Retinoblastoma 1 protects T cell maturation from premature apoptosis by inhibiting E2F1

T lymphocytes are key players in cell-mediated immunity. T lymphocyte progenitors are derived from hematopoietic stem cells (HSCs) and then developed for maturation in the thymus (Kondo et al., 1997). Lymphocytes must be able to produce diverse and specific antigen receptors to fight against invading pathogens. These receptors are coded by sequences with different variable (V) regions. However, this stochastic process is prone to generate antigen receptors that are nonfunctional or self-reactive. To avoid autoimmunity, self-reactive lymphocytes must be eliminated (Kruisbeek and Amsen, 1996).

Apoptosis is one mechanism by which immature self-reactive T lymphocytes are eliminated during maturation (Sohn et al., 2007). The transcription factor E2F1, a well-known retinoblastoma susceptibility (RB1)-interacting protein, influences apoptosis during thymic negative selection (Zhu et al., 1999); By using E2f1^−/− mice, Field et al. determined that loss of E2F resulted in a mature stage T cells (CD4⁺ or CD8⁺ single-positive) increased, presumably owing to decreased apoptosis during negative selection (Field et al., 1996). After that, by using mouse embryo fibroblasts, more-recent studies have shown that RB1/E2F1 is able to bind to the Ets1 and Zeb1 promoters to repress their expression (Dean et al., 2015; Liu et al., 2007). In addition, Zeb1^−/− mice show a decrease number in multiple stage of thymocytes, including intrathymic Kit⁺ T precursor cells in T early stage (Dean et al., 2015; Higashi et al., 1997). In a clinical setting, somatic mutation of RB1 is the most common genomic abnormality in ∼21% of chronic lymphocytic leukaemia (CLL) cases (Ouillette et al., 2011; Puiggros et al., 2014). Thus, previous studies have suggested that Rb1 may be involved in thymocyte development, but there is no direct in vivo evidence to prove a role for Rb1 in the process. Whether Rb1 has effects on thymocytes and whether it is in an E2F1-dependent manner are still unclear. And the stage of the affected thymocytes (immature or mature) also needs to be clarified. Furthermore, the mechanism underlying the apoptosis pathway remains to be uncovered.

In this study, we generated a Rb1 loss-of-function zebrafish mutant (rb1^smu8/smu8) using TALENs (Dee et al., 2016; Langenau and Zon, 2005; Trede et al., 2004). By using the rb1^smu8/smu8 zebrafish model, we show that Rb1 is necessary for cell apoptosis during early T lymphocyte maturation. We found that T cell numbers in the Rb1-deficient mutant were inadequate, whereas development of other hematopoietic cells was unaffected. The reduction of T cells in rb1^smu8/smu8 mutants resulted from premature cell apoptosis mediated by elevated caspase 3 activity. We further demonstrated that downregulation of E2F1 could rescue the inappropriate apoptosis in rb1^smu8/smu8 mutants. Our findings suggest that Rb1 can inhibit E2F1 from triggering the caspase cascade during early T lymphocyte maturation.

To investigate whether Rb1 has roles during T cell maturation, we used TALEN targeting to isolate germline mutations in the zebrafish retinoblastoma susceptibility gene (rb1). Zebrafish embryos were injected with TALEN mRNA targeting rb1 exon 2 with high efficiency, obtaining up to 80% allelic loss in injected F0 embryos (Fig. 1A). We screened one F1 adult for germline transmission rb1 alleles using an 8 bp (referred as rb1^smu8/smu8) frameshift deletion, in which a truncated Rb1 protein lacking all functional domains should be produced (Fig. 1A). As confirmed by western blotting analysis, Rb1 protein was absent in the mutants (Fig. 1B). F2 homozygous larvae failed to develop swim bladders and died at ∼15 dpf (Fig. 1C), similar to the previously reported Rb1 mutant space cadet (Gyda et al., 2012).

