet size by setting minimum and maximum peak duration to exclude merged droplets or droplet fragments. Gate droplets encapsulating both cell types of interest by gating those positive for the fluorescence of each cell dye and the reporter fluorescence (Fig. 2h).

  CRITICAL STEP Establishing the intended behavior of the microfluidic system is essential before performing the droplet sort for enrichment of droplets that activate the reporter assay as part of the CRISPR screen. The parameters for the sorting pulse need to be optimized in each system. A small delay is needed due to droplet travel time. Additionally, the parameters creating the electric field needed for the dielectrophoretic manipulation of droplet trajectories are critical to create accurate sorting; these include pulse voltage, number of cycles and frequency. This will be dependent on each microfluidic setup and need to be optimized before sorting. Check droplet sorting accuracy using a positive control emulsion in which all droplets contain fluorescence. Record a video at a high frame rate to confirm the sorting of all positive control droplets when the electrode pulse is activated. When the sorting pulse is deactivated by the user, all droplets should flow to the waste channel. Adjust the sorting parameters above if needed.

  ◆ TROUBLESHOOTING

• 46Collect the positively and negatively sorted fractions in separate 15-mL tubes and overlay with 200 μL PBS. Freeze the emulsion at −80 °C for at least 24 h.

  PAUSE POINT Tubes can be stored at −80 °C for up to a week.

Droplet breakage and gRNA amplification

● TIMING 4 h

• 47Thaw frozen droplet emulsion at room temperature for 1 h.
• 48Break droplets by adding 1 mL 20% (vol/vol) PFO in HFE-7500 to the emulsion, moving the tube gently to mix. For optimal phase separation, centrifuge at 1,000g for 30 s. Transfer the aqueous layer containing cells to a fresh microcentrifuge tube by pipetting.
• 49Proceed to isolate genomic DNA (gDNA) using the Blood and Tissue DNA isolation kit (Qiagen) according to manufacturer’s instructions. Elute gDNA twice with 200 μL, totaling 400 μL.
• 50For library purification and concentration, perform AMPure XP bead-based purification (Beckman Coulter) at a 2× beads-to-sample ratio following manufacturer’s protocol. Elute in a final volume of 40 μL nuclease-free water.
• 51
  Prepare the following PCR mix for gRNA sequence amplification:

  Reagent	Reagent concentration	Final concentration	Volume (per tube)
  ExTaq reaction buffer	10x	1x	10 μL
  dNTP mix	2.5 mM	200 μM	8 μL
  P5 staggered primer mix	100 μM	0.5 μM	0.5μL
  P7 index primer (unique per reaction)	5 mM	0.5 μM	10 μL
  ExTaq polymerase	5 U/mL	75 mU	1.5 μL
  gDNA from sorted fractions	–	–	40 μL
  Nuclease-free water	–	–	31.5 μL
  Total			100 μL

• 52
  Run the following library amplification program.

  Step	Stage	Temperature	Time	Cycles
  1	Initial denaturation	95 °C	1 min	1
  2	Denaturation	95 °C	30 s	28–35
  	Annealing	53 °C	30 s
  	Elongation	72 °C	30 s
  3	Final elongation	72 °C	10 min	1
  4	Hold	4 °C	–	Hold

  CRITICAL STEP Optimization may be needed to determine the appropriate number of PCR cycles for the amount of genetic material recovered. We amplified samples with less than 2,000 sorted droplets at 35 cycles.

• 53Purify the libraries via AMPure XP bead purification at a 2× beads-to-sample ratio and elute in 40 μL nuclease-free water.

  PAUSE POINT DNA libraries can be stored at −20 °C long term.

Library preparation, sequencing and analysis

● TIMING 5 h

• 54Confirm library size distribution using a Bioanalyzer high sensitivity DNA kit (Agilent). Depending on the plasmid used for gRNA transduction, the fragment size will vary. The Brie library (lentiCRISPRv2) used in the original study will yield a library averaging 285 bp in size.

