Report on the 2008 NW Regional Cytometry Meeting
by Allan Kachelmeier
A regional meeting might range widely in what it is, from users group meeting, with discussions of tools usable right now, to Asilomar-type development sessions on software and hardware that aren’t so immediately applicable to users, to discussions best characterized, not in terms of tools used, but in terms of biology. The 2008 Northwest Regional Cytometry Meeting, “Flow and Imaging for an ~omics era”, held March 13 – 15 at OHSU in Portland, had the full scope. Specific objectives, including desire to solicit vendor input on the proposed ACS standards, bring together a meeting with flow, imaging, and high content analysis on the same page, and spotlight the Oregon Translational Research and Drug Discovery Institute, complimented general objectives to encourage further sharing of analysis tools. The uniting factor that brought this mix of flow, imaging, and high throughput together was the momentum of past flow meetings, combined with a desire to see imaging, including high content analysis, more involved in the same cytometry conversation.
The three day event started with an ISAC / FICCS co-hosted meeting. Ostensively organized to get vendor comments on the proposed Analytical Cytometry Standard (ACS), FICCS4 began with a series of 7 talks. Janet Siebert (CytoAnalytics) discussed how using patterns from individual donor samples helped to unravel ensemble averaged data at the population level. Peter Krutzik (Stanford) and Kevin Banks (LabKey) briefed the meeting on open-source software available to manage high throughput data - CytoBank and commercially available LabKey, while Florian Hahne (FHCRC) and Errol Strain (BD) provided details on new BioConductor packages, flowCore and plateCore, to automate annotation and analysis. Raphael Gottardo (UBC) presented a generalized Gaussian mixture model clustering framework for automated gating, with results equivalent to manual gating. Jim Wood (chair, ISAC Data Standards Committee) gave a presentation on the proposed Analytical Cytometry Standard (ACS), noting how ACS would augment FCS, providing a means to include analysis information, such as gating details in XML format, alongside listmode data. ACS will be addressed again at the international meeting in Budapest. Other items that came up in discussion included the logicle scale, a Stanford patent, and quantitation in imaging.
A Minimum Information Standards panel (Ryan Brinkman – BC Cancer Research Center, Janet Siebert, Bob Murphy – Carnegie Mellon, Shannon McWeeney - OHSU, Mark Collins - Cellomics) opened a dialogue on the strengths and weaknesses of various Minimum Information Standard approaches (MIFlowCyte, MIAME, OME, MIAHA) taken under different platforms.
The next two days, comprising 19 plenary speakers and 6 parallel sessions, were a mix of ‘multiplexing’, automation, drug discovery, kinetics, and, implicitly, modeling. In the paragraphs that follow, I’ve tried to give a few brief snatches. Partly as a result of how successful flow cytometry has been, the field of cytometry is expanding. The meeting was formatted for cross-pollination, and to a certain extent the cross-fertilization was successful, for example, in discussion of how high throughput, automated imaging might be brought up to speed with flow in standards.
Numerous hardware and software developments, and planned developments, interspersed the meeting, a list headed off with J. Paul Robinson’s discussion of advanced analysis of scatter, hyperspectral cytometry, and lessons to be taken from remote sensing. He showed how spectral resolution, combined with classification tools such as principal component analysis, enables separation of highly overlapping dyes. Ger van den Engh (Cytopeia) shared how marine applications are driving machine development at Cytopeia, including improvements in scatter detection, greater use of polarization measurements, and detection without sheath fluid. Bill Ortyn (Amnis) gave an update on the deconvolution algorithm Amnis is now applying to its images. Mike Samuels (RainDance Technologies) discussed a droplet-based microfluidic system, which
includes dielectrophoretic sorting. Applications that can be done in microdroplets include enzyme-amplified cytometry, droplet-based ELISA, and qPCR.
The Nolan lab at Stanford is known for having developed cell bar-coding of phospho-specific epitopes, as well as for contributions in automated inference and data management. In fluorescent cell barcoding (FCB), samples are treated with different concentrations of dye, combined and stained with antibody, giving each sample a unique intensity signature. FCB profiling of kinase pathways, as Peter Krutzik showed in his talk, can facilitate drug discovery, not only at the proteomic level, but at various points along the drug discovery pipeline. Jake Jacobberger (Case Western) expanded on the importance of phosphorylation profiling in systems biology – as he showed in his slides, phased oscillations of cell cycle biochemistry, exemplified in protein modifications, are the basis of finite state expression patterns. Chemical genomics, while including study of protein modifications, leads as well to studying small molecule interventions. Larry Sklar (UNM) pointed out 50% of the NIH Roadmap budget is for ‘molecular libraries and imaging’, highlighting the importance of ligand-receptor and protein-protein interactions. Systems biology requires assays which probe the cellular system in context. Mike Mancini (Baylor College of Medicine) reviewed what’s involved in cell-based assay development, drawing examples from his work on single cell multiplex analyses of gene regulation by estrogen and androgen receptors.
