Special Report on Cell Biology: Cytometry for all
As applications broaden, footprints shrink
The tribe gathers to collect the ritual offerings they will present to the local shaman. They have reached a roadblock in their understanding and need input from the wise man before they can move forward in their explorations.
Everything must be perfect when they make the offering.
Approach the sacred chamber with items in the wrong vessel and they will be rejected out of hand. Arrive at the same time as any of the nearby tribes and acrimony may boil over. Their goal is peaceful co-existence, but they must have answers.
In lock-step, they traverse the causeway to the great sanctum, and gather in hushed tones before the vast entryway.
With swift, determined movements, their leader strikes upon the door, only to involuntarily step back as the great wooden obstruction slides back into the space beyond.
“We seek wisdom,” the leader does her best to state calmly, holding the offerings before her.
“We shall intervene on your behalf,” the acolyte responds, taking the offerings and disappearing into the mysterious void.
The sacred rites, known only to a special few, will take some time, but if all goes as planned, the small gathering will eventually have their answers.
As the collective retreats, one minor member turns to another.
“This is ridiculous,” he whispers. “It shouldn’t be this difficult to get a simple CD4 count.”
Historically, flow cytometers have been the domain of central or core lab facilities where trained technicians and cell biologists would help departmental scientists devise and execute experiments. And while these specialists still remain critical in many centers, instrument automation and improved user interfaces for both experimental design and data analysis are helping to put more of the control over experiments into the hands of end-users.
“We are seeing the transition of technologies going from big expensive boxes with a lot of infrastructure requirements to something that eventually can move away from the core facilities and be used in your typical labs,” says José Morachis, CEO of NanoCellect Biomedical, a relatively recent entrant into the cell analyzer market.
“With traditional cell sorters, even just looking at the instruments, scientists are usually a little bit taken aback or intimidated by using the technology,” he continues, suggesting that the technology should instead support the research, ideally without the scientist needing to become a flow cytometrist.
From his perspective, three things need to happen to make cell analyzers more approachable.
“One is reducing the size and complexity of the instrument,” he presses. “And by doing that, we’re reducing significantly the expense and cost to make these instruments, because you not only need smaller footprints, but you also need to make this more affordable. For most laboratories, the estimated budget for instrumentation is under $100,000.”
“People just want something that is easy to use,” suggests Nicole Ellis-Ovadia, product manager for Bio-Rad, adding Morachis’ third element.
The evolution of flow units has thus been a balance between diminishing footprints and ever more complex applications.
“On the higher end, people who are looking to do more complex experiments, looking at more things at one time, trying to get a deeper understanding of whatever system they’re looking at in terms of like a cellular system, they’re definitely the ones who are taking up to 14, 15, 20 parameters and above.”
From Ellis-Ovadia’s perspective, the expanding demands of cell analysis systems has required not just changes in the physical instrument but also in the supporting fluorophore tags.
“There is a lot of development being done on the dye and conjugate side in order to address that, so people can run the maximum number of colors on a laser without having overlap between the fluorophores,” she offers. “That’s come a long way from where it was maybe five to 10 years ago.”
Because of advances in these areas, she continues, researchers can run six to seven colors off a single laser.
Things on the hardware side have also helped bring high complexity to systems, she adds, pointing to Coherent’s Obis lasers, the compact design of which allow multiple lasers to fit into a smaller footprint.
“So now, you’re not talking about a huge instrument,” she explains. “You’re talking about something that’ll fit on your benchtop where you can have five lasers.”
Morachis concurs that companies, no matter what the technology, are always going to try to push the limits of what is possible. For its part, however, NanoCellect decided to focus its design on broadest practical applicability.
“Instead of designing something with 10 lasers and addressing a small percentage of the market, we realized that 80 percent of the market doesn’t use more than a couple of markers to analyze the cells that they’re looking for,” he says. “In fact, when we talked to a lot of core facilities managers, approximately 50 percent or more are only looking at GFP-positive cells.”
