Autophagy regulates T lymphocyte proliferation through selective degradation of the cell-cycle inhibitor CDKN1B/p27Kip1

Abstract

The highly conserved cellular degradation pathway, macroautophagy, regulates the homeostasis of organelles and promotes the survival of T lymphocytes. Previous results indicate that Atg3-, Atg5-, or Pik3c3/Vps34-deficient T cells cannot proliferate efficiently. Here we demonstrate that the proliferation of Atg7-deficient T cells is defective. By using an adoptive transfer and Listeria monocytogenes (LM) mouse infection model, we found that the primary immune response against LM is intrinsically impaired in autophagy-deficient CD8⁺ T cells because the cell population cannot expand after infection. Autophagy-deficient T cells fail to enter into S-phase after TCR stimulation. The major negative regulator of the cell cycle in T lymphocytes, CDKN1B, is accumulated in autophagy-deficient naïve T cells and CDKN1B cannot be degraded after TCR stimulation. Furthermore, our results indicate that genetic deletion of one allele of CDKN1B in autophagy-deficient T cells restores proliferative capability and the cells can enter into S-phase after TCR stimulation. Finally, we found that natural CDKN1B forms polymers and is physiologically associated with the autophagy receptor protein SQSTM1/p62 (sequestosome 1). Collectively, autophagy is required for maintaining the expression level of CDKN1B in naïve T cells and selectively degrades CDKN1B after TCR stimulation.

Keywords:

• ATG3
• ATG7
• autophagy
• CDKN1B
• proliferation
• T lymphocytes

Abbreviations

Atg	=	autophagy related
CDK	=	cyclin-dependent kinase
CDKN1B	=	cyclin-dependent kinase inhibitor 1B (p27 Kip1)
CFSE	=	carboxyfluorescein succinimidyl ester
cfu	=	colony-forming units
CKI	=	cyclin-dependent kinase inhibitor
LM	=	L. monocytogenes
OVA	=	ovalbumin
RT	=	room temperature
SQSTM1/p62	=	sequestosome 1
TCR	=	T cell receptor

Introduction

The highly conserved cellular homeostasis pathway, macroautophagy (referred to as autophagy thereafter), degrades large protein aggregates, removes damaged or extraneous organelles, recycles nutrients, and promotes cell survival during stressful conditions.^1-3 In the immune system, autophagy is a mechanism to eliminate intracellular pathogens⁴ and deliver endogenous antigens to MHC II-bearing compartments.^5,6 Autophagy plays an essential role in the development and function of T lymphocytes. Autophagy regulates the calcium mobilization and energy metabolism in T cells,^7–9 is critical for effector CD8⁺ T cell survival and memory formation,^10,11 promotes the proliferation and survival of invariant NKT cells^12,13 and tunes down TCR signaling by degrading BCL10.¹⁴ Autophagy starts with a membrane precursor called the phagophore. The phagophore elongates and engulfs cytosolic materials to form a typical double-membrane structure, called an autophagosome. The final maturation step involves the fusion of the autophagosomes with lysosomes to form autolysosomes, with the sequestered macromolecules degraded by lysosomal proteases and lipases. A series of molecules, termed autophagy-related (Atg) genes, regulate the initiation, elongation and maturation of autophagosomes. The formation of autophagosome requires 2 ubiquitination-like systems to generate membrane inserting ATG12–ATG5 conjugates and LC3. During this process, ATG7 functions as an E1-like activating enzyme, and ATG10 and ATG3 function as E2-like conjugating enzymes.¹⁵ Both ATG7 and ATG3 are required for autophagy in T cells.^16-18 Previous studies from other groups and our lab indicated that autophagy-deficient conventional T and invariant NKT lymphocytes fail to proliferate efficiently,^12,18-21 but the mechanism of the proliferation defect is not clear.

TCR activation-induced proliferation is triggered by the interaction between the T cell receptor-CD3 complex (TCR-CD3 complex) and the MHC-peptide complex on antigen presenting cells. This process expands antigen-specific T cell clones and ultimately generates the adaptive immune response. On the cellular level, T cell expansion is regulated by the progression of the cell cycle. In mammalian cells, the cell cycle consists of the transition from G₀ to G₁ then to S-phase, and is controlled by positive regulators, the cyclins and cyclin-dependent kinases (CDKs), as well as negative regulators, the CDK inhibitors (CKIs).^22,23 CCND (cyclin D) forms a complex with CDK4 or CDK6, while CCNE (cyclin E) or CCNA (cyclin A) associates with CDK2. The expression levels of cyclins and CDKs are very low in resting T cells, whereas they are significantly increased in G₁ phase after upon TCR stimulation.²⁴ The negative regulators, CDKIs, include 2 families: the INK4 inhibitors and the Cip/Kip inhibitors. The INK4 family contains CDKN2A/p16/INK4a, CDKN2B/p15/INK4b, CDKN2C/p18/INK4c, and CDKN2D/p19/INK4d, which bind to CDK4 or CDK6, and inhibit kinase activity. The Cip/Kip (CDK-interacting protein/kinase inhibitor protein) family includes CDKN1A/p21/Cip1, CDKN1B/p27/Kip1, and CDKN1C/p57/Kip2, which inhibit all CDKs.²⁵

