On-chip clonal expansion and phenotyping of edited T cells

As previously described, human primary T cells were transfected with Cas9 ribonucleproteins (RNPs) targeting CXCR4, a gene encoding a surface receptor that acts as a co-receptor for HIV⁹. The RNP complex was mixed with a short ssDNA oligonucleotide HDR template designed to replace 12 nucleotides within CXCR4 (Fig. 1b) and impair cell-surface expression. We previously reported up to ~20% HDR efficiency at this locus⁹ based on deep sequencing analysis of a bulk population of edited cells. However, bulk sequencing of alleles from a cell population cannot distinguish the portion of mono- and bi-allelic knock-ins at the single-cell level. To obtain both phenotypic and genotypic data from individual edited clones, T cells were imported onto the chip for one (day 1) or 4 days (day 4) after electroporation with CXCR4 Cas9 RNPs (Fig. 2a). We assessed editing efficiency at these two time-points to identify further timeline compression options. After loading, flow was stopped to keep cells immobile within the main channel, which distributes media to multiple NanoPens (up to 3500/chip) through diffusion. Single cells were automatically selected and trapped into light cages that enable single-cell positioning within the NanoPens, in 17 out of the 18 fields of view that are visualized on the chip (Fig. 2a–c, Supplementary Movie 1). Non-penned cells remaining within the channel were flushed out of the chip. Importantly, we performed a second import with T cells electroporated with RNPs containing a scrambled control gRNA that does not target any locus in the human genome, positioning them in the remaining field of view (Fig. 2b, c, Supplementary Movie 2). After 3 days of culture, during which fresh media was perfused into the main channel, we assessed on-chip clonal expansion. We first identified the pens that were initially loaded with single cells (to ensure clonality) and counted the number of pens that contained >6 cells after 3 days of culture. We established that, across multiple chips, approximately 15 ± 6% of single cells loaded at day 1 (Fig. 2d, blue circles), or 40 ± 9% of single cells loaded at day 4 (Fig. 2d, orange circles) formed a colony. The size of the individual colonies was heterogeneous (Fig. 2b, solid circles). The average doubling time was about 24 ± 2 h (day 1) and 18 ± 2 h (day 4) over 3 days of growth (Supplementary Fig. 1A). The lower off-chip clonal expansion (OCCE) and the longer doubling time of cells loaded at day 1 are likely due to higher cell death caused by the introduction of the RNP complex and the ssDNA in the first 48 h after electroporation. These data strongly suggest that manipulation by OEP does not impair cell viability, and that diffusion of nutrients from the channel to the NanoPens maintains cell growth at expected levels. Importantly, we used NanoPens that were initially empty to track putative on-chip cross-contamination (cell transferred from one pen to another). Fewer than 2% of initially empty pens acquired visible cells within the 3 days of culture in both conditions, indicating >98% on-chip clonality (Supplementary Fig. 1B). This rare cross-contamination that was observed might be explained by the high motility of activated T cells.

Open in a separate window

Fig. 2

On-chip clone expansion, identification, and selection. a Electroporated cells located in the channel are automatically identified and singles are captured within light cages (white squares) and positioned into NanoPens, then cultured for 3 days. Bar = 50 μm. b Schematic representation of OCCE and CXCR4 staining as a function of on-chip positioning. Each circle represents a single colony within a NanoPen, identified by XY coordinates. The diameter of the circle is proportional to the colony size. Clones positive for CXCR4 are depicted as magenta circles. The heavy box indicates the field of view reserved for control cells. c Composite image of the entire chip in the TxRed channel, showing CXCR4 staining. Eighteen fields of view were assembled together. White rectangle shows control cells loaded within a single field of view. The other 17 fields of view contain clones electroporated with CXCR4 Cas9 RNPs. Bar = 3.3 mm. d Dot plot of OCCE in chips loaded 1 (blue circles) or 4 (orange circles) days after electroporation. Bars represent mean ± SD of OCCE in 9 or 7 chips per condition, respectively (>300 clones per chip). e, f On-chip phenotype assessment with fluorescent anti-CXCR4 antibody. Left panel, control cells. Right panel, edited candidates, negative for CXCR4 staining. g Quantification of CXCR4 staining for control and edited cells loaded 1 (blue) or 4 (orange) days after electroporation. Bars indicate mean ± SD of CXCR4 staining in seven chips per condition (>300 clones analyzed per chip). h, i Representative images of the split export process: during Culture Export, the first half of the colony is unpenned using OEP (white bars), pushed into the channel, and exported into a well of a 96-well plate. Remaining cells are pushed back into the NanoPen to ensure clonality. For sequencing export, the second half of the colony is exported into a second 96-well plate for on-target sequencing. Numbers in the upper portion of the panels indicate the duration of a single export (in minutes). Bar = 50 μm

