Vol. 11 •Issue 4 • Page 61
Microbiology Applications of Flow Cytometry
Bacterial detection, identification and susceptibility testing are promising new applications for flow cytometry.
Since the 1980s, flow cytometry has seen increasing application in a number of areas of laboratory medicine. Most hematology analyzers use optical and/or electronic flow cytometric measurements to count and size red and white cells, platelets and reticulocytes, and to perform differential white cell counts. Fluorescence flow cytometry is widely used to count CD4 T-cells in patients with HIV infection, immunophenotype cells from leukemias and lymphomas, assess the degree of malignancy of breast cancer and other tumors and crossmatch organs for transplantation. However, although clinical microbiology has long been regarded as an area in which flow cytometry looked very promising, the technology is still not in routine clinical use.
Improvements in instrumentation and reagents now allow characterization of the genetics and physiologic states of microorganisms with far greater accuracy and precision than was possible 20 years ago, but unexpected complexities revealed by such analyses may make it less straightforward to adapt flow cytometry to clinical microbiology than was originally envisioned. It is timely, therefore, to reconsider the basic needs of the diagnostic microbiology laboratory and where modern cytometric methods—flow-based and otherwise—might fit in.
Microbiology Tasks, Specimens
The first task of the clinical microbiology laboratory is to determine whether pathogenic microorganisms are present in a specimen. If a pathogen is detected, it must next be isolated and grown in pure culture so that it can be identified and its antibiotic sensitivity determined.
The procedures followed are different for various types of specimens. Urine specimens, which account for 30 percent to 50 percent of samples in typical laboratories, should, according to most textbooks, be subjected to examination under the microscope. An extended period of culture is unnecessary; the patient will have served as the incubator until the count of organisms reaches the level of tens of thousands/ml considered clinically significant. However, while microscopy is reliable for identifying bacteriuria, it requires a substantial expenditure of time by a highly trained technician.
When blood is cultured to detect bacteremia, it is essential that the culture be read as positive if only one or a few viable organisms are present in the 10 ml sample. No apparatus can detect such small numbers of organisms directly; instead, the original inoculum is incubated and allowed to proliferate for 10 or more generations, permitting the metabolic products of many thousands of organisms to be detected in positive cultures.
Although urine and blood from healthy individuals contain few or no bacteria, stool contains more than 50 percent bacteria by weight, and the rationale for stool cultures is detection of pathogens not normally found there. This is usually accomplished by incubation in or on media selective for the relevant pathogens.
The procedures for identification and antibiotic sensitivity testing of a pathogen, once one is found and obtained in pure culture, are substantially similar regardless of the nature of the original specimen. Identification generally requires inoculation of aliquots of a pure culture into a variety of media containing different energy sources, other nutrients or metabolic inhibitors, which can reveal different patterns of growth between even closely related genera and species. Antibiotic sensitivity testing is accomplished by determining growth characteristics of the pathogen in one or more concentrations of a number of different antibiotics. The selection of both identification media and antibiotics is influenced by the preliminary classification of the organism as a Gram-positive or negative bacterium or a fungus.
In classical microbiology, single organisms are only sought or examined for purposes of bacterial detection, as was described in connection with examination of urine specimens. Microscopy, while insufficiently sensitive to be reliable for the detection of organisms in uncultured blood, can be used without or prior to culture in analysis of other samples. However, many laboratories forego direct observation in favor of examination of plate cultures after overnight incubation and/or (in the case of urine specimens) chemical screening methods. The simpler methods are less labor intensive and, therefore, more economical.
Classical microbiologic end points for detecting growth are colony formation on solid media and development of turbidity in liquid media. This usually requires culture for at least eight hours. Detection of growth in liquid media by automated methods is typically possible at the lower bacterial concentrations reached after four to five hours’ incubation. Identification and sensitivity testing are typically done on organisms cultured in broth from a single suspect colony grown on a solid medium, and the combined incubation times typically occupy at least 24 hours.
