Facts

  • Pro-survival BCL-2 proteins have overlapping roles in securing the survival of certain cell types, whereas other cell populations rely on the expression of a specific pro-survival BCL-2 family member.

  • Many cell types during embryogenesis as well as in the adult cannot tolerate the loss of the pro-survival protein MCL-1.

  • On-target toxic effects of BH3 mimetic drugs targeting specific pro-survival BCL-2 proteins on healthy cells are generally less severe than the defects seen in gene-targeted mice lacking these proteins.

Open questions

  • Why can the survival dependence of distinct cell types not be explained exclusively by the expression profile of the different pro-survival BCL-2 proteins?

  • Why is MCL-1 critical for early embryogenesis and the survival of so many cell types?

  • Are there apoptosis-unrelated functions of pro-survival BCL-2 proteins, in particular MCL-1, that are critical for the survival of distinct cell types?

  • Do the reported apoptosis unrelated functions of pro-survival proteins play a role in the efficacy and/or on-target toxic effects of BH3 mimetic drugs?

Pro-survival BCL-2 proteins—same, same but different

Mammals have five pro-survival BCL-2 family members: BCL-2, BCL-XL, MCL-1, BCL-W and A1(murine)/BFL-1(human). They regulate mitochondrial apoptosis through interactions with two sub-groups of pro-apoptotic BCL-2 family members [1,2,3]; the BH3-only proteins (PUMA, BIM, tBID, NOXA, BMF, BIK, HRK, BAD) that are critical for apoptosis initiation as well as the multi-BH (BCL-2 homology) domain effectors of apoptosis, BAX and BAK [2]. The BAX/BAK-related multi-BH-domain protein BOK is not regulated by pro-survival BCL-2 proteins [4, 5]. Apoptosis is initiated in response to diverse stresses, such as cytokine deprivation or oncogene activation, resulting in the transcriptional and/or post-transcriptional upregulation of pro-apoptotic BH3-only proteins [6]. BH3-only proteins bind the pro-survival BCL-2 proteins with high affinity and thereby neutralise them. This unleashes BAX and BAK from their restraint by the pro-survival BCL-2 proteins, allowing them to cause mitochondrial outer membrane permeabilisation (MOMP) [2, 3]. Some BH3-only proteins can reportedly also directly activate BAX and BAK [7,8,9]. MOMP results in the release of apoptogenic factors (e.g., Cytochrome c) from the mitochondrial inter-membrane space leading to the activation of the caspase cascade that causes the ordered dismantling of dying cells (Graphical Abstract) [3].

All pro-survival proteins contain four BH domains and interact with both sub-groups of pro-apoptotic BCL-2 family members through their hydrophobic BH3 binding groove (Fig. 1A, B). Even though highly related to each other, different pro-survival BCL-2 proteins have some distinct features. BCL-2, BCL-XL, MCL-1 and BCL-W have a C-terminal transmembrane (TM) domain that allows them to anchor to different intracellular membranes (Fig. 1A). BCL-2 is found at the endoplasmic reticulum (ER), the mitochondrial outer membrane (MOM) and the nuclear envelope (NE) [10,11,12]. MCL-1 is found at the ER, the MOM as well as the mitochondrial inter-membrane space, the NE and in the cytosol [13,14,15,16,17]. BCL-XL is located in the cytosol and at the MOM but has also been detected at the NE and the ER [18,19,20]. BCL-W is cytosolic but can insert into the MOM in response to intracellular stress [21]. A1 lacks a TM domain and is therefore mainly found in the cytosol but it is also detected at the MOM, the ER and the NE [22,23,24] (Fig. 2).

Fig. 1: The regulation of the mitochondrial apoptotic pathway by pro-survival BCL-2 proteins.
figure 1

A Schematic presentation of the structure of the different pro-survival BCL-2 proteins (not to scale). B Binding of the different pro-survival BCL-2 proteins to the different pro-apoptotic BH3-only proteins as well as the apoptosis effectors, BAX and BAK.

Full size image
Fig. 2: Subcellular localisation of the pro-survival BCL-2 proteins.
figure 2

Schematic presentation of the subcellular localisation of BCL-2, BCL-XL, MCL-1, A1 and BCL-W. Superscript numbers relate to the literature reference.

Full size image

MCL-1 has a long N-terminal stretch harbouring 4 putative PEST sequences, and this is in part responsible for its short (20–30 min) half-life [25, 26]. In contrast, BCL-2, BCL-XL and BCL-W have half-lives of >20 h [27, 28]. A1/BFL-1 protein has a half-life of only 10 min [29] (Fig. 1A).

The pro-survival BCL-2 proteins differ in their binding to the apoptosis effectors BAX and BAK. All pro-survival BCL-2 proteins can restrain BAX, but only BCL-XL, MCL-1 and A1 also bind BAK [30, 31]. Furthermore, whereas all pro-survival BCL-2 proteins can bind to the BH3-only proteins BIM, PUMA and tBID, NOXA interacts only with MCL-1 and A1, while BAD selectively binds BCL-2, BCL-XL and BCL-W [9, 30, 32] (Fig. 1B).

