Unparalleled high dimensional analysis of cells

Mass cytometry (and the Helios in particular) is primarily an extension of, or complement to, traditional fluorescence-based flow cytometry analysis. It allows for more parameters to be measured than current flow cytometry techniques.If you’re familiar with flow cytometry techniques and analysis you’ll definitely have a head-start in mass cytometry - the approaches and techniques are very similar. If you have no flow cytometry experience, don’t worry, we’re here to guide you and make your projects a success.

Mass cytometry provides a high-resolution proteomic profile at the single cell level. What does that actually mean? We can measure 40-50 different proteins at on every cell in your samples (hundreds of thousands or even millions of cells). Mass cytometry is a single-cell technique like RNA-seq but because we are measuring proteins via antibodies we can also measure the cell signalling or function (by using phospho-specific antibodies). More and more researchers are integrating mass cytometry into their single-cell assays and pipe-lines to yield more information and combine their data with sequencing and other assays. Mass cytometry gives you a more comprehensive functional and phenotypic characterization of complex systems with unprecedented resolution and a strong complement to your other single cell assays or flow cytometry approaches.

Overall, mass cytometry gives you:

• The ability to detect and quantify more than 40 user-defined markers.
• Additional channels set aside for functional markers like live-dead discrimination, bar-coding, DNA and thymidine analogues. In flow cytometry you have to use standard channels for these markers. In mass cytometry they come for free leaving you a true 40 channels for the markers you are interested in.
• Simultaneous phenotypic and functional assays lead to substantial improvement in the understanding of the cell states and functions - phenotype cells and interrogate signalling pathways at the same time by using phospho-specific antibodies or histone marks.
• Remove the limitations of spectral overlap of fluorescence and the dysfunction of fluorophores inside cells to allow targeting of more intracellular proteins.
• Full signature identification for each intact cell prevents the need to cross-correlate - combine your current separate smaller panels into one large panel.

Throughput

Compared to flow cytometry, mass cytometry is very slow:

• 30ul/min maximum throughput
• 300-500 events/sec optimal recording rate

The fastest rate we can feed sample into the instrument is 30ul/min, which is quite slow. This shouldn’t be a problem as we are still free to increase the cell concentration to make up for the limited volume/min. However, studies have shown that running samples faster than >500 cells/sec causes large increases in doublet formation with very small increases in single cell acquisition. For this reason we do not run the instrument at >500cells/sec and often prefer to run at 300 cells/sec if compatible with your required acquisitions.

No scatter measurements

In flow cytometry we can use light scatter parameters to enhance our ability to discriminate cell types. We don’t have these parameters available in mass cytometry and the analogues we do use are not sensitive enough to discriminate cell types in the same way. If scatter parameters are vital to your ability to discriminate cell types and we can’t find a probe to use in mass cytometry, this technique may not be viable for you.

“Lossi-ness”

The protocols required for mass cytometry result in large amounts of sample loss. There are 2 steps where sample loss occurs:

• Staining techniques are very harsh
  • strong fixation and permeabilization reagents
  • Samples centrifuged upwards of 12 times
• Acquisition
  • Cells are nebulized into a spray chamber before introduction by vacuum into the instrument for analysis. 30-50% of the nebulized sample adheres to the spray chamber walls before it can be drawn into the detector array.

Cost

The cost of doing experiments in mass and flow cytometry are comparable but: * For flow you probably have reagents in your fridge to build larger panels around. If this is your first mass cytometry experiment you will have to buy every probe and reagent. * The large increase in available channels means that you will gain large increases in information per cell but also means that you will be purchasing 30-40 antibodies for mass cytometry compared to the <20 you will need for the largest flow cytometry experiment. Buying 30-40 antibodies rather than 20 is 1.5-2x the price but should provide more than double the information per cell.

Acquisition costs are very low compared to reagent costs (they are comparable or sightly more than the he cost of cell sorting).

Fixation

Samples must be fixed before introduction into the Helios instrument. If your studies require live cells, for instance if you are studying calcium flux or changes in a reporter protein, fluorescent approaches may be better for you.

