Perinatal deletion of hepatic Rps6 stunts growth and induces severe neonatal liver hypoplasia

To delete hepatic Rps6 prior to birth, mice harboring a conditional Rps6^lox/lox allele [34] were bred to mice expressing Albumin-Cre (Alb-Cre) which specifies gene targeting in hepatoblasts of the embryonic liver from ~E15 onwards [35,36]. Progeny with all of the genotypes from matings of Rps6^lox/wt mice to Rps6^lox/wt:Alb-Cre mice were born at the expected Mendelian ratio indicating that targeted deletion of hepatic Rps6 prior to birth in Rp6^lox/lox:Alb-Cre mice (herein referred to as ΔS6 mice) did not cause embryonic lethality. However, monitoring of post-natal body weight showed that ΔS6 mice were significantly smaller than their wild-type (WT, Cre-) and S6^lox/wt;Alb-Cre littermates between ~2–6 weeks of age (Figs 1A, S1A, and S1B) and had livers that were disproportionately small in relation to body size (S1C and S1D Fig). Analysis of the targeted Rps6^lox/lox allele (ΔS6^del) in liver showed that recombination was low at birth, but increased to maximal levels of ~50–60% by ~2 weeks of age, coinciding with the onset of growth retardation and liver hypoplasia (S2A and S2B Fig). Northern blotting (S2C and S2D Fig) and immunohistochemistry of ΔS6 livers at post-natal day 15 (P15) (S2E Fig) confirmed that Rps6 deletion was incomplete, regional and varied between mice with livers expressing 30–50% of the normal amount of Rps6 mRNA. Despite being runted as neonates, ΔS6 body weight gradually increased suggesting that hepatic Rps6 deficiency was delaying, rather than permanently stunting growth (S1A and S1B Fig). Given the hypoplastic nature of ΔS6 livers and the liver’s ability to re-establish functional tissue mass through regeneration, we asked if the catch-up in body weight reflected improved liver function by performing liver function tests (LFTs) on plasma collected from WT and ΔS6 mice at 4–5 weeks of age when ΔS6 mice were at their most runted, and again at 8–9 weeks of age as they neared normal weight. Analysis showed that while biliary function (Alk-Phos and T-Bil) had improved within this period, markers of hepatocellular damage (ALT and AST) remained elevated (S3 Fig) indicating that livers of ΔS6 mice remained functionally compromised well into adulthood.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Perinatal deletion of hepatic Rps6 retards growth and results in cholestatic liver disease.

(A) A WT and runted ΔS6 mouse at postnatal day 28 showing that hepatic Rps6- deficiency retards neonatal growth. (B) Gross appearance of livers from a WT and a ΔS6 mouse at post-natal day 7 (P7). Severe yellowing of the ΔS6 liver is indicative of cholestatic disease. (C) Photomicrographs of H&E stained sections of liver from WT mice at P3, P7 and P15 (a, d and g) and age-matched ΔS6 mice (b-i). Canalicular accumulation of bile (yellow deposits, open arrowheads) is evident in ΔS6 livers at P3 and P7 and feathery degeneration (fd) of hepatocytes is evident at P7. At P15, bile infarcts (_*) resulting from bile leakage due to canalicular or hepatocyte membrane rupture can be seen throughout the parenchyma of ΔS6 livers (Original magnifications, a, b, d, e, g, h (x 125; 50μ scale bars); c, f and i (x 375; 25μ scale bars). (D) Photomicrographs of H&E stained sections of liver from a WT mouse (a and b) and a ΔS6 littermate (c) at 4 weeks of age. In contrast to the WT liver (a and b) which shows an abundance of bile ducts (arrowheads), ΔS6 livers appear to either lack or have a paucity of bile ducts (c) while nodules (n) and areas of biliary hyperplasia (*) are prominent indicating that Rps6 insufficiency has severely disrupted liver architecture. (Original magnifications, a and c (x 32.5); b (x 125)). Scale bars; all 50μ.

https://doi.org/10.1371/journal.pgen.1010595.g001

Hepatic Rps6 deficiency inhibits bile duct development and induces cholestatic liver injury provoking regeneration

The perinatal/neonatal period constitutes the most dynamic period for the liver during which bile duct development is nearing completion at a time when hepatic mass is increasing at its fastest rate while also responding to the dramatic metabolic adaptations that occur at birth and at weaning [37,38]. Given that hepatoblast-specific ablation of Rps6 has the potential to impact hepatocytes and biliary cells, we analyzed ΔS6 livers from the perinatal period to ~10 weeks of age to determine how loss of Rps6 impacted the liver at this crucial time as it transitioned from an immature to fully mature functional organ. Gross inspection of ΔS6 livers revealed marked yellowing by P7 (Fig 1B) consistent with jaundice. Histological evaluation of H&E stained sections of ΔS6 livers showed evidence of canalicular bile accumulation at P3 (panel c of Fig 1C) and cholate stasis at P7 as determined by the presence of widespread feathery degeneration of hepatocytes, a form of non-apoptotic inflammatory-induced cholestatic death characterized by hepatocyte ballooning and flocculent cytoplasmic inclusions, akin to necroptosis [39] (panels e and f of Fig 1C). Foci of dead hepatocytes or bile infarcts were also evident at P15 (panels h and i of Fig 1C), consistent with toxic bile acid-mediated degeneration leading to confluent hepatocyte necrosis. Examination of H&E stained liver sections from young adult mice also indicated that liver architecture was dramatically altered in ΔS6 livers and that they appeared to have fewer bile ducts than their WT counterparts (Fig 1D). To determine if Rps6-insufficiency was interfering with bile duct development or causing the loss of pre-existing bile ducts, we performed IHC of WT and ΔS6 livers from E17, shortly after establishment of the ductal plate, to P15 when biliary development and morphogenesis is complete [40]. Using an antibody specific for Sox9, a transcription factor that marks immature biliary cells and mature bile ducts but not mature hepatocytes, we found that ΔS6 livers had fewer Sox9-positive cells surrounding the portal vein than WT livers by E17 (panels a and b of Fig 2A). Moreover, of the few Sox9-positive cells that were visible, none became incorporated into structures resembling bile ducts as biliary development progressed (panels c-j of Fig 2A), a finding that was confirmed by performing IHC with a pan-cytokeratin (CK) antibody that preferentially recognizes biliary-type CKs (CK7/19) (Fig 2B) Quantitative analysis confirmed that ΔS6 livers had only 25% and 21% of the normal number of Sox9-positive cells at E17 and P1 respectively (Fig 2C) and fewer still at P3 (panel f of Fig 2A), and had either 1 or no bile ducts per portal vein (bds/PV) (mean, < 0.5 bds/PV) in contrast to livers of WT mice which had 1–3 (mean, 1.4 bds/PV) (Fig 2D). Finally, while Sox9- and pan-CK-positive cells that had not been incorporated into bile ducts were rare or absent in WT livers at P7 or P15 (panels g and i of Fig 2A and panel e of Fig 2B), Sox9-expressing cells had begun to proliferate around and radiate out from portal veins in ΔS6 livers (panels h and j of Fig 2A and panel f of Fig 2B) in a pattern that suggested emergence of a nascent ductular reaction (dr) (panels h and j of Fig 2A), which is a hallmark of hepatic progenitor cell (HPC) activation in response to extreme hepatic injury [41].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Hepatic Rps6-deficiency inhibits bile duct development.