The roles of the different pro-survival BCL-2 proteins in embryogenesis

During the initial stages of embryogenesis (days 0–2) the expression of proteins in the zygote is driven from maternal mRNA. Only after several cell divisions (6–8 cell state) around embryonic day 3 (E3) the embryonic genome is activated to allow mRNA and protein synthesis [33].

Among the pro-survival BCL-2 proteins, the poorly studied BCL-B(human)/DIVA(murine) is the most prominently expressed at the mRNA level in oocytes and early embryos [34]. However, loss of BCL-B/DIVA has no impact on oocytes or early embryos [35]. Of note, some reports indicate that BCL-B/DIVA lacks a BH3 domain and its function in regulating cell death is therefore unclear [36,37,38]. Neither A1 nor BCL-W are expressed during early embryogenesis. MCL-1 and BCL-XL are highly expressed from both maternally inherited mRNA as well as embryonic transcripts [39]. BCL-2 expression levels are low during early embryogenesis but increase after the late blastocyst stage (~E4.5). Here, we focus on the most relevant pro-survival BCL-2 family members during embryogenesis, MCL-1, BCL-XL and BCL-2.

BCL-2 (Table 1): BCL-2-deficient mice are born but die soon after weaning due to polycystic kidney disease, a disorder that commences during embryogenesis [40]. This demonstrates an essential role of BCL-2 for renal epithelial cell survival during embryogenesis. BCL-2 levels are high in the central nervous system during embryogenesis but decline after neuronal tube formation [39]. In contrast, high levels of BCL-2 are maintained in the peripheral neuronal system [10, 41,42,43]. BCL-2-deficient mice show normal neuronal development during embryogenesis but exhibit a significant loss of sympathetic, motor and sensory neurons during the early postnatal period [44].

Table 1 Bcl-2 gene targeting in mice.
Full size table

BCL-XL (Table 2): Constitutive absence of BCL-XL (encoded by the Bcl2l1 gene) results in embryonic lethality ~E13.5 due to erythroid and neuronal defects [45, 46]. To further study the role of BCL-XL during development and in adulthood, chimaeric mice were generated through injection of Bcl2l1-/- (in the following referred to as Bcl-x-/-) or control embryonic stem (ES) cells into wild-type blastocysts. No significant differences in the contribution of Bcl-x-/- vs control cells were observed in the heart, kidney or muscle [46], but BCL-XL was indispensable for the survival of embryonic erythroid progenitors (EryP) and definitive erythrocytes (EryD) in the adult [45]. Furthermore, Bcl-x-/- ES cells contributed less to the spleen and thymus compared to control ES cells [45].

Table 2 Bcl2l1 (Bcl-x) gene targeting in mice.
Full size table

Bcl-x-/- embryos present with massive apoptotic death of post-mitotic immature neurons in the developing brain, spinal cord, and dorsal root ganglia [46]. Accordingly, extensive cell death was detected in cultures of telencephalic neurons derived from Bcl-x-/- embryos, and this was rescued by loss of BAX [47, 48]. BCL-XL deficiency specifically in catecholaminergic neurons (tyrosine-hydroxylaseCre-Bcl-xfl/fl mice) resulted in viable offspring, although this neuronal population was reduced by one-third [49]. However, some catecholaminergic neurons without BCL-XL were present [49], indicating that their survival can be safeguarded by additional pro-survival BCL-2 proteins.

MCL-1 (Table 3): Amongst the pro-survival BCL-2 proteins, only MCL-1 is essential for early embryogenesis. Constitutive absence of MCL-1 in mice results in peri-implantation lethality (~E3.5) [50]. The authors of this study failed to detect an increase in apoptotic cells in the Mcl-1-/- blastocysts and therefore hypothesised that the essential role of MCL-1 during implantation is unrelated to its role in inhibiting apoptosis. Further investigations are needed to clarify whether the peri-implantation lethality caused by the absence of MCL-1 is triggered by excess cell death or defects in apoptosis-unrelated functions of MCL-1.

Table 3 Mcl-1 gene targeting in mice.
Full size table

Tissue-restricted gene deletion revealed an essential role for MCL-1 in neuronal and cardiac tissues in embryos. NestinCre-Mcl-1fl/fl mice, with neuronal-restricted loss of MCL-1, developed only until ~E12.5 [42, 51]. A wave of apoptosis was detected at E9.5 in the brainstem and cervical spinal cord and at E10.5 in the forebrain. Interestingly, while loss of BAX was sufficient to protect the majority of neuronal precursor cells from apoptosis, this rescued only ~50% of cells in the dorso-medial brainstem and ventral-thoracic spinal cord [51]. Thus, MCL-1 ensures NPC survival mainly by restraining BAX in the brain stem, while the survival of other brain cells must also rely on the inhibition of BAK. The critical role of MCL-1 in neurogenesis in the forebrain was further demonstrated in Foxg1Cre-Mcl-1fl/fl mice, which all died before E15 [42].