For Morachis, the two parts of the cell analyzer market are evolving at different paces.
“You have flow cytometers that only do analysis, where the market has developed very good ones and they’ve become much more affordable,” he explains. “But the ones that are still very expensive are the flow cytometers that do cell sorting.”
And it is this latter category, the cell sorters, where he sees the greatest opportunities in growth in terms of applications through its connection to downstream technologies.
He offers the example of molecular biologists who are using flow systems as sample prep, suggesting that the researchers are not experts in doing 10-plus color experiments, but simply need to isolate the right cells to continue their studies.
“Overall, the key is being able to isolate the cells and getting that much richer information downstream, whether you’re doing basic sequencing, qPCR, western blot—all sorts of different applications that people require,” he says. “There are all these cool tools downstream, but it seems like people have forgotten that the most critical step is getting your sample ready or getting the right sample.”
As an example, he describes the interesting work he has seen using NanoCellect’s Wolf cell sorter with CRISPR gene editing.
“People engineer their cells very easily and effectively, and use our technology to sort out those positively engineered cells and to isolate those into individual cells per well, like in a 96-well plate,” he describes.
He similarly highlights its applications in antibody development as researchers engineer and ultimately isolate cell lines particularly adept at producing the antibodies of interest.
At the same time, Ellis-Ovadia isn’t yet declaring the end of core facilities and flow managers, despite seeing a future where Bio-Rad’s ZE5 unit sits atop benches in individual labs.
“I think the tricky thing currently is that you could get quickly overwhelmed if you’re a new user,” she cautions. “With large-parameter experiments, as much as we put tools in there and it’s easy just to get the actual data, being able to design an effective experiment at that number of parameters and being able to decipher the data can be very difficult for a novice.”
Thus, she argues, researchers will still probably depend on the core lab to guide them in those types of experiments. The core lab could also offer to run the samples for the end-users, she adds.
“I think what we’re going to see before we see it move to the benchtop in initial labs is to see that the core labs will start saying you can run your own experiments versus us having to run them for you, which really frees up the core labs to be able to do other things,” Ellis-Ovadia continues, pointing to various features designed into the ZE5 platform that facilitate this relationship.
“The ZE5 will actually check and make sure everything is where it’s supposed to be, and it has a spectral viewer that allows you to pick colors and lasers and helps design the experiment,” she comments. “It’s very easy to load a plate or tube. It also allows you to do things while it’s running.”
Alongside developments in fluorophores and lasers, possibly the biggest influence on the shrinking footprint of flow systems and cell sorters is the application of microfluidic technologies.
In a recent review, Gabriel López and colleagues at Duke University and University of New Mexico examined this evolution.
“[Fluorescence-activated cell sorting (FACS)] devices are expensive, large, complicated to use and typically require high operating pressures that may compromise cell function or viability,” the authors noted. “[Magnetic-activated cell sorting (MACS)] devices suffer from entrainment of non-labeled cells in target cell populations and highly nonlinear magnetic forces that can result in poor capture of labeled cells positioned far from the field source.”
The researchers suggested that microfluidic cell sorting devices address many of these issues, and highlighted several important features, including:
- Sorting performance almost on par with FACS and MACS;
- Improved cell viability due to low-pressure or absent sheath fluids;
- Lower unit costs and decreased power consumption;
- Improved user-friendliness with simplified user interfaces and potential for disposability;
- Strong spatial and temporal control of cells; and
- Smaller footprints that reduce reagent costs, enhance analytical timescales and facilitate portability.
They further suggested that the sheer variety of clinical applications for microfluidic cell sorting is a strong impetus for future development, whether fractionating whole blood or isolating rare cells such as circulating tumor cells (CTCs) or hematopoetic stem cells.
The translation of such devices into commercial production, however, has had its challenges (see sidebar Clogged microfluidic market below after the end of this article).
