“Mega” cytometry for a “mega” challenging cell type

Abstract

In this issue (page 302) of Cytometry Part A, Niswander et al. present a novel Imaging Flow Cytometry (IFC)-based approach for analysing Megakaryocyte (MK) phenotype and maturation state. MKs form a distinct lineage of the hematopoietic system and develop from the bone marrow-resident Hemocytoblast progenitor. During the maturation process as they prepare to become “factories” for platelet production, MKs undergo several rounds of DNA replication without completing cytokinesis (endomitosis) resulting in an increase in cellular ploidy and cytoplasmic volume. In response to as yet poorly defined cues, mature MKs form long branch-like processes called protoplatelets that eventually fragment to release platelets into the circulatory system. Platelets play a key role in the blood clotting network and perturbations in this process can have severe consequences for an organism 1. Traditionally, MKs are studied using a combination of conventional flow cytometry (CFC) and microscopy 2, 3. As Niswander et al. correctly state to be the motivation behind their study, the inherent cellular properties of MKs render them technically challenging to analyse by CFC or microscopy alone. To fully understand this statement, one needs to consider the inherent properties of MKs. For example, MKs can be highly diverse in terms of size, reaching upward of 50 μm in diameter. There is also a potential for shape alteration as they start to synthesise and release platelets. With regard to CFC-based analysis, there may be poor correlation between the size and shape of an MK and the forward scatter (FSC) measurement. FSC is not an accurate measure of cell size or morphology as it depends on the nature of the detector, the intrinsic properties of the particle and the refractive index differences between it and the sheath fluid. As such image-based measurements using fluorescence or transmitted bright field light provide a more accurate appraisal of MK status based on morphometric parameters, particularly with respect to identifying those with a high nuclear to cytoplasmic ratio that have shed their cytoplasm as a result of platelet production. Without spatially resolved signals, CFC based approaches will likely suffer from confounding issues caused by aggregates. The efficacy of eliminating aggregated events by CFC firstly relies on how well various pulse-derived parameters represent this state in measurement space and secondly how well the process of gate-based data reduction eliminates them from the analysis. These factors together may contribute to inclusion of false positives as well as the loss of actual MKs from the final analytical population. Moreover, the ability to accurately determine the ploidy state of an MK during the maturational progress of the cell may also be confounded by associated nucleated debris. CFC is a powerful method for determining relative nuclear content using fluorescent DNA binding dyes such as DAPI, PI and DRAQ5 whereas accurate ploidy analysis by traditional microscopy is far more challenging. If a single nucleated cell or piece of nuclear debris becomes attached to an MK this will most likely result in an overestimation of ploidy. Even the most efficient elimination of cell aggregates by CFC is unlikely to circumvent this issue. Certainly, the ability to restrict the measurement of DNA content using dye fluorescence to just the MK component of a cellular aggregate should improve the accuracy and precision of ploidy measurements. Finally, MKs are highly infrequent cells, representing <0.05% of the nucleated bone marrow population 4. While microscopy-based approaches will be better at identifying MKs based on morphology and to eliminate false positives, in statistical terms the number of cells that can be practically analysed is too few to be meaningful. As such the lack of key information afforded by CFC-based approaches is tolerated in exchange for the perceived benefit of sample throughput. Several groups, including our own, have found IFC can play a key role in the analysis of both rare cells and infrequent cellular occurrences. For example, the role and prevalence of asymmetric cell division in the adaptive immune response remains highly controversial. Analysis of rare telophasic events (0.1–0.01%) within defined division rounds traditionally employed a very inefficient, subjective approach requiring cytokinetic inhibitors to obtain more than a handful of events to measure how molecules are distributed over the cytokinetic plane. The high throughput, multiparametric properties of IFC allowed us to obtain statistically robust numbers of telophasic cells without the use of drugs such as cytochalasin 4-6, enabling us to show that such compounds may actually induce an artefactual molecular asymmetry in T cells 7. Without the inherent properties of IFC, this discovery would not have been possible. IFC-based approaches have also been elegantly applied to the analysis of circulating tumor (CTC; 8 and Endothelial cells (CEC; 9. In all cases, the high throughput and spatial information facilitated the accurate and precise identification of the target cells even when potentially confounding debris were attached. IFC provides high-throughput collection of multispectral imagery circumventing several of the limitations encountered when using CFC or classical microscopic approaches to study MKs. In particular, the