Cellular marker profiling
To demonstrate the extraction and analysis features of Cytokit as well as validate the underlying image processing libraries, a series of traditional immunofluorescence experiments were first conducted on human primary T cells. The first of these, shown in Figs. 6 and 7, comprised of primary human T cell samples stained with CODEX oligonucleotide-conjugated antibodies against human CD3, CD4, and CD8 (and a separate HOECHST stain). These slides were then imaged at 20X on a 1.9 mm × 2.5 mm grid of 25 images over an axial depth of 12.5 μm. The resulting 1008x1344x25 (height x width x depth) image volumes were then deconvolved, segmented, and quantified before being gated using Cytokit Explorer to isolate helper (CD4 positive) and cytotoxic (CD8 positive) subpopulations.
Fig. 6 T cell CD3 (blue), CD4 (red), and CD8 (green) intensity. a First row of 5 images in 5 × 5 experiment grid. b Single tile image with corresponding cell and nucleus segmentation, where cells are defined as a fixed radius away from the nucleus in the absence of a plasma membrane stain. c Center zoom on (b) showing co-expression of CD3 and CD4 (magenta) and CD3 and CD8 (cyan) as well as debris in DAPI channel
Fig. 7 T cell gating workflow as annotated Explorer screenshots. a Morphological and intensity gates applied to isolate CD3+ cells. b Cytotoxic and helper cell subpopulations. c Individual cell images matched to subpopulations in (b)
While these CD4 and CD8 positive (i.e. CD4+ and CD8+) populations were easily resolved in this experiment (Fig. 7), we found that dissociated cell samples like this are difficult to prepare without non-trivial amounts of debris and diffuse nuclear staining, usually as a result of lysed cells that did not survive the centrifuge. This is visible in Fig. 6 (c) where a minority of the nuclei segmentations resulting from the CellProfiler U-Net are fixed around roughly circular areas of greater DAPI intensity, albeit at low contrast. This invariance to contrast is generally very desirable, but it also demonstrates the importance of curation in image cytometry as artifacts like this can easily go undetected without a way to relate variations in inferred morphological or expression profiles for cells back to raw images.
A further investigation of the ability of this method to isolate CD4 + CD8- and CD4-CD8+ cell populations was also conducted on experimental replicates and validated against flow cytometry based surface marker profiling. Shown in Fig. 8, population proportions matched closely and verified that dissociated cells quantified in this way can produce results comparable to other methods; however, tools like Explorer and OpenCyto [25] were necessary to reach this degree of parity due to over-saturated image tiles (detected in Explorer) and intensity calibration differences resulting in substantial movement in the modes of the CD4/CD8 populations across replicates and donors. This latter issue was compensated for, in a downstream analysis, through the use of the t-distributed mixture models provided in OpenCyto (via flowClust [26]) that can capture translated distributions regardless of the intensity scale unique to each imaging dataset.
Fig. 8 T cell population recovery comparison (notebook). a CD4/CD8 gating results, as determined by automatic gating functions in OpenCyto [25], over two imaging replicates for each of 4 donors. While all images were collected over a field of view of the same size, samples for donors 40 and 41 were prepared at 3x higher cell concentrations to demonstrate that segmentation and intensity measurements are robust to greater image object densities. b Cell population size for both replicates compared to a single flow cytometry measurement for each donor as well as Pearson correlation demonstrating strong agreement between the two (r > 0.99, P < 0.0001, two-tailed t-test). c Cell images from donor 41 showing (from left to right): DAPI (blue) and PHA (red) stain, DAPI and PHA with cell and nuclei segmentations, and DAPI with CD4 (red) as well as CD8 (green). See supplementary file Additional file 1: Figure S1 for a comparison of these results to those from the same workflow without much of the gating used here to remove invalid cells