It would seem obvious that flow cytometry, which is essentially an automated form of microscopy, could be adapted to detect bacteria; flow cytometric measurements of bacteria were, in fact, first made during World War II in attempts to develop a method to detect biowarfare agents.1 Later studies in clinical environments showed that electronic (Coulter) particle counters could reliably detect clinically significant bacteriuria; however, they failed to distinguish between bacteria and other particles in the same size range present in urine, resulting in an unacceptably high false-positive rate.2
Because the size, mass, nucleic acid and protein content, etc. of bacteria are approximately 1/1,000 the magnitude of the same parameters in mammalian cells, it is harder to make good flow cytometric measurements of bacteria, although instruments have been designed and/or adapted to permit sensitive and precise measurements of small angle (forward) and large angle (orthogonal) light scattering and fluorescence of bacteria and even of large viruses.
Scatter measurements alone are not usually sufficient either for distinction between species or for discrimination of microorganisms from other small particles, although fungi, which are generally larger than bacteria, can be distinguished from them on the basis of scatter signals. Further improvements in cytometric discrimination of microorganisms typically rely on fluorescent staining; nucleic acid stains; fluorogenic enzyme substrates and membrane potential—sensitive dyes have all been used for this purpose. These reagents all produce stronger signals and require less sample preparation time than would be necessary to detect organisms with more specific reagents such as labeled antibodies, lectins or oligonucleotide probes. The specificity of antigens and oligonucleotides may be more useful for identifying organisms when a sample is being screened for one or a very few pathogenic species.
Flow cytometry can detect organisms in whole blood at concentrations as low as 100/mL, but since the lowest concentration of interest is in the range of 1 organism/10 mL, a practical cytometric method for detection of bacteria in blood would still require culture prior to analysis.
When bacteremia is detected, it is clinically beneficial to obtain identification and antimicrobial susceptibility information as rapidly as possible. Cytometric procedures, which could work with relatively low concentrations of organisms and short culture times, might find a niche in the clinical laboratory.
Mixtures of fluorescent dyes have been used to produce what has been described as a cytometric equivalent to the Gram stain. Further improvement in bacterial detection can be achieved using combinations of dyes, with each responding to a different characteristic distinguishing bacteria from other small particles. This requires relatively elaborate instrumentation, but may be justified when dealing with samples containing high particulate backgrounds.
A cytometric apparatus for laboratory bacteriology should, ideally, be inexpensive and simple to operate and use relatively inexpensive reagents. Small flow cytometers using red diode lasers and/or other small, energy-efficient and inexpensive solid-state light sources and capable of bacterial detection are now available. These could potentially be incorporated with suitable sample processing hardware and software for unattended operation into dedicated clinical instruments.
ID, Susceptibility Testing
Bacterial detection can be accomplished using one or a few aliquots of sample, accounting for only a minute or two of instrument time. When a pathogen or suspected pathogen is detected, the next task is generally to isolate and grow it in pure culture for purposes of identification and antimicrobial susceptibility determination.
As previously mentioned, the classical identification method depends on analysis of growth of multiple aliquots of a pure culture in a variety of media. When this is done by ELISA or turbidity measurement of samples in multiwell microtiter plates, the apparatus generally reads many wells simultaneously, analyzing all in less than a minute.
This parallel processing, however, is not readily available in a flow cytometer. Significant time is required just to clear excess sample from the cytometer’s tubing between aliquots, and, even with the newest technology for sample handling, analysis of a 96-well plate takes more than five minutes.
Sample throughput is also an issue in antimicrobial susceptibility testing, which typically follows the growth of multiple aliquots of organisms in one or more concentrations of at least 10 antibiotics. A flow cytometric susceptibility test can be initiated with much lower concentrations of organisms than those required for bulk methods (104 organisms/mL vs. 106) and read after only a few generation times, providing results in an hour for rapidly growing organisms. However, if the flow cytometer can run only one-fifth the number of samples per hour, the plate reader catches up in a relatively short time.
Flow cytometers recently have been used for chemical analyses in which the analyte of interest is bound to plastic beads and detected with a fluorescent antibody or probe. Several assays can be run in a single tube, each using beads that can be discriminated on the basis of their size, color or both from those used in the other assays; this is referred to as multiplexing or multiplex analysis.
Multiplex analysis at the single-cell level might possibly be usable to facilitate flow cytometric microbiologic assays that would be infeasible if multiple aliquots had to be analyzed separately. Antibiotic susceptibility testing provides an obvious example. If the organisms incubated with different antibiotics and at different dose levels were color coded by staining with different covalently bonded dyes or mixtures, multiple aliquots of organisms could be mixed together before introduction into the instrument, which could determine which drug and dose level had been applied to each and use only a single, contrasting stain to assess growth or viability of organisms from all of the aliquots.