One advantage of moving to a microfluidic system is the opportunity to develop units that are more modular in design, with plug-and-play components that not only facilitate use by scientists with a broad range of experience and expertise, but also, as suggested above, increase the opportunities for the use of disposable components.
“In a traditional cell sorter, one of the issues is messing around with all of the fluidic interfaces, and maintaining the fluidic carts, where they have these big pumps and reservoirs,” Morachis explains. “It’s a system that needs to be maintained regularly and that’s why people go to a core facility. There is usually a flow cytometry manager who just runs the samples for them for fear of contamination or that they’ll clog the system or do something to the instrument.”
In contrast, systems such as the Wolf cell sorter or MACSQuant Tyto (Miltenyi Biotec) rely on a disposable cartridge such that there is minimal need to adjust any other components of the system and anything touched by sample is easily removed at the end of the experiment, effectively eliminating cross-contamination.
“It’s also a walk-away feature,” Morachis adds. “When I am done sorting, I just easily toss away that cartridge and don’t worry about post-cleaning or maintaining the instrument.”
“And then that sets up for future applications, making it feasible for potential clinical applications where sterility and contamination are critical,” he extrapolates.
Advances in modularity and ease-of-use are also impacting what might be the biggest technical challenge: the ability to move flow cytometry outside of the lab completely, whether to remote and possibly temporary testing facilities or to particularly resource-limited settings such as makeshift rural clinics in the developing world.
The Auto40 from Apogee Flow Systems, for example, is designed for immunophenotyping experiments such as CD4 T cell measurements. Loaded via a volumetric syringe and stepper motor, the platform provides direct determination of cell numbers per unit volume, eliminating the need for reference materials such as microbeads. As well, sample preparation is minimized as the unit can run no-lyse-no-wash fluorescent antibody-tagged samples.
In 2012, François-Xavier Mbopi-Kéou and colleagues in Cameroon tested the Auto40’s performance alongside the clinical stalwart FACSCalibur (from BD Biosciences), determining that the smaller unit offered similar performance with low intra- and inter-run variability.
Perhaps more importantly, wrote the authors, “the compact low-range single platforms Auto40 and [Sysmex-Partec] CyFlow SL_3 are portable desktop flow cytometers that do not require optical alignment, can run on a 12-volt car battery and can be connected to a laptop computer, suggesting their possible use in healthcare mobile units and thereby making CD4 T-cell enumeration available in remote or hard-to-reach locations.”
In some cases, labs need to miniaturize experiments because the sample is limited, whether due to the cell type being analyzed or in high-throughput experiments where a single sample needs to be spread across dozens or hundreds of wells.
“With the ZE5, the product is actually designed with a unique sample delivery system so that you don’t waste any sample,” offers Ellis-Ovadia. “If you want to, you can have your sample returned back to the sample wells, such that if you want to do some further analysis because you’ve seen something interesting, you’ve actually retained the sample.”
In some cases, researchers and clinicians are not interested so much in entire cells but rather microparticles or extracellular vesicles (EVs) released from cells either as a normal component of intercellular communication or as a signal of disease (see also “Non-invasion of the body snatchers” in the October 2015 issue of DDNews).
As these particles move into the body, they afford clinicians the opportunity to monitor their levels and characteristics from body fluids such as blood or urine that can be biopsied much less invasively than tissue biopsies.
Although EVs typically carry biomarkers that can be tagged, their 100- to 1000-nm diameters can fall below the detection limits of many flow cytometers. To address this, engineers have enhanced flow cytometry optics and improved detector sensitivity, creating nanoscale systems better able to cope with such submicron entities.
In early 2016, Western University’s Hon Leong and colleagues described their use of the A50-Micro platform from Apogee Flow Systems to examine prostate microparticles (PMPs) in retrospectively collected plasma from healthy volunteers and patients diagnosed with benign prostatic hyperplasia or localized or castration-resistant prostate cancer.