CDKN1B is abundantly expressed in T lymphocytes and is the major negative regulator of the cell cycle in T lymphocytes.²⁶ CDKN1B is accumulated in quiescent cells, and the degradation of CDKN1B facilitates the transition from G₁- to S-phase in T lymphocytes after T cell activation. Previous data have shown that CDKN1B is degraded through the ubiquitin-proteasome pathway.²⁷ The increased expression of cyclin D and CDK4 or CDK6 results in the decrease of CDKN1B in G₁ phase, and the activation of CCNE-CDK2. The activated CCNE-CDK2 phosphorylates threonine 187 (T187) of CDKN1B. Phosphorylated CDKN1B interacts with the F-box protein SKP2 (S-phase kinase-associated protein 2, E3 ubiquitin protein ligase).²⁸ SKP2 associates with SKP1, CUL1, RBX1/ROC1 and CKS1B/CKS1 to form the SKP1–CAND1/CDC53-CUL1–F-box protein (SCF) ubiquitin-protein ligase (E3) complex.^29-32 Finally CDKN1B is degraded by the 26S proteasome.²⁷ A SKP2-independent pathway, which is mediated by the ubiquitin ligase KIP1 ubiquitination-promoting complex (KPC), also regulates the degradation of CDKN1B during the G₁ phase.³³ Besides regulating cell cycle, CDKN1B also plays roles in mediating other functions of T cells. For example, CDKN1B negatively regulates the CD4 memory response by promoting the apoptosis and inhibiting proliferation in both effector and memory CD4⁺ T cells.³⁴ CDKN1B also regulates the homeostasis of CD8⁺ T cells and limits the proliferation of effector memory cells.³⁵ Finally, CDKN1B is required for tolerance induction in T cells.³⁶

Here we show that ATG7-deficient T cells cannot proliferate in vitro, similar to what is observed in Atg5- and Atg3-deficient T cells.^18,19 We identify that the impairment of proliferation causes the intrinsic defect in the primary immune response in autophagy-deficient CD8⁺ T cells to infection by L. monocytogenes. Furthermore, we show that autophagy-deficient T cells fail to enter into S-phase after TCR stimulation and correspondingly, the cell-cycle inhibitor, CDKN1B, is abnormally accumulated in autophagy-deficient naïve T lymphocytes and cannot be degraded after activation. The genetic deletion of Cdkn1b restored the proliferative capacity in autophagy-deficient T cells. Interestingly, natural CDKN1B forms polymers that are physiologically associated with the autophagy receptor protein, SQSTM1/p62. Taken together, our data indicates that autophagy regulates the proliferation of T lymphocyte through selectively degradation of the cell-cycle inhibitor, CDKN1B.

Results

The primary immune response is defective in autophagy-deficient T cells

In previous studies, our group and others have found that Atg5- or Atg3-deficient T cells cannot proliferate efficiently.^18-21 To further confirm that autophagy is essential for the proliferation of T lymphocytes, the proliferation of Atg7-deficient (Atg7^f/f Lck-Cre) T cells was analyzed. Proliferation in both CD4⁺ and CD8⁺ Atg7-deficient T cells was severely impaired. The percentage of proliferating T cells in Atg7-deficient T cells was significantly reduced compared to that in floxed control (Atg7^f/f) T cells under all TCR stimulation conditions (Fig. 1A and 1B). Compared to anti-CD3 antibody stimulation alone, costimulation with anti-CD28 could only partially improve the proliferation in Atg7-deficient T cells. Interestingly, the use of PMA and ionomycin, which bypasses the TCR proximal signaling, did not rescue the proliferation defect in Atg7-deficient T cells (Fig. 1A). These data suggest that the defective proliferation in autophagy-deficient T cells is downstream of proximal TCR signaling. These results are consistent with previous data, which indicated that the proximal TCR signaling remained intact in autophagy-deficient T cells.^17,18

Figure 1. Defective proliferation of Atg7-deficient T cells. Splenocytes from Atg7^f/f and Atg7^f/fLck-Cre mice were loaded with CFSE and stimulated with coated anti-CD3 mAb (2C11), soluble anti-CD3 plus anti-CD28 (sCD3+CD28) or PMA together with ionomycin for 72 h. The CFSE-diluted cell populations were analyzed by flow cytometry and all cells were gated on 7-AAD negative cells. These experiments were repeated 3 times. (A) Representative flow cytometry profiles of CD4⁺ or CD8⁺ T cell proliferation from Atg7-deficient T cells. (B) The percentages of CFSE-diluted CD4⁺ or CD8⁺ T cells from Atg7^f/f and Atg7^f/fLck-Cre mice. Each symbol represents one mouse.

The survival of autophagy-deficient T cells is impaired.^10,18-20,37 To exclude the possibility that the proliferation defect is caused by cell death, all cells in the carboxyfluorescein succinimidyl ester (CFSE) dilution assay were gated on 7-AAD negative live cells (Fig. 1A). The death of autophagy-deficient T cells after anti-CD3 stimulation was determined. The survival of autophagy-deficient T cells was improved after TCR stimulation (Fig. S1).

To further analyze the physiological function of autophagy in T cells, primary immune responses of autophagy-deficient T cells were analyzed using adoptive transfer and L. monocytogenes infection. We utilized a recombinant strain of L. monocytogenes expressing chicken OVA (LM-OVA).³⁸ The use of an inducible deletion system, Atg3^f/f ER-Cre, provides an ideal tool since the numbers of mature T cells are comparable between the floxed control and Atg3^f/f ER-Cre after the deletion of Atg3.¹⁸ As shown in Figure 2A, the percentage of Atg3-deficient PTPRC^b/CD45.2⁺ donor cells at d 5 after infection was comparable to that of floxed control donor cells. This indicates comparable short-term engraftment between Atg3-deficient and floxed control T cells. However, the percentage of Atg3-deficient donor cells was significantly lower than that of floxed control donor cells at d 7 after infection, when the immune response against L. monocytogenes infection reaches its peak (Fig. 2B).³⁹ Therefore, both in vitro proliferation assays and in vivo adoptive transfer infection experiments indicate that the autophagy-deficient T cells cannot proliferate efficiently and the primary immune response against L. monocytogenes infection may be defective.

