Memory Cells Can Undergo Asymmetric Cell Divisions
We first generated mice containing a defined population of antigen-experienced CD8+ T cells. A small number of P14 Thy1.1+ T cells were transferred to naïve wild-type mice, which were subsequently infected with LCMVarm. After >60 days, mice were infected with Listeria monocytogenes expressing gp33 (LMgp33) to specifically re-challenge the GP33-specific memory CD8+ T cells in vivo. At 42–46 hrs after re-challenge, Thy1.1+ memory CD8+ T cells were sorted for confocal microscopy.
CD3 and the IFN-γ receptor (IFN-γR) polarize to the immunological synapse (7
) and segregate asymmetrically in mitotic CD8+
T cells recruited into a primary immune response, hereafter referred to as primary responding T cells (3
). Using confocal microscopy, we found both of these proteins co-localized with the microtubule-organizing center (MTOC) of blasting, pre-mitotic memory cells (). In mitotic memory, CD3 and IFN-γR segregated to one side of the plane of division (). Cells with a central memory (CD62Lhigh
) phenotype appeared more likely to exhibit mitotic asymmetry than cells with an effector memory (CD62Llow
) phenotype (), which may be consistent with the suggested division of labor among memory subsets. Effector memory cells preferentially home to non-lymphoid tissues and exert immediate function at sites of pathogen re-entry without needing to divide, Central memory cells retain an intermediate state of differentiation, lymphoid migration, brisk mitotic potential and an apparent capacity to regenerate more memory cells while producing secondary effector cells (10
FIGURE 1 CD8+ T memory cells are capable of undergoing asymmetric cell division. A, WT mice received 5×105 Thy1.1+ P14+ splenocytes and were infected with LCMVarm. At 60+ days p.i., mice were re-challenged with LMgp33. At 42–46h after re-challenge, (more ...)
Memory cells asymmetrically segregate CD25 and T-bet to the same side of the dividing cell
IL-2 is thought to play a role in the re-expansion of memory CD8+
T cells during secondary infection (12
). We found that the alpha chain of the IL-2 receptor, CD25 was polarized in blasting () and mitotic () memory CD8+
T cells, as had been suggested for CD4+ T cell blasts (8
). We also found that the transcription factor T-bet was polarized during mitosis (), as suggested for primary responding cells (8
). Eomes, however, was not asymmetrically partitioned (), suggesting the two homologous transcription factors are regulated differently. Thy1.1 was also evenly distributed during mitosis, suggesting asymmetry is not a feature of all proteins during division ().
The ancestral polarity protein, protein kinase C-zeta (PKC-ζ), has been shown to have a role in T cell migration, activation, and asymmetric division of primary responding T cells (9
), as well as T cell differentiation during an immune response (16
). In pre-mitotic memory cell blasts, we found PKC-ζ polarized to the same side of the cell as the MTOC (), opposite of what was observed in primary responding T cells (3
). Moreover, PKC-ζ was localized to the same side of the cell as CD3 in mitotic memory CD8+
T cells (), also opposite from its localization in primary responding T cells (3
). PKC-ζ localized to the same side of the cell as both CD25 and T-bet (), suggesting one daughter cell could inherit more CD25 and T-bet than the other daughter cell.
FIGURE 2 CD25 and T-bet are polarized to the same side of a dividing memory CD8+ T cell. Cells were prepared for microscopy as in and stained for CD3, CD25 or T-bet (red), and PKC-ζ (purple). A, In interphase blasts PKC-ζ, localized to (more ...)
