Research ArticleT CELLS

Modulation of asymmetric cell division as a mechanism to boost CD8+ T cell memory

See allHide authors and affiliations

Science Immunology  12 Apr 2019:
Vol. 4, Issue 34, eaav1730
DOI: 10.1126/sciimmunol.aav1730

RAPping their way to stemness

One interesting property of stem cells is their ability to undergo asymmetric cell division (ACD), which allows one of the daughter cells to retain stemness, whereas the other daughter cell becomes more differentiated. Akin to stem cells, memory cells in the immune system including memory T cells have the ability to undergo ACD. Borsa et al. have developed a strategy to enhance the ability of memory T cells to undergo ACD. They show that transient inhibition of the mTOR pathway using rapamycin enhances the memory potential of CD8+ T cells and their ability to undergo ACD. Developing methods to enhance T cell memory remains critical on several fronts, from vaccination to T cell–centric immunotherapies.

Abstract

Asymmetric partitioning of fate determinants is a mechanism that contributes to T cell differentiation. However, it remains unclear whether the ability of T cells to divide asymmetrically is influenced by their differentiation state, as well as whether enforcing asymmetric cell division (ACD) rates would have an impact on T cell differentiation and memory formation. Using the murine LCMV infection model, we established a correlation between cell stemness and the ability of CD8+ T cells to undergo ACD. Transient mTOR inhibition was proven to increase ACD rates in naïve and memory cells and to install this ability in exhausted CD8+ T cells. Functionally, enforced ACD correlated with increased memory potential, leading to more efficient recall response and viral control upon acute or chronic LCMV infection. Moreover, transient mTOR inhibition also increased ACD rates in human CD8+ T cells. Transcriptional profiling revealed that progenies emerging from enforced ACD exhibited more pronounced early memory signatures, which functionally endowed these cells with better survival in the absence of antigen exposure and more robust homing to secondary lymphoid organs, providing critical access to survival niches. Our data provide important insights into how ACD can improve long-term survival and function of T cells and open new perspectives for vaccination and adoptive T cell transfer therapies.

INTRODUCTION

CD8+ T cells play a key role in adaptive immunity and contribute to host defense and antitumor responses. After initial activation, CD8+ T cells proliferate and differentiate, acquiring effector functions. After clearance of the infection, most of these effector cells undergo apoptosis during the contraction phase, but a subset of memory precursors survives and differentiates into memory cells, which are long-lived, self-renewing cells that afford increased protection against secondary encounter with the pathogen (1). How differentiation into effector and memory cells occurs is still incompletely understood, although the contribution of specific transcription factors (2), metabolic profiles (37), epigenetic changes during the course of immune responses (8, 9), and asymmetric cell division (ACD) (1014) has been explored.

The contact interface between the antigen-presenting cell (APC) and the engaged T cell is termed immunological synapse. By establishing a stable immunological synapse, T cells undergo an extensive cytoskeleton remodeling to build a global polarization axis (15, 16), leading to the asymmetric partitioning of surface molecules, such as the T cell receptor (TCR), CD4/CD8, cytokine receptors, cell fate determinants such as transcription factors, and metabolism such as mTOR (mammalian target of rapamycin) activity (13, 14). The two emerging progenies exhibit different fate potentials (11) and transcriptional profiles (10, 17), with the proximal daughter adopting an effector cell fate and the distal daughter being destined to become a memory cell (10, 18). Most studies analyzing ACD have concentrated on naïve murine T cells undergoing their first mitosis after TCR engagement. However, with the exception of memory CD8+ T cells, which were also reported to undergo ACD (19), it is largely unknown whether other, more terminally differentiated, antigen-experienced T cell subsets, such as effector or exhausted CD8+ T cells, can undergo ACD. Furthermore, it also remains to be established whether ACD can be modulated with any functional consequence for T cell fate determination.

In this study, we report that ACD is an ability linked to cell stemness, as only naïve and memory CD8+ T cells could divide asymmetrically, whereas terminally differentiated effector cells and exhausted cells, which are unable to form a self-renewing reservoir, divided symmetrically. We identified a strategy to enforce ACD by transient inhibition of mTOR by rapamycin. Progenies emerging from enforced ACD conditions exhibited increased memory potential because they showed improved survival and expansion upon adoptive transfer, resulting in more efficient protection upon viral challenge. Furthermore, transcriptional profiling of progenies derived from symmetric and asymmetric divisions revealed strengthened memory signatures in “memory daughters” emerging from enforced ACD. Functionally, the early memory signature endowed these cells with improved survival and homing to secondary lymphoid organs. Together, these data define ACD as a key mechanism for differentiation and establishment of a memory reservoir in murine and human CD8+ T cells.

RESULTS

ACD is a feature of CD8+ T cells that retain stemness

As previous reports about ACD in lymphocytes focused on naïve and memory T cells, we investigated whether CD8+ T cells in other differentiation stages would also divide asymmetrically. To this end, we adoptively transferred naïve TCR transgenic P14 CD8+ T cells, which specifically recognize the gp33–41 peptide from the lymphocytic choriomeningitis virus (LCMV) glycoprotein, followed by acute or chronic LCMV infection to generate different CD8+ T cell subsets. Effector cells were isolated 8 days after LCMV WE infection, memory cells were isolated at least 30 days after LCMV WE infection, and exhausted cells were isolated 30 days after LCMV C13 infection. For analysis of naïve cells, we directly isolated CD8+ T cells from naïve P14 mice splenocytes. After stimulation on α-CD3–, α-CD28–, and human Fc–ICAM-1 (intercellular adhesion molecule–1)–coated plates for 30 to 36 hours, cells could be identified in different stages of mitosis and analyzed by confocal microscopy. No cell cycle synchronization was used in this work, thus avoiding any perturbation in the cytoskeleton network having an influence on the generation of asymmetry. We used CD8 cell surface distribution as a readout, based on previous reports that have shown the preferential polarization of this molecule toward the proximal daughter (1114, 18), and confirmed that naïve and memory CD8+ T cells could divide asymmetrically. By contrast, cells isolated from the peak of an acute LCMV infection, composed mostly of short-lived effector cells (SLECs), and exhausted cells, derived from a chronic infection, lacked the ability to generate asymmetry upon mitosis (Fig. 1, A to C). As it has been shown that the integrin ICAM-1 is important for establishment of asymmetry, we also submitted naïve cells to in vitro culture stimulation in the absence of plate-bound human Fc-ICAM-1 and confirmed that this resulted in a marked decrease of asymmetry rates (Fig. 1D). Our experimental observations concerning CD8 distribution in symmetric cell division (SCD) conditions led us to arbitrarily classify a mitosis as asymmetric when one of the daughter cells inherited 50% more CD8 than the other one. To clarify whether the in vitro stimulation using agonistic antibodies was equivalent to antigen presentation by APC, we stimulated P14 CD8+ T cells in a coculture with gp33–41 peptide–loaded dendritic cells. We observed comparable ACD rates under both conditions (fig. S1).