ability to still analyse target events that are stuck to contaminant cells by directing the fluorescent measurements to only MK delineated pixels means that fewer cells will be lost during the process of data reduction and analysis. Certainly, IFC has been used previously to show that activated platelets, which share surface phenotype with MKs, will stick to other cells and may generate false positive events 10. All things considered, Niswander et al. rightly propose that IFC offers a unified approach to the analyses of MKs, bringing together the throughput and statistical rigour of CFC with the spatial information afforded by microscopy Another very encouraging aspect to the work by Niswander et al. in this issue (page 302) is the level of methodological detail and metadata they have provided. We recently proposed a set of standards for publishing work containing IFC-derived data that was in effect a modification of the MIFlowCyt guidelines 11. It includes all the existing CFC metadata requirements but also asks that key imaging components such as masking and feature extraction be included 12. Development, evaluation, and reproduction of all IFC-based methods absolutely require full disclosure of how the pixelated imagery has been collected, processed and cytometrically analyzed. The analysis workflow proposed by Niswander et al. is both elegant and logical, however should one need to make future modifications or additions we feel that full methodological disclosure will facilitate this. They provide information on how the masking has been formulated with example overlays on appropriate image sets. They also describe each measurement that has been made on the pixels under the masked areas. Moreover, they employ the concept of a morphometrically relevant biological (MRB) control to provide context to the spatial and morphometrically derived features that they use 12. One key element to IFC analysis is that unlike CFC, there may not be gross changes in the overall intensity of fluorescence signal, rather the spatial or morphometric context may change. The MRB control should be viewed as an extension to fluorescence minus one or absence of biological target controls for CFC 13, whereby the upper and lower limits for any morphometric or spatial measurement can be determined and gates/threshold set appropriately 12. We would argue that IFC could be considered a “paradigm” of cytometry as it facilitates multiparametric, single cell measurements to be made on comparatively large population sizes. Niswander et al. use this fact beautifully in their approach. However, one critique is that although they have managed to analyse ∼400 verified MKs this number is close to the threshold for accurately resolving DNA intensity peaks and determining ploidy. The authors state that this limit was set by the practical challenge of working with large 50,000 event multispectral image files. As the authors then select the CD41+ nucleated events post-acquisition from multiple sample runs to create a single concatenated file they could have overcome this issue of file size by setting a collection criteria on the MK CD41 signal to perform what is in effect a “virtual” sort. In more recent versions of the acquisition software, this can easily be done using gating logic on compensated fluorescence data. In older versions, it requires setting a classifier based on uncompensated values that could, if set inappropriately, lead to the irretrievable loss of data and as such should be practiced with caution. However, they circumvent much of the perceived issues around cell number and ploidy analysis by setting gates on the total DNA signal obtained from all nucleated cells and applying these to the MK populations to establish the location of the 2n and 4n peaks. In terms of additional parameters that could be included in this method we have previously used IFC to analyze the morphological properties of activated human platelets (unpublished data) and mouse dendritic cells 14. This was an interesting application as it goes against the widely accepted convention that a cell will lose all defining morphology in suspension. As such, it may be possible to identify MKs with some kind of alteration in membrane morphology that could correlate with the release of platelets. Moreover, if the cytoplasm of the MK were fluorescently labeled with CFSE-like dyes then one would assume the platelet progeny to also maintain such a label and this could be quantified by IFC. In conclusion, there are almost a limitless number of morphological and spatial parameters that could be derived from IFC-based analysis of MK biology. In the wider biological context, an IFC-based approach to the study of MK/platelet phenotype and function could be used to study various human conditions where numbers and or function of these cells are perturbed. These include a list of prominent disorders such as Essential Thrombocythemia, Congenital Amegakaryocytic Thrombocytopenia, postsurgical/chemotherapeutic complications, HIV infection, myelodysplastic syndromes, and aplastic anemia. IFC will certainly allow studies to be done using primary cells rather than more artificial in vitro models where the biological context may not reflect the true ex vivo state. It is fair to say that IFC is not the only way to study MK biology but it offers “mega” possibilities for these and other potentially challenging rare cells and rare cellular processes.