While this approach sounds promising, the feasibility of the multiplexing scheme has not been determined. Since, in recent years, a number of sensitive but relatively inexpensive scanning laser cytometers have been developed that are capable of analyzing small numbers of cells in multiwell plates, it may be simpler to adapt these instruments for bacterial identification and sensitivity testing than to attempt to develop flow-based multiplexing schemes. The scanning approach could also take advantage of new developments in microfluidic technology that now allow sample mixing, washing and incubation of multiple aliquots to be accomplished in a single, small, disposable unit.
When the suspect pathogen may be any of hundreds of species of organisms, even partisans of cytometry concede that bulk, small-sample methods are more practical for bacterial identification. Classically, organisms are only subjected to antimicrobial susceptibility testing after isolation in pure culture and identification; cytometrists have suggested that it may be more sensible to aim for rapid (presumably cytometric) determination of susceptibility first, providing clinicians with the information needed for therapeutic decision-making, while leaving precise identification of epidemiologic significance until later. This may be the only approach to pathogens engineered by rogue nations and/or terrorist groups for which conventional identification criteria may not be available; rapid determination of therapeutic options would then be essential.
Attempts to develop rapid flow cytometric methods for antimicrobial susceptibility testing of bacteria have typically used only a single fluorescence measurement combined with light scattering measurements. Among the dyes used to determine putative viability of organisms are those which are sensitive to membrane potential and present in metabolically active bacteria but absent in dead cells. Indicators of putative nonviability include nucleic acid dyes such as propidium, SYTOX Green and TO-PRO-3 (the latter two from Molecular Probes Inc., Eugene, OR) not taken up by viable bacteria but which enter dead organisms with damaged membranes.
Bacteria permeable to propidium, TO-PRO-3, etc. were presumed to have zero membrane potential; however, recent simultaneous analyses of membrane potential and nucleic acid dye uptake, done in the author’s laboratory, revealed that antibiotic treatment and certain culture conditions could render bacteria permeable to TO-PRO-3 while they retained normal membrane potential. Other researchers have also recently begun to do multiparameter flow cytometric analysis of nucleic acid dye uptake, membrane potential, and other functional parameters of bacteria, and it now appears that there is considerable variation in behavior from species to species.3 There may also be equally substantial differences among drugs in their cytometrically determined effects on bacterial physiology. All of this makes it unlikely that the one or two measurements will be able to determine susceptibility of all or a wide variety of bacterial species to a wide variety of antimicrobial agents.
However, cytometric bacterial counts can still provide a highly reliable, if somewhat low-tech, indicator of antimicrobial susceptibility. These can be determined directly (in instruments that use a volumetric pump for sample feed) and indirectly (by adding beads at a known concentration to each aliquot of sample). Counts also have an advantage over membrane potential and dye uptake measurements; both must be done on unfixed, potentially infectious organisms, while counts can be done on fixed, noninfectious samples.
This strategy has been used to develop a safe, rapid sensitivity test for Mycobacterium tuberculosis; counts of particles matching the scatter signals of the organisms, in samples fixed (and rendered sterile) with paraformaldehyde after incubation with antimycobacterial agents, provided reliable indicators of susceptibility. The cytometric test can be done after only a few days’ incubation; conventional methodology requires weeks, suggesting that cytometry could easily establish a niche for work with Mycobacteria and other slow-growing organisms, whether or not it is more generally used for other species.
Dr. Shapiro is president, Howard M. Shapiro, MD, PC, West Newton, MA. He has participated in the development of instrumentation and methodology for flow cytometry of both eukaryotic cells and microorganisms and is the author of Practical Flow Cytometry.
1. Gucker FT Jr, O’Konski CT, Pickard HB, Pitts JN Jr. A photoelectronic counter for colloidal particles. J Am Chem Soc 1947;69:2422-2431.
2. Alexander MK, Khan MS, Dow CS. Rapid screening for bacteriuria using a particle counter, pulse-height analyser, and computer. J Clin Pathol 1981;34:194-198.
3. Shapiro HM. Microbial analysis at the single-cell level: Tasks and techniques. J Microbiol Methods 2000;42:3-16.