Fig. 1 ACD is a feature of CD8+T cell subtypes that retain stemness.

(A) Confocal images from fixed samples of CD8+ T cells 32 to 36 hours after in vitro (re-)activation on (+/-) Fc-ICAM-1–, α-CD3–, and α-CD28–coated plates. Naïve cells (n = 63) were obtained from wild-type C57BL/6 or P14 mice. Effector (n = 60), memory (n = 61), and exhausted (n = 24) cells were obtained upon sorting of adoptively transferred P14 cells after LCMV (WE or C13 strain) infection. Mitotic cells were identified on the basis of nuclear and β-tubulin structures and imaged during late anaphase or cytokinesis, where daughter cells could be distinguished. dpi, days post infection. (B) ACD rates are higher in subsets that retain stemness, namely, naïve and memory cells (30 days after LCMV WE infection). CD8 staining was quantified in both daughter cells, and P1 was arbitrarily set up as the pole (daughter cell) with higher amounts of CD8. Mitosis was classified as asymmetric when the amount of CD8 was 50% higher in one daughter cell in comparison with the other one, defining the threshold of 0.2 (dotted line in the graph). Data are represented as mean ± SEM. (C) Percentage of asymmetrically dividing cells from (B). (D) Stimulation in the absence of Fc-ICAM-1 leads to symmetric mitosis. Naïve CD8+ T cells were activated in α-CD3– and α-CD28–coated plates in the presence or absence of human Fc-ICAM-1. Data are represented as mean ± SEM. (E) Further subdivision of differentiation subsets revealed that SLECs and PD-1hi/low cells lack the ability to divide asymmetrically. Different cell subsets were obtained upon sorting of adoptively transferred P14 cells 8 days after LCMV WE infection (SLECs, n = 38; MPECs, n = 30), 30 days after LCMV WE infection (EM, n = 23; CM, n = 26) and >30 days after LCMV high-dose C13 infection (PD-1int, n = 47; PD-1hi, n = 32). (F) ACD rates were higher in subsets that retain stemness, MPECs, and MPs. Data are represented as mean ± SEM. (G) Percentage of asymmetrically dividing cells from (F). Statistical analysis was performed using the unpaired two-tailed Student’s t test. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 (see also fig. S1).

Because we had used bulk populations in our first experiments, we further analyzed whether specific subpopulations within these bulk populations would differ in ACD rates. Specifically, effector P14 cells from 8 days after LCMV WE infection were sorted into SLECs [killer cell lectin-like receptor subfamily G member 1 (KLRG1) interleukin-7 receptor alpha chain (IL-7Rα)] or memory precursor effector cells (MPECs; KLRG1 IL-7Rα+); P14 cells from 30 days after LCMV WE infection were sorted into effector memory cells (EM; CD44+ CD62L) or central memory cells (CM; CD44+ CD62L+); and P14 cells from 30 days after chronic LCMV C13 infection were sorted into PD-1int (CD44hi) exhausted cells that still retain moderate proliferative capacity and are known to be reinvigorated upon programmed cell death protein 1 (PD-1): programmed death ligand 1 (PDL1) blockade or terminally exhausted PD-1hi (CD44lo) cells (20). We observed that within the population of effector cells found 8 days after LCMV infection, ACD rates were higher for MPECs in comparison with SLECs. Furthermore, within the memory pool of cells, CM CD8+ T cells divided more asymmetrically than EM CD8+ T cells, whereas for PD-1int or PD-1hi, no asymmetry was observed (Fig. 1, E to G). These results support the concept that ACD correlated with cellular stemness, a feature of naïve and memory cells.

Transient mTOR inhibition enhances and restores ACD capability

As we observed a correlation between cell stemness and the ability of a CD8+ T cell to undergo ACD, we investigated whether we could modulate asymmetry rates in the different subsets of CD8+ T cells. mTOR inhibition has been reported in T cells to benefit memory formation (21, 22), and the inhibition of the homologous TOR pathway leads to higher asymmetry rates in yeast cells (23). mTOR signals via two different complexes, mTORC1 and mTORC2 (24), containing distinct scaffold-associated proteins, Raptor and Rictor, which define the downstream targets of each complex. mTORC1 activity is typically measured by the phosphorylation of S6 kinase β-1 (P-S6), whereas mTORC2 activity is commonly quantified by phosphorylation of the kinase Akt at serine 473 (S473) (25). We therefore transiently inhibited mTOR by treatment with rapamycin (preferential inhibitor of mTORC1) or Akt-kinase inhibitor (which specifically inhibits phosphorylation of Akt at S473, downstream of mTORC2) (fig. S2).

In activated naïve cells, we observed increased CD8 and T-bet asymmetry in mTOR-inhibited daughter cells, under both transient rapamycin and Akt-kinase inhibitor treatments (Fig. 2, A to C, and fig. S3). mTOR inhibition did not induce ACD in the absence of plate-bound Fc-ICAM-1, suggesting that the polarization initiated by the immune synapse, which is disturbed in the absence of ICAM-1, is necessary to generate the asymmetry that is maintained during mitosis and enhanced by transient mTOR inhibition. Transient inhibition after initial TCR activation was crucial to promote ACD rates, because continuous mTOR inhibition partially prevented mitosis and, within mitotic cells, did not increase ACD rates in comparison with untreated cells (fig. S4A). Similar results to continuous rapamycin treatment were obtained when CD8+ T cells from CD4-cre Raptorfl/fl or CD4-cre Rictorfl/fl mice (lacking mTORC1 and mTORC2 activity in T cells) were stimulated in vitro under ACD conditions (fig. S4B). The transient nature of mTOR inhibition was further confirmed by washing out rapamycin from the cell culture medium 30 to 36 hours after activation and by evaluating P-S6 levels 8 hours later, when we observed reestablishment of its expression with preserved asymmetry (fig. S4C).

Fig. 2 Transient mTOR inhibition enhances and restores ACD capability.

(A) Confocal microscopy images from naïve C57BL/6 or P14 mice CD8+ T cells 32 to 36 hours after stimulation on α-CD3– and α-CD28–coated plates in the presence or absence of Fc-ICAM-1 [−ICAM, n = 48; −ICAM R, n = 28; +ICAM, n = 90; +ICAM R, n = 80; +ICAM Akt i (Akt-kinase inhibitor), n = 69]. (B) Percentage of asymmetrically dividing cells from (C). (C) ACD rates are higher in cells that were submitted to transient mTOR inhibition with both rapamycin or Akt-kinase inhibitor. Data are represented as mean ± SEM. (D) Confocal microscopy images of mitosis from SLECs (n = 12), MPECs (n = 25), EM (n = 13), CM (n = 20), PD-1int (n = 32), and PD-1hi (n = 24) CD8+ T cells after restimulation. (E) Percentage of asymmetrically dividing cells from (F). Frequencies of symmetrically and asymmetrically dividing cells from conditions with no pharmacological intervention (+ICAM) are depicted, respectively, in white and gray. Frequencies of asymmetrically dividing cells upon transient rapamycin treatment (+ICAM R) are shown in black. (F) ACD is increased upon transient mTOR inhibition in MPECs, EM, and CM CD8+ T cells, and reestablished in PD-1int CD8+ T cells. Asymmetry rates shown in Fig. 1F (white) are depicted on the left side of data from the same subsets obtained upon transient mTOR inhibition (black). Data are represented as mean ± SEM. (G) Confocal microscopy images of mitoses of PD-1int TCF-1lo or TCF-1hi cells submitted or not to transient mTOR inhibition during stimulation (TCF-1lo, n = 16; TCF-1lo R, n = 10; TCF-1hi, n = 15; TCF-1hi R, n = 11). Statistical analysis was performed using the unpaired two-tailed Student’s t test. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 (see also figs. S3 and S4).