(A) Photomicrographs of SOX9 IHC of WT (a, c, e, g, i) and ΔS6 (b, d, f, h, j) livers from E17 to P15 (Original magnifications, all x 125; 50μ scale bars) (AEC chromogen (red/orange), no counterstain). Note the gradual disappearance of SOX9-positive cells from the ductal plate in ΔS6 livers between E17-P3 and the abnormal position and expansion of SOX9-positive cells throughout the parenchyma from P7 onwards. (B) Photomicrographs of pan-CK IHC of WT (a, c, e) and ΔS6 (b, d, f) livers from P1 to P7 showing that the number of pan-CK-positive biliary cells in ΔS6 livers is also reduced at P1 and P3 (b and d) while their expansion at P7 (f) mirrors the increase in the number of SOX9-positive cells (A (h)) signifying emergence of a nascent ductular reaction. (Original magnifications, all x 250; 50μ scale bars) (AEC chromagen (red/orange), no counterstain). (C) Graph depicting quantitative analysis of the number of SOX9-positive cells around portal veins (PVs) in WT and ΔS6 livers at E17 and P1. ΔS6 livers have 20–25% of the normal number of SOX9-positive cells (mean values 7.8 vs 30.7 at E17; P < .0001 and 9.8 vs 47.5 at P1; P < .0001); 2-tailed unpaired Student’s t-test. (D) Graph showing average number of bile ducts per portal vein (PV) (bds/PV) in WT and ΔS6 livers at P1, P3, P7 and P15. While WT livers have an average of 1–2 fully formed bds/PV, ΔS6 livers have an average of < 0.5. Data are mean ± SEM; **** P < .0001; 2-tailed unpaired Student’s t-test.

https://doi.org/10.1371/journal.pgen.1010595.g002

Examination of ΔS6 livers as mice reached adulthood showed that despite the gradual normalization of body weight, livers remained small and were discolored and uneven with a “cystic-like” appearance (Fig 3A and 3B). Histological evaluation of livers at P28 revealed that in contrast to the typical, well ordered lobular architecture of WT liver (Fig 3C), ΔS6 livers contained an abundance of regenerative nodules comprised of cells that resembled immature hepatocytes interspersed by cords of small oval-shaped cells that were attempting to organize into ducts consistent with induction of a full-blown dr (Fig 3D, 3E and 3F) which can facilitate secondary bile duct development [42]. IHC profiling of ΔS6 livers using a panel of antibodies directed against proliferating cell nuclear antigen (PCNA) and a variety of liver cell markers that are selectively expressed in mature and immature liver cells showed that nodules contained highly proliferative immature hepatocytes (S4F Fig) that expressed abundant AFP (S4H Fig) and HNF4α (S4J Fig). This was in contrast to the majority of ductular cells which expressed a variety of biliary (HNF1β, SOX9, (pan)-cytokeratin(CK)) (S4L, S4N and S4P Fig) and HPC (TROP2 and EPCAM) (S4R and S4T Fig) markers, but not HNF4α or AFP. IHC for β-catenin, whose level of expression and sub-cellular localization is an indicator of wnt signaling status, showed that although HPCs and immature nodular hepatocytes both expressed β-catenin, expression was significantly enriched in the ductular cell population compared to nodular hepatocytes (S4V Fig). Moreover, retention of β-catenin at the membrane of nodular hepatocytes, rather than cytoplasmic or nuclear localization (S4V Fig) and the absence of glutamine synthetase expression (GLUL), a classical wnt target within nodules (S4X Fig), indicated that hyperactive wnt signaling was not driving nodular growth in ΔS6 livers. Evaluation of ΔS6 livers at ~8–10 weeks of age showed a marked reduction in the dr and restoration of normal liver architecture (Fig 3G). However morphological and nuclear heterogeneity within hepatocytes persisted indicating that ΔS6 livers remained compromised even after regenerating (Fig 3H). Taken together, these results show that hepatoblast-specific deletion of Rps6 profoundly disrupts perinatal liver development by compromising hepatocyte survival and interfering with bile duct development by limiting the pool of Sox9-expressing biliary precursors within the ductal plate, all of which leads to a ribosomopathy-like phenotype characterized by hepatic hypoplasia, sub-lethal hepatic failure and regeneration.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Perinatal deletion of hepatic Rps6 results in hypoplastic livers and triggers regeneration.

(A) Gross appearance of livers from a WT and ΔS6 mouse at postnatal day 28 (P28). The ΔS6 liver is smaller than the WT liver, is discolored and has an uneven mottled appearance. (B) Close-up image of ΔS6 liver in a) highlighting “cystic-like” nodules on the surface of the liver. (C-H) Photomicrographs of H&E stained sections of liver from a WT mouse (C) and ΔS6 mice at P28 (D-F) and P65 (G and H). At P28, ΔS6 livers display an abundance of regenerative nodules (n) and a prominent ductular reaction (dr) signifying dynamic regeneration in response to severe injury. By P65, the absence of regenerative nodules indicates that the regenerative response has largely dissipated. However, hepatocyte morphology is heterogeneous and remnants of the dr persist as seen by the presence of irregular luminal structures in the vicinity of portal veins (g, h). (Original magnifications, C (x 112); D, G (x 32); E, H (x 125) and F (x 250)). Scale bars correspond to 50μ for C, E, F and H and 200μ for D and G.