CkmmCre-Mcl-1fl/fl mice, in which MCL-1 is removed from cardiomyocytes and skeletal muscle cells die around post-natal day 10 due to myocardial degeneration, demonstrating that MCL-1 is essential for cardiomyocyte survival [52]. The loss of MCL-1 had only minor impact on skeletal muscle cells, at least within the short lifetime of the pups [52], suggesting that their survival is safeguarded by additional pro-survival BCL-2 proteins.

The roles of pro-survival BCL-2 proteins in adult mice

Some healthy cells in the adult are protected from apoptosis by mostly one pro-survival BCL-2 family member, whereas others are safeguarded by two or more (Fig. 3). Moderate to high expression of MCL-1 is found in most adult tissues, including the brain, digestive system, liver, kidney, reproductive organs (male and female), smooth and skeletal muscle and cardiomyocytes [53,54,55]. BCL-2 expression is reported as low in the adult heart, liver, skeletal muscle but moderate to high in the colon, male and female reproductive organs, skin, colon and kidneys [53,54,55]. Moderate to high levels of BCL-XL have been observed in many adult tissues, however, others such as cardiac and smooth muscles as well as the skin lack BCL-XL [53,54,55]. BCL-W is expressed in the brain, spinal cord, colon, testes, pancreas, liver, heart, spleen and mammary glands of pregnant mice [21]. A1 is only found in haematopoietic cells [53,54,55] (Fig. 3).

Fig. 3: Expression profile of pro-survival BCL-2 family members in solid tissues.
figure 3

A mRNA expression of the pro-survival BCL-2 family members BCL2 (encoding BCL-2), BCL2L1 (encoding BCL-XL), MCL1 (encoding MCL-1), BCL2L2 (encoding BCL-W), BCL2A1 (encoding BFL-1) in human tissues. Data shown are from the Human Protein Atlas representing pooled Consensus Normalized eXpression (NX) levels created by combining the data from the three transcriptomics datasets (HPA, GTEx and FANTOM5) using an internal normalization pipeline [54]. B Protein expression profile of pro-survival BCL-2 family members in human tissues, quantified in a range from not detected to high expression. Data are from the Human Protein Atlas [54].

Full size image

BCL-2 (Table 1): BCL-2-deficient mice die ~30 days after birth (C57BL/6 background) due to polycystic kidney disease, a disorder that starts during embryogenesis [40, 56]. BCL-2-deficient mice also present with substantial reductions in mature B and T lymphocytes and melanocytes, the latter causing premature greying [40, 56,57,58]. All defects caused by the absence of BCL-2 could be prevented by concomitant loss of the pro-apoptotic BH3-only protein BIM [59, 60]. This demonstrates that BCL-2 safeguards the survival of several cell types by preventing BIM-induced apoptosis.

BCL-2-deficient mice also exhibit an abnormal reduction of sympathetic, motor and sensory neurons during the early postnatal period [44]. NestinCreERT2-Bcl-2fl/fl mice were generated to facilitate inducible deletion of Bcl-2 specifically in NPCs in adult mice. This revealed that BCL-2 is essential for the survival of doublecortin-expressing immature neurons in the central and peripheral nervous systems [61].

BCL-2 is also critical for endothelial cell survival. Abnormally decreased numbers of endothelial cells and pericytes were observed in retinas from BCL-2-deficient mice, resulting in decreased retinal vascular density [62].

BCL-XL (Table 2): The loss of only one allele of Bcl2l1 (encoding Bcl-x) impairs male fertility, although with incomplete penetrance [63]. A critical role of BCL-XL in male and female reproductive organs was confirmed in a Bcl-x hypomorphic mouse strain which exhibited reduced germ cell survival during gonadogenesis. All Bcl-x hypomorphic males were sterile and had abnormally small testes. Interestingly, 25% of the Bcl-x hypomorphic females were fertile but had abnormally small litters [63].

CreERT2-induced whole-body deletion of Bcl-x in Bcl-xfl/fl;RosaCreERT2 mice was fatal at ~25 days due to severe anaemia and thrombocytopenia caused by the loss of erythroid progenitors and platelets [64]. Unexpectedly, deletion of BCL-XL in all cells other than haematopoietic ones also caused fatal anaemia and thrombocytopenia, in this case due to severe kidney damage resulting in the loss of the red blood cell-stimulating and megakaryocyte-stimulating hormones erythropoietin (EPO) and thrombopoietin (TPO), respectively [64]. This identifies an essential role of BCL-XL for the survival of mitochondria-rich renal tubular cells in the kidneys [64].