A bit of hybridization of techniques between cytometry platforms is bringing the biology into better focus, as well as providing better tools for multiparametric comparisons. Dorothy Lewis (Baylor College of Medicine) explained how flow and microarray analysis together give a better picture of CD8 cell death in HIV, work which has implicated by-stander effects, as well as CD28 and Granzyme B. Ed Walker (EACRI, Providence), analyzing cancer vaccine trials, discussed how clustering techniques, borrowed from microarray analysis, enables more efficient data reduction, and easier identification of differentially expressed Gp100-specific T cell subphenotypes. Bob Murphy (Carnegie Mellon), speaking on the relevance of automated learning to flow, observed that machine learning is not applied as often to flow data as to image data, as flow experiments are often designed to simplify the underlying biological reality, and hence the data analysis, to create ‘positives’ and ‘negatives’. Better to ‘model and embrace complexity’, he suggested, than to skew the real biology and select reagents/conditions because they produce better binary phenotypes. David Basiji (Amnis) spoke about sampling and measurement error, and proposed a resolution metric Rd, the difference of the means divided by the sum of the variances, which would enable simpler comparison of imaging and flow cytometers. Bob Zucker (US EPA), while reviewing the similarities between flow and confocal microscopy for QA, made special note of the relative lack of standards in microscopy.
Probes are common ground, and all cytometry platforms have a vested interest in improvements. Iain Johnson (Invitrogen/Probes) talked about a few of the newer probes, including Click-iT EDU (replacing BrdU), BacMam, and Quantum dots, while addressing the larger question of how quantitative the expression of fluorescence is. Alan Waggoner (Carnegie Mellon) remarked on how fluorescence detection had gotten to be a billion dollar business. He discussed work at the Molecular Biosensor and Imaging Center on fluorogen-activating single chain antibodies (scFvs). scFvs are small (<30kDa) antigen recognition molecules, which combine with fluorogens, and only fluoresce when bound. Chris Geddes (Institute of Fluorescence) discussed developments in fluorescence spectroscopy and plasmonics, particularly the orders of magnitude improvements in sensitivity promised by metal-enhanced and microwave augmented fluorescence.
High content analysis has aspects like flow cytometry, in that both are high-throughput, cell-based technologies. Ability to revisit cells after interrogation adds a critical dimension, as does, in the case of imaging, morphology, but at the cost of added complexity. Rick Monsma (Schering-Plough) gave a quick tour of drug discovery by HCA, from the larger picture of target identification, lead discovery, chemical optimization, and clinical testing, to specifics, such as the parametric changes in cells one can measure. Doug Auld (NIH Chemical Genomics) described NCGC’s role as ‘technology accelerator’, providing information on ways to reduce and quantify ‘off-target’ artifacts. For example, cell aggregation causes enzyme inhibition, with potential false
positive readings for a putative inhibitor. The NCGC routinely runs titrated series of samples, partly to reduce artifacts, but also to collect data on structure-activity relationships.
In addition to plenaries, parallel sessions offered a wide array of topics. These included vendor talks on new probes, equipment, and software, as well as talks on taking better measure of cell signaling, modeling scatter, imaging at the single protein level, FRET, and university-based drug discovery programs.
The 2008 NWRCM, for a regional program, was broadly formatted. $19,000 in vendor table fees covered some speaker expenses, lunches, and venue, ISAC support of the FICCS meeting amounted to $1,700, an additional estimated $15,000 in plenary speaker expenses were born directly by vendors, with perhaps on the order of $75,000 spent for vendor displays and demos, and an estimated $120,000 in expenses incurred by registrants (travel, hotel) – the basic program consisted of 26 plenary speakers, including FICCS4, but excluding the Amnis dinner talks, 44 parallel session speakers, 32 vendor supporters, as well as one academic sponsor. In general, the ‘flow’ discussions were well attended, while imaging sessions Saturday afternoon were less well attended, as the meeting may have been seen primarily as a ‘flow’ meeting. Considering the ‘ecumenical’ objective, it might have been better to have had a mixed itinerary of ‘flow’ and ‘imaging’ talks. In the same vein, a more explicit effort to bring together informatics and biological expertise on real problems might have initiated more action. But, overall, most took cognizance of ‘fair warning’ – the program was packed, and it offered plenty of opportunities, particularly for the proactive.
Sponsors included Oregon Hearing Research Center, BD Biosciences (incl. spon. Rick Monsma), Beckman Coulter (co-hosted the Friday evening buffet), Invitrogen (incl. spon. Alan Waggoner and Iain Johnson), PerkinElmer/Evotec (incl. spon. Ghislain Bonamy), Amnis Corp. (incl. spon. David Basiji), Cytopeia (incl. spon. Ger van den Engh), Dako, CompuCyte (spon. Jake Jacobberger), TTPLabtech (incl. spon. Doug Auld), Thermo Fisher Scientific, Accelrys (incl. spon. Mike Mancini), Accuri Cytometers, Applied Cytometry Systems, Bay Bioscience, BioLegend, Chi-Square Works, Cell Signaling Technology, Cytek Development, De Novo Software, FloCyte Assoc. (spon. Dorothy Lewis), Guava Technologies, iCyt Mission Technology, IntelliCyt Corp. (spon. Larry Sklar), Miltenyi Biotec, MDS Analytical Technologies, Olympus Advalytix, Partec, Phoenix Flow Systems, StemCell Technologies, Tree Star, Union Biometrica, and Verity Software House.
Allan Kachelmeier, 2008 NWRCM coordinator
Manager, confocal microscopy core
Oregon Hearing Research Center
OHSU