https://doi.org/10.1371/journal.pgen.1010595.g003

Hepatic dysfunction in ΔS6 livers reflects loss of Rps6 in hepatocytes and biliary cells

In addition to being a constituent of the 40S ribosomal subunit, Rps6 is unique among the rps in that it is also a phospho-protein whose phosphorylation is regulated by the rapamycin-sensitive branch of the mTOR/S6K signaling (mTORC1) pathway, a conserved nutrient and energy-sensing pathway that positively regulates metabolism, growth, proliferation and survival [43]. Given that Rps6 phosphorylation has been shown to be required for a diverse array of physiological and pathological functions [44], we were interested in establishing the normal distribution of Rps6 in WT liver and assessing its phosphorylation status in any Rps6-expressing cell populations to determine if the loss of hepatocyte viability or the bile duct defect simply reflected cell-autonomous loss of Rps6 expression in each cell-type or whether they also involved the loss of additional phosphorylation-dependent functions. We therefore performed IHC with an antibody that recognizes Rps6 irrespective of its phosphorylation status (total Rps6) and another that recognizes Rps6 only when phosphorylated on the mTOR-dependent sites Ser^235/236 (phospho-Rps6). IHC with the total and phospho-specific Rps6 antibodies showed that normal hepatocytes expressed cytosolic Rps6 protein in a decreasing periportal (pp)-to-central vein (cv) gradient (panel a of S5A Fig), with the phospho-Rps6 antibody predominantly staining periportal hepatocytes consistent with the highest degree of phosphorylation being in periportal zone 1 (panel b of S5A Fig). This contrasted with bile ducts which, despite expressing abundant Rps6 (panel c of S5A Fig), failed to demonstrate any phospho-specific Rps6 staining (panel d of S5A Fig). This finding indicates that although Rps6 is expressed in both functional compartments of the liver, both cell types differ with respect to their Rps6 phosphorylation/mTOR activation status at least on Ser^235/6. Thus, while loss of hepatocyte viability and the cholangiopathy in ΔS6 livers likely reflects the cell-autonomous impact of Rps6 loss on each cell type, it is possible that the loss of phosphorylation-specific functions of Rps6 also contribute to the hepatocyte, but not the biliary defect in ΔS6 livers.

Regeneration in ΔS6 livers is mediated via the proliferation of 2 different Rps6-expressing cell types: HPCs and mTOR-activated immature hepatocytes

In light of incomplete Alb-Cre-mediated deletion of Rps6 during the perinatal period (S2 Fig), we performed IHC with the same two Rps6-specific antibodies to determine if ΔS6 livers were regenerating via the proliferation of residual Rps6-expressing cells or other cell types, and if so, whether Rps6 was phosphorylated (and thus mTOR activated) in such cells. IHC with the total-Rps6 antibody showed that regenerating ΔS6 livers contained two different Rps6-expressing cell populations; namely Sox9-positive dr/HPC cells and AFP-positive nodular hepatocytes (panel e of S5A Fig). However, only nodular hepatocytes demonstrated phospho-Rps6 immunoreactivity indicating that both cell types differed with respect to their Rps6^235/6 phosphorylation status (panel f of S5A Fig). Immunoblotting confirmed that another mTOR target, eIF4E binding protein 1 (4E-BP1), was also hyper-phosphorylated in ΔS6 livers (S5B Fig), while Akt, an effector of PI3K signaling that lies upstream of mTOR failed to show any significant increase in phosphorylation above basal levels, indicating that mTOR had been activated in regenerating ΔS6 livers and that its activation was independent of PI3K/Akt. To definitively show that Rps6 phosphorylation in nodular hepatocytes reflected mTOR activation, we treated a cadre of ΔS6 mice with the mTOR inhibitor rapamycin and assessed total and phospho-Rps6 expression by IHC (S5C Fig) and immunoblotting (S5D Fig). As expected, livers of vehicle-treated ΔS6 mice retained strong staining of nodular hepatocytes and dr cells with the total Rps6 antibody reflecting Rps6 expression in both cell types irrespective of its phosphorylation status (panels a, c, and e of S5C Fig). However, rapamycin treatment of ΔS6 mice abolished both total and phospho-Rps6 staining in all but a few solitary parenchymal cells, confirming that all of the Rps6 in nodular hepatocytes was phosphorylated in an mTORC1-dependent manner (panels b, d, and f of S5C Fig). This was in direct contrast to the Rps6-expressing ductular cells whose staining was unaffected by rapamycin, consistent with the absence of a Ser^235/236-phosphorylated form of Rps6 in these cells (panel fof S5A Fig). Thus, regeneration in ΔS6 livers is mediated by 2 different Rps6-expressing cell types; HPCs without mTOR activation and mTOR-activated, AFP-expressing immature hepatocytes suggesting that Rps6-expression confers a survival and/or proliferative advantage in the context of hepatic Rps6 deficiency.

Hepatic Rps6-deficiency activates p53, disrupts rRNA processing and activates a transcriptional program indicative of de-differentiation/regeneration, cell cycle arrest/senescence and inflammation

In light of previous reports documenting p53 stabilization and activation of the p53-dependent checkpoint in other mouse models of Rps6 deficiency [34,45,46], we analyzed the p53 status of ΔS6 livers. Immunoblotting of liver lysates from WT and ΔS6 livers with a p53-specific antibody showed that it recognized 2 proteins; a faster migrating, non-specific protein that was present in all samples and a slower migrating protein representing bona fide p53 that was abundant in ΔS6 livers and in A431 cells that express high levels of mutant p53, but which was absent from WT livers and p53-deficient Saos2 cells confirming that p53 had been stabilized in ΔS6 livers (S6A Fig). To establish the spatio-temporal pattern of p53 stabilization following Alb-Cre-mediated deletion of Rps6, we performed IHC with the same p53-specific antibody on sections of WT and ΔS6 livers at E17 and P7 after validating it on livers that express SV40 large T-Antigen (TAg) [47], an oncoprotein that binds and stabilizes p53 in the nucleus (panel a of S6B Fig). While the paucity of biliary cells in ΔS6 livers precluded us from determining if p53 had been stabilized in cholangiocytes, IHC showed abundant nuclear p53 in a subset of hepatoblasts in ΔS6 livers at E17 (panels d and f of S6B Fig) and in the majority of hepatocytes undergoing feathery degeneration and in neighboring hepatocytes at P7 (panels e and g of S6B Fig) indicating that p53 had undergone rapid stabilization following Alb-Cre-mediated deletion of Rps6.

