Cell cycle regulation in NAFLD: when imbalanced metabolism limits cell division

Resection of 2/3 of the liver activates an acute phase response followed by a profound rearrangement of the entire cellular signaling network [7]. Based on the consecutive activation of a diverse group of transcription factors (TFs), the process of liver regeneration is conceptually divided into three phases: priming, progression, and termination (Fig. 2). The initiation phase is characterized by a fast activation of early-response genes [2] with a dramatic increase in blood pressure [8]. Collectively, these initiating events have been associated with the release of IL-6 and activation of the IL-6-gp130-Stat3 axis in Kupffer cells [9]. IL-6 from Kupffer cells activates the IL-6 receptor gp130 in hepatocytes which in turn will phosphorylate STAT3, causing increases in transcriptional activity (Fig. 2). Early response TFs promote the activation of a second group known as late-response genes that undergo de novo transcription and translation [2]. Combined with other TFs activated by growth factors like hepatocyte and epidermal growth factors (EGF, HGF), the IL-6-gp130 axis activates expression of key cell cycle genes [10‐15].

Cell cycle genes drive the progression phase until the onset of the termination phase [16]. Within 1–2 weeks after resection, any remaining cellular division will eventually cease in murine livers. As such, two weeks after resection, the liver will adjust to the initial pre-resection size and function while in other mammals the process may take longer time. The final stage of regeneration is characterized by the termination phase. Here, the liver-to-body weight ratio is carefully adjusted. The ratio is regulated by the rate of hepatic division and apoptosis until the liver reaches an adequate size and all normal physiological functions are restored [17, 18].

Another distinctive feature of liver regeneration is the dynamic alterations in cellular hypertrophy and division [19]. Normally, after hepatocytes are primed by injury, they temporarily increase in size with a subsequent activation of cell cycle genes and cellular division [17]. However, we and others have shown that an increase in hepatocyte size alone is sufficient to recover organ mass in the absence of cell division in mice. For example deletion of the cell cycle genes Skp2 or Cdk1 [20‐23] does not impair regeneration. In addition, early observations indicated that components of the mTOR-AKT signaling are essential to activate an alternative regenerative response to promote tissue recovery by compensatory cellular hypertrophy [19, 24, 25]. Nevertheless, we also observed that compensatory cellular hypertrophy is sufficient for mass recovery, but liver functions are reduced [26]. On top of that, our recent unpublished data suggest that in the long-term, increases in the size of hepatocytes may induce activation of an inflammatory response and fibrosis. In conclusion, in the short term cell hypertrophy may offer an alternative form of tissue regeneration, but may not be sufficient for recovery of liver functions.

During liver regeneration, hepatocytes display an equivalent pattern of expression of cell cycle genes comparable to that observed in cell culture models in vitro (for an exhaustive review on cell cycle regulators refer to Refs. [27, 28] and Fig. 2). For example, before liver resection, most genes involved in cell division like Cdk1 and cyclin B are downregulated. Instead genes involved in quiescence like p21^Cip1/Waf1 and p27^Kip1 are highly expressed [12, 29‐31]. During the initiation phase of liver regeneration, the expression of genes involved in G1 progression like cyclin D and Cdk4 is induced [29]. Finally, during the progression phase of regeneration most cell cycle genes are actively expressed [32, 33] (Fig. 3). However, there are several exceptions to this rule. For example, the murine cyclin B transcript is regulated at different time points during liver regeneration, with a peak of expression that coincides with the peak of DNA replication (S-phase, ≈ 36 h after resection) [34, 35]. Additionally, in regenerating livers cyclin A and cyclin B are observed in the cytoplasmic, nuclear, and membrane containing fractions [36, 37]. In the same way, Cdk1 is expressed as early as 30 h after hepatectomy, which is the period of DNA replication in regenerating hepatocytes [38]. These observations suggest that Cdk1 may display novel functions outside of mitosis, which is supported by recent data [26, 38] but was not observed in vitro models. Collectively, these results suggest that Cdk1 and its partners cyclin A and cyclin B, may possess non-canonical functions in the liver, different to the ones described in cell culture systems. We and others have shown that during liver regeneration Cdk1 controls cell size and cell division [39], cell metabolism [26], and DNA replication [38, 40]. However, little data is available on the partners cyclin A and cyclin B. Hence, further studies may be essential to understand the roles of these cyclins during liver regeneration.