Sample Destruction

All cells undergoing mass cytometry analysis are completely destroyed (ionized by plasma) so there is no possible way to use the analysed cells for downstream applications.

Identifying appropriate reagents

With optimisation, we can get any probe working for your mass cytometry study but there is a definite hierarchy of preferences when it comes to sourcing reagents

It’s always better to use antibodies that have previously been used in mass cytometry studies or, if that’s not possible, at least in flow cytometry studies - we KNOW they work in the types of application we will be applying them to. Furthermore, monoclonal antibodies are preferred for these studies. Monoclonal antibodies conjugate more efficiently and generally behave better in the harsh protocols required to label cells and prepare them for CyToF analysis.

However, it’s not always possible for novel studies and we can make antibodies that have been used for Western or immune-blotting work. We have successfully labelled polyclonal antibodies so all is not lost if there is no monoclonal available. Basically, we can get anything to work but validated mAbs will save you time and tears in the long run as they will get you up and running more quickly and more easily and work better from lot to lot over longer projects.

There are some other things that you should bear in mind when purchasing antibodies for mass cytometry from different vendors:

• Vendors are staring to provide mass cytometry-ready formulations (look for CyToF-ready or mass cytometry ready)
• If there are no mass cytometry-specific formulations, look for in-vivo ready products (like LEAF from Biolegend or NA-LE from BectonDickinson)
• Like with other applications there are good vendors and bad vendors for antibodies. We have had good results with antibodies from RnD, BD and Biolegend, whilst we cannot recommend reagents from Sana Cruz.
• __Regardless of your vendor there are specific things you should look for when sourcing reagents:
  • NO BSA - this severely inhibits conjugation reactions
  • NO Carrier Protein - sometimes vendors substitute other carrier proteins instead of BSA, these can also inhibit the conjugation reactions
  • Glycerol, Glycerin and Azide in small concentrations are okay
  • If the only formulation you can source contains BSA, abcam have a BSA removal kit or Thermo have an antibody clean-up kit.

Which channels are available to assign antibodies to

There are approximately 44 user assignable channels available across the 100+ channels on the instrument

Lanthanides are available in the mass ranges 89 and 141-176 (excluding 157; it exists but is prohibitively expensive to use due to its rarity), Bismuth can be used in channel 209 and Fluidigm have recently released 7 Cadmium isotopes in mass ranges between 106 and 116.

and which channels have the highest sensitivity

With this information we can generate a panel using:

The 2 basic rules of panel design

Assigning antibodies to labels (metals or fluorophores)

Like in flow cytometry we measure our labelled probes in specific “channels”. For flow cytometry each channel is a laser/filter/fluorochrome/PMT combination. In mass cytometry we have mass ranges from 141-176.

Like in flow each of these channels are not created equally.

In flow, for instance, we have brighter or more efficient fluorochromes such as BV421, PE or APC. We assign the probes that are most interesting to us or whose marker is expressed the least to the higher sensitivity channels, which gives us the best opportunity to pick up differences in low-expressed markers or those we think might show differences we are interested in the best.

In mass cytometry there is a similar principle at work. We don’t have differences in our metal labels - we are just measuring mass which is invariate. However the machine is tuned to be more sensitive in detecting those masses in a specific range see the figures above).

For the Helios, the area of the mass range that is most sensitive is around 153-165. These channels should be used for our most important or lowest-expressed antigens of interest . Furthermore, there are also differences at either end of the mass range.

Around 140 is the area of least sensitive detection for lanthanides. The area around 170+ is lower sensitivity than around 160 but better than at 140. Additionally, we now have the cadmium metal conjugates to worry about. The masses of the cadmium isotopes are 106-116. As you can imagine, if the mass range around 140 is low sensitivity, the mass range 30 lower is very low sensitivity. If we are to use the cadmium isotopes in a very large panel we must assign very highly expressed, very easy to distinguish positive-from-negative markers (think things like CD4 or CD8) or it won’t be worth using these probe-metal combinations at all!

Important Notes about Sample Contamination

When performing flow cytometry experiments we take certain precautions to improve how samples are interpreted on the machine - most notable of these is not to expose our stained samples to too much light so that we don’t damage or photobleach our fluorophores.