To investigate the effect of Rb1 loss on T lymphocyte development, we examined the expression of T-cell markers in rb1^smu8/smu8 mutant using whole mount in situ hybridization. We examined the expression of ikaros (ikzf1) the marker for both immature and mature lymphocytes (Willett et al., 2001), and found that its expression was slightly reduced in the thymus of the mutants at 5 dpf (Fig. 2A), suggesting that thymic lymphocytes are impaired. Levels of the T-cell-specific tyrosine kinase gene (lck), which is expressed in thymic T cells in both immature and mature thymocytes (Langenau et al., 2004), were markedly decreased in the thymus of mutants compared with siblings (Fig. 2A). Furthermore, the expression of rag1, which encodes the recombinase responsible for recombination of the V(D)J and T- and B-cell antigen receptor genes (Wienholds et al., 2002), was also severely downregulated at 5 dpf in the mutants (Fig. 2A), suggesting that immature T cells are reduced in rb1^smu8/smu8 mutants. To validate this observation, by recording the number of lymphocytes in the thymus of Tg(rag2:DsRed) transgenic zebrafish from 3 dpf to 8 dpf, we found that the number of the DsRed⁺ T cells was greatly decreased in mutants (Fig. 2B). Consistently, the qPCR analysis also showed that levels of rag2 and cd4 (cd4-1) were much lower in the mutant thymus (Fig. 2C), indicating the block of early T cell maturation in the rb1^smu8/smu8 mutant. In addition, levels of the thymic epithelial cell marker foxn1 (Schorpp et al., 2002) and the HSC marker myb in the thymus were not altered in rb1^smu8/smu8 mutants (Fig. 2C-E), suggesting the T-cell deficiency is not caused by a HSC defect or a failure in thymus development. Except for T-cell markers, expression of other hematopoietic markers [myb for HSPCs (Zhang et al., 2011), gata1 and βe1 (hbbe1) for erythrocytes (Belele et al., 2009) and pu.1 (spi1b), mfap4 and lyz for myelocytes (Kitaguchi et al., 2009; Zakrzewska et al., 2010)] is unaltered in rb1^smu8/smu8 embryos (Figs S1A-D and S2A-F). Collectively, these data indicate that early T-cell development is disturbed in the absence of Rb1, whereas other hematopoietic lineages are not affected. To determine why rb1 plays specific role in T cells, we further compared rb1 expression in lymphocytes with that in other hematopoietic lineages. As expected, we found that rb1 was highly expressed in rag2:DsRed⁺ T cells, but expressed at lower levels in lyz:DsRed⁺ granulocytes, mpeg1:DsRed⁺ macrophages and globin:DsRed⁺ erythrocytes (Fig. S2H), suggesting its specific function in the T-cell lineage.

Subsequently, we attempted to determine whether the impaired expression of lymphocytic markers in the thymus arose from the deficiency of Rb1 in T cells. The rb1 expression construct driven by the lck or rag2 promoter was injected into rb1^smu8/smu8 embryos to rescue the quantity of T cells. As shown by rag1 whole-mount in situ hybridization or rag2:DsRed fluorescence, T cells in rb1^smu8/smu8 mutants were markedly rescued by rb1 restoration (Fig. 2F and Fig. S3). These rescue experiment showed that delivery of rb1 to T cells can rescue T cell loss in rb1^smu8/smu8 embryos effectively, suggesting a cell-autonomous role for Rb1 in early T-cell development. This result confirms that the defect in development of early T lymphocytes is caused by the absence of Rb1.

Several possibilities could explain the reduction of early T lymphocytes, including reduced proliferation and increased cell death. When compared with siblings, the number of proliferating T cells in the thymus was not decreased in rb1^smu8/smu8 mutants, as indicated by the bromodeoxyuridine (BrdU)/rag2:dsRed incorporation assay (Fig. 3A). On the other hand, as indicated by the TUNEL assay, the number of apoptotic early T lymphocytes was significantly increased in rb1^smu8/smu8 embryos compared with siblings (Fig. 3B), demonstrating that the reduction in the number of early T lymphocytes in rb1^smu8/smu8 is attributed to increased apoptosis. We further examined the apoptotic T-cell numbers in lck:rb1-injected rb1^smu8/smu8 mutants to see whether the increased apoptosis could be rescued. Results show that the apoptotic T-cell numbers in rb1^smu8/smu8 mutants were markedly reduced by rb1 restoration (Fig. 3C), indicating that the Rb1 loss-induced apoptosis is indeed the reason for T-cell loss. Taken together, these results reveal that Rb1 is involved in the early immature T lymphocyte development by inhibiting their apoptosis.