Figure 2. Impaired primary T cell immune response in autophagy-deficient T cells. (A) Analysis of autophagy-deficient T cells in primary response against the infection of L. monocytogens through adoptive transfer assay. One pair of Atg3^f/f OT-I and Atg3^f/fER-Cre OT-I mice were used to prepare the donor cells. Purified CD8⁺ cells were transferred to 3-5 PTPRC^a/CD45.1 host mice. The blood was withdrawn at d 5 and d 7 after infection, and the frequency of antigen-specific CD8⁺ T cells was analyzed by gating on the PTPRC^b/CD45.2⁺ CD8⁺ cells. The experiments were repeated 3 times. (B) The frequencies of antigen-specific CD8⁺ T cells (PTPRC^b/CD45.2⁺ Dimer X⁺ CD8⁺) pooled from 3 mice that received Atg3^f/f OT-I and 5 mice that received Atg3^f/fER-Cre OT-I cells.

To directly test whether an impaired primary immune response was due to the incapability of autophagy-deficient T cells to proliferate, the division of antigen-specific CD8⁺ T cells responding to LM-OVA was analyzed in vivo. CFSE-labeled OT-I CD8⁺ T cells or Atg3-deficient OT-I T cells were transferred to normal host mice. Recipient mice were infected with LM-OVA at 24 h post transfer. The division of antigen-specific T cells was analyzed at 48 h after infection. While control OT-I T cells divided very well as expected, autophagy-deficient T cells failed to proliferate normally (Fig. 3A). The cell numbers in each generation after generation 4 were reduced in autophagy-deficient OT-I cells comparing to floxed control T cells (Fig. 3B), indicating that the proliferation of autophagy-deficient T cells is impaired.

Figure 3. Defective in vivo proliferation of autophagy-deficient T cells. Atg3^f/f OT-I and Atg3^f/fER-Cre OT-I mice were injected with tamoxifen to induce the deletion of Atg3. Purified CD8⁺ T cells were loaded with CFSE and then transferred to 3 to 5 normal host mice for each donor cell population intravenously. The next day after transfer, host mice were infected with L. monocytogenes-OVA (1×10⁴cfu/mouse). At 48 h after infection, the splenocytes were harvested. Live, CFSE⁺ CD8⁺cells were analyzed for CFSE dilution. This experiment was repeated twice. (A) Representative data of flow cytometry profiles from a pair of mice that received CFSE-labeled Atg3^f/fOT-I or Atg3^f/fER-CreOT-I T cells are shown. (B) The cell numbers of each generation from 3 pairs of mice that received Atg3^f/fOT-I or Atg3^f/fER-CreOT-I were analyzed and calculated by Flowjo.

Autophagy-deficient T cells fail to enter into S phase after TCR stimulation

Since autophagy-deficient T cells have obvious deficiencies in proliferation after TCR stimulation, cell-cycle analysis of autophagy-deficient T cells was performed after TCR stimulation in both Atg3- (Fig. 4) and Atg7- (Fig. S2) deficient models (Atg3^f/f Lck-Cre or Atg7^f/f Lck-Cre). As shown in Figure 4A, 20% of floxed control CD4⁺ and more than 40% floxed control CD8⁺ T cells entered into S-phase after stimulation with soluble anti-CD3 in the presence of B cells. However, there were reduced percentages of autophagy-deficient CD4⁺ and CD8⁺ T cells entering into S-phase.

Figure 4. Autophagy-deficient T cells fail to enter into S-phase after TCR stimulation. (A) Splenocytes from Atg3^f/f and Atg3^f/fLck-Cre mice were stimulated with soluble anti-CD3 plus anti-CD28 antibodies overnight. Cell cycle was analyzed by flow cytometry. The statistical analysis in the lower panel was derived from 3 pairs of Atg3^f/f and Atg3^f/fLck-Cre mice (mean±SD). The experiment was repeated twice. (B) CDKN1B is accumulated in autophagy-deficient T cells. Atg3^f/fOT-I and Atg3^f/fER-Cre OT-I mice were inducibly deleted the Atg3 through tamoxifen injection. At d 6 or d 35 after the first injection, the CD8⁺ T cells were purified and cell lysates were prepared. The expression level of CDKN1B and CDKN1 were analyzed by western blot. The numbers are the ratios of the intensity of target molecule to the loading control ACTB/actin. The normalized intensities from 3 pairs of Atg3^f/f OT-I and Atg3^f/fER-Cre OT-I mice are shown in the right panel (mean±SD). (C) Impaired degradation of CDKN1B in autophagy-deficient T cells after TCR-mediated activation. Splenocytes were stimulated with anti-CD3 plus anti-CD28 antibodies or without any stimulation overnight. Total T cells were purified and cell lysates were prepared. The expression levels of CDKN1B and CDKN1 were analyzed by western blot. The normalized intensities from 3 pairs of Atg7^f/f and Atg7^f/f Lck-Cre mice are shown in the lower panel (mean±SD).