The liver is a vital center for homeostasis of glucose, lipid, and overall metabolism [41]. Therefore, acute tissue damage imposes significant stress to the general metabolic performance of the whole organ. Nevertheless, during liver regeneration after 2/3 liver resection, hepatocytes are able to remodel their metabolic network to promote regeneration while maintaining overall systemic metabolic balance. These early observations suggest that liver metabolism is robust. In this sense, metabolic pathways compensate for the loss of many individual enzymes, which is supported by mouse genetics [42‐45]. However, liver specific deletion of some enzymes including nicotinamide phosphoribosyl transferase (Nampt) and several members of the P450 cytochrome reductase family leads to lethality after partial hepatectomy [46‐48].

The changes in metabolic demands usually induce small rearrangements in metabolism through local mechanisms without substantial rewiring of the overall metabolic network [41]. However, 2/3 partial hepatectomy leads to both loss of tissue mass and glycogen, a major energy reservoir for systemic glucose and energy metabolism [49]. Major resection of glycogen in the liver, leads to a substantial reduction in blood glucose. This is particularly apparent during the initiation phase of liver regeneration when there is a significant increase in the oxidative status as a response to a whole body induction of catabolism [49]. Because glucose supply via glycogenolysis is limited, the energy required to promote liver regeneration may be partly acquired from alternative sources that are actively transported into hepatocyte, such as lipids [50]. In fact, deletion of the Caveolin1 (Cav1) gene induced hepatic failure during early phases of liver regeneration, indicating that lipid import into hepatocytes is essential to drive regeneration [51]. CAV1 is essential to transport lipids into hepatocytes. Therefore, these observations are in agreement with temporary hepatic steatosis being a key feature of the early stages of liver regeneration [52, 53]. Interestingly, physiological steatosis does not overlap with cell division during liver regeneration, which may imply that lipid metabolism and cell division are strictly temporally separated. However, these ideas need to be further tested.

During the termination phase of liver regeneration, hepatic gluconeogenic functions regain homeostatic levels, marked by a sharp increase in expression of Akt [44]. These events are followed by the activation of biosynthetic pathways to promote tissue homeostasis, accumulation of glycogen as well as normal glucose and lipid metabolic processes. Based on the main two primary sources of energy, we would like to propose a division of metabolic-related events during liver regeneration in two phases (1) oxidative and (2) biosynthetic (Fig. 4). The oxidation phase is distinguished by oxidative metabolism correlated to increases in cell size. This would last from the early minutes after liver resection to 30 h, when hepatic DNA synthesis starts to peak (Fig. 4). After that, the biosynthetic pathways will be active until metabolic homeostasis is reached.

Reduction in blood glucose represents one of the primary metabolic markers of hepatic regeneration [26]. The characteristic short-term hypoglycemia is required for adequate liver regeneration. This is supported by clinical data indicating that patients with hyperglycemia typically experience an adverse regenerative outcome [54]. Another cause for decreased concentration of glucose in serum is the reduced expression of metabolic enzymes during the priming and progression phases [17]. We have recently published similar findings in mice employing a combination of transcriptomic, metabolomic and advanced molecular imaging [26].

During cellular division CDK1 phosphorylates specific subunits of the mitochondrial oxidative phosphorylation complex I [55, 56]. In this sense, CDK1 may couple cell division and energy metabolism. In post-mitotic tissues, CDK1 regulates gluconeogenic pathways by phosphorylation of FOXO-1 [57] and glycolytic genes like PFKFB3 [58, 59]. Following this line of evidence, we published that ablation of CDK1 in hepatocytes led to a significant reduction in NADH concentration because mitochondrial function is suboptimal [60]. These observations are in agreement with recent reports suggesting that deletion of Nampt reduced hepatocyte proliferation during regeneration [47]. Collectively, these results suggest an interplay between the oxidative state of hepatocytes and CDK1. Deletion of Cdk1 in the liver led to a significant reduction in mitochondrial metabolic processes during S-phase [26]. A standard explanation for this is that DNA replication and oxidative metabolism are not compatible. Reactive oxygen species (ROS) generated as a side-product of oxidative metabolism would be detrimental when the maintenance of DNA fidelity is most critical [61, 62].