Previous studies in cancer cells have shown that lack of Rb1 promotes chromosome segregation errors, whereas loss of p53 (tp53)allows tolerance and the continued proliferation of these unstable aneuploid cells (Manning et al., 2014). To examine whether P53 is required to limit apoptosis after Rb1 loss, we examined the expression of rag1 in tp53^M214K/M214K and tp53^M214K/M214K rb1^smu8/smu8 zebrafish double mutants. We found that the expression of rag1 was not restored in embryos with rb1 and p53 double-knockout larvae at 5 dpf (Fig. 4A). Consistently, the expression of p53 and its downstream targets [p21 (cdkn1a) mdm2 and ccng1] in the thymus region of rb1 mutants remained unchanged (Fig. 4A), indicating that early T-lymphocyte deficiency in rb1 knockout zebrafish mutants is independent of the p53 pathway.

Caspases, a family of cysteine proteases, are highly conserved throughout vertebrates and are mostly known as executioners of apoptosis (Chowdhury et al., 2008). To determine whether the Rb1 loss-induced apoptosis depends on the caspase cascade, we used the pan-caspase inhibitor Z-VAD-FMK (Vandenabeele et al., 2006) to rescue the lymphocyte apoptosis in the mutants. We found that Z-VAD-FMK can significantly restore early T lymphocyte numbers in rb1^smu8/smu8 mutants (Fig. 4B), suggesting that Rb1 knockout-induced apoptosis is mediated by caspases. Furthermore, we found that both the gene expression and the protein product activity of casp3 were markedly increased in thymus region in mutants compared with siblings (Fig. 4C). These results reveal that the apoptosis of early T lymphocytes in Rb1-deficient mutants is dependent on the activation of caspase 3.

Previous studies have suggested that E2F1 is able to trigger cell apoptosis (Lin et al., 2001), and e2f1^−/− mice exhibit an excess of mature T lymphocytes owing to apoptosis deficiency (Field et al., 1996). We first detected the expression level of e2f1 mRNA in thymus region and found it specifically upregulated in rb1^smu8/smu8 larvae but not in wild-type larvae (Fig. 5A). To investigate whether e2f1 is essential for cell apoptosis regulation during T lymphocyte maturation in zebrafish, knockdown experiments were conducted by injection of e2f1 splice antisense morpholino (MOsp) into wild-type embryos (Fig. 5B,C). The MOsp is predicted to bind to the exon-intron boundary of e2f1 mRNA to block its expression, and the expression level of e2f1 is indeed downregulated in morphants (Fig. 5B). As expected, the number of rag2:DsRed⁺ T cells was increased in the thymus of e2f1-knockdown morphants (Fig. 5C). These data suggest that e2f1 negatively regulates T cell numbers in zebrafish. To further test whether E2F1 acts downstream of Rb1 in T lymphocyte development, we performed e2f1 knockdown experiments in rb1^smu8/smu8 embryos to examine the T-cell development. After e2f1 knockdown, rag2:DsRed⁺ T cells were partially restored owing to the reduced apoptotic T-cell numbers in rb1^smu8/smu8 mutants (Fig. 5D-F). Consistently, T-cell marker genes [rag2, tcra (trac), tcrb2 and cd3 (ighv1-2)] were upregulated in e2f1 morphants and, as expected, the expression of these markers was downregulated in rb1^smu8/smu8 mutants and could be restored by e2f1 knockdown in rb1^smu8/smu8 mutants (Fig. 5G). These data suggest that e2f1 knockdown could partially restore T-cell differentiation. In addition, we also examined the expression of casp3 in e2f1 morphants and showed that it is downregulated (Fig. 5G). Moreover, the elevated casp3 expression in rb1^smu8/smu8 mutants could be restored to normal by e2f1 knockdown (Fig. 5G). The above results reveal that the apoptosis of T lymphocytes in rb1^smu8/smu8 mutants is E2F1 mediated, and casp3 may act downstream of Rb1-E2F1 axis-mediated T-lymphocyte development.

