Image analysis goes with the flow

A new technology combining imaging with conventional flow cytometry has arrived in Australia. Kate McDonald spoke to the inventor, Amnis Corporation's Dr David Basiji.

The great benefit of flow cytometry technology over microscopy is the ability to count and analyse thousands of cells per experiment. The drawback with flow cytometry is its inability to discriminate between differing cell morphologies or fluorescence signal distributions.

The addition of imaging to flow cytometry, however, would obviously solve many of those problems, with the additional benefit of improving data quality relative to standard flow cytometry.

That's exactly what American company Amnis Corporation has done, developing a technology called ImageStream that allows for multi-spectral imaging of cells directly in flow and adds the qualities that make flow cytometers such an integral part of cell analysis.

Amnis, which was established in Seattle in 1999, has sold its first two international systems to the University of Sydney and the Queensland Brain Institute (QBI).The ImageStream system combines high-speed imaging with the statistical power of a flow cytometer, making it possible to gather detailed quantitative morphologic and fluorescence information from tens of thousands of cells and use that information to define and analyse cell subpopulations.

"People have tried in the past to mount cameras on a flow cytometer and get transmitted light imagery," Amnis' CEO and co-inventor, Dr David Basiji, says. "You can do that but it's very difficult to get sensitive fluorescence imagery.

"We took a different approach: our system is an ultra-sensitive imaging technique - once you have the fluorescence images, you can analyse them in various ways to produce the data that a flow cytometer would.

"The sample handling is almost identical to flow cytometry - the samples are in suspension and the flow system is very similar to a conventional flow cytometer. It uses a technique called hydrodynamic focusing, so the cells are in a single-file line. The current system is strictly an analyser but we expect to have a sorter out in about three years."

Basiji was in Australia recently as a guest of local distributor Millennium Science and to oversee an upgrade of the systems being used at Professor Nick King's laboratory at the University of Sydney's Department of Pathology. The second is installed in Dr Geoff Osborne's lab at QBI, where he is using the technology to look at processes extending from neuronal cells.

"One scientist is using the machine to look at microparticles attached to dendritic cells, which is a difficult thing to do on a flow cytometer because the microparticles are tough to detect," Basiji says. "With imaging you have a better sensitivity with smaller objects, so it works very well."

Applications

Basiji says there are many applications of the ImageStream technology, which he divides into three classes. "There are the applications you can't do with flow cytometry. A flow cytometer will tell you how much of any given molecule is present on a cell or in a cell but it doesn't tell you where. Imaging gives you that information.

"In particular, being able to look at multiple fluorescence images and multiple markers at the same time allows you to co-localise, so we can do both relative and absolute localisation. We can tell if a signal is uniformly distributed in a cell or if it has moved to one side and we can quantify exactly how far and to what degree.

"We can also provide a measure of whether two different signals are in the same location in the cell or not."

The second class of applications he characterises as improved flow cytometry - the ability to do standard flow cytometric assays but use the image information to, for instance, discriminate against false positives.

"One application is a tunnel assay for measuring apoptosis," he says. "It's quite common in this sort of assay that once a cell is apoptotic it sheds its nuclear contents and these apoptotic bodies are floating in your sample. They can adhere to a cell that's not apoptotic, but the flow cytometer then measures it as apoptotic. When you have the image you can see that there's a thing stuck to the outside and it's probably not something that you want to count."

The third class is rare event imaging, he says. "In relation to imaging microscopes, one of ImageStream's great strengths is that that it has the statistical power of flow cytometry. You can look at tens or hundreds of thousands of cells, whereas on a microscope you're typically looking at no more than a couple of hundred.

"So if you want to look at a highly heterogenous sample and find a very rare subpopulation, by which I typically define as 0.1 percent, in our system in 10 minutes you can image a hundred thousand cells and find a hundred of the target that you are looking for.

"You still get very good statistics with 100 cells even though it's a very small subpopulation. You can do it in a reasonable period of time and that is something you just can't do on a microscope."

Development

Basiji, who did his undergraduate biophysics training at Berkeley before moving north to the University of Washington to study bioengineering for his PhD, describes himself as a stereotypical "propeller head", interested in getting his hands dirty in visionary projects.

"Just as I was finishing (his masters) a famous scientist came to the university - Lee Hood, who developed the DNA sequencer," Basiji says. "He set up a department of molecular biotechnology to develop technology for the Human Genome Project."

Inspired by his experiences, Basiji undertook a project to develop an ultra-sensitive DNA sequencing and protein analysis technology using fluorescence imaging. Mass spectrometry and capillary electrophoresis, however, overtook the potential of this project so he and his colleague Bill Ortyn, now COO of Amnis and co-inventor of the ImageStream, had a little brainstorming session and came up with the concept of imaging in flow.

"The idea was conceived in 1998-1999, we had feasibility prototype in November 2000, and sold a later prototype to Amgen Corporation, which had taken a very strong interest in this technology. We sent a machine to Amgen in 2003 and we started selling commercial systems in January 2005. The first two international systems were shipped to Australia."

Basiji says imaging-in-flow is a concept with long-standing appeal but had always faced the problem of adequate fluorescence sensitivity. "The ImageStream system solves the sensitivity problem by using a unique charge coupled detector (CCD) camera and a powerful image detection technique called time-delay-integration (TDI).

"This has been implemented in the ImageStream system to enable imaging-in-flow with a sensitivity comparable or superior to conventional flow cytometry."

How it works

The ImageStream system captures six high-resolution images of each cell directly in flow. Imagery includes brightfield, darkfield and up to four fluorescence colours, all acquired simultaneously at over 10,000 cells per minute.

This new capability, combined with easy to use data analysis software, greatly improves traditional flow cytometric applications and broadens the scope of flow analysis to include signal localisation and co-localisation, morphologic assessment and assay multiplexing.

The system enables advanced applications in apoptosis, nuclear translocation of transcription factors such as NFkB, receptor capping, co-localisation of cell surface markers and intracellular molecules, automated morphologic cell classification, sub-cellular compartmentalisation, T cell/APC conjugate analysis, morphologic and phenotypic cell classification, shape change, cell cycle and mitosis.