In the previous sections, we highlighted that hepatic accumulation of lipid droplets is associated with the initiation of regeneration and not cellular division. Therefore, we speculate that the presence of either lipid droplets or their components are incompatible with proliferation. A lipid droplet is composed by triglycerides, lipoproteins, cholesterol and other components [63]. Cholesterol represents the fundamental structural component and thus is essential for its formation. In mammals there are two main sources of cholesterol, (1) gastrointestinal digestion and (2) biosynthesis. Regardless of the source, cholesterol is stored and metabolized as esterified cholesterol in pericentral hepatocytes [64]. As mentioned earlier, during liver regeneration the metabolic capacity of hepatocytes is heavily reduced, and it is expected that cholesterol production may be hindered as well. In fact, our preliminary data show that concentration of serum total, free and esterified cholesterol are significantly reduced during liver regeneration (unpublished).

In mice, hepatic Srebp-1c is an essential gene for the transcriptional activation of enzymes involved in cholesterol metabolic processes. Deletion of Srebp-1c in hepatocytes led to increased accumulation of cholesterol [65], with no changes in the composition of other lipids. When mice without Srebp-1c were subjected to 2/3 partial hepatectomy, hepatocyte division was significantly enhanced in comparison to wild type mice [65]. This was in agreement with previous observations that showed reduced expression of CDK1 and less mitotic cells upon cholesterol starvation [66, 67]. Together, these studies further support the idea that there might be a relationship between cholesterol metabolism and cellular division. We have shown that reduced division rate of cells during liver regeneration, led to a significant reduction in expression of enzymes from cholesterol and mevalonate pathway, suggesting an interaction between the machinery that regulates cell cycle and that of cholesterol synthesis. This is further supported by experiments indicating that during cell cycle progression, cells regulate their lipid composition and localization [68]. We are currently dissecting the molecular mechanism by which CDK1 may regulate cholesterol synthesis in hepatocytes. Forthcoming work should address the association between the formation of lipid droplets during physiological steatosis and cellular division during liver regeneration independently.

NAFLD evolves from the response of altered (1) lipid transport into hepatocytes; (2) increased production of lipids from hepatocytes; as well as (3) reduced export and consumption of lipids [63]. In diabetes and obesity, NAFLD is accompanied by changes in insulin metabolism [5]. In hepatocytes, insulin promotes the biosynthesis of lipids by inhibiting fatty acid oxidation, with leptin being essential for the insulin response [73]. Murine models with mutations in either leptin or the leptin receptor gene led to body fat accumulation, and the onset of obesity and diabetes mellitus. Initial observations using these mouse models demonstrated impaired cell cycle progression of hepatocytes after liver regeneration. For example, hepatocytes from Zucker rats arrested in G1, with a concomitant lower expression of cyclin D1 after hepatectomy [74]. These initial results led to the hypothesis that accumulation of fat droplets may negatively regulate hepatic cell division and reduce cell proliferation. Besides, in ob/ob mice survival rate after liver resection was significantly also lower than in wild type. These mice do not produce leptin, and displayed significant hypoglycemia as well as impaired hepatocellular proliferation after liver resection [75]. These effects occur with no difference in serum concentration of pro-proliferative hormones during liver regeneration like TNFα, IL-6, and insulin. In a third model, fa/fa rats displayed elevated concentration of leptin in blood and a similar reduction in the survival after hepatectomy [76]. Nonetheless, leptin replacement restored TNFα and IL-6 release and induced cyclin D1, suggesting that leptin may play a central role in the interaction between cell cycle regulators and lipid metabolism [77]. A conflicting result arises from the fact that leptin supplementation improved hepatic division, but did not reduce the onset of liver failure [78, 79]. In addition, intraperitoneal injection of leptin in wild type mice with no obesity increased mitotic counts during liver regeneration [78]. Collectively, these results suggest that adipokines play a significant role in hepatocellular proliferation. Similar results were observed in a model of fatty liver in ALPPS rats [80]. In summary, leptin is essential for hepatic division and together with other factors may act in recovery after hepatectomy [81]. Subsequent research should address the role of adipokines like leptin in regulation of hepatic cell cycle genes.

