Nuclei multiplexing with barcoded antibodies for single-nucleus genomics

Single-nucleus RNA-seq (snRNA-seq) enables the interrogation of cellular states in complex tissues that are challenging to dissociate or are frozen, and opens the way to human genetics studies, clinical trials, and precise cell atlases of large organs. However, such applications are currently limited by batch effects, processing, and costs. Here, we present an approach for multiplexing snRNA-seq, using sample-barcoded antibodies to uniquely label nuclei from distinct samples. Comparing human brain cortex samples profiled with or without hashing antibodies, we demonstrate that nucleus hashing does not significantly alter recovered profiles. We develop DemuxEM, a computational tool that detects inter-sample multiplets and assigns singlets to their sample of origin, and validate its accuracy using sex-specific gene expression, species-mixing and natural genetic variation. Our approach will facilitate tissue atlases of isogenic model organisms or from multiple biopsies or longitudinal samples of one donor, and large-scale perturbation screens.


Introduction
Single-nucleus RNA-seq (snRNA-Seq) has become an instrumental method for interrogating cell types, states, and function in complex tissues that cannot easily be dissociated (1-3). This includes tissues rich in cell types such as neurons, adipocytes and skeletal muscle cells, archived frozen clinical materials, and tissues that must be frozen to register into specific coordinates. Moreover, the ability to handle minute frozen specimens (4) has made snRNA-seq a compelling option for large scale studies from tissue atlases (5,6), to longitudinal clinical trials and human genetics. However, to maximize the success of such studies there is a crucial need to minimize batch effects, reduce costs, and streamline the preparation of large numbers of samples.
For single cell analysis, these goals have recently been elegantly achieved by multiplexing samples prior to processing, which are barcoded either through natural genetic variation (7), chemical labeling (8,9) or DNA-tagged antibodies ("cell hashing") (10). These methods have improved technical inter-sample variability by early pooling, lower the cost per sample by overloading cells per microfluidic run -due to an increased ability to detect and discard coencapsulated "cell multiplets" sharing the same bead barcode -and reduce the number of parallel processing steps in large studies.
Here, we follow on these studies by developing a sample multiplexing method for nuclei ("nucleus hashing"), using DNAbarcoded antibodies targeting the nuclear pore complex. Unlike methods leveraging natural genetic variation (7), barcoded antibodies allow pooling of isogenic samples, such as from isogenic mouse models, multiple specimens from the same human donor, or tissues sampled and preserved from a given donor over time.

Results
We isolated nuclei from fresh-frozen murine or human cortical tissues, stained them with antibodies carrying a sample-specific DNA barcode, and pooled samples prior to droplet encapsulation for single-nucleus RNA-Seq   Figure 1a), and achieved a similar number of genes expressed per nucleus as the non-hashed control (Supplementary Figure 1b), whereas a PBS based buffer (used in cell hashing (10)   Next, we validated our hashtag based demultiplexing with Demuxlet (7), an approach based on natural genetic variation.
We observed excellent agreement between the two methods for the 8 human cortex

Data availability
All mouse data will be available from the    visualization on the filtered count matrix as previously described (16,17). Specifically, we selected highly variable genes as previously described (18)  $]\^≥ 0.1a |. If this number is 1, the droplet is a singlet.

Materials and Methods
Otherwise, it is a multiplet.

Generation of the "species-mixing" plot.
We only used RNA UMIs that were confirmed by at least 2 reads to generate the "species-mixing" plot. Requiring at least 2 reads to confirm a RNA UMI helps filter potentially erroneous RNA UMIs produced from PCR and sequencing errors.

Tissue lysis and homogenizing
Nuclei were extracted as previously Proceed with standard 10X protocol for cDNA sequencing library preparation.

Library preparation for mRNA-derived cDNA <300bp (supernatant fraction)
Purify Hashtags using two 2X SPRI purifications per manufacturer protocol: • Add 1.4X SPRI to supernatant to obtain a final SPRI volume of 2X SPRI.
• Transfer entire volume into a low-bind 1.5mL tube.
• Incubate 10 minutes at room temperature.
• Place tube on magnet and wait ~2 minutes until solution is clear.
• Carefully remove and discard the supernatant.
• Add 400 µl 80% ethanol to the tube without disturbing the pellet and stand for 30 seconds (only one ethanol wash).
• Carefully remove and discard the ethanol wash.
• Centrifuge tube briefly and return it to magnet.
• Remove and discard any remaining ethanol.
• Perform another round of 2X SPRI purification by adding 100 µl SPRI reagent directly onto resuspended beads.
• Mix by pipetting, and incubate 10 minutes at room temperature.
• Place tube on magnet and wait ~2 minutes until solution is clear.
• Carefully remove and discard the supernatant.
• Add 200 µl 80% ethanol to the tube without disturbing the pellet and let stand for 30 seconds (first Ethanol wash).
• Carefully remove and discard the ethanol wash.
• Add 200 µl 80% ethanol to the tube without disturbing the pellet and let stand for 30 seconds (second Ethanol wash).
• Carefully remove and discard the ethanol wash.
• Centrifuge tube briefly and return it to magnet.
• Remove and discard any remaining ethanol and allow the beads to air dry for 2 minutes (do not over-dry beads).
• Mix vigorously by pipetting and incubate at room temperature for 5 minutes.
• Place tube on magnet and transfer clear supernatant into PCR well. • Incubate 5 minutes at room temperature.
• Place tube on magnet and wait 1 minute until solution is clear.
• Carefully remove and discard the supernatant.
• Add 200 µl 80% ethanol to the tube without disturbing the pellet and let stand for 30 seconds (first ethanol wash).
• Carefully remove and discard the ethanol wash.
• Add 200 µl 80% ethanol to the tube without disturbing the pellet and let stand for 30 seconds (second ethanol wash).
• Carefully remove and discard the ethanol wash.
• Centrifuge tube briefly and return it to magnet.
• Remove and discard any remaining ethanol and allow the beads to air dry for 2 minutes.
• Pipette mix vigorously and incubate at room temperature for 5 minutes.
• Place tube on magnet and transfer clear supernatant to PCR tube.

Quantify library
Quantify library by standard methods (QuBit, BioAnalyzer