As defective rRNA processing is a hallmark of ribosomal stress we also assessed the relative abundance of rRNA intermediates in RNA isolated from WT and ΔS6 liver by Northern Blotting using rRNA-specific radiolabeled probes homologous to regions within intervening sequence 1 and 2 (ITS1 and ITS2) of the 47S rRNA. Hybridization with the ITS1-specific probe showed accumulation of 30S rRNA and a corresponding decrease in the abundance of 21S rRNA in ΔS6 livers (S6C Fig) while hybridization with the ITS2-specific probe showed a reduced abundance of 17S rRNA (S6C Fig) confirming that loss of Rps6 had disrupted rRNA processing. Re-probing of both blots with an 18S rRNA-specific probe showed that the rRNA processing defect was not sufficient to diminish steady state levels of 18S rRNA.

Finally, to obtain an unbiased overview of the transcriptional changes that occurred in response to hepatic Rps6 deficiency, we performed gene expression microarray analysis using mRNA isolated from livers of 5 week old WT and ΔS6 mice which showed that loss of Rps6 perturbed the expression of a large number of mRNAs (S7A Fig). Applying a ≥ 8-fold up or down cut-off for differential mRNA expression, we determined that Rps6-deficiency resulted in the differential expression of 235 mRNAs, of which 184 were upregulated and 52 were downregulated (S1 Table). While the most highly expressed mRNAs in ΔS6 livers included oncofetal and imprinted genes (eg. H19, Bex1 and Igf2) or mRNAs that are known to be expressed in immature hepatocytes or liver progenitor cells (eg. Afp, Nope, Cd24a, Tacstd1/Epcam, Sox9 and Krt19), mRNAs that were downregulated included mature liver genes such as cytochrome P450s and major urinary proteins (Mups), consistent with a shift to a more de-differentiated, immature liver. Classical p53 targets including the cell cycle inhibitor p21/Cdkn1a, Sox4, a protein required for stabilization of p53 during checkpoint activation, and Noxa and Peg3, which mediate p53-dependent cell cycle arrest and apoptosis respectively, were also among the upregulated mRNAs. However, an additional group of mRNAs signifying activation of innate immunity and the senescence-associated secretory phenotype or SASP that included chemokines (Cxcl13, Cxcl14), damage associated molecular patterns or DAMPs (S100a8, S100a9, S100a6, S100a11 and S100a14) and NF-κB-associated genes (Dmbt1, Tff3, Muc1 and Sprr2a) was also upregulated in ΔS6 livers. Bioinformatics analysis using Ingenuity Pathway Analysis software identified signaling networks associated with cell cycle arrest/senescence (p21/Cdkn1a, MAPK), regeneration (Jun/Spp1) and inflammation/innate immune system activation (NF-κB) as the most prominent networks associated with altered gene expression in ΔS6 livers (S7B Fig). These results confirm that loss of hepatic Rps6 activates p53 and disrupts rRNA processing and that regeneration in ΔS6 livers reflects an attempt to re-establish homeostasis in the face of cell cycle arrest/senescence, injury and inflammation.

Rps6 is required for hepatocyte survival in adult liver

Our results showing that loss of Rps6 in normal non-regenerating hepatocytes compromised cell viability contrasts with previous reports of Rps6-deficiency blocking hepatocyte proliferation post partial-hepatectomy [34] suggesting that the fate of Rps6-deficient hepatocytes is context- or stimulus-dependent. Moreover, because hepatocytes can be irreversibly damaged by the detergent action of hydrophobic unconjugated bile acids that accumulate in cholestatic disease [48], it was important to determine if the robust hepatocyte death that occurred in neonatal ΔS6 livers reflected an intrinsic dependency on Rps6 for survival, bile-acid induced hepatotoxicity, or a combination of both. To eliminate bile-acid mediated hepatotoxicity as a key mediator of hepatocyte death in ΔS6 livers, we adopted a strategy that allowed us to specifically delete Rps6 in adult hepatocytes while sparing the neonatal period and bypassing biliary development. This was done by breeding Rps6^lox/lox mice to a line of bi-genic mice that co-expresses a modified reverse tetracycline transactivator (rtTA_M2) [49] driven by the liver-specific regulatory elements of the ApoE gene [50] (ApoE-rtTA_M2) and Cre recombinase under control of the minimal CMV promoter and tetracycline responsive elements (S8 Fig) such that Cre could be activated in hepatocytes by providing mice with doxycycline (dox). Empirical determination of the minimal dose of dox required to achieve efficient hepatocyte-specific recombination of the Rps6^lox/lox allele (ΔS6^del) showed that inclusion of dox at a concentration of 200–250 μg/ml in drinking water beginning at 5–6 weeks of age was sufficient to achieve 80–90% recombination in as little as 3–7 days, a significantly higher degree of recombination than had been achieved with Albumin-Cre, although recombination decreased over time (Fig 4A and 4B). Analysis of livers from WT and Rps6^lox/lox:ApoE-rtTA:TRE2-Cre mice (herein referred to as ΔS6^dox mice) provided with drinking water without dox or supplemented with 200–250μg/ml dox for various lengths of time showed that, without dox, livers of ΔS6^dox mice were mildly discolored, but otherwise normal despite exhibiting a low level of “leaky” (dox-independent) recombination (Fig 4A and 4B). This was in contrast to livers of ΔS6^dox mice, which began to show evidence of cellular stress after just 7 days of dox treatment. While bile ducts were normal, hepatocytes appeared crowded due to the narrowing of sinusoidal spaces with some hepatocytes also appearing to lack nuclei suggestive of apoptotic cell death (panel b of Fig 4C). TUNEL analysis of livers from ΔS6^dox mice provided with dox for different lengths of time confirmed that hepatocytes were indeed apoptotic with death peaking ~1 week post-dox initiation and continuing at a lower level thereafter (panel d of Fig 4C and Fig 4D). Despite the low level of dox-independent recombination, livers of 8–9 week old ΔS6^dox mice without dox remained normal (panel a of Fig 4E). This was in sharp contrast to livers of ΔS6^dox mice provided with dox which showed extensive hepatocyte vacuolization and focal drop-out after ~5 weeks (panel b of Fig 4E), which triggered a full-blown regenerative response involving nodular growth (panel c of Fig 4E) and emergence of a dr (panel d of Fig 4E) that often co-existed with the presence of ballooning hepatocytes and piecemeal necrosis consistent with ongoing hepatocyte death (panel e of Fig 4E). Finally, to determine if the progressive decline in recombination of the ΔS6^del allele reflected repopulation of ΔS6^dox livers with Rps6-expressing cells to replace dying hepatocytes, we performed IHC with PCNA-, phospho^Ser235/6-Rps6- and pan-CK-specific antibodies. Analysis showed that, like ΔS6 livers, regenerating ΔS6^dox livers also contained nodules comprised of PCNA-positive and phospho^Ser235-236-RpS6-positive hepatocytes and a dr composed of phospho-Rps6-negative oval cells (panels a, b, and c of Fig 4F). While these results show that neonatal and adult hepatocytes both require Rps6 for survival, our results are consistent with the idea that the rapid and near-catastrophic hepatocyte death in neonatal ΔS6 livers reflects Alb-Cre- mediated depletion of Rps6 from both functional compartments of the liver which compounds disease due to exacerbation of an intrinsic susceptibility to death caused by exposure to toxic bile acids resulting from a failure to complete bile duct development.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Deletion of Rps6 in adult liver induces hepatocyte apoptosis and triggers regeneration.