In T lymphocytes, the cell cycle is primarily regulated by the cyclin-dependent kinase inhibitor CDKN1B, which is degraded after TCR stimulation and IL-2 production. The levels of CDKN1B and other Kip/Cip family members were quantified in naïve T cells, in either an inducible deletion system of autophagy related gene or a Lck-Cre system. As shown in Figure 4B, the protein level of CDKN1B was comparable between autophagy-competent and autophagy-deficient T cells within 1 wk of Atg3 deletion. The protein expression level of CDKN1B in naïve floxed control and autophagy-deficient T cells was relatively low, while the CDKN1 expression level was extremely low (Fig. 4B). However, the expression level of both CDKN1B and CDKN1 increased significantly in autophagy-deficient T cells if the deletion of autophagy related genes was induced for a longer time period (Fig. 4B, right panel). Similarly, the expression levels of CDKN1B and CDKN1 were also increased in autophagy-deficient Atg7^f/fLck-Cre T cells compared to floxed control T cells (Fig. 4C), in which Atg7 was deleted at DN3 stage of thymocyte development. These results suggest that autophagy constitutively regulates the expression level of CDKN1B and CDKN1. It has been shown that the degradation of CDKN1B after TCR-mediated T cell activation facilitates T cell entry into the S phase and proliferation.²⁶ As shown in Figure 4C, the protein level of CDKN1B was decreased in floxed control T cells after TCR stimulation, while the protein level of CDKN1B in autophagy-deficient T cells remained unchanged. This suggests that autophagy is required for the degradation of CDKN1B after TCR stimulation. It has been reported that concanavalin A and IL-2 stimulation transcriptionally decrease the abundance of CDKN1B.⁴⁰ However, there was no difference in the expression of Cdkn1b at the mRNA level between floxed control and autophagy-deficient T cells (data not shown). IL-7 promotes T cell proliferation through the regulation of stability of CDKN1B at the protein level, which is dependent on PRKCQ/PKC-θ (protein kinase C, theta).⁴¹ But IL-7 fails to rescue the proliferation defect in autophagy-deficient T cells (our unpublished data). Interestingly, the protein level of CDKN1 was slightly increased in floxed control T cells after TCR stimulation, but it was strongly elevated in resting and activated autophagy-deficient T cells (Fig. 4C). CDKN1 may be regulated at the level of both expression and degradation in activated T cells. Taken together, the proliferation defect in autophagy-deficient T cells may be due to a degradation defect of CDKN1B and CDKN1, resulting in the incapacity of autophagy-deficient T cells to enter into S-phase.

The defect in proliferation is caused by the abnormal accumulation of CDKN1B in autophagy-deficient T cells

The increased protein levels of CDKN1B and CDKN1, along with the impaired proliferation in autophagy-deficient T cells after TCR activation, suggest that CDKN1B and CDKN1 may control the cell cycle and proliferation in T lymphocytes after activation. To test whether the decreased level of CDKN1B can restore proliferation, Atg3^f/fER-Cre mice were bred to Cdkn1b^-/- mice to generate atg3^-/-Cdkn1b^+/- T cells. Consistent with our hypothesis, deletion of just one allele of Cdkn1b in autophagy-deficient T cells restored the capability to enter into S-phase and permitted proliferation (Fig. 5A and 5B). The percentage of proliferating T cells from Atg3^f/fER-CreCdkn1b^+/- mice was comparable to floxed control cells in both CD4⁺ and CD8⁺ compartments (Fig. 5A and 5B). As expected, no obvious accumulation of CDKN1B was observed in autophagy-deficient Cdkn1b^+/- T cells (Fig. 5C). These results suggest that the abnormal accumulation of CDKN1B in autophagy-deficient T cells blocked their proliferation.

Figure 5. Deletion of CDKN1B restores proliferative capacity in autophagy-deficient T cells. (A) Atg3^f/f, Atg3^f/fER-Cre, Cdkn1b^+/-, Atg3^f/fER-Cre Cdkn1b^+/- and Atg3^f/fCdkn1b^+/- mice were injected with tamoxifen to inducibly delete Atg3. Splenocytes from Atg3^f/f, Atg3^f/fER-Cre, Atg3^f/fER-Cre Cdkn1b^+/- and Atg3^f/fCdkn1b^+/- mice were isolated, loaded with CFSE, and stimulated with soluble anti-CD3 mAb, or soluble anti-CD3 plus anti-CD28 antibodies for 72 h. Live cells were analyzed for CFSE dilution. All cells were gated on CD4⁺7-AAD^- or CD8⁺7-AAD^- populations. There were 3 or 4 mice in each group in 2 independent experiments. (B) The percentages of CFSE-diluted CD4⁺ or CD8⁺ T cells after stimulation with soluble anti-CD3 plus anti-CD28 were analyzed (n=3 or 4). (C) The expression levels of CDKN1B in purified T cells from the above mice by western blot. ACTB/actin serves as loading control. The numbers are normalized expression levels for CDKN1B.

Autophagy selectively degrades CDKN1B

Previously published data has demonstrated that CDKN1B is degraded through the proteasomal pathway.²⁷ The accumulation of CDKN1B in autophagy-deficient T cells suggests that autophagy may be involved in maintaining homeostatic levels of CDKN1B. Interestingly, CDKN1B formed polymers in nonreducing conditions as analyzed by western blot (Fig. 6A). Under reducing conditions, the CDKN1B bands showed up at 50 kDa and 27 kDa. Under nonreducing conditions, in which the disulfide bonds remained intact, CDKN1B migrated at larger molecular masses in addition to the other 2 relative weak bands (Fig. 6A).