After establishing an efficient method of increasing asymmetry in naïve CD8+ T cells, the same approach was applied to the CD8+ T cells in other differentiation stages, using rapamycin as mTOR inhibitor (Fig. 2, D to F). CD8 asymmetry rates were increased in memory precursor cells, EM, and CM, but SLECs continued to divide symmetrically. However, more unexpected was the fact that among the exhausted cells, PD-1int CD8+ T cells gained the ability to divide asymmetrically upon transient mTOR inhibition. As TCF-1 was reported to be expressed in a population of memory-like T cells that sustains T cell responses in chronic infections (26), we sorted PD-1int CD8+ T cells into PD-1int TCF-1lo and PD-1int TCF-1hi cells. We observed that TCF-1hi cells can divide asymmetrically and show increased asymmetry rates upon transient rapamycin treatment during stimulation (Fig. 2G). Moreover, similarly to what was observed for naïve CD8+ T cells, higher T-bet asymmetry rates were also observed upon transient mTOR inhibition for MPECs and PD-1int CD8+ T cells (fig. S3). Together, these results show that ACD can be modulated in asymmetrically dividing cells by short-term mTOR inhibition during mitosis.

Higher ACD rates lead to better memory potential and viral control

Having established a mechanism to modulate ACD rates, we investigated whether higher asymmetry rates would have any functional consequences. We stimulated naïve P14 cells under asymmetric or symmetric conditions with or without transient rapamycin treatment before transfer into naïve C57BL/6 recipient mice. Recipients were infected with LCMV WE 30 days after transfer (Fig. 3A). We monitored the in vivo expansion of the adoptively transferred P14 cells in the blood and spleen of recipient mice, reporting on both survival of the cells in the absence of antigen before challenge and their secondary expansion potential. We consistently observed higher frequencies of progenies derived from cells that had been stimulated in vitro in the presence of human Fc-ICAM-1 and transiently treated with mTOR inhibitors (Fig. 3, B and C). This increase was partially accompanied by phenotypic changes, with P14 cells submitted to transient mTOR inhibition showing significantly lower frequencies of KLRG1hi IL-7Rαlo cells after in vivo recall, and a trend toward higher frequencies of KLRG1lo IL-7Rαhi cells (Fig. 3D). However, when effector cytokine production was assessed, we observed no differences in frequencies of interferon-γ (IFN-γ) or tumor necrosis factor (TNF)–expressing P14 cells. The higher numbers of total cytokine producers were a consequence of the higher numbers of P14 cells after in vivo recall (Fig. 3E).

Fig. 3 Higher ACD rates lead to better memory potential and viral clearance.

(A) Experimental setup. Naïve P14 cells were stimulated under different conditions, harvested after 36 hours, and adoptively transferred into wild-type recipients. Each recipient mouse received 10,000 cells and was infected intravenously with LCMV WE 30 days later. ffu, focus forming units. (B) Frequency of P14 cells within the CD8+ T cell population in the blood of recipients that were infected with LCMV WE 30 days after adoptive transfer. (C) Frequency and numbers of adoptively transferred P14 cells within CD8+ T cells in the spleens of recipient mice. (D) Representative plots of KLRG1 and IL-7Rα expressing splenic P14 cells stimulated on human Fc-ICAM-1–, α-CD3–, and α-CD28–coated plates in the presence or absence of rapamycin for 30 to 36 hours before adoptive transfer (left); percentages of KLRG1+ IL-7Rα and KLRG1 IL-7Rα+ P14 cells in the spleens of recipient mice (right). (E) Frequencies and numbers of IFN-γ– and TNF-producing cells. (F) Top: Experimental setup. P14 naïve CD8+ T cells were stimulated under different experimental conditions for 36 hours and further adoptively transferred to wild-type recipients. Each recipient mouse received 10,000 cells and was infected intravenously with LCMV WE or LCMV C13 30 days after transfer. Bottom: Increased memory potential derived from higher ACD rates led to better viral clearance upon LCMV infection. LCMV titers obtained 4 days after infection in spleens of recipient mice that were infected with LCMV WE and 12 days after infection in blood and spleens of recipient mice that were infected with LCMV C13. (G) Frequency and numbers of adoptively transferred P14 cells within CD8+ T cells in the spleens of recipient mice. (A to C) Data pooled from five independent experiments. (D) Data pooled from three independent experiments. (E) Representative data from one of two experiments. (F and G) Representative data from one of four experiments. Data are depicted as mean + SEM. Statistical analysis was performed using the unpaired two-tailed Student’s t test or, when data did not pass the normality test, the unpaired Mann-Whitney U test. *P < 0.05; **P < 0.01; ***P < 0.001 (see also figs. S4 to S6).

Next, we screened for other drugs that target downstream effectors of mTOR (fig. S5A). We found that similar to rapamycin, inhibition of hypoxia-inducible factor–1α (HIF-1α) enhanced ACD rates (fig S5A). Conversely, FTY720, a sphingosine-1-phosphate receptor agonist well known for altering lymphocyte trafficking (27, 28) but also a preferential inhibitor of long fatty acid ceramide synthesis (29), decreased ACD rates (fig. S5A). As expected, lower ACD rates resulted in impaired re-expansion potential when activated P14 cells were transferred after in vitro activation to naïve C57BL/6 hosts, followed by LCMV infection (fig. S5, B and C).

We then tested the protective capacity of progenies from “regular” and enforced ACD after exposure to acute or chronic LCMV infection (Fig. 3F). We found that P14 CD8+ T cells that were transiently exposed to rapamycin upon in vitro stimulation conferred significantly better protection against infection with both the WE and the C13 strains of LCMV in comparison with untreated progenies (Fig. 3F). The enhanced protection correlated with higher frequencies and numbers of P14 cells in the spleens of infected recipient mice (Fig. 3G).

