CyTOF

In 1994 Tsutomu Nomizu and colleagues at Nagoya University performed the first mass spectrometry experiments of single cells. Nomizu realized that single cells could be nebulized, dried, and ignited in plasma to generate clouds of ions which could be detected by emission spectrometry.^[7] In this type of experiment elements such as calcium within the cell could be quantified. Inspired by Flow cytometry, in 2007 Scott D. Tanner built upon this ICP-MS with the first multiplexed assay using lanthanide metals to label DNA and cell surface markers.^[8] In 2008 Tanner described the tandem attachment of a flow cytometer to an ICP-MS instrument as well as new antibody tags that would allow massively multiplexed analysis of cell markers.^[9] By further optimizing the detection speed and sensitivity of this flow coupled to ICP-MS they built the first CyTOF instrument.^[1]

The CyTOF instrument was originally owned by the Canadian company DVS Sciences but is now the exclusive product of Fluidigm after their acquisition in 2014 of DVS sciences. In 2022 Fluidigm received a capital infusion and changed its name to Standard BioTools.^[10] There have been 4 iterations of the CyTOF apparatus named CyTOF, CyTOF2, Helios™^[5] and CyTOF XT.^[11] The successive improvements were largely in increased detection range and software parameters with the Helios instrument able to detect from metals ranging from yttrium-89 to bismuth-209 and throughput and analyze 2000 events per minute.

The Lanthanide group of elements are used for tagging antibodies, as the background in biological samples is very low.^[12] When choosing the appropriate isotope for the biomarker, low expression biomarkers should be paired with an isotope that has high signal intensity.^[13] If a less pure isotope must be used, it should be paired with a low expression biomarker, to minimize any non specific binding or background.

Isotope polymers are constructed using diethylenetriaminepentaacetic acid (DTPA) chelator to bind ions together.^[13] The polymer terminates with a thiol or a maleimide that links it to reduced disulfides in the Fc region of the antibody.^[14] Four to five polymers are bound to an antibody, resulting in about 100 isotope atoms per antibody.^[14] Tagged antibodies may be in solution, conjugated to beads, or surface immobilized. The cell staining follows the same procedures as in fluorescent staining for flow cytometry.^[14]

To distinguish between live and dead cells, cells can be probed with rhodium, an intercalator which can only penetrate dead cells. Then all cells are fixed and stained with iridium, which penetrates all cells, to be able to visualize which are alive.^[14]

The cell introduction method of the mass cytometer is an aerosol splitter injection.^[13] The cells are then captured in a stream of argon gas, then transported to the plasma where they are vaporized, atomized, and ionized. The cell is now a cloud of ions, which passes into the ion optics center. Then a time of flight analyzer is used to measure the mass of the ions.

Ions are accelerated through the spectrometer in pulses. The electron cloud generated from a single cell is typically 10-150 pulses. The output of a Helios™ run is a binary integrated mass data (IMD) file that contains electron intensities measured from the ions for each mass channel. The continuous pulses must be resolved into individual cell events corresponding to the ion cloud generated from one cell. Each bin of between 10-150 pulses that passes the user set lower convolution threshold, is considered a cell event by the Helios™ software.^[1][5] The lower convolution threshold is the minimal ion count that must be reached across all ion channels to be considered a cell event. The value for this parameter increases with the number of ions being measured and thus more counts are required to define a cell event when more labels are used.^[5]  

For data analysis, the IMD file is converted into the flow cytometry standard (FCS) format. This file contains the total ion counts for each channel for every cell arranged in a matrix and is the same file generated during flow cytometry.^[5] Manual gating of this data can be performed as is done for flow cytometry and most of the tools available for flow cytometry analysis have been ported to CyTOF (See flow cytometry bioinformatics).^[6] CyTOF data is typically high dimensional. To delineate relationships between cell populations dimensionality reduction algorithms are often used. Several multidimensional analysis clustering algorithms are common. Popular tools include tSNE, FlowSOM, and the diffusion pseudo time (DPT).^[6] The downstream analysis methods depend on the research goals.