Figure 6. Autophagy selectively degrades CDKN1B in T lymphocytes. (A) CDKN1B forms polymers naturally. Cell lysates of purified T cell from wild-type mice were prepared. The cell lysates were mixed with reducing sample buffer, which contains 2-ME or nonreducing sample buffer without any 2-ME. The western blot was probed with anti-CDKN1B antibody. (B) CDKN1B interacts with SQSTM1. 293T cells were cotransfected with Sqstm1-HA and Cdkn1b-FLAG constructs. Then tagged proteins were immunoprecipitated with an anti-FLAG antibody and probed with an anti-HA antibody (upper panel). The interaction between endogenous of CDKN1B and SQSTM1 is shown in the lower panel. Atg3^f/f and Atg3^f/fER-Cre mice were injected with tamoxifen to delete Atg3. Splenocytes from Atg3^f/f and Atg3^f/fER-Cre mice were isolated and cell lysates were prepared. CDKN1B was immunoprecipitated with anti-CDKN1B antibody and probed for SQSTM1 interaction. (C) Ubiquitination of CDKN1B in splenic T cells. Splenocytes from normal mice were stimulated with soluble anti-CD3 plus anti-CD28 antibodies for 6 h. Then CDKN1B was immunoprecipitated with anti-CDKN1B mAb and the western blot membrane was probed with an anti-ubiquitin antibody. (D) CDKN1B localized in the same autophagy-related membrane structures as that of LC3-II and SQSTM1. Splenocytes from 3 normal mice were stimulated with anti-CD3 antibody in RPMI with choloroquine (50 µg/ml) for 6 h. The homogenates were prepared with protease inhibitors and choloroquine (50 µg/ml) using Dounce homogenizers. Postnuclear subcellular compartments were added into ultracentrifuge tubes on the top of prepared gradients. The distributions of target molecules in each fraction were detected by western blot in reducing condition. The arrow represents the concentration of the gradients from lower to higher densities. P, cell pellet; L, cell lysate. (E) The degradation of CDKN1B was analyzed in the presence of lactacystin or bafilomycin A₁ treatment. Splenocytes from normal mice were stimulated with soluble anti-CD3 plus anti-CD28 antibodies with or without lactacystin or bafilomycin A₁ treatment. The cells were collected at different time points and the T cells were purified. The expression levels of CDKN1B in T cells were analyzed by western blot.

Selective autophagy can degrade specific targets. Several autophagy receptor proteins facilitate the loading of targets into phagophores.⁴² The receptor molecule SQSTM1 is one of these autophagy receptors. SQSTM1 can associate with LC3 during the induction of autophagy and at the same time binds ubiquitinated proteins.⁴³ To test the interaction between SQSTM1 and CDKN1B, recombinant HA-tagged SQSTM1 and FLAG-tagged CDKN1B were coexpressed in 293T cells. CoIP indicated that SQSTM1-HA could interact with CDKN1B-FLAG (Fig. 6B upper panel). The endogenous association between CDKN1B and SQSTM1 was analyzed as well. As shown in the lower panels of Figure 6B, the endogenous SQSTM1 interacts with CDKN1B in both floxed control and ATG3-deficient splenocytes. Autophagy-deficient CD4⁺ and CD8⁺ T cells have relatively higher ubiquitination levels compared to floxed control T cells (Fig. S3). To directly determine its ubiquitination status, CDKN1B was isolated from normal T cells by immunoprecipitation in the presence or absence of TCR activation, followed by western blot under both reducing and nonreducing conditions. Under reducing condition, ubiquitinated CDKN1B shown as multiple bands with higher molecular weight was detected (Fig. 6C). Under nonreducing conditions, some ubiquitinated CDKN1B could also be detected (Fig. 6C). TCR stimulation did not obviously enhance ubiquitinated CDKN1B (Fig. 6C). To further confirm that CDKN1B can translocate to phagophores, a subcellular fractionation assay was performed. CDKN1B was localized within the same fractions (fraction: 4 to 8) in which the autophagosome marker LC3-II was fractionated (Fig. 6D). Therefore, our results indicate that autophagy selectively maintains the expression level of CDKN1B and degrades CDKN1B when T cells are activated.

Our data do not exclude the functions of ubiquitin/proteasome pathway in the degradation of CDKN1B. The kinetic degradation of CDKN1B in T cells after TCR stimulation was analyzed. After stimulation, CDKN1B started being degraded at 3 h, further decreased at 5 to 10 h, and was mostly degraded at 10 h or more post stimulation (Fig. 6E). The proteasome inhibitor lactacystin inhibited and delayed the degradation of CDKN1B (Fig. 6E). Bafilomycin A₁, which is an endosomal and lysosomal acidification inhibitor that can inhibit the fusion of autophagosomes and lysosomes, also delayed the degradation of CDKN1B (Fig. 6E). Another autophagy inhibitor, 3-MA, could more clearly inhibit the degradation of CDKN1B in activated T cells (Fig. S4). 3-MA did not affect the activation of T cells as detected by the expression of murine IL2RA/CD25 and CD69 (data not shown). Considering the published data that the proliferation of T cells from CDKN1B^T187A knockin mice is decreased, but can still be promoted by IL-2,⁴⁴ there may be a proteasome-independent pathway for the degradation of CDKN1B. Our data shown here from mouse genetic models demonstrate that autophagy is required for CDKN1B degradation in both naïve and activated T cells.