Hepatocyte-specific loss of BCL-XL in AlbCre-Bcl-xfl/fl mice caused hepatocyte apoptosis and liver fibrosis [65]. RIPCre-Bcl-xfl/fl mice, with specific loss of BCL-XL in pancreatic β-cells, are viable and show normal development of the pancreatic islets but these cells are hypersensitive to apoptosis-inducing agents [66]. Approximately 50% of mice in which BCL-XL is absent in lung epithelial cells (SftpcCre-Bcl-xfl/fl) die soon after birth. Interestingly, the analysis of the viable offspring revealed that BCL-XL expression is not essential for lung development, but the pulmonary epithelial cells are hypersensitive to apoptotic stimuli [67].

MCL-1 (Table 3): Loss of MCL-1 in cardiomyocytes is fatal in adults [19, 52, 68]. MyhCreER-Mcl-1fl/fl as well as MerCreMer-Mcl-1fl/fl mice died around 3–4 weeks after induced deletion of Mcl-1 [52, 68, 69]. Interestingly, while CkmmCre-Mcl-1fl/fl as well as tamoxifen-treated (to activate CreERT2) MyhCreERT2-Mcl-1fl/fl mice developed until late adulthood in the absence of BAX and BAK (CkmmCre-Mcl-1fl/fl;Baxfl/fl;Bak-/- and MyhCreERT2-Mcl-1fl/fl;Baxfl/fl;Bak-/- mice), defects in mitochondrial dynamics and oxygen consumption were observed in these triple knockout mice [52]. It was therefore proposed that loss of an apoptosis-unrelated function of MCL-1 in mitochondrial dynamics and energy production is at least partially responsible for the cardiac defects observed [52, 68]. Specifically, it was reported that a proteolytically cleaved form of MCL-1 is imported into the mitochondrial intra-membrane space where it regulates mitochondrial fusion and mitochondrial respiratory complex assembly [70]. However, the CRE-mediated deletion of the floxed Mcl-1 and Bax alleles in CkmmCre-Mcl-1fl/fl;Baxfl/fl;Bak-/- mice as well as tamoxifen-treated MyhCreERT2-Mcl-1fl/fl;Baxfl/fl;Bak-/- mice was achieved simultaneously [52]. Given the large difference in the half-life of MCL-1 (~20 min) [25, 26] vs BAX (24 h) [71], it cannot be excluded that the observed defects in mitochondrial ultrastructure and oxygen consumption were a consequence of BAX-mediated induction of MOMP, because these cells would experience several hours when MCL-1 is no longer present but BAX is still there. In order to avoid this complication and be able to better investigate the importance of the postulated apoptosis-unrelated function of MCL-1, a mouse model that allows the deletion of Mcl-1 and Bax independently is needed (e.g., first deletion of Bax using the FRT/flp system and then deletion of Mcl-1 using the CreERT2/loxP system).

The deletion of MCL-1 in neuronal cells of adult mice was achieved by stereotaxic injection of NestinCre-expressing plasmids into the lateral ventricles of adult Mcl-1fl/fl mice. This demonstrated a critical role of MCL-1 in adult NPCs of the subventricular zone [19]. MCL-1 is also needed for hepatocyte survival [72]. AlbCre-Mcl-1fl/fl mice are born with abnormally small livers due to high rates of hepatocyte apoptosis, resulting in chronic liver damage and hepatic pericellular fibrosis [72]. Other studies revealed that MCL-1 and BCL-XL cooperatively maintain the integrity of hepatocytes in developing and adult mice [73]. Accordingly, AlbCre-Bcl-xfl/flMcl-1fl/fl mice display a massive reduction in hepatocytes and abnormally small livers and die within 1 day after birth [73].

A critical role for MCL-1 was reported for epithelial cells, including thymic epithelial (TECs) as well as intestinal epithelial cells (IECs) [74, 75]. In contrast, loss of BCL-2 or BCL-XL had no impact on the survival of TECs or IECs. Foxn1Cre-Mcl-1fl/fl mice, in which MCL-1 is absent in TECs produce viable offspring. However, these mice are severely immunocompromised due to early thymic atrophy and T-cell lymphopenia, with near complete loss of thymic tissue by 2 months of age [75]. The loss of MCL-1 in IECs caused increased apoptosis leading to severe entero-colopathy. The increased IEC apoptosis was associated with hyperproliferative crypts, driven by compensatory proliferation of cells that had not recombined Mcl-1fl, and this led to epithelial barrier dysfunction, chronic inflammation and accumulation of DNA damage in hyperproliferating IECs. This caused the development of intestinal tumours with morphological and genetic features of human adenomas and carcinomas [74]. MCL-1 is also essential for mammary epithelial cell survival and lactation using K5Cre-Mcl-1fl/fl, MMTVCre-Mcl-1fl/fl and WAPiCre-Mcl-1fl/fl mice [76] as well as the survival of endothelial cells. Even though some endothelial cell specific Mcl-1-deleted mice develop until adulthood, most Tie2Cre-Mcl-1fl/fl pups are lost before weaning. They exhibit abnormally high rates of apoptosis in the angiogenic vasculature and a decline in vessel density [77].

