Best of both worlds

Microscopy provides high information content that can be used to discriminate cells based on their appearance but analysis is often subjective and lacks statistical power. Combining the information content of microscopy with the speed of flow cytometry in a single platform offers statistically quantitative image analysis for microscopy applications

Microscopy offers detailed cellular images and morphologic information, which are often essential for the study of cellular function. However, manual image acquisition is laborious and often subjective, making rigorous statistical quantification of microscopy assays difficult. Flow cytometry on the other hand is excellent for quantitative phenotyping and yields statistically robust results by rapidly interrogating large numbers of cells. However, flow cytometry lacks the per-cell information content common to microscopy, so sub-cellular localisation and cell function are measured indirectly.

[caption id="attachment_28278" align="alignright" width="200" caption="Figure 1: Composite imagery of erythroid lineage cells. FITC-CD71, PE-glyA, and DRAQ5 nuclear dye"][/caption]

Overcoming the limitations of both techniques by combining microscopy with flow cytometry using ImageStream technology enables quantitative statistical analysis of internalisation, morphology and cell signalling, even in rare events of immunophenotypically defined cells.

The following example shows the analysis and quantitation of cell morphological changes associated with erythroid stem cell differentiation. As hematopoietic stem cells differentiate through the erythroid lineage, they undergo a drastic change in size and eventually enucleate while progressing towards becoming RBC. These cells can also be immunophenotypically characterised by staining with anti-CD71 (transferrin receptor) and anti-glycophorinA (glyA) antibodies as in Figure 1. During the mid-phase of the differentiation program these cells express both markers. As the differentiation program comes to an end, erythroblasts continue to gain expression of glyA, but continually and gradually lose expression of CD71. To date, it has been difficult to adequately correlate the morphologic changes with immunophenotyping, especially since CD71+/glyA+ erythroblasts have highly variable morphologies.

[caption id="attachment_28279" align="alignleft" width="200" caption="Figure 2: Example data from erythroid lineage cells at day 10, 13 and 16 demonstrating the major points of this study: Day 10: Flow-based imaging enables high-content analysis of non-adherent cells. Day 13: The image analysis includes morphology as well as traditional fluorescence based measurements. Day 16: Rare subpopulations can be easily identified by localisation of various fluorophores. Left images are FITC-CD71 and the right images are composites of PE-glyA and DRAQ5"][/caption]

In this experiment, human CD34+ early hematopoietic cells were cultured to promote differentiation into erythroid lineage cells¹ and during the differentiation program cells were sampled at 10, 13, or 16 days (Figure 2) followed by staining the cells with a nuclear dye and for expression of CD71 and glyA.  The cells were then analysed immunophenotypically and morphologically on the ImageStream. The resulting data showed an initial reduction in both cellular and nuclear area from day 10 to day 13, followed by a loss of CD71 expression from day 13 to day 16. Many CD71?/glyA dim cells were either bare nuclei or were in the process of enucleating. This experiment demonstrates the unique ability of this method to objectively and conclusively classify thousands of immunophenotypically defined cells using a combination of intensity, morphology, and location-based parameters.

Live cell events with normal size (brightfield area) and scatter intensity are gated (R1) in Figure 3. Events with zero or one nuclear spot are gated (R2) in Figure 4 to include anucleated reticulocytes and single nucleated events. While some CD71+/glyA- erythroblasts are still present at day 10 (Figure 5), most cells have progressed to the double positive erythroblast phase. By day 16, many cells have lost expression of CD71 as they progress to the reticulocyte phase and then to mature RBC. A large population of double negative bare nuclei is seen as well.

[caption id="attachment_28285" align="alignright" width="200" caption="Figure 3: Identification of single nucleated cells using brightfield area and scatter intensity, R1"][/caption]

As hematopoietic stem cells progress to mature RBC, they gain expression of glycophorin A and lose CD71 expression by the end of the differentiation program. Immunophenotypically defined erythroblasts (CD71+/glyA+; Figure 6) are a heterogeneous group and include basophilic, polychromatophilic, and orthochromatic subsets, historically identified using histochemical stains^{2, 3}. A variety of morphologically distinct cell types can be distinguished in Figure 6: as the culture progresses, the cells and nuclei become smaller, and by day 16, some cells have begun to extrude or have already extruded their nuclei.

To quantify the percentage of enucleating cells, we used the Deltra Centroid XY feature, which measures the distance between the centres of the DRAQ5 and glyA images (Figure 7). If the nucleus is in the middle of the cell, the value will be small. If the nucleus is being extruded, the value will be larger than the radius of the cell.

The use of flow cytometry in combination with microscopy offers a completely new solution to studying, analysing and imaging erythroid lineage cells among others.  In this article, we have demonstrated how a platform combining the two technologies can be used for novel studies of internalisation, morphology and cell signalling.

[caption id="attachment_28286" align="alignleft" width="200" caption="Figure 4: Identification of single nucleated cells and anucleated reticulocytes, R2. Composite imagery of brightfield and DRAQ5 nuclear dye"][/caption]

[caption id="attachment_28287" align="alignright" width="200" caption="Figure 5: Early and late populations of interest. Day 10 shows a histogram of glyA intensity. Images of sample cells with low glyA intensity are shown. Day 16 shows bare nuclei and enucleated/enucleating cells in relation to the glyA intensity. Sample images are examples of bare nuclei"][/caption]

[caption id="attachment_28288" align="alignleft" width="200" caption="Figure 6: Histograms of CD71 intensity and corresponding CD71/glyA/DRAQ5 staining profiles at day 10, 13 and 16"][/caption]

[caption id="attachment_28289" align="alignright" width="200" caption="Figure 7: Quantitation of nuclear size and enucleating cells. Histogram of delta centroid XY of DRAQ5/glyA. Representative images of internal nucleus and extruding nucleus are shown"][/caption]

The Authors:

Thaddeus George, Raymond Kong and Haley Pugsley of EMD Millipore, and Amittha Wickrema and Hui Liu of the University of Chicago

Contact:

hpugsley@amnis.com +1 (0) 206-576-6866

References

1. Kang J-A., Zhou Y., Weis TL., Liu H., Ulaszek J., Satgurunathan N., Zhou L., Besien KV., Crispino J., Verma A., Low PS., Wickrema A. Osteopontin regulates actin cytoskeleton and contributes to cell proliferation in primary erythroblasts.  JBC 283:6997-7006 (2008).

2.  Wickrema A., Krantz SB., Winkelmann JC., Bondurant MC.  Differentiation and erythropoietin receptor gene expression in human erythroid progenitor cells.  Blood 80(8):1940-9 (1992).

3.  McGrath KE, Bushnell TP, Palis J.  Multispectral imaging of hematopoietic cells: where flow meets morphology.  J Immunol Methods336(2):91-7 (2008).