In mass cytometry their are similar things me MUST do to protect our samples from contamination that can affect how samples are analysed. Most of the metals we use as probes do not exist in nature, which makes contamination almost impossible BUT their are at least 3 contaminants that can be measured by the mass cytometer that can make sample acquisition almost impossible:

• Lead - Lead is used extensively in glassware manufacture and can leech into any samples. You should not use any glass when preparing samples for mass cytometry.
• Barium - Barium is a component or contaminant of dish washing solutions and powders. You should not use any lab-ware that has been “glass-washed”
• Iodine - Iodine is a major component in cell separation mediums and centrifuge gradients such as lympho-paque. If you must use these media, be very conservative when removing the cell layer as not to take up any of the media in you sample preparations.

In mass cytometry their are similar precautions we must take to minimize contamination of our samples

Single cell suspension Preparation

As with flow cytometry this can be the most important aspect of sample preparation. Finding an optimal method to generate single cell suspension from your culture or tissue will require substantial optimization to reduce losses through cell death and reduce the amount of debris. Things to consider include: * Accutase instead of trypsin * GentleMACS to disrupt tissues (increases yield and reduces cell death * Debris and/or dead cell removal kits

Surface Staining

Surface staining should closely replicate your flow cytometry protocols (if you have them!) - you may already know what works well from these applications. If you haven’t done any flow cytometry at all or on this particular cell-type or tissue, no problem; we can follow best practises:

• It’s often best to do the surface stain on live cells before any fixation occurs. Most of the CD antibodies we use were developed to stain cells for flow cytometry and were often developed for sort applications, using live cells that could be cultured or analysed downstream.
• Are you bar-coding your samples? IF yes you may already have fixed your cells at this stage so the question of fixation is already answered.

Fixation

Fixation is absolutely necessary in mass cytometry samples, regardless of the type of staining undertaken - mass cytometry samples are acquired in pure or near pure ddH₂0. If you are only surface staining you could fix at the end but otherwise fixation early on in your staining protocol will happen. I generally recommend doing surface staining on live cells and then fixing after. The cells are then ready for permeabilization and further staining, processing or acquisition.

Crosslinking fixatives

The primary crosslinking fixative used in biological assays is formaldehyde (generally 1.6%-4% w/v solutions are used).

Formaldehyde works by forming covalent bonds between amino acids, primarily lysines, causing proteins to be crosslinked into cage-like structures.

There are very niche applications where 1% gluteraldehyde might need to be used (running erythrocytes is an example of this where even with PFA fixation significant haemolysis occurs). Gluteraldehyde generally crosslinks “better” or more “rigidly” because it is a larger molecule (with 2 reactive aldehyde groups) and so can crosslink proteins over greater distances. However, the increased size means that gluteraldehyde often takes longer to get into cells and tissues so take this into account if you are thinking about using it.

In general using 2-4% PFA will work for all common applications (but be prepared to test alternative concentrations)

The major advantage of crosslinking fixatives is their action retains the secondary and (often) tertiary structure of proteins, which may be important for some conformational epitopes that antibodies react with.

Agglutination/Precipitating fixatives

This class of fixatives includes alcohols (Methanol/Ethanol) and compounds like acetone. They predominantly work by agglutinating proteins, causing their precipitation (by disrupting the hydrophobic core of proteins). They do not retain the secondary structure of proteins due to their denaturing action. This can help some antibodies and hurt others.It’s most often employed in mass cytometry protocols when we are using phospo-protein antibodies to probe signalling chains.

Saponin

Saponin is a mild detergent and the mildest permeabilization protocol that we can use. Its action of permeabilization functions by reacting with and removing cholesterol molecules from lipid membranes. This creates small “holes” in the membrane through which antibodies can pass (and small proteins can escape the cell if they are insufficiently cross-linked via your fixation protocol.