The researchers noted significantly lower levels of PMPs in healthy patients compared to the others, and that only patients with Gleason Scores ≥8 demonstrate significantly elevated PMPs versus those with other Gleason Scores. As well, PMP levels diminished in patients following prostatectomy from preoperative levels.
Total fluorescence doesn’t necessarily tell the whole story of what is happening within a cell. Rather, the intracellular localization of that fluorescent signal or morphological changes within subpopulations of cells can provide valuable information about changes in cellular characteristics and behaviors.
Previously, efforts to identify and understand such changes were the purview of imaging methods such as fluorescence in-situ hybridization and microscopy. But these techniques were inherently low-throughput and relied significantly on user input to identify qualitative changes.
Improvements in image acquisition and data analysis, however, have allowed researchers to combine imaging with flow cytometry, dramatically increasing throughput without losing important information. The method is known as imaging flow cytometry (IFC).
Briefly, IFC replaces the photomultiplier tubes commonly found in flow systems with CCD cameras that capture images of cells and particles as they pass through the laser beam, overlapping features from the front- and side-scatter (brightfield and darkfield) with whatever fluorescent signals are also available. The composite images allow for the more thorough interrogation of many biological features such as protein migration and co-localization, as well as dynamic morphological changes.
In 2015, for example, Christopher Parris and colleagues at Brunel University London and St. Bartholomew’s Hospital used the ImagestreamX Mark II from EMD Millipore to monitor DNA damage in cells from healthy individuals and those defective in double-strand break (DSB) repair exposed to the ionizing radiation.
As they reported, DSBs in DNA trigger the phosphorylation of the minor histone protein H2AX, which the researchers identified using fluorescently tagged antibodies. They hoped to develop a method to qualify DNA repair in radiotherapy patients, and thereby predict patients’ clinical responses.
“For the quantitation of γ-H2AX foci, the ability to image foci at higher magnification (up to 60X) together with a focus stacking capability provided by the extended depth of field (EDF) function potentially allows a more accurate assessment of foci number throughout the complete nuclear region,” the authors suggested.
The EDF function is a combination of specialized optics and image-processing algorithms that effectively combine images from multiple planes into a single crisp image, improving the resolution of subcellular and surface cellular structures to levels, said the authors, achieved with in-situ microscopy.
More recently, Shuqun Cheng and colleagues at Shanghai’s Second Military Medical University described their efforts to apply IFC to characterize CTCs in hepatocellular carcinoma (HCC) patients, correlating their findings with microvascular invasion (MVI), tumor recurrence and survival.
Rather than rely on surface markers, however, the researchers focused their attention on the relative areas of cellular nuclei to cytoplasm, what they dubbed the karyoplasmic ratio.
“Traditional methods for CTC detection in peripheral blood are based on CTC surface markers, such as CD133 and EpCAM,” the authors explained. “However, even in tumor tissues, only up to 20 percent of cells express the three markers. Furthermore, owing to the [epithelial-mesenchymal transition] effect, many tumor cells lose these surface markers as they are shed into circulating blood.”
“Our approach was based on morphological properties that are shared among all tumor cells, resulting in a high sensitivity,” they continued. “Thus, compared with previous methods that detect approximately 1–10/100,000 CTCs in the peripheral blood of cancer patients, our method can detect several times more CTCs.”
The scientists examined blood samples from healthy subjects, non-cancer patients and HCC patients, noting that the incidence of high karyoplasmic ratio (HKR) cells was significantly higher (almost 10-fold) in the last group compared with the former groups. Similarly, HCC patients with MVI had double the HKR cells of HCC patients without.
They then compared patients with high or low HKR cell counts (using a threshold of 57.3 HKR cells/100,000 cells) and noted that over a 12-month follow-up period, patients in the high HKR group were more likely to experience HCC recurrence or death. And using blinded blood samples, they found HKR cell levels were highly diagnostic of HCC with and without MVI.
