Reporting on Fluidigm's single-cell user meeting at the Sanger Institute

The Genomics community is pushing ahead fast on single-cell analysis methods as these are revolutionising how we approach biological questions. Unfortunately my registration went in too late for the meeting running at the Sanger Institute this week (Follow #SCG16 on Twitter), but the Fluidigm pre-meeting was a great opportunity to hear what people are doing with their tech. And it should be a great opportunity to pick other users brains about their challenges with all single-cell methods.

Imaging mass-cytometry: the most exciting thing to happen in 'omics?

Mark Unger (Fluidigm VP of R&D) started the meeting off by asking the audience to consider the two axes of single-cell analysis: 1) Number of cells being analysed, 2) what questions can you ask of those cells (mRNA-seq is only one assay) - proteomics, epigenetics, SNPs, CNVs, etc.

Right now Fluidigm has the highest number of applications that can be run on single-cells with multiple Fluidigm and/or user developed protocols on the Fludigm Open App website; 10X Genomics only have single-cell 3' mRNA-seq right now, as do BioRad/Illumina and Drop-seq. But I am confident other providers will expand into non 3'mRNA assays...I'd go further and say that if they don't they'll find it hard to get traction as users are likely require a platform that can do more than one thing.

There are three sessions over the two days:

• Session I: Single-cell heterogeneity, classification and discovery
• Session II: Immunotherapy in oncology—new insights at single-cell resolution
• Session III: Single-cell functional biology 

Session I: Single-cell heterogeneity, classification and discovery

Achieve new insights through single-cell biology. Candia Brown, Director Strategic Marketing, Fluidigm

Candia asked the audience "what are we trying to do with single-cell genomics methods?" She focussed her brief introductory presentation on understand biological mechanisms and pathways, cell differentiation, cell lineage, etc, and for biomarker discovery, therapeutics...or even in the clinic in the future? Much of the initial work has been done on identifying cell types within populations and to understand heterogeneity. Moving beyond this kind of classification requires more complex methods and analyses. Ultimately we'll need to be using spatio-temporal methods such as in-situ sequencing of carefully prepared samples, and combination analyses with data from RNA, DNA and proteins. We need to detect from single cells (this was a hot topic for Fluidigm at the beginning of 2016) and Candia shoewed examples of population classification and discussed how we might move past relatively "simple" atlasing studies to more complex experiments that aim to make mechanistic insights. Fluidigm aim to present all the latest updates on their tech during this meeting for the C1, Biomark, Helios and Polaris systems.

Dissecting cerebral organoids and fetal cortex using single-cell RNA-seq. Gray Camp, a post-doc in Svante Pääbo's group at Max Planck Institute for Evolutionary Biology, Germany. Gray is also collaborating closely with the Treutlein lab.

Cerebral organoids make biological experimentation easier in the same way that tumour organoids are better informing cancer biology. The group are deconstructing cellular heterogeneity in cerebral organoids using single-cell RNA-seq compared to bulk analysis. Now using organoids developed from patients to generate samples that recapitulate periventricular neuronal heterotopia. Also following the reprogramming of fibroblasts into induced neurons (recently published in Nature and in their News and Views). This great editorial in Development discusses the impact that organoids are having on biological research.

Becoming a new neuron in the cerebral cortex. Ludovic Telley, University of Geneva, Switzerland.

Ludovic is also talking about cells in the brain, single-cell methods are having a huge impact on brain biology. His talk focussed on the "L4 neurons" the main recipient of sensory input into the brain. Using a novel technology called FlashTag to visualise and isolate neurons during their development, see Science 2016 paper. Isolated neurons are then profiled using Fluidigm single-cell RNA-seq to track neuronal transcriptional programs. They found that waves of transcriptional activity are seen as each neuron progresses from proliferative to migratory and finally to connectivity phases.

A cost-effective 5’ selective single-cell transcriptome profiling approach. Pascal Barbry, Institut de Pharmacologie Moléculaire et Cellulaire, France.

Pascal's group are using Fluidigm single-cell methods to investigate Mucocilliary differentiation. Today he describes the modified SMART-seq method they developed incorporating on-chip barcoding and UMIs. This is somewhat similar to STRT-seq published in 2011, but now on the Fluidigm IFC. Pascal spent some time describing the impact of UMIs (Unique Molecular Identifiers), showed the figure from Cellular Research's PNAS paper, and mentioned one of the four methods to reduce RNA-ligation biases. After processing cDNA is fragmented and 5' fragments are isolated by the biotin tag before completion of library prep and sequencing. Showed data on performance and reproducibility of the assay: reads are very biased to the 5' end of transcripts (but have not copared directly to CAGE data), saw about 25% efficiency of ERCC cloning, data suggest that more than 1 million reads per cell is unnecessary. Interestingly they saw a correlation of 0.9 for a C1+IonProton versus Drop-seq+Illumina, but with a reasonable number of genes that appear to be present in only one method! The script will appear on Fludigm's Open App site after publication!

