Cytometry of single bacteria still represents one of the most advanced measurement methods available to microbiologists, particularly when applied to statistically relevant numbers of single cells. However, progress in the field has been hampered by difficulties in signal measurement and interpretation of the data. With their small size microorganisms are at the edge of detection sensitivity of flow cytometers thus requiring extra care to avoid and minimize interference factors. Another problem comes from the physiological variability of microorganisms. The problem was discussed as part of the 2008 meeting on Microbial Analysis at the Single Cell Level (http://qbab.dbs.aber.ac.uk/sc2008) to get consensus and broader acceptance of microbial cytometry method and data standardization.
Standardization can be applied in three domains, the sample manipulation, the measurement itself and the data interpretation as summarized in Figure 1. At the meeting in Bad Schandau, we started off an initiative to drive this forward. We will attempt to characterize a number of food grade/pharmaceutical grade nonhazardous reference stains against a base panel of stains between a number of laboratories to achieve agreement over the staining methods and the respective patterns. Although it is expected that due to the nature of diversity some organisms will behave differently than the reference strains, this should facilitate identification of situations in which the problem does not come from the staining method or the measurement end but can be attributed to the diversity of the biology under investigation.
As biological standards, we will try to establish a couple on nonpathogenic organisms between laboratories. The current aim is to use lyophilized starter culture to which we will just add a fixed volume of reconditioning solution. We have to come up with a simple and easy reproducible condition for the cells as the effect of nutrients, electrolytes, redox potential, and pH is important for the physiological behavior as is the growth stage of the microorganisms.
Particular care has to be taken about storage of stains. Sometimes stains susceptible to water are stored in DMSO. The degree of hydration of the DMSO can be deducted from its ability to freeze in the refrigerator (1). Also solvents and detergents used to solubilize/stabilize stains can influence dye uptake and they can also form interfering micelles. Propidium iodide (PI) and ethidium bromide (EB), for example can be dissolved in distilled water up to 1 mg/ml when stored at room temperature.
With regards to membrane permeabilization stains, PI remains the most reliable point of reference. Still, limitations of temporary permeabilization or physical stress (Fig. 2) or chemical stress including the presence of pore formers like sublethal concentrations of nisin have to be considered as well as other physiological exceptions (2).
There are several issues with regards to sample processing. Apart from the necessity to remove interfering bits from the sample matrix it is important to detach adhered cells first. At the same time, it is crucial to disaggregate the cells in order to generate single cell suspensions. Mechanical disruption by ultrasound is a convenient method but highly variable due to changes in the geometry of the arrangement, sample container or sample concentration (e.g., numbers of cells or other energy absorbing materials). It can lead to a substantial increase in cell counts by flow cytometry as well as in culture (3). However, it can destroy filamentous organisms altogether. Sonication has to be optimized for maximum counts under close observation of cell injury and should be performed before stains are added (see Fig. 2).
Cell aggregation can be due to chain formation as part of the growth phase or to carbohydrate-based interactions. The latter can sometimes be inhibited with EDTA or by adding appropriate sugars. Re-aggregation also has to be considered in the choice of washing steps based on filtration or pelleting.
Fixed cells are often used as a control/standard for dye uptake. However depending on the fixative used (heat, alcohol, or cross-linkers), the signal intensity of the fixed cells can be misleading when they are used as controls and compared with cells depolarized or permeabilized by other causes. Still they are important for instrument setup and should not be dismissed.
The key requirement is that the fluidic system is free of particles. This is best achieved by in line filtration close to the flow cell to minimize reintroduction of bubbles or particles from biofilms or disintegrating tubing itself. Commercial sheath solutions tend to contain bactericidal compounds such as sodium azide or sodium fluoride and detergents potentially interfering with the measurements. Salinity can affect cell physiology of bacteria as well as the light scatter response of the instrument as a difference in refractive index between sheath and sample can act like a lens system. When using small sample volumes, small volumes of sheath liquid coming into contact with the sample can affect the cell physiology.
Optics and Electronics
Because of the variations between instruments from different manufacturers and even within instruments of the same type there are limits to the amount of standardization possible. The best course is to characterize/calibrate the systems response using particle standards. Variations can come from the excitation side of the equipment (wave length, beam size and shape/profile, energy level at intercept, polarization) and from the emission side (filter selection, position of field stops/apertures) and the characteristics of the detectors as well as the electronics. Because of the different spectral responses between instruments the calibration to fluorochrome equivalents can only be done with particles with fluorochrome equivalent spectra.
Probably, the most critical part in the process is the signal discrimination as signals from bacteria and nanoparticles are very close to the instrument noise. A big problem in the published literature is the use of forward scatter as a trigger/discriminator. Whilst fairly robust for leukocyte detection it is the most variable signal between systems and it is most alignment critical. It is affected by refractive index mismatches between sheath and sample, beam geometry, polarization, beam stop position, and collection angle. In some cases the relative forward scatter position of particles of different sizes does not follow their relative order in physical size. In jet in air sorters the beam geometry and the jet undulation at the intercept are critical factors whereas in cuvette-based instruments these tend to be dirt on the optical surfaces and slight rotation of the flow cell to the beam axis.
To get a snapshot about instrument performance between different manufacturers I made up a mixture of particles that I tested on a lot of machines myself, sent out to some labs to kindly measure them for me on their instruments and even got some manufacturers to run them on their machines at the last ISAC meeting in Budapest on the stand. Interestingly but perhaps not surprisingly, all systems show good correlation with regards to right angle light scatter (RALS) or side scatter response. This is in part because all instruments aim for maximal light collection and thus are designed to achieve similar numerical apertures. The absence of the horizontal beam stop and the better optical coupling give a slight advantage to cuvette-based instruments. Figure 3 shows the response of a variety of commercial instruments with regard to side scatter of a mixture of latex beads normalized to the 660 nm particles that were part of the set. All instruments tended to be able to detect 380 nm latex beads whereas the 190 nm beads sometimes had to be separated from interfering noise signals by their slight green fluorescence. With analyzers, where critical alignment of forward scatter usually can not be performed by the user RALS is the more robust trigger/discriminator. Still discrepancies between particle light scatter and bacterial light scatter remain, as shown in Figure 4.
DATA ANALYSIS AND PRESENTATION
The subject of standardization of data analysis and presentation is worth an article in its own right and ISAC has a whole committee working on it. The minimum information about a flow cytometry experiment (MIFlowCyt) required for documentation has recently been addressed in this journal (4). One requirement is to describe the instrument details discussed above (optics, fluidics, and electronics configuration). These are keys to understand the data analysis process. I am also a serious advocate of the suggestion to make the list mode data available to reviewers for independent reanalysis and data validation. Failing that at least a comprehensive data display for all samples should be given for the benefit of the reviewer. The power of flow cytometry in general is the ability to correlate a variety of different measurements from a single cell. Therefore, it is good practice to look at least the correlation of fluorescence to light scatter for all collected signals which I would recommend to every user as a set of primary displays or even better “everything against everything,” as shown in Figure 5. This approach helps to identify interfering signals easily and also to identify the primary population of interest as even DNA stains can interact with interfering particles.
As usual, standardization requires a group of people to come together. So thanks and compliments to the organizers and participants of Microbial Analysis at the Single-Cell Level meetings as these serve as a useful platform to discuss such issues and help us to put forward an initiative to take this task on in the near future.