(A) Ethidium-stained gel showing recombination of the ΔS6 allele in livers of ΔS6^dox mice with and without dox (250μg/ml) for different lengths of time. (B) Graph of quantitation of % ΔS6 recombination from gel shown in A). While there is a low level of recombination (~18%) in the absence of dox (lane 1), it increases to ~85% when mice are provided with 250μg/ml dox representing loss of Rps6 in a majority of hepatocytes, the most abundant cell type in the liver (lanes 3–8). After ~3 weeks, the extent of recombination begins to decrease reflecting the loss of Rps6-deficient hepatocytes and their replacement with Rps6-expressing cells (lanes 9–11) (see F below). Note that the degree of recombination achieved in livers of Rps6^lox/lox:ApoE-rtTA:TRE-Cre mice (provided with dox) before widespread hepatocyte loss occurs (lanes 3–8) is consistently greater than that achieved in livers of Rps6^lox/lox:Alb-Cre mice (lane 12). (C) Photomicrographs of H&E staining (a and b) or TUNEL staining (c and d) on sections of ΔS6^dox livers provided with water (a, c) or water containing 250μg/ml dox (b, d) for 7 days (Original magnifications, all x187; scale bars 50μ). While bile ducts are present and appear normal in ΔS6^dox livers (arrow heads), hepatocytes appear disordered, show mild vacuolization and some lack nuclei (arrows). (D) Graph showing quantitation of number of TUNEL-positive cells/field in liver sections of WT and ΔS6^dox mice provided with dox for different lengths of time. (E) Photomicrographs of H&E stained sections of ΔS6^dox livers provided with water (a) or 250μg/ml dox for different lengths of time (b-f). After ~35 days on dox, hepatocyte blebbing and focal drop-out is visible (b), followed by a full-blown regenerative response involving the emergence of regenerative nodules (n) and a ductular reaction (dr) (c-e). Image in (e) is a higher magnification of the area in (d) bounded by the dotted line (Original magnifications, a, c, d (x31.25); b, e (x125)). Scale bars: a), c) and d), 100μ; b) and e), 50μ. (F) Photomicrographs of IHC for PCNA, phospho-RPS6^(Ser235/6) and pan-cytokeratin (CK) on livers of ΔS6^dox mice provided with dox for 54 days showing that regenerating nodules contain PCNA- and phospho-RPS6^(Ser235/6)-positive immature hepatocytes while pan-CK-positive cells comprise the dr. Original magnifications, all x125; scale bars, 50μ.

https://doi.org/10.1371/journal.pgen.1010595.g004

Chronic hepatic Rps6 deficiency predisposes to liver overgrowth and tumor development

The transition from a hypoplastic, cytopenic state to a hyperplastic, cancerous state is a poorly understood aspect of the ribosomopathies. However, ribosomal protein haploinsufficiencies in model organisms have often resulted in overgrowth or cancer phenotypes indicating that mechanisms that normally regulate tissue homeostasis are lost, overridden or actively disabled in the context of chronic RiBi dysfunction. Because liver mass in the mouse is maintained at ~5% of body weight throughout adult life, we took advantage of this strict size control to determine if livers of ΔS6 mice overgrew or developed cancer as they aged. Analysis of liver mass in cohorts of WT and ΔS6 mice at ≥ 6 months showed that in contrast to WT mice, all of which had livers that remained within normal % Liver/Body Weight (%L/BW) range, ~60% of livers from ΔS6 mice displayed hepatomegaly ranging from mild to extreme (Fig 5A and 5B). In most cases, ΔS6 livers were grossly misshapen due to the aberrant growth of one or more lobes, many of which displayed nodular growth indicative of pre-malignant or malignant conversion (Fig 5C). Histological analysis of enlarged ΔS6 livers revealed the presence of biliary malformations such as bile duct hamartomas (panel b of Fig 5D) as well as adenomas that almost invariably contained PCNA- and phospho-Rps6^Ser235/6-positive cells (panels c, d, and e of Fig 5D). However, most striking of all was our finding that malignant tumors, most of which were moderately differentiated hepatocellular carcinomas (HCCs) with solid or trabecular growth patterns that stained strongly for PCNA and phospho-Rps6^Ser235/6 (panels f, g, and h of Fig 5D), developed in ~50% of ΔS6 livers by ~1 year of age (Fig 5E). Notably, although hepatomegaly only developed in 2/14 (14%) of ΔS6^dox mice maintained on dox for longer than 15 weeks, both cases (%L/BWs of 6.7% and 8.1%) occurred in mice that had received continuous dox treatment for 30 or 38 weeks respectively (Fig 5F) suggestive of a trend towards liver overgrowth as a function of increasing age or continued depletion of Rps6. Moreover, tumors developed in 6/14 (43%) of mice maintained on dox for >15 weeks, 4 (75%) of which developed in mice that had received dox for >26 weeks (Fig 5G). These results suggest that loss of Rps6 from adult hepatocytes disrupts liver homeostasis and predisposes to overgrowth and tumor development, the extent to which appears to depend on developmental context and timing of Rps6 depletion.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Rps6-deficient livers overgrow and are predisposed to spontaneous tumor development.