Why PKC-ζ localizes to the opposite side of a dividing memory CD8+
T cell than has been observed in primary responding CD8+
T cells is not yet clear. It has been suggested that PKC-ζ is part of a transcriptional signature shared between memory T and B cells and hematopoietic stem cells (17
). It is possible that preformed PKC-ζ protein must be segregated to the putative memory daughter of a primary responding naïve T cell in order to catalyze establishment of the memory cell fate. In re-activated memory cells, it may be unnecessary to donate greater PKC-ζ protein to maintain a less differentiated daughter if enhanced transcription of the gene encoding PKC-ζ is already an established, heritable trait of the memory parent cell. Other differences in the current model system, such as the primary challenge having been viral rather than bacterial may account for the difference in PKC-ζ localization. PKC-ζ function appeared critical for the asymmetry of T-bet (9
), yet the present findings suggest that T-bet is still asymmetrically inherited in memory cells with PKC-ζ localized on the opposite side of the cell as it was in naïve cells. This suggests that the critical, T-bet-positioning activity of PKC-ζ is independent of the precise localization of PKC-ζ protein, another mammalian atypical PKC (probably PKC-λ/ι) may subserve this function, or that the mechanism for T-bet polarization is not analogous between naïve and memory T cells.
T cell division upon re-challenge yields two distinct phenotypic cell subsets
To further investigate the early phenotype of memory CD8+
T cell progeny, CFSE-labeled Thy1.1+
memory cells were transferred secondarily to naïve mice that were subsequently infected with LMgp33. In uninfected recipients, transferred memory cells displayed heterogeneity of CD62L expression but remained undivided, CD25low
, and T-betlow
(). At the earliest point at which division could be detected, first generation memory daughter cells contained differing CD25, CD62L, and T-bet levels (). CD25high
cells had higher levels of CD8, higher side scatter (SSC), and lower CD62L levels compared to CD25low
cells (Supplemental Fig. 1A
), as has been observed in primary responding CD8+
T cells (3
cells also contained higher amount of T-bet, (Supplemental Fig. 1A
), which is consistent with the co-localization of CD25 and T-bet in mitotic memory cells (). It is, therefore, possible that CD25 and T-bet may be unequally inherited during memory cell mitosis.
FIGURE 3 Heterogeneity in the earliest divisions of memory CD8+ T lymphocytes upon re-challenge. Spleens were harvested at 60+ days after LCMV infection from mice that had received Thy1.1+ P14+ splenocytes. Cells were labeled with CFSE and transferred to naïve (more ...)
At slightly later times, we still detected two distinct populations of daughter cells with differential CD25 levels in the spleen (). Generally, cells that had undergone more than two rounds of division were CD25high
(), but a population of CD25low
cells that had undergone fewer than three divisions remained detectable (). Later generation CD25high
cells also contained higher T-bet, lower CD62L, and higher SSC than CD25low
cells (Supplemental Fig. 1B, 1C
). The observed heterogeneity in daughter cells was induced specifically by antigen-driven division since memory cells transferred into uninfected Rag1−/−
and wild-type recipients remained CD25low
daughter and parent cells, respectively (Supplemental Fig. 2
). These data suggest that, within the spleen, antigenic activation of memory cells results in two populations. CD25high
cells, which may represent transit amplification and differentiation of secondary effector cells. CD25low
cells may represent renewal of a less-differentiated memory pool. The observed bias in the localization of CD25low
daughter cells in lymph nodes () and bone marrow () may be consistent with their role as a regenerated memory cell pool.
The present data support a model wherein a resting memory CD8+ T cell may up-regulate markers of effector differentiation, such as CD25 and T-bet, upon re-encountering antigen. If the re-activated memory cell is capable of asymmetric, self-renewing division, it might beget one daughter cell that contains higher levels of CD25 and T-bet, divides more, and produces the majority of the secondary effector pool. The other, self-renewing daughter cell might inherit low levels of CD25 and T-bet, which facilitates less division and differentiation, thereby replenishing a central memory reservoir. The present data do not exclude conversion of CD25low to CD25high cells, which might even be necessitated if antigen or inflammation persists. The present findings provide a mechanistic basis for how the continual selection of validated clonotypes can accommodate the two mutually opposing demands of adult stem cells, terminal differentiation and self-renewal. Understanding how the process of self-renewal is maintained in infrequent re-challenges and stressed during chronic infection may offer new strategies for immunotherapy.