Previous studies have shown that E2F1 can upregulate the transcription of casp3 to induce cell apoptosis (Müller et al., 2001). To further determine whether caspase 3 acts as an executor in Rb1-E2F1 axis-mediated T-lymphocyte development, we overexpressed e2f1 in rb1^smu8/smu8 T lymphocytes and examined casp3 expression. We found that the T lymphocyte marker rag1 was downregulated by e2f1 overexpression, which mimics the low rag1 levels in rb1-deficient embryos (Fig. 6A). The expression of casp3 was upregulated by e2f1 overexpression and was further elevated in rb1-deficient larvae (Fig. 6A). Meanwhile, Z-VAD-FMK was used to inhibit caspase activity in e2f1-overexpressed/rb1-deficient embryos, and T lymphocyte numbers were counted. As expected, overexpressing e2f1 can downregulate the number of rag2:DsRed⁺ T lymphocytes, and the T lymphocyte loss can be recovered by suppressing the caspases activity (Fig. 6B), suggesting that caspase 3 acts downstream of E2F1 during the apoptosis of immature T cells in rb1^smu8/smu8 mutants. In summary, we demonstrated that Rb1 inhibits apoptosis during early immature T-lymphocyte development by repressing the activity of E2F1 to downregulate casp3 expression (Fig. 6C).

To investigate the function of Rb1 in the regulation of T-cell development, we generated Rb1-deficient zebrafish. Zebrafish rb1^smu8/smu8 mutants display a significantly decreased number of T cells, with other hematopoietic cell development being largely unaffected. The decreased T cell number appears to be due to excessive apoptosis that occurs in the immature T cells, which is mediated by caspase 3 but independent of P53. We further showed that E2F1 knockdown could rescue Rb1 deficiency-induced apoptosis, suggesting a crucial role for the Rb1-E2F1-caspase axis in the regulation of immature T-lymphocyte apoptosis.

Our data suggest that during the embryonic stages of T-cell development in zebrafish, Rb1 inhibits apoptosis by repressing the activity of E2F1 and downstream caspase activation. It is possible that, under normal circumstances, in the apoptosis of nonfunctional or self-reactive thymocytes, the dissociated E2F1 from Rb1 plays a crucial role in promoting casp3 expression. However, when E2F1 is overexpressed or there are insufficient Rb1 to bind up all the E2F1, derepression of E2F1 would promote inappropriate apoptosis in early immature thymocytes. It is worth unveiling the mechanism, but there are several challenges in the current study. Antibodies specific for zebrafish T-cell sub-populations have yet to be developed, which makes the identification of distinct developmental stage T cells impossible at present. Nonetheless, we believe that this problem is likely to be solved with the creation of new transgenic zebrafish.

Rb1-null mice studies showed that Rb1 loss induced not only disturbed proliferation, but also excess apoptosis in the neural system and deregulated the maturation of erythrocytes (Chau and Wang, 2003; Lee et al., 1992). Rb1 absence also has been reported to be harmful to many cellular processes, including differentiation (Korenjak and Brehm, 2005; McClellan and Slack, 2007), survival (Delston and Harbour, 2006; Hallstrom and Nevins, 2009), senescence (Ben-Porath and Weinberg, 2005; Liu et al., 2004) and genome stability (Knudsen et al., 2006). However, none of these studies has underlined the exact function of Rb1 in the development of T lymphocytes. This may be because zebrafish develop rapidly ex utero (Langenau and Zon, 2005) and lymphopoiesis occurs in the thymus by 3 dpf (Jagannathan-Bogdan and Zon, 2013). Compared with Rb1-null mice, which die before E16 with multiple defects (Lee et al., 1992), Rb1-null zebrafish can survive to 15 dpf. Interestingly, the thymic cellularity of E2f1^−/− mice was noticeably increased at 4-6 weeks. These data indicate that Rb1 may also be necessary for T-cell development in mice through repression of E2f1.