BCL-W: BCL-W (encoded by the Bcl2l2 gene)-deficient mice develop normally and most tissues from BCL-W-deficient mice are indistinguishable from those of wild-type controls [21, 78]. However, BCL-W-deficient males are infertile, demonstrating a critical role for BCL-W in spermatogenesis [78, 79].

A1: BFL-1 is expressed from a single gene in humans (BCL2A1) but there are four A1 genes in the mouse genome [80]. Bcl2a1a, Bcl2a1b and Bcl2a1d are expressed and almost identical in sequence, whereas Bcl2a1c is a pseudogene. Mice lacking all three functional A1 genes develop and age normally and only exhibit minor abnormalities in certain haematopoietic cell subsets [80, 81]. This is not surprising, given that A1 expression is largely restricted to haematopoietic cell lineages (Fig. 4) [80].

Fig. 4: mRNA expression profile of pro-survival BCL-2 family members in haematopoietic cell populations.
figure 4

A Expression profile of the different pro-survival BCL-2 family members Bcl2 (encoding BCL-2), Bcl2l1 (encoding BCL-XL), Mcl1 (encoding MCL-1), Bcl2l2 (encoding BCL-W), Bcl2a1a-d (encoding A1 proteins) in different murine haematopoietic cell populations. Data are derived from the Haemopedia Mouse RNA-Seq atlas and shown in log2 transcripts per million [84]. B Expression profile of the different pro-survival BCL-2 family members BCL2 (encoding BCL-2), BCL2L1 (encoding BCL-XL), MCL1 (encoding MCL-1), BCL2L2 (encoding BCL-W), BCL2A1 (encoding BFL-1) in different human haematopoietic cell populations. Data are derived from Novershtern et al. [85] and shown as log2 normalised expression. Data for both heatmaps were sourced from www.haemosphere.org [84]. MPP – multi potential progenitor; ST-HSC - short term haematopoietic stem cell; CMP – common myeloid progenitor; GMP – granulocyte macrophage progenitor; MEP – megakaryocyte erythroid progenitor; cDC – conventional dendritic cell; pDC – plasmacytoid dendritic cell; NK cell – natural killer cell.

Full size image

The roles of the different pro-survival BCL-2 proteins in haematopoiesis

Haematopoiesis describes the production and differentiation of all mature blood cell subsets from haematopoietic stem and progenitor cells (HSPCs). HSPCs have self-renewal capability and give rise to multipotent progenitor cells (MPPs) which in turn give rise to the lineage-committed progenitor cells, including the common lymphoid progenitors (CLPs) and common myeloid progenitors (CMPs). The latter further differentiate into megakaryocyte-erythroid progenitor (MEPs) or granulocyte-macrophage progenitors (GMPs). CLPs are the origin of all lymphoid cells, including B as well as T lymphocytes and natural killer (NK) cells. MEPs give rise to erythroid cells and megakaryocytes, which shed platelets, while granulocytes (e.g., neutrophils, eosinophils, monocytes) originate from GMPs.

The survival of immature and mature haematopoietic cells is safeguarded by pro-survival BCL-2 proteins with distinct family members being critical in different cell subsets (Fig. 4). MCL-1 is highly expressed in HSPCs and many mature haematopoietic cell subsets in both mice and humans [82,83,84,85]. In contrast, the expression of BCL-XL is moderate throughout haematopoiesis, but with relatively high levels found in erythroid cells and megakaryocytes [82,83,84,85]. BCL-2 is most prominently expressed in certain B and T lymphoid populations as well as NK cells and moderate expression is also found in stem and progenitor populations, including HSCs, MPPs, CMPs and CLPs, but only low expression is seen in erythroid cells [82,83,84,85]. A1 is found in antigen receptor stimulated T and B lymphocytes, dendritic cells, neutrophils, eosinophils and monocytes [82,83,84,85]. BCL-W is expressed at comparatively low levels in the haematopoietic compartment (Fig. 4) [82,83,84,85].

Conditional gene deletion studies in mice have identified the essential roles of distinct pro-survival BCL-2 proteins in the survival of different haematopoietic cell types.