(A) Graph of % liver/body weights (% L/BW) in WT and ΔS6 mice at ≥ 6 months of age showing that many ΔS6 livers have grown to exceed normal size. Data are mean (± SEM); **** P < .0001; 2-tailed unpaired Student’s t-test. (B) Graph showing % of WT and ΔS6 mice at ≥ 6 months of age with % L/BWs in normal range (3.5–6%) or larger than normal (> 6%). (C) Photomicrographs of livers from 2 ΔS6 mice at ~8 months of age. Livers have grown to 2–3 times normal size and are grossly misshapen due to abnormal nodular growth. (D) Photomicrographs of H&E stained slides and IHC of livers from WT (a) or ΔS6 mice at ≥ 6 months of age (b-h) showing bile duct hamartomas (b) and tumors with compact nodular (c) or trabecular growth patterns (d). IHC for PCNA (d and g) and phospho-RPS6^(Ser235/6) (e and h) show that tumors are comprised of highly proliferative cells that display activation of mTOR. AEC chromagen (red/brown); hematoxylin counterstain (blue). Original magnifications; a, c, x 62.5; b, d-h, x 125 (scale bars, all 50μ). (E) Kaplan-Meier curve showing % of WT and ΔS6 mice that are tumor-free at ≥ 6 months of age. 50% of ΔS6 mice have developed at least 1 tumor by 372 days. P = .0008 (Log-rank (Mantel-Cox) test). (F) Graph of %L/BWs of WT and ΔS6^dox mice showing that some ΔS6^dox livers (2/7 (28.6%)) have also overgrown after being maintained on dox for >26 weeks of age. While none of the dox treatment durations result in significantly different % L/BWs between WT and ΔS6^dox mice, there is a trend towards hepatomegaly in ΔS6^dox mice with increasing time on dox (2-tailed unpaired Student’s t-test). (G) Photomicrographs of livers from a WT mouse provided with 250μg/ml dox for 7 months (3.8% L/BW; left), a ΔS6^dox mouse provided with water (no dox) for 6.5 months (3.6% L/BW; middle) and a ΔS6^dox mouse provided with 250μg/ml dox for 7 months (8.1% L/BW; right (top of liver) and far right (underside of liver)) that shows hepatomegaly and aberrant nodular growth extending out from a lobe.

https://doi.org/10.1371/journal.pgen.1010595.g005

The hepatomegaly and spontaneous tumor development in ΔS6 livers prompted us to further explore the idea that hepatic Rps6 deficiency was a priming event for tumor development. We therefore lowered the threshold for malignant conversion in ΔS6 livers by co-opting the Alb-Cre transgene used to delete Rps6 to also delete the tumor suppressor Pten, a mild, yet reliable oncogenic stimulus, that by itself, results in delayed-onset liver tumor development in mice by ~1 year of age [51,52]. Analysis of recombination of the Rps6^lox/lox (ΔS6^del) and Pten^loxlox (ΔPTEN) alleles in livers of WT, ΔS6, ΔPTEN and doubly deficient ΔS6:ΔPTEN mice demonstrated that Alb-Cre transgene expression was not limiting for recombination of either allele when both were present in the homozygous (lox/lox) state (S9A Fig). Western blotting of liver lysates from each of the parental strains using phospho-Rps6^Ser235/6- and phospho-Akt^Ser473-specific antibodies also confirmed that mTOR, but not Akt, was activated in ΔS6 livers, Akt, but not mTOR, was activated in ΔPTEN livers and that both pathways were active in ΔS6:ΔPTEN livers (S9B and S9C Fig). Monitoring of cohorts of WT, ΔS6, ΔPTEN and ΔS6:ΔPTEN mice up to ~1 year of age for signs of hepatomegaly or tumor development showed that, as previously documented, livers of ΔS6 mice that had initially been smaller than normal began to grow and exceed normal size after ~30 weeks of age (Fig 6A and S2 Table). Conversely, ΔPTEN livers, which demonstrate PI3K/Akt driven metabolic changes that promote steatotic hepatocyte hypertrophy, were consistently larger than normal from the outset. However, the growth of ΔS6:ΔPTEN livers began to diverge from each of the parental strains and accelerate at ~8–12 weeks of age, at which time livers became peppered with pale colored nodules that covered the surface (Fig 6B). Sampling of livers from mice of all genotypes at different ages revealed that loss of both Rps6 and Pten accelerated tumor development relative to each of the parental strains to the extent that 50% of ΔS6:ΔPTEN livers had developed at least 1 tumor by ~20 weeks age (Fig 6C) with 1 mouse developing tumors as early as 8 weeks of age. ΔS6:ΔPTEN livers also displayed other hyperproliferative lesions and abnormalities including bile duct hyperplasia and dysplasia, biliary hamartomas, duct ectasia, bile infarcts, altered hepatic foci and fatty adenomas, many of which were present in mice <6 months of age (Fig 6D). Notably, in contrast to ΔS6 livers which typically developed HCCs, ΔS6:ΔPTEN livers developed a wider spectrum of tumors including cholangiocarcinomas and hepatocholangiocellularcarcinomas (S3 Table and Fig 6D). Thus, Rps6-deficiency disrupts normal hepatic homeostasis and primes the liver for malignant conversion and accelerates liver tumor development in the context of hyperactivated PI3K/Akt-mediated growth factor signaling.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Rps6-deficiency lowers the threshold for tumor development in PTEN-deficient livers.

(A) Graph of %L/BWs in WT, ΔS6, ΔPTEN and ΔS6ΔPTEN mice at different ages showing accelerated growth of ΔS6ΔPTEN livers from ~12 weeks of age. For mean %L/BWs and P values, please see S2 Table. (B) Gross appearance of WT, ΔS6, ΔPTEN and ΔS6ΔPTEN livers at 4 and 8 weeks of age. Note the dramatic change in appearance of the ΔS6ΔPTEN liver between 4–8 weeks age as fatty nodules develop. (C) Kaplan-Meier curve showing % of liver tumor-free WT, ΔS6, ΔPTEN and ΔS6ΔPTEN mice. Age at which 50% of ΔS6, ΔPTEN and ΔS6ΔPTEN mice develop at least 1 liver tumor is 367, 365 and 150 days respectively. ***, P = .0001; ****,P < .0001. (Log-rank (Mantel-Cox) test). (D) Photomicrographs of H&E stained liver sections from a) a 3 month old ΔS6 mouse showing regenerative nodules and remnants of the dr, b) a 5.5 month old ΔPTEN mouse showing typical pericentral steatosis, c) a 5.5 month old ΔS6ΔPTEN mouse showing a nodule (n) compressing the parenchyma adjacent to an area displaying bile duct hamartomas (*), d) a 5.5 month old ΔS6ΔPTEN mouse with a bile duct hamartoma, e) a 10 week old ΔS6ΔPTEN mouse with fatty adenomas, f) an 8 week old and g) a 3 month old ΔS6ΔPTEN mouse with cholangiocarcinomas, and h) an 11 month old ΔS6ΔPTEN mouse with a trabecular HCC (f). Dashed lines denote nodule boundaries. Original magnifications; a, b, c, d (x 62.5); e, f (x 56.25); g, h (x 125); Scale bars, all 50μ.