Discussion

We have previously reported that autophagy developmentally regulates the homeostasis of organelles, such as the ER and mitochondria, and promotes T cell survival.^18,19 Here we demonstrate that Atg7-deficient T cells behave like Atg5- or Atg3-deficient T cells and cannot proliferate efficiently in vivo and in vitro. Interestingly, a recent report shows that autophagy-deficient invariant NKT cells also exhibit defective proliferation.¹² However, the molecular mechanism underlying the defective proliferation in both types of T cells is not clear. The primary response of autophagy-deficient CD8⁺ T cells against the LM infection is severely impaired due to proliferative effects both in vitro and in vivo. Consistently, autophagy-deficient T cells fail to enter into S phase after TCR stimulation. The main negative cell-cycle regulator, CDKN1B, is accumulated in naïve autophagy-deficient T cells and CDKN1B cannot be degraded after T cell activation in the absence of autophagy. Furthermore, genetic deletion of a single CDKN1B allele restores the proliferative capacity in autophagy-deficient T cells. Although both CDKN1B and CDKN1 are accumulated in autophagy-deficient T cells, the genetic restoration by deleting a single CDKN1B allele suggests that CDKN1 is not the primary inhibitor of proliferation of autophagy-deficient T cells. Moreover, we found that CDKN1B is ubiquitinated and physiologically associated with an autophagy receptor molecule, SQSTM1. Most importantly, natural CDKN1B forms large polymers. This is the first evidence to indicate that autophagy regulates the cell-cycle negative regulator, CDKN1B, in naïve T cell and reduces its abundance when T cells are activated to enter into the cell cycle.

The level of CDKN1B remains higher during the early to middle of the G1 phase to prevent cell-cycle progression by inhibiting the kinase activity of CCNE- or cyclin A-CDK2 complex. Intensive studies have been published about the degradation of CDKN1B during cell-cycle regulation. Previous reports showed that CDKN1B is degraded by the ubiquitin/proteasome system upon cell-cycle entry. This proteasome degradation of CDKN1B is initiated by CCNE-CDK2-dependent CDKN1B phosphorylation at T187 and then recognized by the SCF-SKP2 ubiquitin ligase complex.^27,45 CDKN1B is also phosphorylated at a highly conserved tyrosine 88 (Y88) by the Src-family kinase LYN or the oncogenic product, BCR-ABL. The phosphorylation of Y88 facilitates the phosphorylation of T187 and is followed by SKP2-dependent degradation.⁴⁶ CDKN1B localizes to the nucleus to inhibit cell-cycle progression and can be redistributed to the cytosol for degradation upon cell-cycle entry. This cytoplasmic relocation is dependent upon the phosphorylation of serine 10. Although CDKN1B is generally degraded in the cytosol, it can also be degraded in the nucleus.⁴⁷ The degradation of CDKN1B can also be accomplished through a SKP2-independent ubiquitination pathway demonstrated in SCF (SKP2)-deficient cells.^48,49 Another study indicates that CDKN1B is downregulated through endo-lysosomal degradation, mediated by an interaction with endosomal protein SNX6 (sorting nexin 6).⁵⁰

Although it has been demonstrated that CDKN1B is degraded by the proteasome, our genetic mouse model provides evidence that the intracellular homeostasis of CDKN1B is maintained by autophagy in both naïve and activated T lymphocytes as CDKN1B is not degraded efficiently in the absence of autophagy. Why does CDKN1B require autophagy to maintain its basal levels in naïve T cells and to be further degraded upon activation to enter into S phase? The reason may be that native CDKN1B forms large polymers which require autophagy for breaking down. Our data suggest that autophagy is a downstream step for degradation of CDKN1B.

The proteasome and autophagosome are the main organelles to degrade misfolded proteins or protein aggregates.⁴² Ubiquitination of target proteins designates them as a substrate for degradation. Indeed, CDKN1B is ubiquitinated. K48-linked ubiquitin chains target the protein to the proteasome, while the K63-linked ubiquitin chain proteins are shuttled into the autophagy pathway.⁵¹ Whether the ubiquitination on CDKN1B is K48- or K63-linked remains to be determined. We examined the role of proteasome-mediated degradation of CDKN1B by testing CDKN1B levels in the presence of proteasome inhibitors in normal T cells. Interestingly, proteasome inhibitors delayed but not completely prevented CDKN1B degradation. This result is consistent with a role of autophagy in CDKN1B degradation. Knockdown of the CDKN1B E3 ligase SCF^Skp2 caused cell-cycle arrest and increased apoptosis in which the cells displayed characteristics of both activation and autophagy.⁵² The energy-sensing pathway regulates the phosphorylation of threonine 198 (T198) of CDKN1B and increases its stability. Ectopic expression of the phosphomimetic T198 to aspartic acid mutation (T198D) induces the activation of autophagy.⁵³ This suggests that autophagy and the cell cycle may regulate each other, especially during stressful conditions, including proliferation.

Our finding that autophagy selectively degrades CDKN1B may provide new insights for regulating immune responses through autophagy inhibition or induction. Previous studies have provided examples of selective targeting of ubiquitinated BCL10 in T cells by autophagy.¹⁴ A recent study also demonstrates that chaperone-mediated autophagy regulates T cell response by degrading RCAN1.⁵⁴ Two recently published findings indicate that autophagy is critical in the regulation of memory CD8⁺ T cells using virus infection models.^10,11 Autophagy is reduced in T cells from aging mice and enhancing autophagy could restore CD8 memory response.¹¹ Our data provides new clues and explanations for these published findings. Also, selective degradation of CDKN1B by autophagy may be a mechanism of action of chloroquine in the treatment of autoimmune diseases. Considering the function of CDKN1B in the regulation of cancer cell growth, inhibition of autophagy may be used as a strategy to treat cancer. Our preliminary experiments indicated that the autophagy inhibitor, 3-MA, could inhibit the proliferation of a leukemia cell line Jurkat (data not shown). Above all, our data suggests that autophagy regulates the abundance of the cell-cycle inhibitor, CDKN1B, and selectively degrades CDKN1B upon cell activation and entry into the cell cycle.