Saponin-plus

The major disadvantage of Saponin permeabilization is the inability to use it for intra-nuclear staining. Saponin works through interaction with cholesterol. Cholesterol is rare in the nuclear membrane. To enable us to stain for intra-nuclear proteins like transcription factors, we can use Saponin but add a stronger, more reactive detergent. Triton-X and Tween are popular. They are often used at concentrations of 0.1-0.5% in 0.5% Saponin buffers. Many of the proprietary buffers that are recommended for intranuclear staining have a formulation similar to that described in this section. These stronger buffers have a broader spectrum of action reacting with more lipids than cholesterol (but also some proteins)

Methanol

Methanol has been used as a way to permeabilize cells and enable staining for phosphoproteins and some transcription factors (with low background). It may be that many of the antibodies used for phosphoproteins and some transcription factors were developed for Western and immunoblotting where denaturation is typical. Therefore, staining under perm conditions that recreate denaturing conditions (Methanol) is sensible. Using very cold Methanol works well for many of these proteins and is often an essential component of a staining protocol if you are analyzing many phosphoproteins simultaneously.

Intracellular Staining

Once the fix-perm conditions have been optimized, intracellular staining is as simple as adding the correct titrations of the relevant antibodies.

Cell Acquisition Solution

Traditionally cells are acquired in ddH₂0. This leads to cell loss, even in fixed cells. To reduce losses associated with putting your samples in water, at OHRI we acquire your samples in Cell Acquisition Solution (CAS). CAS is a 0.1% salt solution (likely of ammonium nitrate) that helps stabilize the cells.

FlowJo

FlowJo has become the standard software to analyse flow cytometry data and is increasingly trying to offer techniques that allow its expansion in to mass cytometry data. FlowJo has built dimensionality reduction (via tSNE) into the base software package while the new Plug-in system allows users to utilize a small suite of packages such as FlowSOM, Phenograph and flowClean to analyse high dimensional data

Cytobank

Cytobank is a browser-based software solution that provides tSNE dimensionality reduction and FlowSOM, SPADE and CITRUS for clustering/additional analysis. being browser-based it has all the advantages (the power of your local computer has no bearing on the speed of the analysis) and disadvantages that that brings (windows can be slow to update and reaching the servers can sometimes be a challenge).

R and python

R is a programming language with a strong focus on statistics that has become a favourite platform for flow and mass cytometry analysis. It’s relatively easy to learn and provides the greatest power and flexibility to analyse mass cytometry data through the packages developed to handle mass cytometry data. Packages like flowCore allow for the importation of fcs files, which can then be analysed through the other packages in myriad ways. If you’re more familiar with Python, there are similar packages available in that language too.

Dimensionality Reduction

Dimensionality reduction is a method that “flattens” high parameter data from n dimensions down to an easier to understand 2 dimensions

Dimensionality Reduction techniques used in genomics like Principle Component Analysis can be used with some success but Cytometry data is not normally distributed and is not necessarily a vector (increased expression/decreased expression) meaning PCA is not always a good fit for analyzing mass cytometry data.

t-Distributed Stochastic Neighbour Embedding

tSNE and the associated viSNE were the original ways used by cytometrists to visualize high dimensional data in 2D. It is a powerful technique for non-linear data like fcs files. However, beware reading too much into the layouts it produces. tSNE produces even islands that can look very like clusters but there is no useful information in how the tSNE plot is laid out - you must rely on clustering algorithms to provide that information.

Uniform Manifold Approximation and Projection

A newly applied algorithm to high-dimensional cytometry data. It competes well with tSNE - it computes much faster and on larger data sets. Furthermore, because UMAP retains more of the global features throughout the algorithmic process the final 2D appearance of the map can provide some cluster-like information.

Hierarchical Stochastic Neighbour Embedding

Similar to tSNE but runs faster and with lower computational requirements - a strong update to tSNE making it more competitive against newer algorithms like UMAP and PARC.

Clustering

Dimensionality reduction techniques like tSNE or UMAP let us visualize the data in a simple way. However, we often perceive patterns in the data that are just random noise. Clustering techniques define similar groups and populations in the data in an unbiased and unsupervised way. There are many available methods to do this including:

Trajectory Analysis

Trajectory analysis, which adds probabilities of the kinetic trajectory of populations i.e. probabilities of which populations precede or are antecedent. Combined with kinetic studies this type of algorithm can be very powerful in defining developmental or kinetic studies. Useful packages include SPRING and FlowMAP.