Next, we asked whether an increase of ACD rates by transient rapamycin treatment would also lead to enhanced memory potential in P14 cells in other differentiation states. We therefore stimulated CM P14 cells in vitro with α-CD3, α-CD28, and human Fc-ICAM-1 in the presence or absence of transient rapamycin treatment, followed by adoptive transfer into naïve wild-type recipients and LCMV rechallenge 30 days later (fig. S6A). Similar to naïve P14 cells, transient rapamycin treatment not only increased ACD rates but also conferred enhanced memory potential to CM, reflected by the increased frequencies within CD8+ T cells and total numbers of P14 cells in the blood, spleen, and lungs of recipient mice after acute LCMV infection (fig. S6, B and C). Also in line with what was observed for naïve cells, frequencies of cytokine-producing cells were similar among progenies from all in vitro experimental conditions (fig. S6D). Together, these data provide strong evidence that the enhanced frequencies and numbers of P14 cells after recall antigen exposure correlate with enforced ability to undergo ACD.

The polarisome is required for ACD modulation by transient mTOR inhibition

To provide a mechanistic link between (enforced) ACD and memory potential, we sought interventions that would abolish ACD and would allow us to evaluate its functional consequences. Polarisome activity and establishment of polarity depend on protein kinase Cζ (PKCζ) activity (10, 12, 30). We therefore abrogated polarisome activity by inhibiting PKCζ with aurothiomalate (AuTM) (31, 32). Naïve CD8+ P14 T cells were activated in vitro with α-CD3, α-CD28, and human Fc-ICAM-1 in the presence or absence of rapamycin and/or AuTM (Fig. 4A). We observed that inhibiting PKCζ, rapamycin treatment was not able to increase asymmetric distribution of both CD8 and T-bet (Fig. 4, B and C). We then tested whether abrogation of ability to divide asymmetrically due to polarisome disruption would affect memory potential, even in cells that were exposed to transient rapamycin treatment. For that, P14 naïve CD8+ T cells were stimulated under enforced ACD conditions in the presence or absence of AuTM, and their progenies were adoptively transferred to wild-type recipients, followed by LCMV WE infection after 30 days (Fig. 4D). We observed significantly higher frequencies and numbers of P14 cells in the spleens and lymph nodes of mice that had received progenies from enforced ACD conditions in the absence of AuTM compared with mice that had received progenies from ACD conditions in the presence of AuTM (Fig. 4E). These data provide strong evidence for a causal link between memory potential and the ability to undergo ACD.

Fig. 4 The polarisome is required for ACD modulation by transient mTOR inhibition.

(A) Inhibition of phosphorylated extracellular signal–regulated kinase (P-ERK) was used as a readout for AuTM efficiency on PKCζ inhibition. (B) Confocal microscopy images from stimulated naïve CD8+ T cells obtained from P14 mice (+ICAM R, n = 52; +ICAM AuTM, n = 33; +ICAM AuTM R, n = 32). (C) CD8 and T-bet asymmetry upon transient PKCζ inhibition. Data are represented as mean ± SEM. (D) Experimental setup. P14 naïve CD8+ T cells were stimulated under different experimental conditions for 36 hours and adoptively transferred to wild-type recipients. Each mouse received 10,000 cells and was infected intravenously with LCMV WE 30 days after transfer. (E) Frequencies and numbers of adoptively transferred P14 cells within CD8+ T cells in the spleens and lymph nodes (LNs) of recipient mice 10 days after LCMV infection. Data are depicted as mean + SEM. (A to C) Data pooled from two independent experiments. (D and E) Representative data from one of two experiments. Statistical analysis was performed using the unpaired two-tailed Student’s t test or, when data did not pass the normality test, the unpaired Mann-Whitney U test. *P < 0.05; **P < 0.01; ****P < 0.0001.

Increased ACD and memory potential can be induced in exhausted CD8+ T cells expressing intermediate levels of the co-inhibitory receptor PD-1

The robust improvement in memory CD8+ T cell potential in naïve and CM by enforced ACD led us to investigate whether we could also reestablish memory potential in PD-1int cells, because this subset responded to transient rapamycin treatment by gaining, to some extent, the ability to divide asymmetrically. P14 PD-1int cells were stimulated in vitro with α-CD3, α-CD28, and human Fc-ICAM-1 in the presence or absence of transient rapamycin treatment, and the resulting progenies were transferred into new naïve mice that were rechallenged with LCMV WE 30 days later (Fig. 5A). We found that the reestablishment of ACD resulted in a considerable gain of re-expansion potential upon LCMV infection (Fig. 5, B and C). Of relevance, this re-expansion potential was only seen in cells that were stimulated in the presence of plate-bound Fc-ICAM-1 upon transient mTOR inhibition, also the only condition where PD-1int cells were able to undergo ACD. Besides a drastic numeric increase, these cells also exhibited increased memory markers, characterized by the expression of IL-7Rα (Fig. 5D), and higher frequencies of effector cytokine (IFN-γ, TNF)–producing cells (Fig. 5E). Thus, the restoration of effector functions in a previously exhausted T cell population not only corroborates the idea that installing ACD leads to an improved memory potential but also opens new perspectives for therapies that could reinvigorate effector responses in chronically infected individuals.

Fig. 5 Exhausted CD8+T cells expressing intermediate levels of the co-inhibitory receptor PD-1 show recovered memory potential when ACD capacity is reestablished.

(A) Experimental setup. P14 exhausted CD8+ T cells (PD-1int CD44hi cells isolated from spleens of recipient mice >30 days after LCMV C13 infection) were stimulated under different experimental conditions for 36 hours and further adoptively transferred to wild-type recipients. Each mouse received 10,000 cells and was infected with LCMV WE 30 days after transfer. (B) Frequency of P14 cells within the CD8+ T cell population in the blood. (C) Frequency and numbers of adoptively transferred P14 cells within CD8+ T cells in the spleens and lungs. (D) Representative plots of KLRG1 and IL-7Rα expressing splenic P14 cells stimulated on human Fc-ICAM-1–, α-CD3–, and α-CD28–coated plates in the presence or absence of rapamycin for 30 to 36 hours before adoptive transfer (left); percentages of KLRG1+ IL-7Rα and KLRG1 IL-7Rα+ P14 cells in the spleens of recipient mice (right). (E) Frequencies and numbers of IFN-γ– and TNF-producing cells. Data from two pooled experiments. Data are depicted as mean + SEM. Statistical analysis was performed using the unpaired two-tailed Student’s t test or, when data did not pass the normality test, the unpaired Mann-Whitney U test. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

Human naïve and CM CD8+ T cells show higher ACD rates upon transient mTOR inhibition

We next examined whether human lymphocytes would respond to transient mTOR inhibition similarly to murine cells. Using identical in vitro conditions for stimulation and cell culture as previously described for CD8+ T cells from mice, human naïve (CD45RA+, CD62L+) (Fig. 6, A to C) and CM (CD45RO+, CD62L+) (Fig. 6, D to F) CD8+ T cells showed increased asymmetry rates upon transient mTOR inhibition when cultured in the presence of plate-bound Fc-ICAM-1. Increased ACD rates led to better survival when cells were kept in in vitro culture under limiting IL-15 conditions (fig. S7). Thus, transient rapamycin treatment can modulate asymmetry also in human CD8+ T cells, being a potentially useful tool to enhance memory T cell generation in adoptive transfer settings.