Pascal briefly mentioned their work on the 800 cell IFC, they're pretty happy so far. But would like to be sequencing on Next-seq, which needs lots of PhiX to be added due to the need to read through the oligo-dT sequence. He suggested starting sequencing from the 5' end instead.

Single-cell analysis of clonal dynamics and tumour evolution in childhood ALL. Virginia Turati, Enver lab UCL, UK.

ALL is the most common childhood cancer with 1 in 2000 affected and around 500 cases per year in UK. ALL was one of the first disease where branching evolution was described. Using Fluidigm C1 single-cell in a "mouse clinic" from primary patient tumour material, where treatments can be monitored over time. Analysis during chemotherapy of PDXs shows no impact on intratumour heterogeneity i.e PDXs recapitulate the patient tumour. Single cell WGS was much more difficult than RNA-seq! But an average of 37 CNV were found in each cell. They are generating around 10 million reads per cell to generate a coverage of around 0.2x. Saw multiple variants around CDKN2a locus.

Virginia presented some data that shows how small numbers of cells (Freddy) overlap transcriptomes with resistant cells, suggesting that these are evolving towards resistance. Understanding this process is key to improving outcomes for patients. They are aiming to identify a signature of resistant cells to use in the clinic.

See more with the C1™: explore the breadth of applications available on the C1 platform for single-cell genomics. Shaun Cordes, Senior Product Manager, Fluidigm.

Shaun gave an overview of the different methods users can run on the C1 system. He also confirmed the 10,000 cells coming soon, as is a Fluidigm automated imaging system which includes a cloud based software toolkit. New applications coming include single-cell protein analysis with two anti-bodies carrying probes that allow qPCR analysis (read more about the Proximity Ligation Assay approach in the Science 2015 paper).

Session II: Immunotherapy in oncology—new insights at single-cell resolution

Mass Cytometry applications from Fluidigm. Gary Impey, Director, Product Management - Mass Cytometry, and Robert Ellis, Director, Product Management, Fluidigm.

About half the audience are either using mass-cytometry already, or are considering using it. A search on PubMed for "mass-cytometry" or "CyToF" results in 196 papers - a pretty high number given how new this method is. Gary is talking about how Fluidigm's Helios system can be used to interrogate cells for immunogenic markers. Gary referenced a Wall Street Journal article: Immunotherapy and cancers super survivors. David Lane (formerly Chief Scientist at CRUK) was quoted as saying “It’s the most exciting thing I’ve ever seen”. To get real insights we need highly-dimensional single-cell methods - Fluidigm's Helios CyToF is one tool that can help.

Fluidigm currently have 50 high-purity metal isotope tags which allow almost generation of data with minimal biological or technical noise. Metals are tagged to antibodies and these are used to tag cell surface or intra-cellular markers.

Robert is presenting an overview of a new method called imaging mass-cytometry (see the figure at the top of this post - it may be the most exciting thing to happen in 'omics in a while). This allows spatial resolution of proteomic data from tissues in-situ. The system requires a new box to be bolted onto the Helios instrument to perform imaging, a UV laser vaporises tissue by scanning across the section one line at a time (approximately 1um per pixel), and the ionised tissue goes into the mass-cytometer for semi-quantitative analysis. It works with fixed or frozen tissue on standard microscope slides. The process takes approximately 1 hour to get a region 0.5mm square - highly detailed but highly focused (spatially). Robert presented software developed in the Bodenmiller group at ETH, Zurich. You can do LCM-style selection and pick defined regions.

Robert showed some wonderful images of imaging mass-cytometry compared to IHC or FISH. Alos some lsides from David Hedley's group at Toronto. You can label your own antibodies using a kit from Fluidigm, but Robert showed a slide of their Immuno-Onc panel with a broad concentration range for different anitbodies- just how much empirical work tis required to get the balance right is unclear!

Imaging Mass Cytometry—about proteins, tissues and biomedical research. Valerie Dubost and Markus Stoeckli (also on the SAB of Imabiotech a CRO for mass-cytometry imaging), Novartis, Switzerland.