BCL-2 (Table 1): Examination of chimaeric mice with a BCL-2-deficient haematopoietic compartment revealed that BCL-2-deficient mature T cells developed normally in vivo but had abnormally short lifespan in vitro and increased sensitivity to glucocorticoids and γ-irradiation compared to control T cells [86]. Antigen receptor stimulation enhanced the in vitro survival of T cells lacking BCL-2 [86], presumably because this causes NF-kB-driven up-regulation of BCL-XL [87]. In these chimaeric mice BCL-2-deficient T and B cells disappeared from the bone marrow, thymus, and peripheral lymphoid organs by 4 weeks of age [86, 88]. Characterisation of a Bcl-2 YFP reporter mouse strain revealed that BCL-2 is critical for the survival of effector and memory T lymphocytes [89, 90]. Notably, the survival defects of all lymphoid subsets caused by the absence of BCL-2 could be rescued by the concomitant loss of only a single allele of pro-apoptotic Bim [60]. This demonstrates that the function of BCL-2 in mature B and T lymphocytes is mainly to oppose BIM-induced apoptosis.

The absence of BCL-2 had no impact on the numbers of myeloid as well as erythroid cells and platelets. Notably, even cells that express moderate to high levels of BCL-2 (e.g., pro-myelocytes, normoblasts, megakaryocytes) are present at normal numbers in BCL-2-deficient mice [88]. Tissue-restricted deletion of floxed Bcl-2 revealed that BCL-2 contributes to NK cell survival (Ncr1Cre-Bcl-2fl/fl mice) [91] but is dispensable for the survival of megakaryocytes and platelets (Pf4-Cre-Bcl-2fl/fl mice) [92].

BCL-XL (Table 2): Even though moderate to high expression of BCL-XL has been reported for most haematopoietic cell populations (Fig. 4), the absence of BCL-XL only impacts the survival of some. The role of BCL-XL in haematopoiesis was studied in chimaeric mice that were generated by injection of Bcl-x-/- ES cells into wild-type or Rag1-/- blastocysts, the latter unable to give rise to mature T or B cells [46]. This revealed impaired survival of immature lymphocytes, such as CD4+CD8+ thymocytes, but not mature lymphocytes [46, 93]. Conditional deletion of Bcl-x specifically in T lymphoid cells from an early stage of differentiation (LckCre-Bcl-xfl/fl mice) revealed that the development of effector and memory T lymphocytes was not impacted by the loss of BCL-XL [94].

Analysis of the aforementioned chimaeric mice revealed that Bcl-x-/- ES cells did not contribute to circulating EryD cells in the peripheral blood, demonstrating that BCL-XL plays an important role in erythroid progenitor survival. In vitro differentiation analysis confirmed that BCL-XL is critical for the survival of both primitive (EryP) and definite erythroid (EryD) progenitor cells [45]. Interestingly, the differentiation of Bcl-x-/- and wild-type ES cells in vitro yielded similar numbers of EryP and EryD cells, however, prominent apoptosis of Bcl-x-/- EryP and EryD occurred upon further maturation. This demonstrates that BCL-XL is critical at later stages of erythropoiesis. Accordingly, MMTVCre-Bcl-xfl/fl transgenic mice that express the CRE recombinase in various secretory tissues but also the haematopoietic system [95,96,97,98] develop fatal anaemia [99]. Notably, the loss of only a single allele of Bcl-x or point mutations that reduce BCL-XL protein half-life (Bcl-xPlt16/Plt16 and Bcl-xPlt20/Plt20 mice) cause a significant reduction in platelets [100]. Platelets specifically depend on BCL-XL to inhibit BAK-mediated apoptosis. When platelets are shed from megakaryocytes, they no longer produce much protein. Therefore, the relative levels of BCL-XL vs BAK set up a timer of platelet lifespan and consequently a reduction in BCL-XL (e.g., in platelets from Bcl-x+/- mice) reduces platelet survival [100]. Conditional deletion of Bcl-x in megakaryocytes (Pf4-Cre-Bcl-xfl/fl mice) demonstrated that BCL-XL is dispensable for the development and survival of these cells [101]. In fact, megakaryocyte survival is safeguarded by the combination of BCL-XL and MCL-1, with only the absence of both causing their depletion [102, 103].

MCL-1 (Table 3): A reduction in lymphocytes, red blood cells (RBCs) but not platelets was observed in Mcl-1+/- mice in which MCL-1 protein levels are reduced by ~40% compared to wild-type cells [104]. Conditional Mcl-1 gene-targeted mice were generated to identify the role of MCL-1 in the survival of different haematopoietic cell populations.

LckCre-Mcl-1fl/fl mice with MCL-1 loss from early stages of T lymphocyte development lack all immature and mature T cell populations, with their development arrested at the progenitor (CD3-4-8- triple-negative (TN)) stage [32]. CD19Cre-Mcl-1fl/fl mice with loss of MCL-1 in B cells present with an arrest of B lymphocyte differentiation at the pro-B cell stage [32]. Although these studies identify a critical role for MCL-1 in the survival of early T and B cell progenitors, they do not provide insight into possible roles of MCL-1 at later stages of T and B cell differentiation. Inducible deletion of MCL-1 using MxCre-Mcl-1fl/fl mice was able to reveal a critical role of MCL-1 for the survival of mature B and T cells in culture [105]. Moreover, MCL-1 is required for the survival of activated B cells and the formation of the germinal centre [106].