Fig. 6 Human naïve and CM CD8+T cells show higher ACD rates upon transient mTOR inhibition.

(A) Confocal images from fixed samples of human naïve CD8+ T cells 32 to 36 hours after in vitro stimulation. Transient mTOR inhibition (12 hours after activation until fixation) was achieved by treatment with 20 nM rapamycin. Stimulation was done on α-CD3– and α-CD28–coated plates in the presence or absence of Fc-ICAM-1 (−ICAM, n = 11; −ICAM R, n = 10; +ICAM, n = 37; +ICAM R, n = 42). (B) ACD rates in the presence or absence of transient mTOR inhibition with rapamycin. Data are represented as mean ± SEM. (C) Percentage of asymmetrically dividing cells from (B). (D) Confocal images from fixed samples of human CM CD8+ T cells 32 to 36 hours after in vitro stimulation. Transient mTOR inhibition (12 hours after activation until fixation) was obtained upon treatment with 20 nM rapamycin. Stimulation was done on α-CD3– and α-CD28–coated plates in the presence or absence of Fc-ICAM-1 (−ICAM, n = 7; −ICAM R, n = 7; +ICAM, n = 15; +ICAM R, n = 19). (E) ACD rates in cells that were submitted to transient mTOR with rapamycin. Data are represented as mean ± SEM. (F) Percentage of asymmetrically dividing cells from (E). Pooled data from peripheral blood mononuclear cells of four different healthy donors. Statistical analysis was performed using the unpaired two-tailed Student’s t test. *P < 0.05; **P < 0.01 (see also fig. S5).

ACD generates progenies with strengthened memory signatures

To experimentally assess whether the better maintenance and/or proliferation potential exhibited by progenies derived from enforced ACD conditions resulted from higher ACD rates or as a “side effect” of rapamycin treatment, we sorted P14 CD8hi and CD8lo daughter cells after their first mitosis following in vitro stimulation with or without transient mTOR inhibition. Cells were then transferred to naïve C57BL/6 recipients, which were LCMV-infected 30 days later. Quantification of P14 frequencies 1 to 3 weeks after infection in the blood of recipients showed that CD8lo cells expanded significantly better than CD8hi cells, irrespective of previous rapamycin treatment, confirming that the increased ACD rates imposed by transient rapamycin treatment inferred enhanced memory potential to CD8+ T cells (Fig. 7A). These results corroborate previous observations on a better memory potential inherited by CD8lo daughter cells (11, 13).

Fig. 7 ACD generates progenies with strengthened memory signatures.

(A) Naïve P14 cells were stimulated on human Fc-ICAM-1–, α-CD3–, and α-CD28–coated plates in the presence or absence of rapamycin (initiated 12 hours after stimulation) for 30 to 36 hours. Cell trace yellow (CTY) dilution was used to identify first daughter cells, from which CD8lo and CD8hi cells were sorted and transferred to recipient mice. Thirty days after adoptive transfer, recipients were infected with LCMV WE. Frequencies of P14 cells in the blood of mice (bottom); CD8 expression in first daughter cells that were submitted or not to rapamycin treatment before adoptive transfer (top). Representative data from one of two experiments. Data are depicted as mean ± SEM. Statistical analysis was performed using unpaired two-tailed Student’s t test or, when data did not pass the normality test, the unpaired Mann-Whitney U test. ns, not significant. (B) Differential expression of genes among CD8lo and CD8hi first daughter cells (top 100 differentially expressed genes based on the lowest P value) presented as a heatmap. (C) Expression of selected transcription factors previously associated with CD8+ T cell differentiation into effector or memory cells. (D) Expression of selected genes associated with survival and homing to secondary lymphoid organs, features of long-lived memory cells. (E) Principal components analysis of 50 selected genes previously associated with CD8+ T cell differentiation. Green, effector gene signatures; blue, memory gene signatures. (F) Differential expression of the 50 selected genes among CD8lo and CD8hi first daughter cells derived from SCD, standard ACD, and enforced ACD.

To gain mechanistic insights into the processes that endow CD8+ T cells with increased memory potential upon enforced ACD, we compared the transcriptional profiles of CD8lo (potential memory fate) and CD8hi (potential effector fate) first daughter cells derived from progenies of naïve CD8+ T cells that were stimulated under SCD (no Fc-ICAM-1) or ACD (in the presence of Fc-ICAM-1) conditions in the presence or absence of transient mTOR inhibition. The distinct gene expression profiles presented by these eight different daughter cell populations reflected their original CD8 expression profiles, classifying them as “effector daughters” (CD8hi) or “memory daughters” (CD8lo), but also whether plate-bound Fc-ICAM and transient rapamycin treatment were present or not during stimulation. Within CD8lo and CD8hi cells, gene signatures were shared despite the nature of the stimulation they originated from. However, differences between CD8lo and CD8hi daughters were more pronounced under ACD stimulation conditions, confirming the contribution of this mechanism to generate diversity (Fig. 7B). To address whether distinct memory signatures would be enriched in CD8lo daughters and particularly within CD8lo daughters emerging after enforced ACD, we quantified the expression of transcription factors classically associated with effector and memory CD8+ T cell fates. As expected, CD8lo daughters showed higher expression of Tcf7 and Bcl6, whereas CD8hi cells expressed more Tbx21 and Hif1a (Fig. 7C). The differences between CD8lo and CD8hi cells were stronger among daughter cells emerging from ACD conditions compared with those emerging from SCD conditions, and CD8lo daughter cells derived from enforced ACD showed the most pronounced memory phenotype. Furthermore, expression of genes linked to homing and survival, features of memory cells, followed a similar trend, with CD8lo cells originating from enforced ACD conditions showing higher expression of Il7r, Sell, Ccr7, and Cxcr3 (Fig. 7D). Sell and Cxcr3, as a number of other genes, were highly influenced by rapamycin treatment, defining a synergistic impact of ACD and mTOR inhibition on the gene signature of CD8lo cells originating from enforced ACD conditions, explaining their better memory potential. ACD also led to distinct profiles of cyclin expression. CD8lo cells expressed considerably lower levels of Ccnd2, Ccnd3, and Cdc25c compared with CD8hi cells (fig. S8). This gene expression profile was supported by functional data, as the proliferation rates observed for progenies of naïve CD8+ T cells coming from enforced ACD conditions were slower than the ones observed in standard ACD conditions or SCD conditions (fig. S8).