Valerie is talking about her early access results from the imaging mass-cytometry methods presented by Robert. Valeri is a histologist so her perspective will be an interesting one, and potentially give insights into how likely this technology is to make it int the clinic. Novartis haev moved quickly to build a cross-functional team to focus on mass-cytometry imaging technology application and development. Using FFZN and FFPE tissue, incubate a panel of up to 30 antibodies, slides loaded into imaging mass-cytometer for laser ablation and analysis.

Data presented included validation of the antibodies - this is critical and too many scientific papers are messed up by the use of poorly characterised antibodies. Comparison of IHC to IMC looked excellent. She showed beautiful images of cell segmentation by Voronoi boundaries. The need to carefully consider cellular architecture is important in interpretign results from IMC - you are still going to need a pathologist to help interpret this kind of data. Pathology:Molecular Pathology:IMC Pathology is going to increase our understanding of tissue architecture, and possibly interactions.

Session III: Single-cell functional biology 

An introduction to single-cell functional biology. Simon Margerison, Senior Manager, Application Support, Fluidigm 

Simon gave an overview of how the Helios and Polaris systems can be used to investigate functional single-cell biology. We heard lots about the Helios yesterday and Simon showed some Cancer data using panels where 10 markers were used for phenotpying and 30+ markers used to investigate functional biology.

However Simon spent a little more time describing the Polaris system which was not really mentioned yesterday. This is a system that allows selection of  48 single-cells, and culture them for up to 24 hours while modulating the environment  - this is automated cell culture and I'm hoping Polaris is the first of many such systems that will allow highly parametric experiments to be performed where instead of a simple A vs B, treated and untreated experiment, we'll do A,B,C,D,E,F & G, treated at different doses and times all without being messed up in the tissue culture lab.

A holistic view of the mucosal immune system: identification of tissue- and disease-specific cellular networks. Frits Koning, Leiden University Medical Center, Netherlands 

Frits is presenting work published recently in Immunity. His lab has built a mass-cytometry panel to look at heterogeneity of the adaptive and innate immune compartment, applied to Human intestinal samples (Coeliac disease). He presented data from an initial cohort of 44 patients. 8 months to generate the data, 6 months to analyse it - a common bioinformatics challenge! He showed a merge scatterplot of all 2.5 million cells from all 44 patients, the different cell types clearly separate into the canonical immume cell populations. However the different samples (PBMC vs colon) and individual patients show very different enrichments for cell populations.

They were able to distinguish distinct mass-cytometry signatures that divide patients from controls, and were able to detect patients with mucosal lymphoid malignancies. His group has been working hard on developing computational methods to analyse these huge datasets quickly, all 5.2 million cells in 1 hour on a 32Gb laptop! See the Cytosplore website for more details. Frits was very bullish about the use of mass-cytometry in the clinic and finished by saying "we are moving towards an unbiased diagnostic tool".

The nature and nurture of cell heterogeneity: single-cell functional analysis, temporal single-cell sequencing and imaging of gene edited macrophages. Esther Mellado, Wellcome Trust Centre for Human Genetics, UK.

Esther's work is the focus of a spotlight article on Fluidigm's website. She is running the Polaris system at the WTCHG and presented her work isolating single cells and perturbing them to understand the role of macrophages in HIV pathology. And in particular cells with mutations in SAMHD1 gene and the effect of this mutation on HIV latency. They used multiple microenvironmental conditions in early and late activation so adjusted dosing for either 1 or 8 hours, comparing mutant and wild-type macrophages across 10 replicates. They performed high-resolution imaging off the Polaris to investigate morphology and behaviour. They saw that knockout of SAMDH1 has important paracrine signalling effects.

The WTCHG team call the Polaris their "10 Postdocs in a Box". It allows much mire complex experiments to be performed than an individual in the lab can realistically manage. As I said above I'm hoping Polaris is the first of many automated cell culture systems - and ideally we'd see instruments that can handle bulk cells too.

Understanding cellular heterogeneity. Sarah Teichmann, Wellcome Trust Sanger Institute and EMBL-EBI, UK

Sarah is presenting her groups work on cellualr heterogeneity, it turns out that much of this is of functional significance. She stumbled upon this when doing bulk RNA-seq could not relate the abundance of transcripts to counts of single-molecule RNA-FISH. Bulk RNA is limiting, single-cell rocks!

She presented data from a new publication just deposited on the BioRxiv: Temporal mixture modelling of single-cell RNA-seq data resolves a CD4+ T cell fate bifurcation. They used temporal modelling of single-cell RNA-seq to analyse development of Th1 and Tfh cell populations in mice infected with Plasmodium, and show that a single cell gives rise to both cell types. I'd really suggest reading the paper.