Next, we established an on-chip phenotypic assay to identify clones that had undergone successful CXCR4 editing. Fluorescently labeled anti-CXCR4 antibody was imported into the chip, and media flow was interrupted to allow diffusion of the antibody into the NanoPens. After 45 min of incubation, the chip was continuously flushed for 30 min with fresh media, to remove excess free antibody. Fluorescent images of the entire chip were taken (Fig. 2c, e, f) and the number of colonies positive for CXCR4 surface expression was quantified. Among the colonies formed by control cells across all chips, roughly 95% (day 1) and 85% (day 4) of clones were positive for CXCR4 (Fig. 2e, g). Strikingly, for CXCR4-edited cells loaded 1 day after electroporation, only 20% of the colonies showed presence of CXCR4 on the cell surface. In cells from healthy donors loaded 4 days post-electroporation, the number of colonies positive for CXCR4 staining dropped to around 5% (Fig. 2f, g). Importantly, each single pen was assessed for colony formation and fluorescence signal and a report was automatically generated to identify the NanoPens containing the clones of interest (Fig. 2b, c).

Cas9 RNPs electroporation

A two-component gRNA system was used- crRNAs targeting either CXCR4 or no human genomic sequence (for gRNAs sequences, see Supplementary Data 1) were synthesized (Dharmacon) and resuspended in 10 mM Tris-HCl pH 7.4 with 150 mM KCl to a final concentration of 160 µM. tracrRNA was similarly synthesized and resuspended. The crRNA and tracrRNA were mixed 1:1 by volume and incubated for 30 min at 37 °C to produce 80 µM gRNA. Forty micrometer SpCas9 (QB3 Macrolab) was added at 1:1 by volume to the gRNA (a 1:2 molar ratio of Cas9 to gRNA) and incubated for 15 min at 37 °C to yield a 20 µM RNP. RNPs were prepared immediately before electroporation into T cells. A short ssDNA HDR template (ssODN) to insert a defined 12 bp sequence into CXCR4 was chemically synthesized (IDT) and resuspended in nuclease-free H₂O at 100 µM. The same CXCR4 targeting HDR template was used for both CXCR4 and scrambled gRNAs (for ssDNA sequence, see Supplementary Data 1). Two days following stimulation, T cells were harvested and resuspended in P3 electroporation buffer (Lonza) at a concentration of 1 million cells per 20 µL of buffer. Five microliter of RNP (100 pmols) were added to 20 µL of cells (1 million T cells) along with 1 µL of HDR template (100 pmols) were mixed and electroporated in a single well of a Lonza nucleofection cuvette on a Lonza Nucleofector 4D X-Unit device using pulse program EH-115. Immediately following electroporation, 80 µL of pre-warmed culture media were added directly to the cuvette and the cells were allowed to rest in a 5% CO₂ 37 °C incubator for 15 min in the cuvettes before being stimulated and transferred out for further culture (see human T-cell isolation and culture).

Preparation of cell suspension for penning in Optoselect chip

T cells were cultured for 1 day or 4 days after electroporation in culture media [RPMI–1640 (Gibco) supplemented with 2 mmol L^–1 Glutamax (Gibco), 10% (vol/vol) FBS (Seradigm), 2% Human AB serum (ZenBio) and 50 IU mL^–1 IL-2 (R&D Systems), in the presence of Anti-CD3/CD28 Dynabeads (Gibco)]. Prior to loading onto the chip, cells were resuspended in culture media supplemented with 10 ng mL^–1 IL-7 and IL-15 (PeProTech) at a final density of 5e6 cells mL^–1.

Conditions for automated cell penning

Experiments were conducted on Berkeley Lights, Inc. platforms and Optoselect chips. After priming, chips were washed twice with de-ionized water and flushed six times with culture media. Cells were imported onto the chips and loaded as single cells into NanoPens using OEP with the following parameters—nominal voltage: 4.5 V; frequency: 1000 kHz; cage shape: square; cage speed: 8 µm s^–1; cage line width: 10 µm. Loading temperature was set to 36 °C. Brightfield images of each chip were acquired automatically at the end of the loading process and a BerkeleyLights, Inc. proprietary algorithm was used to detect and count cells by pen.