Distinct gene signatures of fully formed effector and memory CD8+ T cells have been extensively described. Recently, defined expression clusters being up- or down-regulated in effector cells (KLRG1+IL-7Rα, 8 days after LCMV infection) in comparison with memory cells (CD44+CD62L+, 60 days after LCMV infection) were reported (33). We compared the transcriptional profile of first CD8lo and CD8hi daughter cells emerging from SCD, standard ACD, and enforced ACD stimulation conditions with the expression profiles of fully differentiated effector and memory cells. CD8hi daughter cells presented a transcriptional profile more similar to that of effector CD8+ T cells than their CD8lo counterparts (fig. S9, left). Among the CD8lo cells, the ones derived from ACD exhibited gene signatures closer to memory cells, especially the ones derived from enforced ACD conditions (fig. S9, right). It needs to be emphasized, however, that the differentially expressed genes defining “effector” and “memory” clusters were established by a comparison between CD8+ T cells that were much further in their differentiation than first daughter cells (33). As such, it is remarkable that already first daughter cells tend to show transcriptional similarities to their future to be differentiation states. The only available transcriptional profiling of first daughter cells previously reported used single-cell RNA sequencing data (17), which make direct comparisons technically challenging owing to significant differences in sequencing depth. To obtain a better insight into an early (potential) memory gene signature among the CD8lo daughter cells, we selected 50 genes linked to T cell differentiation that were differentially expressed both in our dataset (first daughter cells) and in the one from He et al. (33) (fully formed effector and memory cells). Among the CD8hi cells, the ones deriving from standard ACD conditions had a gene signature more similar to effector cells, whereas enforced ACD gave rise to CD8lo progenies with a stronger memory signature (Fig. 7E). Moreover, the differences in gene expression between CD8lo and CD8hi cells were generally higher between daughters deriving from ACD conditions (Fig. 7F), confirming a role for ACD in the generation of diversity and cells with enhanced memory potential.

Enforced ACD generates daughter cells with improved homing features

The transcriptional profiling of CD8lo and CD8hi daughters originating from distinct stimulation conditions highlighted the increased expression of genes associated with survival and homing to secondary lymphoid organs in CD8lo cells, particularly after enforced ACD.In vitro, progenies of transiently mTOR-inhibited cells survived better under limiting IL-15 conditions (fig. S7). To functionally corroborate whether this would also reflect in improved T cell homing and access to survival niches in vivo, we isolated CD8lo and CD8hi daughter cells from SCD, standard ACD, and enforced ACD stimulation conditions, and equal numbers of each type of cells were transferred into naïve hosts. After 6 to 7 hours, we harvested secondary lymphoid organs and peripheral organs from the recipient animals (Fig. 8A). We consistently observed significantly higher frequencies and numbers of CD8lo cells in secondary lymphoid organs in comparison with their CD8hi counterparts (Fig. 8B). In the peripheral organs analyzed (lungs and liver), this was also observed but to a lower extent. CD8lo cells from enforced ACD conditions had the best performance concerning homing to secondary lymphoid organs. To confirm whether the homing would also reflect in access to survival niches, we analyzed sections of spleens from the recipient mice. CD8hi cells could not be found in most of the spleen sections analyzed, and when present, they were mostly localized in the red pulp. CD8lo cells were found in the white pulp, localized mostly in the T cell zone, and were there in significantly higher numbers when originating from enforced ACD (Fig. 8C). To experimentally address whether better homing to secondary lymphoid organs also reflected in improved maintenance of donor cells, we transferred equal numbers of P14 cell progenies originating from symmetric, asymmetric, or enforced ACD stimulation conditions into naïve wild-type recipients with no further viral challenge. After 4 weeks, we observed significantly higher frequencies of enforced ACD progenies in the spleen and lymph nodes of recipient mice (Fig. 8D). Our results strongly suggest that ACD contributes to the rise of CD8lo cells endowed with homing potential to T cell zones of secondary lymphoid organs, a feature that is improved upon enforced ACD and that contributes to their survival and the observed increased re-expansion upon antigen reencounter.

Fig. 8 Enforced ACD generates daughter cells with improved homing features to T cell zones of secondary lymphoid organs.

(A) Experimental setup. Naïve P14 cells were stimulated for 30 to 36 hours on α-CD3– and α-CD28–coated plates, in the presence or absence of plate-bound Fc-ICAM-1 and/or transient mTOR inhibition by rapamycin (starting 12 hours after activation). CTY dilution was used to identify first daughter cells, from which CD8lo and CD8hi cells were sorted and transferred to recipient mice. Each recipient mouse received 1 × 105 to 3 × 105 P14 cells. Organs were harvested 6 to 7 hours after transfer. (B) Frequencies of P14 cells within the CD8+ T cell population in spleen, inguinal lymph nodes, lungs, and liver of recipient mice. (C) Right: Confocal images from 10-μm splenic sections. Tissues were stained for the localization of metallophilic macrophages (CD169), B cells (B220), and CD8+ T cells (CD8). P14 cells could be identified by their preserved CTY staining. Left: Quantification of P14 cells present in the imaged sections of T cell zones from different recipients (−ICAM CD8lo, n = 35; +ICAM R CD8lo, n = 50; −ICAM CD8hi, n = 30; +ICAM R CD8hi, n = 25). (D) Naïve P14 cells were stimulated under different conditions and harvested after 36 hours, and progenies were adoptively transferred into wild-type naïve recipients (1 × 105 cells per mouse). (E) Frequencies and numbers of transferred cells were assessed 4 weeks later in spleen and lymph nodes of recipients. (A and B) Pooled data from five independent experiments. (C) Representative data from one of two experiments. (D and E) Representative data from one of three experiments. Data are depicted as mean + SEM. Statistical analysis was performed using the unpaired two-tailed Student’s t test or, when data did not pass the normality test, the unpaired Mann-Whitney U test. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

DISCUSSION

CD8+ T lymphocytes are an integral part of the adaptive immune system and are equipped with highly variable TCRs, which convey antigen specificity. There is already an immense repertoire of different naïve CD8+ T cells before the antigen encounter, but for an immune response to be effective, both clonal expansion and differentiation must take place (1). The observation that an individual naïve cell can generate different progenies, in a diversification degree comparable to polyclonal T cells responses, supports the idea that fate is not predetermined (3436). Furthermore, previous studies have established transcriptional profiles (2, 37), epigenetic control of gene expression (8, 9), usage of preferential metabolic pathways (7), and ACD (3840) as mechanisms that contribute to differentiation. Generation of diversity through ACD is a conserved mechanism of special relevance for development and cellular adaptation, being reported not only in different metazoan cell types besides lymphocytes but also in a wide range of other organisms. Asymmetry relies on the establishment of a polarization axis, which is attained in CD8+ T lymphocytes through the interaction with APCs in the immunological synapse. The orientation of the polarity axis and the composition of the two opposite poles are guided by polarity proteins, and asymmetric distribution is built in different layers after T cell activation, leading to the rise of two progenies that differ in phenotype, transcription factors composition, metabolic status, and transcriptional profile (10, 11, 13, 14, 18) and, as a result, inherit different potential fates. These data are supported both by microscopy observations and by transcriptional profiling of single cells during immune responses, where an early bifurcation of gene expression is detected already after one cycle of mitosis (17).