https://doi.org/10.1371/journal.pgen.1010595.g006

Overexpression of Myc preserves hepatocyte viability and eliminates the need for neonatal ΔS6 livers to regenerate but fails to prevent malignant progression

Many studies have established bi-directional crosstalk between c-Myc and ribosomal proteins in the regulation of ribosome biogenesis, cell growth and cancer [53–58]. Indeed, decreased expression of c-Myc has been implicated as a driver of pancreatic hypoplasia in a mouse model of Schwachman-Diamond Syndrome (SDS) (OMIM #260400) [59]. Because c-Myc is highly expressed in fetal hepatoblasts of the developing liver yet drops to virtually undetectable levels by birth [60], we were unable to determine if hypoplastic neonatal ΔS6 livers expressed less c-Myc than their WT counterparts. However, array analysis did indicate that Myc mRNA was elevated in ΔS6 livers during the regenerative phase despite failing to reach the 8-fold cut-off to be included in the list of differentially regulated genes (S1 Table). Northern blotting confirmed that Myc was modestly elevated (~3-5-fold) in ΔS6 livers (S10A and S10B Fig), with IHC with c-Myc-, pan-CK and Rps6-specific antibodies revealing the source of increased c-Myc to be pan-CK-positive ductular cells rather than Rps6-expressing nodular hepatocytes suggesting that nodular growth in regenerating ΔS6 livers was not being driven by an increase in c-Myc (S10C and S10D Fig). Given c-Myc’s role in positively regulating ribosome biogenesis and that rp haploinsufficiency constrains Myc-dependent oncogenesis in vivo [53], we asked how increasing the level of c-Myc in the liver altered its response to Rps6-deficiency either during the early hypoplastic phase or the later tumor-prone phase. This approach involved breeding ΔS6 mice to Albumin-c-Myc (Alb-c-Myc) transgenic mice [61] which express a modest level of c-Myc (~8-fold above normal) in postnatal hepatocytes (S10A and S10B Fig) that is sufficient to stimulate ribosome biogenesis and promote hepatocyte hypertrophy in livers of young mice, but unable to drive fully penetrant HCC development before 1 year of age. Analysis of progeny from matings arranged to generate S6^lox/lox:Alb-Cre:Alb-c-Myc mice (herein referred to as ΔS6:c-Myc mice) confirmed that ΔS6:c-Myc mice were born at the expected frequency. The first indication that elevated c-Myc was altering the liver’s response to Rps6 deficiency came after analyzing recombination of the ΔS6^del allele which showed that it had increased from ~50% in ΔS6 livers to ~80% in ΔS6:c-Myc livers (S11A Fig), an effect that was not seen in livers of ΔS6:ΔPTEN mice (S11B Fig). Northern blotting confirmed that this increase in abundance of the recombined ΔS6^del allele in ΔS6:c-Myc livers translated into a further reduction in the level of Rps6 mRNA causing it to fall below the level required to sustain normal levels of 18S rRNA (S11C Fig). Given that ΔS6:c-Myc livers expressed less Rps6 than their ΔS6 counterparts, we expected neonatal ΔS6:c-Myc mice to be just as small, if not smaller than ΔS6 mice. However, to the contrary, ΔS6:c-Myc mice were visually indistinguishable from their WT or Alb-c-Myc littermates (Figs 7A, and S12A and S4 Table). Moreover, livers of young adult ΔS6:c-Myc mice were neither small nor mottled (Figs 7B and S12B) suggesting that c-Myc was suppressing, rather than enhancing the neonatal growth defect and hepatic dysfunction in ΔS6 mice. Histological evaluation of ΔS6:c-Myc livers at P7 showed the complete absence of cholestatic hepatocyte degeneration indicating that c-Myc was exerting a hepatoprotective effect by neutralizing the hepatocyte death caused by loss of Rps6 (Fig 7C). Consequently, ΔS6:c-Myc livers no longer needed to regenerate as seen by the absence of regenerative nodules or evidence of a ductular reaction (Fig 7C). Unexpectedly, we found that in contrast to Rps6 mRNA levels which were decreased in ΔS6:c-Myc livers (S11D Fig), RPS6 protein levels remained unchanged (Fig 7D), suggesting that one mechanism by which c-Myc could be neutralizing hepatocyte death in ΔS6 livers was by influencing the post-transcriptional processing of Rps6, either at the level of splicing or translation. In searching for molecular correlates of c-Myc-dependent hepatoprotection, analysis showed that c-Myc blunted Rps6-phosphorylation/mTOR activation (Fig 7D) and normalized the expression of mRNAs associated with activation of NF-κB and induction of the innate immune response in ΔS6 livers (Figs 7E and S11D). However, it failed to normalize the expression of oncofetal genes such as H19 or Igf2 or the classical p53-dependent targets p21/Cdkn1a and Sox4 (Figs 7E and S11D). Finally, to determine if c-Myc overexpression altered progression to hepatomegaly or malignancy in ΔS6 livers, we monitored cohorts of ΔS6:c-Myc mice at ≥ 6 months of age. Analysis showed that livers of ΔS6:c-Myc mice had a higher propensity to develop moderate to severe hepatomegaly than their ΔS6 or Alb-c-Myc counterparts (S13A and S13B Fig) and that tumors developed slightly earlier in ΔS6:c-Myc mice relative to each of the parental strains (S13C Fig), but not as early as ΔS6:ΔPTEN mice (Fig 6C). Thus, while a modest increase in c-Myc alters the fate of neonatal hepatocytes in ΔS6 livers by preserving hepatocyte viability and eliminating the need for ΔS6 livers to regenerate, it is unable to overcome all of the derangements caused by Rps6 insufficiency and fails to prevent liver overgrowth or tumor development as mice age.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Overexpression of c-Myc rescues the growth defect and eliminates the requirement for ΔS6 livers to regenerate by preserving hepatocyte viability.