  ◆ TROUBLESHOOTING

• 55Quantify DNA concentration of the libraries using a KAPA library quantification kit following the manufacturer’s protocol.
• 56Pool indexed libraries normalizing the concentration to 4 nM each and perform NGS at a read depth of at least 100 reads per gRNA (sorted droplet).
• 57For analysis, trim sequencing reads using Cutadapt⁴⁵ to find gRNA sequences that match the genome-wide library (-g ACACCG … GTTTTAG) and quantify using MAGeCK⁴⁶.

  CRITICAL STEP Several methods for gRNA ranking exist⁵ and researchers are advised to design their analysis to incorporate quality controls, including the sufficient gRNA representation, coherence between replicates and depletion of essential genes. Further filtering steps in the analysis include the use of RNA-seq data from similarly treated cells in vitro or in vivo to validate expression of the gRNA targets in the cells and biological context of interest.

Troubleshooting

Troubleshooting advice can be found in Table 1.

Table 1 |.

Troubleshooting table

Step	Problem	Possible reason	Solution
36	Average co-encapsulation rates are incorrect	Incorrect cell counting. Cells may settle in syringe if OptiPrep concentration is too low	Use automated or manual cell counting system appropriate for cell type of interest. Resuspend cells thoroughly. Optimize concentration of density gradient medium OptiPrep (between 18% and 20%)
37	Droplet formation is irregular	Dust or debris might be obstructing the channels of the microfluidic device. Aquapel-treated devices can be used for up to 3 months, after which Aquapel coating degrades and can change flow properties	Prime device by running droplet generation oil at 10,000 μL/h. If problem persists, use new device. Treat devices with Aquapel shortly before performing experiment
42	Droplets merge during coculture or reinjection	Coalescence may occur when droplets are exposed to air. In addition, anionic surfactants help to stabilize the phase interface. Other media components, such as the pH indicator phenol red, may negatively impact droplet stability	Carefully layer PBS on top of collected droplets before coculture. Add 1% Pluronic F-68 to stabilize emulsion. Optimize culture media for droplet stability over the coculture timeframe
45	Sorting pulse is ineffective at sorting droplets	Timing and characteristics of the sort pulse may be incorrect	Optimize sorting parameters such that pulse delay is appropriate for the position of the laser beam and dielectrophoretic force sufficiently attracts droplet into the ‘positively sorted’ channel
54	gRNA library prep results in flat BioAnalyzer trace	Insufficient number of droplets collected, or insufficient DNA library amplification	Collect more droplets during the screening experiment. Increase number of PCR amplification cycles. Use carrier RNA to promote recovery