ACD strongly relies on the formation of a stable immunological synapse. Both absence of ICAM-1 and presentation of low-affinity peptides by APCs do not induce ACD (11, 18), and we have made similar observations when stimulating CD8+ T cells in the absence of Fc-ICAM-1. Thus far, most studies investigating ACD in T lymphocytes were performed with naïve cells, with one exception reporting memory cells to undergo ACD (19). Here, we unraveled a correlation between ACD and memory potential. SLECs and terminally PD-1hi exhausted CD8+ T cells were unable to divide asymmetrically, whereas cells that retain stemness, i.e., the potential to differentiate into subsets with a different fate, exhibited ACD. Considering the regulatory role of mTOR on CD8+ T cell differentiation and the beneficial impact of mTOR inhibition for memory formation (21, 22), we hypothesized that mTOR inhibition could potentially enhance or reestablish the ability of CD8+ T cells to divide asymmetrically. This hypothesis was also based on studies in yeast, where TOR inhibition led to higher ACD rates (23). We found that transient mTOR inhibition through rapamycin or Akt-kinase inhibitor treatment resulted in higher ACD rates in naïve and memory cells. In addition, it installed the ability to undergo ACD in PD-1int exhausted cells. It is important to highlight that the modulation of ACD was restricted to transient mTOR inhibition after TCR engagement. When CD8+ T cells from CD4-cre Raptorfl/fl or CD4-cre Rictorfl/fl mice, constitutively lacking mTORC1 and mTORC2 activity, were stimulated in vitro under ACD conditions, no increase in ACD rates was observed. This observation corroborates previous findings, where CD8+ T cells from T-Rheb−/− or T-Tsc2−/− mice presenting constitutively low and high amounts of mTOR activity, respectively, showed similar CD8 asymmetry to wild-type cells (13). Being a key regulator of T cell metabolism (24, 41, 42), it is reasonable to hypothesize that mTOR inhibition has multiple downstream consequences. However, we could prove that changes in asymmetry are detached from any permanent change in metabolism because mTOR activity was reestablished few hours after removal of drug treatment.

A further link between (enforced) ACD and the ability to generate progenies with (enhanced) memory potential was obtained by selective polarisome disruption during the first mitosis, as CD8+ T cells stimulated under conditions that normally support ACD and installation of memory potential would fail to do so when PKCζ, a component of the polarisome, was inhibited. The polarisome complex is key for the establishment of asymmetry, and ablation or silencing of PKCζ leads to a general defect in establishing polarity, resulting in defective T cell memory (10, 30). Thus, our results suggest that transient mTOR inhibition is not able to create polarity but benefits the establishment and maintenance of asymmetry in polarizing conditions.

By transiently inhibiting mTOR activity and, consequently, increasing ACD rates, adoptively transferred naïve, memory, and PD-1int exhausted P14 CD8+ T cell progenies showed increased expansion upon LCMV challenge compared with nontreated cells, supporting the idea that asymmetry favors the rise of CD8lo cells that can survive in the absence of antigen and thereafter respond to antigen rechallenge. The transcriptional profiling of CD8lo and CD8hi populations originating from distinct stimulation conditions (SCD, ACD, and enforced ACD) revealed the influence of both ICAM-1 and rapamycin as variables that can affect gene signatures. The role of ICAM-1 engagement in providing a stable immunological synapse and prolonged APC–T cell interactions and the consequences of mTOR inhibition as a tool to improve homing of CD8+ T cell to survival niches and benefit memory differentiation have been previously reported (21, 4346). Our data support the notion that ICAM-1 and mTOR inhibition potentially contribute to alterations in T cell differentiation through ACD modulation, as we observed a synergistic effect of these two components on T cell fate determination. Enforced ACD generated CD8lo daughters presenting higher expression of genes linked to memory differentiation. mTOR inhibition has been previously reported as a tool to improve homing of CD8+ T cell to survival niches (43, 44). Here, we could extend this knowledge and determine the role of first CD8lo daughter cells in this process, with progenies of enforced ACD conditions featuring best survival rates and most efficient homing to T cell zones in secondary lymphoid organs. Mechanistically, we suggest that ACD gives rise to more diverse progenies compared with SCD, which could be confirmed by the more distinct gene expression profile differences between CD8lo and CD8hi daughter cells emerging from stimulation conditions that benefit ACD. Enforcing ACD by transient rapamycin inhibition imprinted CD8lo progenies with an earlier and stronger memory signature, increasing their access to survival niches and their memory potential. As we could also show that human CD8+ T cells increase ACD rates upon transient rapamycin treatment, these results open new perspectives in the context of vaccinations and possibly adoptive transfer therapies to resolve tumor and chronic infections. Even preterminally exhausted PD-1int CD8+ T cells exhibited increased ACD rates and memory potential when transiently treated with rapamycin during reactivation, resulting in recovered expansion potential and effector functions. These results open new opportunities of enforcing ACD as a means to improve memory potential of exhausted CD8+ T cells, with relevance for “rejuvenation” of exhausted CD8+ T cells in the context of chronic infections and cancer.

We acknowledge that despite the potential opportunities of ACD modulation as a tool for boosting memory responses or rejuvenating exhausted CD8+ T cells, the specific mechanisms that drive increased polarization upon transient mTOR inhibition remain to be unraveled. In this respect, there is a definitive need to further unravel the molecular details that link mTOR inactivation and increased cell polarity via PKCζ, potentially allowing the discovery of more specific ways of promoting ACD without altering other cellular pathways. This quest might not be trivial as mTOR complex activity orchestrates many downstream pathways that affect various aspects of cellular physiology, including cellular metabolism, lipid biosynthesis, cell migration, survival and differentiation, proliferation, and protein translation. However, our observation that transient mTOR inhibition during mitosis of T cells retaining stem cell–like potential promotes memory potential, which is linked to increased ACD, provides new opportunities to endow this functional trait to T cells in the context of vaccinations and T cell–based immunotherapies.

MATERIALS AND METHODS

A detailed description of materials and methods is provided in the Supplementary Materials.

Study design

This study aimed to evaluate the impact of the differentiation status of a CD8+ T cell on its ability to divide asymmetrically and to determine the potential of ACD modulation as a mechanism to increase or install memory potential on T cells. For that purpose, we standardized an experimental protocol to quantify polarization of different molecules upon CD8+ T cell mitosis. We used the LCMV infection model as a tool to generate diversity and isolate different T cell subsets. We implemented adoptive transfers of CD8+ T cell progenies, following first mitosis after (re-)activation, as a readout to measure memory potential imprinted by ACD. All experiments were performed at least twice. Every group consisted of at least two mice. No outliers were excluded from the data analysis, and no randomization or blinding was used.