Ncr1Cre-Mcl-1fl/fl mice that lack MCL-1 in NK cells are completely deficient in this cell population [107], identifying an essential role of MCL-1 for NK cell survival. The characterisation of CD11c-Cre-Mcl-1fl/fl revealed a critical role of MCL-1 in the survival of conventional as well as plasmacytoid dendritic cells [108]. A prominent role for MCL-1 was also reported for erythroid progenitors. EpoR-Cre-Mcl-1fl/fl embryos die ~E13.5 due to severe anaemia [109]. Interestingly, MCL-1 is only required during early stages of definitive erythropoiesis but is dispensable for the survival of later stage erythroid progenitor cells that instead depend on BCL-XL [45, 99, 109].

MCL-1 has a less prominent role in myeloid cells and megakaryocytes. LysMCre-Mcl-1fl/fl mice have normal numbers of monocytes and macrophages but display an abnormal reduction in neutrophils [110, 111]. It is, however, noteworthy that the LysMCre transgene is not as effective at recombining floxed genes as other Cre transgenes; therefore the importance of MCL-1 in myeloid cell survival may have been underestimated. Examination of Pf4-Cre-Mcl-1fl/fl mice revealed that MCL-1 is dispensable for megakaryocyte development and survival as well as platelet lifespan [102, 103] with megakaryocyte survival safeguarded by both BCL-XL and MCL-1.

BCL-W: Moderate BCL-W expression was detected in most haematopoietic cell populations [21, 112] (Fig. 4). Nevertheless, BCL-W-deficient mice had normal distributions of all immature as well as mature haematopoietic cell types [78]. Enforced expression of BCL-W renders lymphoid and myeloid cells refractory to diverse cytotoxic conditions [112]. Therefore, it was proposed that BCL-W may play a role in the development of haematological cancers. However, initial reports [113, 114] that BCL-W is a driver of lymphomagenesis in various B cell lymphomas could not be reproduced [115].

A1: A1 is expressed in certain haematopoietic cell types but only minor defects are observed in A1-deficient mice, including a small reduction in TCRγ/δ T cells, antigen-experienced conventional as well as regulatory CD4+ T cells in vivo and impaired survival of conventional dendritic cells (cDCs) in culture [80]. The absence of A1 did not impair T cell responses during viral infection in mice [81]. A1 expression is induced by inflammatory cytokines, suggesting a role in the survival of inflammatory cells [116,117,118,119]. However, no defects in neutrophil survival during infection were observed in A1-deficient mice [120].

Targeting pro-survival BCL-2 family proteins with BH3 mimetic drugs for cancer therapy

In many human cancers apoptosis is dysregulated, often as a result of the overexpression of BCL-2, BCL-XL or MCL-1, and many cancers rely on their aberrant expression for sustained growth [121]. Accordingly, BH3-mimetic drugs that target distinct pro-survival proteins have been developed and the BCL-2 inhibitor Venetoclax is approved for CLL and AML. Given the essential roles of pro-survival BCL-2 proteins for the survival of many non-malignant cells, these compounds may cause undesirable on-target side effects to healthy tissues that may be predicted from the analysis of gene knock-out studies in mice (Table 4). All on-target toxicities observed in patients and mice treated with BH3 mimetic drugs targeting BCL-2, BCL-XL (e.g., thrombocytopenia) or MCL-1 (cardiac, intestinal and haematopoietic toxicity) were also seen in mice deficient for these proteins. Obversely, some of the damages to tissues caused by constitutive or cell type-restricted knockout of pro-survival BCL-2 proteins were not seen in patients or mice treated with BH3 mimetic drugs targeting these proteins. This is not surprising since in contrast to genetic deletion, drug-mediated inhibition is transient and likely does not result in a complete loss of function of a pro-survival BCL-2 protein. Therefore, it is expected that the on-target drug-induced toxicities are milder than the defects caused by constitutive loss of a pro-survival BCL-2 protein upon its genetic deletion. Here, we compare and discuss the consequences of genetic deletion vs drug-mediated inhibition of pro-survival BCL-2 proteins in major tissues.

Table 4 Impact of pro-survival gene targeting versus pro-survival protein inhibition using BH3 mimetic drugs.
Full size table