To address the issue of Rb1 specificity in regulating T-cell development more adequately, we examined the expression of rb1 in different hematopoietic cell types and found that rb1 is more abundant in T lymphocytes than in other blood cells (Fig. S2H). Likewise, e2f1 levels are much higher in the thymus compared with the tail region when Rb1 is mutated (Fig. 5A). These data indicate that Rb1-E2F1 pathway plays a crucial role in T-cell development in the thymus, in which 95% of immature thymocytes are eliminated via apoptosis (Kappler et al., 1987). We believe that a relative high level of Rb1 in developing T cells is crucial for inhibiting e2f1 activity, thereby preventing normal developing T cells from inappropriate apoptosis.

As shown previously, both Rb1-deficient mice and Rb1-deficient zebrafish display severe neuronal defects (Clarke et al., 1992; Gyda et al., 2012; Lee et al., 1992). Consistent with previous findings, we also found that rb1 is highly expressed in the brain tissue (data not shown) and neuron apoptosis is significantly increased in rb1^smu8/smu8 (increased neuronal apoptosis could be rescued by a caspase inhibitor and casp3 was upregulated in brain in rb1^smu8/smu8, data not shown), suggesting that Rb1 plays a similar role in neuron and T-cell development. Interestingly, during both T-cell development and neurogenesis, a large numbers of undesired cells must undergo apoptosis. We speculate that Rb1 may play an essential role in preventing desired cells from inappropriate apoptosis.

Chromosome instability (CIN) and aneuploidy are a common feature of tumour cells, and studies have shown that Rb1 inactivation could promote CIN and aneuploidy (Manning et al., 2010). Clinically, RB1 deletions are frequently associated with additional acquired chromosomal copy number changes in individuals with CLL (Ouillette et al., 2011). This mis-segregation of chromosomes causes eventual death in cells that lack Rb1 function. However, we suspect that when thymocytes are not regulated, such genomic changes potentially promote the evolution of CLL. Combined with our finding that immature lymphocytes that lack Rb1 have enhanced cell apoptosis, removal of Rb1-deficient cells may be important for organ homoeostasis when treating CLL.

In summary, our results provide the first functional assay of Rb1 in early T-cell development and show that Rb1 inhibits E2F1 to trigger the caspase cascade during early T-lymphocyte maturation. Given the fact that somatic deletion of RB1 is the most frequent chromosomal abnormality in CLL, elucidating the mechanism behind the regulation of immature T-cell apoptosis regulation by Rb1 provides an intriguing link between tumour suppression and lymphatic system development that needs to be further investigated.

Zebrafish were maintained at 28.5°C in a 14 h light and 10 h dark cycle. Embryos were collected by natural spawning and raised at 28.5°C. To prevent the formation of melanin pigment, embryos were incubated in egg water containing 0.045% 1-phenyl-2-thiourea (PTU, Sigma, P7629) after gastrulation stage. The embryos were collected at the desired stages (Westerfield, 2000). The following strains were used: AB, tp53^M214K (Berghmans et al., 2005), Tg(rag2:Dsred) (Ma et al., 2012), Tg(mpeg1: loxP-DsRedx-loxP-GFP) (Ellett et al., 2011), Tg(lyz:Dsred) (Hall et al., 2007), Tg(globin:Dsred) and rb1^smu8/smu8 mutants.

Synthesis of digoxigenin-labelled antisense RNA probes and whole mount in situ hybridization were performed as described previously (Jin et al., 2016; Liu et al., 2017). The probes were listed as follows: rag1, myb, lck, ikaros, foxn1, pu.1, lyz, mfap4, gata1 and βe1.

Western blotting was performed as described previously (Huang et al., 2002). The anti-Rb1 antibody was obtained from Proteintech (17218-1-AP).