Timing

Steps 1–14, Preparation of microfluidic devices: 4 d

Steps 15–28, Preparation of CRISPR screening: 14 d

Steps 29–38, Cell encapsulation and coculture: 1–3 d

Steps 39–46, Droplet reinjection and sorting: 2–8 h

Steps 47–53, Droplet breakage and gRNA amplification: 4 h

Steps 54–57, Library amplification, sequencing and analysis: 5 h

Anticipated results

SPEAC-seq experiments⁷ are expected to generate a dataset consisting of gRNA candidates associated with the regulation of the biological function of the reporter circuit. We previously used this method to investigate microglia–astrocyte interactions involved in the regulation of neuroinflammation. Microglia are known to directly regulate autoimmune disease-promoting inflammatory functions of astrocytes through various pathways⁴⁷. Thus, to systematically investigate this dynamic intercellular relationship, we performed SPEAC-seq with the genome-wide perturbation of microglial genes, using a reporter of NF-κB activation to detect neuroinflammatory activities in astrocytes⁶ (Fig. 3a). As per the SPEAC-seq protocol, we first transduced primary microglia with a genome-wide CRISPR–Cas9 lentiviral perturbation library and selected successful transductants via antibiotic selection (Steps 15–28). The p65^EGFP reporter strain contains an enhancer with p65 binding sites that drives enhanced green fluorescent protein (EGFP) expression downstream of NF-κB activation⁴⁸. To optimize reporter detection in a coculture setting, we first determined the subthreshold concentrations of inflammatory cytokines IL-1β and TNF that prime p65^EGFP astrocytes and ensured sufficient reporter sensitivity to activation by intercellular interactions with microglia (Fig. 3b) (Step 29). Furthermore, microglia were primed to an inflammatory state by treatment with LPS (Step 29). Next, using a co-encapsulator device we generated droplets containing cell pairs by setting λ_Microglia = 0.1 and λ_Astrocytes = 0.5 (Box 1) (Steps 29–37). We measured the viability of astrocytes in droplets and determined 24 h to be the optimal time for coculture, allowing sufficient time for cell–cell interactions to occur without confounding effects of low cell viability (Fig. 3c). Furthermore, we optimized the culture medium to maintain droplet stability over the culture time, finding that the pH indicator phenol red substantially impacts droplet stability (Fig. 3d). Finally, we observed that transcriptional differences in microglia could be largely attributed to treatment with LPS or vehicle as opposed to the culture format in droplets or in plates (Fig. 3e). Differentially expressed genes from LPS treatment compared with vehicle drove 29.6% of the overall variance, whereas the genes from droplet versus plate culture contributed to only 16.4% of variance, suggesting that cultivation of microglia in droplets probably recapitulates in vitro phenotypes.

The primary readout from SPEAC-seq is gRNA enrichment in the droplet populations defined by fluorescent reporter cell activation (Step 57). Thus, we initially calculated fold-change enrichment in gRNA detection in GFP⁺ versus GFP⁻ droplets using MAGeCK⁴⁶ (Fig. 3f) (Step 57). To further validate the identification of true regulatory targets in microglia that control NF-κB activation in astrocytes, we filtered gRNA candidates by their expression in RNA-seq of LPS-treated microglia (Fig. 3g) (Step 57). Finally, although SPEAC-seq functions to screen for potential regulators of cell–cell communication, further experimental validation of candidate regulators at the biological level is ultimately required to mechanistically elucidate communication pathways⁶.

BOX 1. Considerations for calculating experimental scale.

To successfully scale up SPEAC-seq for the investigation of large CRISPR screening libraries, it is important to consider the number of input cells needed to successfully analyze a sufficient number of perturbed cell pairs. We provide our example calculations here. Since all aqueous volume forms a droplet surrounded by oil, we can control the average rate of cell encapsulation λ by changing the input cell density. This is given by

$$\frac{\lambda }{V_{\text{drop}}}=\frac{\text{cells}}{V_{\text{Aqueous}\text{input}}}$$

For instance, to encapsulate an average of λ = 0.1 cells per droplet in droplets of 65 μm diameter (~143 pL volume), we can use a cell suspension of ~6.95 × 10⁵ cells/mL. However, because two aqueous cell suspensions are used as input at equal flow rates, the cell concentrations need to be doubled accordingly to 1.39 × 10⁶ cells/mL. For our application, we decided to encapsulate the CRISPR cell type at 0.1 cells per drop (1.39 × 10⁶ cells/mL) and the reporter cell type at 0.5 cells per drop (6.95 × 10⁶ cells/mL). Poisson loading estimates that 2.74% of all droplets will contain a cell pair with exactly one CRISPR cell and one reporter cell (Fig. 2d). Considering our CRISPR library size of ~80,000 gRNAs, the number of droplets needed to encapsulate one of each gRNAs in a paired droplet equals 80,000/0.0274 = 2.92 × 10⁶ droplets, and these will on average contain 2.92 × 10⁵ CRISPR cells in total. Therefore, considering the library size and the co-encapsulation probabilities we can calculate the time it takes to capture one complete library; at cell suspension flow rates of 600 μL/h, it would take 21 min to flow 2.92 × 10⁵ cells of a 1.39 × 10⁶ cells/mL CRISPR cell suspension and encapsulate one full library. For your particular application, use these calculations to conservatively estimate the number of cells and droplets needed to capture a sufficient number of droplets for analysis of gRNA enrichment.

Data availability

All data are available in the primary study describing SPEAC-seq (ref. 6). Sequencing data are available in GEO under the SuperSeries accession number GSE200457.