ABT-737 is the first BH3-mimetic compound described, followed by its orally available derivative ABT-263/Navitoclax that was tested in clinical trials [122, 123]. Both compounds are potent inhibitors of BCL-2, BCL-XL and BCL-W [124,125,126,127]. Despite high potency in killing several types of malignant cells in pre-clinical tests, clinical trials of Navitoclax in chronic lymphocytic leukaemia were halted because of platelet toxicity [123]. The thrombocytopenia and anaemia observed in Navitoclax-treated patients [122, 123] aligns with the essential role of BCL-XL in platelets and erythroid progenitor cells identified in gene-targeted mice [99, 100]. Accordingly, toxic effects on platelets were also observed in mice treated with BCL-XL-specific inhibitors [128,129,130,131]. This will likely complicate clinical development of such agents unless they can be appropriately scheduled to minimise dose limiting toxicity or modified to preferentially target cancerous cells, for example by coupling to an antibody that binds to malignant cells. Another promising approach is the PROTAC-based degradation of BCL-XL, targeting E3-ubiquitin ligases that are not present in platelets but highly expressed in cancerous cells [132,133,134]. Prolonged treatment with Navitoclax did not cause liver or kidney toxicities that are caused by genetic deletion of BCL-XL [135]. Of note, Navitoclax and ABT-737 were both shown to preferentially inhibit BCL-2 rather than BCL-XL in cells in vivo [136]. This may limit conclusions regarding the spectrum of toxicities related to specific pro-survival BCL-2 proteins from clinical studies of Navitoclax. As platelets express only low levels of BCL-2 and BCL-W, BCL-XL can be inferred to be the primary target of Navitoclax in these cells. In contrast, cells in the kidney express significant levels of BCL-2, BCL-XL and MCL-1 (Fig. 3), and thus BCL-XL might not be efficiently targeted by Navitoclax in this tissue. MCL-1 was shown to safeguard hepatocyte survival in the absence of BCL-XL in genetic knock-out studies [73], perhaps explaining why Navitoclax did not cause liver toxicity. Since no BCL-XL-specific inhibitor has entered clinical trials, it cannot be excluded that consequences of genetic loss of BCL-XL might actually occur in patients treated with such a drug.

To prevent dose-limiting thrombocytopenia in cancer patients, the BCL-2-specific inhibitor Venetoclax/ABT-199 was developed [137]. Treatment with Venetoclax is well-tolerated in patients, with no toxicity in major organs, such as the kidney or liver [138]. This agrees with observations from BCL-2-deficient mice, showing that even though BCL-2 is essential for kidney development in the embryo, it is dispensable for the function of major adult organs [40, 60]. In line with what was seen in BCL-2-deficient mice, many patients develop transient neutropenia (~40%) [139, 140]. While investigators have attributed thrombocytopenia and anaemia as adverse events to single agent Venetoclax, this occurred with a large proportion of enroled patients with relapsed/refractory CLL having cytopenia at baseline [139, 140]. As most haematologic adverse events occurred early in treatment and decreased or resolved over time, these observations are most likely secondary to the high burden of disease rather than a direct effect of Venetoclax. Importantly, the neutropenia in Venetoclax-treated patients could be managed with growth factor support [139, 140]. In contrast to the genetic deletion of BCL-2, treatment with Venetoclax does not cause neuronal toxicity in patients [138]. This may be explained by the poor blood-brain barrier penetration of Venetoclax, with drug concentration in the central nervous system only reaching 0.1% of that observed in the plasma [141].

Despite the essential role of MCL-1 for the survival of many critical cell types, pre-clinical studies in mice have identified a therapeutic window for MCL-1-specific inhibitors [142,143,144,145]. This is in line with observations that loss of even a single allele of Mcl-1 obliterates the expansion of MYC-driven lymphomas in mice, unless they carry a mutation in p53 [146], while loss of one allele of Mcl-1 is well-tolerated in mice with only minor reductions in mature B lymphocytes and erythroid cells [104]. Conclusions from pre-clinical observations prompted the initiation of phase 1 clinical trials with several MCL-1 inhibitors. Trials of AMG-397 (Amgen) and AZD5991 (AstraZeneca) were paused due to cardiac toxicity signals [147,148,149,150]. In contrast, phase 1 clinical dose escalation studies for S64315/MIK665 (Servier/Novartis) have been completed without major adverse events being reported. The cardiotoxicity of AMG-397 and AZD5991 is in line with the findings from genetic studies that MCL-1 is essential for cardiomyocyte survival in mice [52, 68]. MCL-1 is also the major survival factor for HSPCs, and accordingly, AMG-176 induced neutropenia and severe anaemia in patients, which may be caused by a reduction in HSPCs [151]. Consistent with this idea, human HSPCs were depleted during in vitro treatment with S64315/MIK665 [151]. MCL-1 loss results in intestinal epithelial cell apoptosis [74] and, accordingly, treatment with AMG-176 caused gastro-intestinal toxicity in patients. Pre-clinical data demonstrate that the combined inhibition of MCL-1 and BCL-XL causes liver toxicity [152], a scenario that could be predicted from gene-targeting studies, showing that MCL-1 and BCL-XL collectively ensure hepatocellular survival [65, 72, 73].

In conclusion, mouse models have been invaluable to identify the critical roles of the different pro-survival BCL-2 proteins for the survival of a broad range of non-transformed cell types during embryogenesis and in the adult. This highlights the importance of using genetic mouse models to form the foundation for developing effective and safe new treatments for cancer patients.