Materials and Methods

Mice

Atg3 floxed mice (Atg3^f/f) were generated in our lab and described in detail.¹⁸ Atg7 floxed (Atg7^f/f) mice have been previously described.⁵⁵ The Lck-Cre (012837), ER-Cre (004847), Cdkn1b^+/- (B6.129S4-Cdkn1btm1Mlf/J; 002781), OT-I (003831) and B6.SJL-Ptprc^a Pepc^b/BoyJ(CD45.1, 002014) mice were purchased from The Jackson Laboratory. All mice were housed in a specific pathogen-free facility at Duke University and used at 6 to 8 wk of age. All animals were used according to protocols approved by the Duke University Institutional Animal Care and Use Committee.

Proliferation assay

Splenocytes from Atg7^f/f, Atg7^f/fLck-Cre, Atg3^f/f, Atg3^f/f ER-Cre or Atg3^f/fER-Cre Cdkn1b^+/- mice were incubated with CFSE (5 μM; Life Technologies, C34554) for 5 min at room temperature (RT) in phosphate-buffered saline (PBS; Life Technologies, 10010–023) containing 5% fetal bovine serum. Then the CFSE loaded cells were washed 3 times with 5% fetal bovine serum-PBS. After washing, the splenocytes were stimulated with plate bound anti-CD3 (5 μg/ml in PBS, 2C11; 100302), soluble anti-CD3 (5 μg/ml) plus anti-CD28 (2 μg/ml; 122003), or PMA (10 ng/ml; Sigma, P8139) plus ionomycin (300 ng/ml; Sigma, I9657) for 72 h. CFSE diluted cells were analyzed by flow cytometry (FACSCanto II, BD, San Jose, CA). All antibodies are from Biolegend. ANXA5/Annexin V-FITC is from BD Bioscience (556420).

Adoptive transfer and Listeria monocytogenes infection

Atg3^f/f OT-I or Atg3^f/fER-Cre OT-I mice were injected with 0.4 mg per mouse tamoxifen (Sigma, T5648) intraperitoneally (i.p.) once a day for 3 d. Three wk after the first injection, the splenocytes were harvested. Using an EasySep mouse CD8 T cell negative-selection kit (StemCell Technologies, 19753), the purified CD8⁺ OT-I T cells (2×10⁴ per mouse) from Atg3^f/f OT-I or Atg3^f/fER-Cre OT-I mice were transferred to PTPRC^a/CD45.1 through i.v. injection. The following day, recombinant L. monocytogenes which express chicken OVA (LM-OVA 1×10⁴ colony-forming units [cfu] per mouse) was injected i.v.. Peripheral blood was drawn at d 5 and d 7 after infection. Antigen-specific CD8⁺ T cells recognizing OVA were stained with Dimer X (BD Bioscience, 550750) loaded with OVA peptide 257–264 (American Peptide Company, 74-1-16A). The donor cells (PTPRC^b/CD45.2⁺) and host cells (PTPRC^a/CD45.1⁺) were distinguished by anti-PTPRC^a/CD45.1-APC (BioLegend, 110713) and anti-PTPRC^b/CD45.2-APC/Cy7 antibodies (BioLegend, 109823).

In vivo proliferation assay

CD8⁺ T cells were purified from splenocytes of tamoxifen inducibly deleted Atg3^f/fER-Cre OT I mice or Atg3^f/f OT I control mice. The purified CD8⁺ T cells were loaded with CFSE and injected into normal host mice intravenously. Twenty-four h later, the adoptively transferred mice were infected with LM-OVA (1×10⁴ cfu/mouse in 0.1ml) intravenously. The splenocytes were harvested 48 h after infection. CFSE dilution was analyzed in CD8⁺Dimer X⁺CFSE⁺ cells. The cell number of each generation was analyzed and calculated by Software Flowjo (TreeStar Inc., Ashland, OR).

Cell-cycle analysis

Splenocytes from Atg3^f/f, Atg3^f/f Lck-Cre or Atg7^f/f, Atg7^f/fLck-Cre mice were stimulated with anti-CD3 (2C11, 5 µg/ml) plus anti-CD28 (2 µg/ml) overnight. Stimulated and control cells were harvested and fixed with 70% ethanol, incubated on ice for 15 min, washed, and stained with propidium iodide (PI) solution (50 µg/ml PI [Sigma, P4170], 0.1 mg/ml RNAse A [Sigma, R6513], 0.05% Triton X-100 [Sigma, X100] in PBS) for 30 min at 37˚C. After washing, the cell cycle was analyzed by flow cytometry (FACScan or Canton II, BD Bioscience, San Jose, CA).

