This correlation data indicate that when CD45RA down-regulates at the end find more of the naïve stage, CCR7 is indeed down-regulated, while CD28 is minimally up-regulated (see

Fig. 4B, blue hatched arrows). Our data are not consistent with the supposition that there is an extra stage as determined by CD45RA−CCR7+CD28+ ( Appay et al., 2002). Events with this phenotype captured by a gating strategy are most likely a mixture of naïve and CM events as defined by this analysis. The CD8+ average model also supports the hypothesis that when CD28 is down-regulated, CD45RA begins to be up-regulated (red arrows, r = 0.56, p < 0.01 with a difference of 1.9 (NS)). The last EF stage, is defined as the point at which the up-regulation of CD45RA has ended. During the developmental progression of memory and effector T cells, a subset of cells may begin to preferentially express markers that might not be expressed in the remaining cells. In PSM, the heterogeneous expression of markers can be visualized with branching expression profiles (see Fig. 5). Fig. 5A shows a progression schematic similar to Fig. 1 but includes a simple branch involving feature C. In this example, when cells reach the checkpoint where feature B is up-regulated, 70% of the cells also up-regulate feature C, while 30%

do not. Fig. 5B delineates the three probability state model EPs that model this simple branch (top = feature A, middle = feature B, and bottom = feature C). Fig. 5C summarizes this progression in the probability state model progression plot, which includes the branching of feature C (see the CB label). Fig. 5D

shows the associated probability Quisqualic acid state model surface dot plots for Venetoclax cell line feature A vs. B (top), feature A vs. C (middle), and feature B vs. C (bottom). Note that branches are not always visible in dot plots, which is why they have been traditionally difficult to detect. Branches are relatively easy to determine with PSM since non-branched EPs are incompatible with branched data, resulting in a dramatic loss of classified events and poor fitting. In this simple example, the branch point is at the end of Stage 2. However, when modeling T-cell branches, the location might be elsewhere along the progression axis. An averaged model featuring 22 samples from healthy donors was used to identify branched markers. Each sample was stained with antibodies against CD3, CD4, CD8, CD45RA, CD28, CCR7 (CD197), CD27, CD62L, CD57, and CD127. Fig. 6 shows the stratification expression profiles of CD45RA, CCR7, and CD28 as well as four branched EPs for CD62L, CD27, CD127, and CD57. Here, CD62L (l-selectin) has a 77% (9%) chance of down-regulating slightly before the end of the naïve stage and correlates best with the down-regulation of CCR7 (blue hatched arrows, r = 0.81, p < 0.00001, diff = − 4.23, NS). CD27 slightly down-regulates with CD45RA and CCR7 at the end of the naïve stage and then has a 75% (17%) chance of fully down-regulating in the middle of the CM stage.