Culturing conditions and cell expansion quantification

Chips were maintained at a temperature of 36 °C during culture. CO₂-buffered culture media was perfused through the chip at a flow rate of 0.01 µL s^–1. For primary cell growth assessment and automated counting, Brightfield images of the chips were taken at distinct time points to quantify on-chip clonal expansion (OCCE), defined as the percentage of NanoPens containing a single-cell that grew into a colony of six or more cells after 72 h of culture. Cross-contamination across each chip was determined as the percentage of initially empty pens that acquired cells during culture.

On-chip T-cell staining

Cell surface staining was performed with αCXCR4–PE (12G5; BioLegend). The antibody was imported into the chip at 1:250 dilution in culture media and incubated for 45 min at 36 °C. After staining, chips were perfused for 30 min with culture media media, to remove the excess antibody, and then images were acquired in Brightfield (25 ms) and Texas Red (1000 ms) channels.

Split export of edited clones

Three to 4 days after loading, clones containing >10 cells that showed negative staining for CXCR4 were sequentially exported for off-chip culturing and genotyping. Forty-eight clones and 48 blanks were exported per chip. In the first step of the split export (culturing export), roughly half of each clone (5–20 cells) was transferred from the NanoPen to the channel using light bars generated by OEP, with the following parameters—nominal voltage: 4.5 V; frequency: 1000 kHz; bar speed: 5 µm s^–1; bar line width: 10 µm. Export temperature was set to 36 °C, export was performed in culture media and cells were flushed in a 20 µL package volume into a barcoded round-bottom, tissue culture treated 96-well plate containing 100 µL of culture media supplemented with 10 ng mL^–1 IL-7 and IL-15 per well. The plate was kept in an incubator at 36 °C and 5% CO₂ for the entire duration of the export. After 7 days of culture, images of the plates in the Brightfiled channel were collected with Cytation 3 Cell Imaging Multi-Mode Reader (BioTek), and wells where cells grew into sizeable clones were counted.

For the second step of the split export (genotyping export), culture media was replaced with export buffer [PBS (Gibco), 5 mg mL^–1 BSA (Fisher Scientific), 0.1% Pluronic F-127 (Life Tech)] by flushing the chip ten times before starting the export. Then, the remaining cells from the previously exported pens were transferred to the channel by OEP with the following parameters—nominal voltage: 5 V; frequency: 1000 kHz; bar speed: 5 µm s^–1; bar line width: 10 µm. Export temperature was set to 36 °C, export was performed in export buffer and cells were flushed in a 5 µL package volume into a barcoded 96-well PCR plate (Eppendorf) containing 20 µL of mineral oil (Sigma–Aldrich) and 5 µL of Proteinase K buffer [(10 mM Tris-HCl pH 8, 0.1 M NaCl, 1 mM EDTA, 200 µg mL^–1 proteinase K (Ambion AM2546)] per well. The PCR plate was maintained at 4 °C for the entire duration of the export.

Sample processing for next-generation sequencing

Genomic DNA was extracted from exported clones by incubating in Proteinase K buffer (0.1 M NaCl, 10 mM Tris-HCl pH 8.0, 1 mM EDTA) for 30 min at 55 °C, then for 20 min at 80 °C to inactivate Proteinase K. The genomic region around the CRISPR/Cas9 target site for CXCR4 gene was amplified by PCR with primers positioned outside of the HDR repair template sequence (positioned to avoid amplification of exogenous template) for ten cycles using KAPA HiFi Hotstart ReadyMix (Kapa Biosystems, KR0370) according to the manufacturer’s protocol (PCR primers listed in Supplementary Data 1). Primers contained inline sample-specific barcodes. Barcoded samples from each plate were pooled to concentrate and remove mineral oil using Zymo DNA Clean and Concentrator Column (Zymo research, D4004). Excess PCR primers were removed by incubating with Exonuclease I (NEB, M0293S) in 1 × exonuclease reaction buffer (NEB, B0293S) for 1 h at 37 °C, followed by enzyme inactivation for 20 min at 80 °C. Amplicon pools were re-amplified by PCR for 15 cycles using a universal primer to add the sequencing adaptor and secondary barcodes to allow parallel sequencing of multiple amplicon pools (Primer sequences are listed in Supplementary Data 1). PCR products of the expected size were isolated with Select-A-Size DNA Clean and Concentrator (Zymo research, D4080) as sequencing libraries. Pooled barcoded libraries were sequenced with 300 bp paired-end reads on a MiSeq (Illumina) instrument using the 300 cycles v3 reagent kit (Illumina).