Western blot

Total T cells, CD4, or CD8 T cells were purified by EasySep mouse CD4 or CD8 T cell negative-selection kits (StemCell Technologies, 19752, 19753). The purity was more than 95%. Cell lysates were prepared. Samples were separated by SDS-PAGE and transferred to Immobilon-FL polyvinylidene difluoride (PVDF) membranes (Millipore, IPVH00010). The PVDF membranes were blocked with 5% nonfat milk for 1 h at RT. The primary antibodies were incubated with working concentration in 3% BSA (Sigma, A2153)-PBS (Life Technologies, 10010023) solution with 0.02% Tween 20 (Sigma, P1379) at 4˚C overnight. After washing 3 times, the membranes were incubated with antibodies specific for the pertinent species and labeled with Alexa Fluor 680- (Life Technologies, A-21059 or A-21096) or IRDye 800- (Rockland Immunochemicals, 605–732–002) secondary antibodies at RT for 1 h. After washing, the blots were visualized using an Odyssey Infrared Imaging System and analyzed using Odyssey software (LI-COR Bioscience, Lincoln, NE). Anti-BCTB/Actin antibody (sc-1616) and anti-CDKN1B mAb (sc-1641) are from Santa Cruz Biotechnology, Inc. Anti-ubiquitin mAb (646301) is from BioLegend.

Subcellular fractionation

Subcellular fractionation was performed with OptiPrep (Axis-Shield PoC, 1114542), according to the literature and protocols recommended by the manufacturer with some modifications.⁵⁶ The working solution of 50% (w/v) iodixanol was prepared by diluting the OptiPrep with buffer (130 mM KCl, 25 mM NaCl, 6 mM EGTA, 150 mM Tris-Cl, pH 7.4). Splenocytes from 3 normal unmanipulated mice were suspended in RPMI with chloroquine (50 µg/ml; Sigma, C6628) and stimulated with anti-CD3 antibody (2C11) at 37^oC for 6 h to activate autophagy. The activation of T cells was confirmed by CD69 and IL2RA/CD25 expression. The cells were harvested and washed with precooled homogenization buffer (130 mM KCl, 25 mM NaCl, 1 mM EGTA, 25 mM Tris-Cl, pH 7.4), and suspended in homogenization buffer with chloroquine and protease inhibitors (Sigma, P1860). The homogenate was prepared manually by dounce tissue grinder with glass pestle (1 ml capacity; Wheaton, 357538) on ice. The homogenates were centrifuged at 1000 g for 5 min to remove the nuclei. The supernatant fraction (containing all subcellular membrane structures) was transferred to another centrifuge tube (Beckman, Brea, CA) and adjusted to 30% (w/v) iodixanol with the working solution of OptiPrep. Different concentrations of OptiPrep gradients were layered in the following amount and order: 1 ml 30%, 1.5 ml 25%, 2 ml 20%, 2 ml 15%, 1.5 ml 10% and 1 ml 5%. The adjusted supernatant fraction (2 ml) was layered at the top of centrifuge tube and the swing-bucket 40Ti rotor (Beckman) was used to centrifuge at 48,000 g for 18 h at 4˚C. One ml of each fraction was collected. The expression levels of target proteins in each fraction were analyzed by SDS-PAGE and western blot.

Affinity isolation assay

The Cdkn1b and Sqstm1 were cloned by RT-PCR with the cDNA from mouse T cells and inserted into a pMI expression vector. The cloning primer sequences are: Cdkn1b forward (F), GCGGATCCATGTCAAACGTGAGAGTGTCTAACGGGAGC, Cdkn1b reverse (R), GCGAATTCCTTGTCGTCATCGTCTTTGTAGTC CGTCTGGCGTCGAAGGCCGGGCTTC, Sqstm1 F, GCGGATCCATGGCGTCGTTCACGGTGAAGGCCT, Sqstm1 R, GC GTCGACAGCGTAATCTGGAACATCGTATGGGTACAATGGTGGAGGGTGCTTCGAATACTGGAT. The reverse primer contains the FLAG or HA sequences. CDKN1B was tagged with FLAG and SQSTM1 was tagged with HA. After 293T cells were cotransfected with CDKN1B-FLAG and SQSTM1-HA, the cell lysates were prepared in lysis buffer (25 mM Tris-HCl, pH 7.4, 150 mM NaCl, 5 mM EDTA, 1 mM PMSF, 1 mM sodium vanadate (Sigma, 72060), 2 mM sodium pyrophosphate (Sigma, P8010),10 mM β-glycerol phosphate (Sigma, 50020), 10 µg/ml leupeptin (Sigma, 62070), 10 µg/ml aprotinin (Sigma, A1153), and 1% Triton X-100). The immunoprecipitation was performed with anti- FLAG antibody (Biolegend, 923601) using Dynabeads Protein G beads (Life Technologies, 10003D). After SDS-PAGE and western blot, the membranes were probed with anti-HA antibody (Sigma, H3663). The endogenous association was analyzed with the cell lysates from splenocytes from Atg3^f/f, Atg3^f/fER-Cre mice and inducibly-deleted Atg3 by tamoxifen injection. Anti-SQSTM1 antibody is from MBL (PM045).

Statistics analysis

The 2-tail distribution Student t test was used to analyze the significance through Excel. A p value < 0.05 was considered statistical significance.

Disclosure of Potential Conflicts of Interest

The authors declare no conflicts of interest.

Acknowledgments

We thank Dr. Qi-Jing Li for critical discussion and suggestions. We thank Dr. Michael Cook, Lynn Martinek, Nancy Martin and Dr. Bin Li for helping with the flow cytometry.

Supplemental Material

Supplemental data for this article can be accessed on the publisher's website.

Funding

This work is supported by NIH grants AI074754 and AI074944 to YWH.