The researchers suggested that not only was the HKR method more accurate that traditional CD133 and EpCAM methods, it was also easier and potentially safer for patients.
Requiring only red blood cell lysis buffer and DAPI staining, the method was economical. And only 1 mL of peripheral whole blood was needed to establish preoperative diagnosis of MVI, which they said can be vital for patients who experience high blood loss during surgery.
As any digital photography hobbyist knows, however, the imaging capability of IFC comes with a cost, as IFC data files take up a lot of storage space and bandwidth when sharing.
Yu-Hwa Lo and colleagues at University of California, San Diego, recently described the data challenges in a review of IFC.
“Compared to the data format in conventional flow cytometry, including integral, peak and width of light intensity, cell images produced by IFC are much more complex,” the authors explained. “Since IFC can produce thousands of multi-spectral cell images per second, files generated by IFC can tremendously burden the digital image transportation and processing realized by the back-end data handling unit.”
The researchers crunched some numbers, finding that with a field of view of 40 μm by 40 μm (100 pixels by 100 pixels), data generates at 100 MB/sec at a throughput of 10,000 cells/sec.
“Therefore, a test of a few minutes can easily create a data file beyond 10s of GB,” they suggested.
Furthermore, they continued, all image analysis is currently performed offline. For IFC to realize its true potential, real-time image construction and analysis are required.
“Possible approaches to extract cell characteristics in real time include the use of field-programmable gate arrays (FPGA) or graphics processing units (GPU) to implement various image processing and machine learning algorithms,” the authors offered.
Addressing these challenges, they concluded, was just one area in which IFC needed to make progress to become both a research and a clinic workhorse.
Don’t label me
For all the advances in fluorophore and dye chemistries, there is often concern about how those compounds and their conjugates (e.g., antibodies) may impact the biology of the labelled cells.
For applications strictly involving cell population enumeration, such artifacts may not be important. When cells are being analyzed for active biological functions and/or sorted for further manipulation, however, researchers may wish or need to avoid external labeling strategies.
Fortunately, a variety of label-free methodologies have been developed that provide a wealth of information, not just about a cell population’s current state, but also about its potential biological and clinical fates.
In 2016, for example, Swansea University’s Paul Rees and colleagues described a label-free workflow for high-throughput IFC to examine cell cycle.
Using both brightfield and darkfield images, the researchers extracted 213 cell features that they summarized into five categories: size and shape, granularity, intensity, radial distribution and texture. These features were then used for supervised machine learning against a subset of cells with known DNA content and mitotic phase based on manual identification or fluorescent labelling.
The researchers were able not only to determine cell cycle status in live Jurkat cells and yeast, but also to identify cells undergoing mitotic block on exposure to nocodazole. The researchers suggested the method could also be used retrospectively, allowing scientists to interrogate data of cells without the appropriate stains for phenotypic identification.
“While current imaging flow cytometers do not have physical cell-sorting capabilities, and for now our approach is suited to experimental contexts where samples are analysed only, this approach may offer the possibility to entirely avoid any fluorescent stain and opens up the perspective for a new generation of image flow cytometers that could operate without fluorescence channels,” the authors concluded.
More recently, University of California, Los Angeles’ Madokht Masaeli and colleagues performed a similar label-free analysis including not only morphological features but also mechanical signatures, using a platform they called deformability cytometry.
Alongside standard high-speed imaging and automated image analysis, they wrote, the assay is based on microfluidic hydrodynamic stretching of cells.
“Briefly, using inertial focusing, single cells arrive at a junction where they are uniformly stretched while being imaged,” they described. “Here we implement deformability cytometry as a high-throughput automated tool to assay 15 biophysical properties of cells, including cell size and strain, time-dependent mechanical properties and morphologies across length scales.”
Testing their platform on stem cells they monitored through differentiation, the scientists found high correlations between the mechanical measurements and the expression of commonly used pluripotency biomarkers. This suggested the platform’s potential as a label-free assay for pluripotency.