Sequencing data analysis and HDR/indel identification

All computational and statistical analysis were performed using Python 2.7 and Unix-based software tools. Quality of paired-end sequencing reads (R1 and R2 fastq files) was assessed using FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc). Reads with sample-specific inline barcodes were demultiplexed using our home-brew python script for FASTQ files splitting. Reads were then mapped on both the wild-type sequence and the expected HDR-edited sequence of CXCR4 using bwa version 0.7.15¹⁷ with default parameters. Alignments files were sorted and indexed using samtools version 1.3.1^18,19. Variants were called using freebayes version 1.0.2²⁰, a Bayesian haplotype-based polymorphism discovery tool. Genotypes were determined for each colony based on the number of reads matching either the wild-type sequence, the HDR sequence or containing variants to these two sequences with a quality above 30. Python scripts implementing the demultiplexing, alignment, and genotyping are available from the authors upon request.

Optoselect technology and chip overview

The OptoSelect^TM platform takes advantage of the OptoElectroPositioning (OEP^TM) technology, which enables light-controlled cell manipulation. OEP is enabled by the generation of a dielectrophoretic force (DEP), which occurs when a polarizable particle is suspended in a non-uniform electric field.

The proprietary OptoSelect nanofluidic device consists of a top transparent electrode and a bottom silicon substrate with a fluidic chamber in between. The substrate is fabricated with an array of photosensitive transistors. When light shines on the transistors, and if voltage is applied, a non-uniform electric field is locally generated in the fluidic channel. This imparts a negative DEP force that repels particles (including cells) using light-induced OEP. In the absence of targeted light, no force is generated; when light is shined on the photoconductive material, DEP force is generated and particles trapped inside light “cages” can be moved across the chamber. The chip contains a main fluidic channel and 3500 individual NanoPens chambers, which hold a 0.5 nL volume each. Media is perfused through the channel by a fluidic system, bringing nutrients to the NanoPens and carrying away waste, with the movement of nutrients and waste between the channel and NanoPens occurring via diffusion. Cells are loaded into the channel through an import needle, from a sample tube or from a well plate, and using light cages they are moved into the NanoPens at a speed of 5–15 µm s^–1 (Fig. 1a). Single cells loaded into pens are isolated from each other, and perfusion of CO2-buffered media through the chip during culturing at 36 °C enables the in-pen expansion of clones over time. The chip is placed on a 3-axis robotic stage and an upright microscope mounted on top of the stage allows image collection of the entire chip area at 4 × or 10 × magnification in brightfield and fluorescent channels, to monitor cell growth, morphology, and to perform phenotypical analyses. After characterization, selected clones can be exported off the chip for further processing. The export is the reverse of the import process, where desired cells are moved using OEP from single NanoPens into the main channel and flushed into a target well of a 96-well plate positioned inside a CO2- and temperature-controlled incubator.

The imaging system can provide both brightfield and fluorescent imaging. A 360 nm LED source is used to illuminate the background and a 400–700 nm white light lamp combined with a digital micromirror device (DMD) is used to structure light in desired patterns for light actuated dielectrophoresis (DEP). The system uses an upright microscope with an automated lens changer to image at 4 × and 10 × magnifications and a linear cube slider to collect fluorescent images in wavelengths corresponding to Cy5, FITC, and TxRED fluorescent channels. More information are available at https://www.berkeleylights.com/contact–us/.

We thank Berkeley Lights employees Yelena Bronevetsky and Natalie Marks for help with the establishment of T-cell culturing conditions on chip. We thank Vincent Pai and Kai Szeto for help with optimization of the split export process. We thank members of Marson lab. Al.M. holds a Career Award for Medical Scientists from the Burroughs Wellcome Fund and is a Chan Zuckerberg Biohub Investigator. We relied on the Flow Cytometry Core at UCSF, supported by the Diabetes Research Center grant NIH P30 DK063720. T.L.R. and J.H. were supported by the UCSF Medical Scientist Training Program (T32GM007618). T.L.R. was supported by the UCSF Endocrinology Training Grant (T32 DK007418).