To promote lipid overloading independently of a genetic mutation, many experiments provide mice with a high-fat diet (HFD). These models are more clinically-relevant compared to mice with mutations in enzymes. HFD leads to high calorie intake, body weight increase, liver steatosis and reduced regeneration caused by metabolic surplus [82, 83]. Liver regeneration in mice fed with HFD was accompanied by increased TNF-α potentially amplified by leptin which promoted increased levels of IκBα preventing NF-κB activation. Consequently, expression of NF-κB target genes, including cyclin D1 was reduced [82] (Fig. 5) leading to a decrease in proliferation of hepatocytes. Mice fed with HFD display a delayed incorporation of BrdU after surgery [84], indicating that steatosis not only inhibits cell division but also delays DNA replication. Expression of cyclin D1 and cyclin E2 are usually higher before partial hepatectomy in mice fed with HFD. However, there was no significant difference in expression between mice fed with HFD and Chow diets during liver regeneration [84]. In contrast, cyclin A2 and cyclin B1 are significantly reduced during liver regeneration in mice fed with HFD, further suggesting that cell division might be impaired [82]. In conclusion, these observations suggest that leptin-related steatosis due to HFD may regulate cell cycle progression differentially by downstream control of the inflammatory pathway NF-κB, which is supported by studies done in mice with lipodystrophy [85]. Lipid synthesis and degradation, as well as the correct localization of the corresponding enzymes are essential for cell division [68]. To the same extent, cell cycle effectors regulate directly or indirectly lipid synthesis in hepatocytes. To follow up on these topics, in the next section we will summarize current data on how cell cycle regulators control lipid synthesis. We will particularly focus on a novel axis of molecular signaling that we describe as the p21-CDK1-E2F1-lipid synthesis axis.

Recent findings have indicated that NAFLD livers display lower expression of Cdk1 [88]. CDK1 is essential for mitotic division, but in hepatic cells CDK1 may have other functions beyond mitosis. For example, we have shown that deletion of Cdk1 in hepatocytes leads to increases in nuclear and cell size [21]. DNA re-replication by endoreduplication may be the underlying cause of this effect. During endoreduplication, hepatic cells may generate oxidative stress leading to chronic liver damage [88]. In this setting, p21^Cip1/Waf1 negatively regulates CDK1, and therefore increased levels of p21^Cip1/Waf1 in liver disease may correlate with reduced activity of CDK1 [88]. Therefore, it may be therapeutically feasible to reduce the expression of p21^Cip1/Waf1 in hepatocytes to reduce NAFLD. However, the pleiotropic functions of p21^Cip1/Waf1 are likely to make it an “undruggable” target. Interestingly, it was recently shown that senotherapeutic drugs improved liver regeneration when p21 was overexpressed further suggesting that SASP may inhibit liver regeneration [89].

An important function for p21^Cip1/Waf1 is the transcriptional regulation of cell cycle genes. Typically, p21^Cip1/Waf1 binds with high affinity to and inhibits CDK2/cyclin E activity [39] which leads to hypo-phosphorylation of the RB protein [90]. Hypophosphorylated RB remains bound to E2F1, repressing its activation and cell cycle progression. Recent results highlighted a number of novel metabolic functions of E2F1 [91]. For example, E2F1 mRNA and protein are increased in NASH livers [92]. On top of that, hepatic deletion of E2f1 significantly decreased fibrosis as a consequence of downregulation of the Egr-1 gene [93, 94]. Moreover, E2F1 regulates the expression of essential genes in lipid synthesis such as Acaca, Fasn, Scd1, Srebp1c, and Chrebp [92]. The interaction of p21^Cip1/Waf1 with CDK1 may lead to inactivation of E2F1. E2F1 also drives the expression of Cdk1 [94]. In this fashion, p21^Cip1/Waf1 regulates cell cycle progression (Fig. 5), cellular division, senescence and metabolism through the expression of numerous genes, including lipogenic enzymes {for review see [95]}.

Finally, the transcription of p21^Cip1/Waf1 is activated by p53. In response to DNA damage, p53 promotes the expression of p21^Cip1/Waf1, leading to cell cycle arrest and senescence [22, 96, 97]. Simultaneously, p53 regulates the transcription of many genes involved in glycolysis, lipogenesis and other metabolic pathways [98]. Particularly interesting is the discovery of frequent SNPs variants in p53 that predispose patients to develop obesity and NAFLD [99] (Fig. 5). When these variants were introduced in mice fed with HFD, they caused significant increases in body weight with liver steatosis, inflammation, and fibrosis [100]. These mice also displayed a remarkable increase in the expression of p21^Cip1/Waf1 and inflammatory genes. Moreover, under fasting conditions, hepatic cells accumulate lipid droplets which correlate with increased expression of both p53 [99, 100] and p21 [86, 87]. Collectively, these observations suggest that p53 may fulfill essential roles upstream of p21 ^Cip1/Waf1 and Cdk1 in NAFLD.