“Assessment of pluripotency in human stem cells currently requires teratoma formation in mice,” the researchers explained. “This is not a quantifiable assay, and merely provides a binary yes-no result with significant time investment and cost.”
They also noted that the platform could be used to monitor the cytotoxicity of drugs as cells undergo mechanical changes in response to chromatin changes. Within a day of exposing Jurkat cells to apoptosis-inducing trichostatin A, for example, the researchers were able to see the cell pool separate into two subgroups based on size and deformability.
“Current cytomorphological analysis is labor-intensive and qualitative, creating a need for automated, quantitative alternatives,” they continued. “Using purely physical properties of cells, we showed accurate detection of malignant cell sub-populations in multiple mixed samples.”
Although their designs have not yet taken them there, the researchers suggested there was no reason why these same mechanical parameters couldn’t also be used to sort cells in the absence of dyes or antibodies.
Thus, despite its advancing age, flow cytometry and cell sorting continue to evolve as a technology with improvements at all levels—fluidics, optics, electronics, informatics—allowing researchers and clinicians to further expand their understanding of biology and medicine.
“I see a great trend in this new biology age where technologies are merging and fields are merging,” concludes Morachis. “Technologies that used to be traditionally for one particular field are now being exploited and used to address and enhance other fields like genomics and molecular biology.”
Clogged microfluidic market
Despite an ever-expanding repertoire of microfluidic cell sorting technologies, very few have managed their way into the clinic or have found widespread use in research, according to Gabriel López and colleagues at the University of New Mexico and Duke University.
“Microfluidic devices are significantly smaller than their forerunners, fluorescence-assisted cell sorting (FACS) and magnetic-assisted cell sorting (MACS) devices, yet they provide equal and sometimes enhanced cell processing capabilities,” the researchers wrote in a recent review. And yet, there is little sign that these microfluidic innovations will soon supplant their forerunners.
They identified several barriers to commercialization, dividing them into device- and commercialization-related factors.
On the device front, the authors suggested that there is typically a greater interest for innovators in producing new technologies, often at the expense of practicality. Engineered from novel materials with multiple complex and sometimes custom components, integration of the new devices with existing lab infrastructure may be difficult.
As well, the very nature of microfluidic devices makes them more prone than their larger counterparts to issues of clogging by anything from cell debris to biofluid residue to simply air bubbles. This can significantly reduce the life span of the device.
On the business side, they suggested, the extensive number of existing patents making broad claims are making it difficult for innovators to operate, fearing increased risk of litigation. This in turn discourages investment, whether for startup development or licensing of developed technologies.
Furthermore, the transition from benchtop systems to microfluidic devices represents a significant investment of time, money and effort to adjust the infrastructures of both the manufacturing and the end-user communities. Thus, to surmount the market entry barrier, an innovative system cannot simply represent an incremental advance, the researchers argued, but rather must offer a significant competitive advantage, whether in terms of overall performance, reduction in costs or, ideally, both.
“We believe the most significant impediment to translation of microfluidic cell sorting devices is associated with philosophy of innovation,” the authors pressed. “Microfluidic systems are usually developed first and then adapted to fit a potential market need.”
“However, this approach must be rethought and reversed in order to enhance translation toward a significant commercial or clinical impact,” they suggested. “Innovators must identify a biological problem that represents a significant unmet need, and then work to develop microfluidic tools that address that need.”
These challenges acknowledged, the authors were quick to note that none of these issues was insurmountable and offered several insights to address each. They pointed to cartridge-based technologies such as that found in the MACSQuant Tyto from Miltenyi Biotec to simplify use, as well as advances in parallelization to facilitate scale-up as needs change. They also mused about how 3D printing technologies and the like might reduce overall costs.
“As these changes take place, we believe microfluidic cell sorters will soon revolutionize the ways in which cell-containing fluids are processed in the laboratory and the hospital,” they concluded.