CD8+ T cell in vitro stimulation and adoptive transfer

CD8+ T cells were cultured in T cell medium (RPMI-1640,BioConcept), 2 mM l-glutamine (BioConcept), 2% penicillin-streptomycin (Sigma-Aldrich), 10% fetal bovine serum (OMNILAB), 25 mM Hepes (Gibco, Life Technologies; Zug, Switzerland), 1× non-essential amino acids (Sigma-Aldrich), 50 μM β-mercaptoethanol (Gibco), and 1 mM sodium pyruvate (Gibco) supplemented with self-made human IL-2 and stimulated on α-CD3 (5 μg/ml) (145-2C11, BioLegend)–coated and α-CD28 (5 μg/ml) (37.51, BioLegend)–coated plates in the presence or absence of plate-bound human Fc-ICAM-1 (50 μg/ml) (R&D Systems, Bio-Techne AG; Zug, Switzerland) for 30 to 36 hours before confocal microscopy analysis or adoptive transfer experiments. mTOR modulation was conducted by adding 20 nM rapamycin (Santa Cruz Biotechnology, Labforce AG; Muttenz, Switzerland), 1 to 2 μM Akt-kinase inhibitor (Sigma-Aldrich), 1 μM c-Myc inhibitor (10058-F4, Sigma-Aldrich), 1 μM HIF-1α inhibitor (FM19G11, Sigma-Aldrich), 1 μM FTY720 (Sigma-Aldrich), or 10 μM AuTM (Sigma-Aldrich) 12 hours after stimulation. For adoptive transfer, 1 × 104 purified CD45.1 P14 cells (stimulated in vitro or not) were intravenously injected into naïve C57BL/6 CD45.2 recipient mice.

Immunofluorescence staining and confocal microscopy

CD8+ T cell lymphocytes previously (re-)activated in vitro were washed in phosphate-buffered saline (PBS) and added on poly-l-lysine (Sigma-Aldrich)–treated coverslips, followed by incubation for 30 min at 37°C. Cells were then fixed with 2% paraformaldehyde (Sigma-Aldrich) for 10 min, permeabilized with 0.1% Triton X-100 (Sigma-Aldrich) for 10 min, and blocked in PBS containing 2% bovine serum albumin (GE Healthcare) and 0.01% Tween 20 (National Diagnostics) for 1 hour at room temperature. The following antibodies were used to perform immunofluorescence staining in murine cells: mouse α-β-tubulin (Sigma-Aldrich), α-mouse immunoglobulin G Alexa Fluor 488 (Abcam, Lucerna-Chem AG; Luzern, Switzerland), α-CD8 APC (53-6.7, BioLegend), α-T-bet phycoerythrin (PE) (4B10, BioLegend), and α-phospho-S6 PE (D57.2.2E, Cell Signaling Technology). For human cells, we performed tubulin staining with the same antibodies, but α-CD8 PE (SK1, BD Biosciences) was used. Mitotic cells (late anaphase to cytokinesis) were identified by nuclear morphology, the presence of two microtubule organizing centers, and a clear tubulin bridge between two daughter cells. Asymmetry rates were calculated on the basis of CD8 quantification (volume and fluorescence intensity) in each hemisphere. P1 was defined as the pole containing higher amounts of CD8, and P2 was defined as the opposite pole. Mitoses were considered asymmetric when CD8 enrichment was 1.5-fold greater in one daughter cell in comparison with the other [(P1 − P2)/(P1 + P2) > 0.2]. For splenic tissue sections, 10-μm-thin sections were prepared from frozen spleens embedded in optimal cutting temperature compound (Sakura). Samples were fixed in 2% paraformaldehyde for 1 hour at room temperature and blocked in PBS containing 1% of rat serum for 1 hour at room temperature. Staining was performed with the following antibodies (diluted 1:100 in PBS with 0.1% rat serum) for 1 hour at room temperature in the dark: α-CD8 BV421 53-6.7, α-CD169 Alexa Fluor 647 3D6.112, and α-B220 fluorescein isothiocyanate RA3-6B2 (BioLegend). 4′,6-Diamidino-2-phenylindole (Sigma-Aldrich) was used to detect DNA. Mowiol (Calbiochem, Merck-Millipore; Schaffhausen, Germany) was used as mounting medium. Twenty to thirty Z-stacks were acquired using a Visitron Confocal System (inverse confocal microscope, Visitron Systems GmbH) with 10× or 100× magnification (10× objective, type: Plan-Neofluar, aperture: 0.3; immersion: air, contrast: Ph1; 100× objective, type: Plan-Neofluar, aperture: 1.3; immersion: air, contrast: Ph3) coupled to Evolve 512 EMCCD cameras (Photometrics). Data were analyzed using Volocity software (version 6.3., PerkinElmer).

Statistical analysis

To test whether data point values were distributed in a Gaussian distribution, we performed D’Agostino-Pearson normality test. For statistical analysis, we performed two-tailed Student’s t test or nonparametrical analysis Mann-Whitney U tests using GraphPad Prism Software (version 7.0). Statistical significance was determined with *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. For animal experiments, sample size varied from two to seven mice per group for each individual experiment.

SUPPLEMENTARY MATERIALS

immunology.sciencemag.org/cgi/content/full/4/34/eaav1730/DC1

Material and Methods

Fig. S1. Stimulation of P14 CD8+ T cells on α-CD3–, α-CD28–, and Fc-ICAM-1–coated plates or in a coculture with dendritic cells leads to comparable ACD rates.

Fig. S2. Transient mTOR inhibition was obtained by treatment with rapamycin or Akt-kinase inhibitor.

Fig. S3. Transient mTOR inhibition modulates T-bet asymmetry.

Fig. S4. Transient but not continuous mTOR inhibition is a requirement for ACD modulation in CD8+ T cells.

Fig. S5. Decreased ACD rates in P14 CD8+ T cells submitted to FTY720 treatment led to reduced proliferation upon LCMV challenge.

Fig. S6. CM CD8+ T cells produce progenies with enhanced memory potential after enforcing ACD.

Fig. S7. Murine and human rapamycin-treated naïve CD8+ T cells produce progenies that can survive better in the presence of limited amounts of IL-15.

Fig. S8. CD8lo and CD8hi daughter cells inherit distinct cyclin expression profiles.

Fig. S9. ACD generates progenies with strengthened memory signatures.

Table S1. Raw data files.

References (4755)

REFERENCES AND NOTES

Acknowledgments: We thank N. Oetiker and F. Wagen for excellent technical assistance and P. Romero for provision of the CD4-cre Raptorfl/fl and CD4-cre Rictorfl/fl mice. We thank members of the Oxenius, Joller, Barral, and Jessberger groups for helpful discussions. Funding: This work was supported by ETH Zurich, the Swiss National Science Foundation (grant no. 310030_166078 to A.O.; grant nos. 310030_53074 and 310030B_173338 to A.H.), and the Novartis Foundation for Biomedical Research. Author contributions: M.B., Y.B., and A.O. designed the experiments; M.B., I.B., N.S.B., K.P., F.G., N.B., and A.H. performed the experiments; RNA sequencing data analysis was carried out by A.Y., M.B., and I.S.; M.B. and A.O. analyzed the experiments; and M.B. and A.O. wrote the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: Raw sequencing data can be found at ArrayExpress (E-MTAB-7651).
View Abstract

Navigate This Article