The coding sequences of zebrafish rb1 and e2f1 were amplified by PCR and spliced into the lck or rag2 promoter-containing pTol vector using XmaI/BamHI or AgeI/BamH1 digestion, respectively. The lck and rag2 promoters have been described previously (Jessen et al., 2001; Langenau et al., 2004). pTol-lck:rb1, pTol-lck:e2f1 or pTol-rag2:rb1 with transposase mRNA were injected into one-cell stage AB embryos at a dose of 100 pg/embryo.

Embryos at 4 dpf stage were incubated in 10 mmol/l bromodeoxyuridine (BrdU, Sigma-Aldrich, B9285) solution (0.5% DMSO in egg water) for 4 h and subsequently fixed in 4% paraformaldehyde. After 30-min treatment with 2 N HCl, the embryos were stained using primary mouse anti-BrdU (Roche, 10875400; 1:50, at 4°C overnight) and rabbit anti-DsRed (Clontech, 632496; 1:400, at 4°C overnight) antibodies, and finally were visualized with Alexa Fluor 555 donkey anti-mouse (Invitrogen, A31572) and Alexa Fluor 488 donkey anti-rabbit (Invitrogen, A21206) antibodies.

The paraformaldehyde-fixed embryos (5 dpf stage) were further incubated in a PBST solution containing 0.1% Triton X-100 and 0.1% sodium citrate for 15 min followed by three rinses in PBST. The embryos were subsequently soaked in the terminal deoxynucleotidyl transferase dUTP nick end labelling mix using the in situ cell death detection kit (Roche, 12156792001) at 37°C overnight and stained using the anti-DsRed antibody.

Total RNA extraction and complementary DNA synthesis were performed as described previously (Lin et al., 2016). Quantitative reverse-transcription PCR (qRT-PCR) was performed using the light cycler Nano Real-time PCR system (Roche) with an SYBR Green Master mix (Roche, 06402712001). The housekeeping gene ef1a was used as the internal control. A least 30 embryos were included in each experiment. The primer sequences are described in Table S1.

rb1^smu/smu8 mutants were genotyped by PCR followed by BclI (Thermo, ER0722) digestion. The wild-type PCR products were digested using BclI into two fragments of 198 bp and 176 bp, respectively, whereas mutant PCR products were resistant to the BclI digestion. Primers for rb1 genotyping were as follow: FP, 5′-GCCACTGCTAAACACTAAAGA-3′; RP, 5′- GCTCCATGCCAGCAATAAAA-3′.

The design and injection of e2f1 MOs was performed as previously reported (Bill et al., 2009). e2f1 MOsp (5′-TTTTAGTAATCATTCATACCTCTGG-3′) targeting protein translation were obtained from Gene Tools and injected into zebrafish embryos at the one-cell stage (0.5 pmol per embryo). The number of T lymphocytes was quantified at 5 dpf.

Z-VAD-FMK (V116, 200 μM) was purchased from Sigma-Aldrich and dissolved in egg water with DMSO.

Caspase 3 activity was determined using the Caspase 3 Activity Kit (Beyotime, C1115). Ten thymus excised from the larvae were collected and pooled as one sample. The larvae were washed twice with phosphate-buffered saline (PBS) and then homogenized in 100 μl of lysis buffer on ice for 5 min. The lysate was centrifuged at 16,000 g at 4°C for 15 min. The supernatants were collected and immediately measured for total protein concentration and caspase 3 activity. For the caspase 3 activity assay, 10 μl of supernatant was placed in a 96-well plate containing 80 μl reaction buffer and 10 μl of caspase 3 substrate (Ac-DEVD-pNA). The plate was incubated at 37°C in the dark for 30 min, and enzyme activity was determined through measuring the optical density of each sample at 405 nm using TECAN infinite M200 Absorbance Reader. Total protein concentration was determined using a Bradford assay (Beyotime, P0006).

The calculated data were recorded and analysed using prism software. The unpaired two-tailed Student's t-test for comparisons between two groups and one-way analysis of variance (ANOVA; with Bonferonni or Dunnett T3 post-test adjustment) among multiple groups. P<0.05 was deemed significant.

We thank Dr Bo Zhang (Peking University, China) for providing us with TALEN reagents and protocol.