(A) Picture of a 28 day old WT, ΔS6, c-Myc and ΔS6:c-Myc mouse. The ΔS6:c-Myc mouse (far right) is indistinguishable from the WT mouse (far left) in terms of size and lacks the jaundiced (yellowed) coat of the ΔS6 mouse (second from left). (B) Gross appearance of livers from 28 day old WT, ΔS6, c-Myc and ΔS6:c-Myc mice. Note the smooth and less jaundiced appearance of the ΔS6:c-Myc liver relative to the ΔS6 liver. (C) Photomicrographs of H&E stained liver sections of 8 and 28 day old WT, ΔS6, c-Myc and ΔS6:c-Myc mice showing the absence of feathery degeneration at P8 and lack of regenerative nodules or a dr at P28 in livers of ΔS6:c-Myc compared to ΔS6 mice. Dotted lines depict nodule boundaries. Asterisks (*) denote characteristic regions of hepatocyte hypertrophy in Alb-c-Myc livers (Original magnifications, x 62.5; scale bars 50μ). (D) Western blot showing that mTOR-dependent phosphorylation of RPS6 is suppressed by overexpression of c-Myc in ΔS6:c-Myc livers. GAPDH; protein load control. (E) Heat map showing that overexpression of c-Myc normalizes the innate immunity molecular signature comprising NF-κB target genes and DAMPs (boxed areas), but not imprinted genes, in ΔS6 livers.

https://doi.org/10.1371/journal.pgen.1010595.g007

Hepatoblast-specific expression of p53^QS mimics the biliary, but not the hepatocyte defect in ΔS6 livers

Having confirmed that p53 had been stabilized in ΔS6 livers, we set out to address p53’s role in the failure of ΔS6 livers to form bile ducts or maintain hepatocyte viability by generating mice in which p53 is artificially stabilized in hepatoblasts of the developing liver. This was done using a strain of mice that expresses a conditional (lox-stop-lox (LSL)) knock-in mutant allele of Trp53 harboring substitutions at amino acids L25Q and W26S (p53^LSL-Q25S26) that abolishes binding to p53’s negative regulator Mdm2 [62]. Although this mutant is hypomorphic for transactivation of a select cadre of p53 target genes, it retains DNA binding capability [63] and has been shown to be sufficient to phenocopy the pigmentation defects that develop in Dsk mouse mutants harboring naturally occurring mutations in Rps19 and Rps20 or genetically engineered mice with keratinocyte-specific deletion of Rps6 [64]. Successive rounds of breeding of p53^LSL-Q25S26 mice to Alb-Cre mice produced mice in which both copies of the p53 mutant were targeted to hepatoblasts in an otherwise p53-null background (herein referred to as p53^QS mice). Monitoring of litters from birth showed that female mice that were homozygous for the mutant p53 allele irrespective of Alb-Cre status were underrepresented in litters, consistent with previously reported lethality associated with trp53-nullizygosity in females [65,66]. Male p53^QS mice were, however viable and although body weights trended lower than their WT counterparts, statistical significance was not reached indicating that hepatic expression of p53^QS did not stunt neonatal growth (S14A Fig). IHC (Fig 8A) and immunoblotting (Fig 8B) confirmed that p53^QS was robustly expressed in >95% of hepatocytes, the majority of which was localized to the nucleus with a smaller amount present in the cytoplasm reflecting nucleo-cytoplasmic shuttling of p53 between both compartments. While gross inspection of livers of neonatal p53^QS mice revealed discoloration indicative of mild jaundice, livers were otherwise unremarkable in that they were neither small (S14B Fig) nor mottled. Moreover, histological evaluation showed that in contrast to ΔS6 livers which were already losing hepatocytes by 2 weeks of age (panels c and f of S14C Fig), p53^QS-expressing hepatocytes remained viable (panels b and e of S14C Fig), with small bile-acid induced infarcts only appearing after mice reached adulthood (panel d of Fig 8C and panel h of S14C Fig). Biochemical analysis of liver function in p53^QS mice revealed evidence of hepatocellular and biliary dysfunction at 4–5 weeks of age, both of which had improved, but not completely resolved by 7–9 weeks of age (S15 Fig). In stark contrast to ΔS6 livers in which liver function had improved commensurate with nodular growth and induction of a robust dr (panel i of S14C Fig), liver function in p53^QS improved without any histological evidence of a regenerative response suggesting that livers were able to tolerate and adapt to p53^QS expression much better than loss of Rps6 (panel h of S14C Fig). In determining the basis for the biliary dysfunction in p53^QS livers, we performed Sox9 IHC which showed that p53^QS livers had fewer Sox9-expressing ductal plate cells at P8 and P15 than WT livers, and that of those that were visible, none were being incorporated into bile ducts (S16A-S16C Fig). However, in contrast to ΔS6 livers which had already begun to show signs of a nascent Sox9-positive dr by P15 (Fig 2A), this response was both delayed and muted in p53^QS livers (S16D Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Hepatoblast-specific expression of p53^QS results in chronic liver failure preceded by induction of senescence and innate immunity and activation of mTOR.

(A) p53 IHC performed on liver sections from a WT (left) and p53^QS expressing mouse (right) showing robust nuclear expression of the p53^QS mutant in hepatocytes. Original magnifications, x125. (AEC Chromogen (orange/red), no counterstain). (B) Western blot of fractionated cytoplasmic (C) and nuclear (N) proteins isolated from the livers of 2 WT mice (lanes 1, 2, 5 and 6) and 2 p53^QS mice (lanes 3, 4, 7 and 8) with a p53-specific antibody. Immunoblotting of the same lysates using antibodies specific for the nuclear protein UBF (upstream binding factor) and the cytoplasmic protein GADPH confirm enrichment of nuclear and cytoplasmic proteins after fractionation. (C) H&E stained sections of WT (a, b) and p53^QS livers (c-h) showing age-dependent progression of disease in p53^QS livers. Regional hepatocyte vacuolization, evident at P11 (c) is followed by focal hepatocyte necrosis (biliary infarcts, *) by ~5 weeks of age (d). After ~3 months, hepatocytes show increasing heterogeneity across the lobule (e, f, g). Karyomegaly (enlarged nuclei (arrowheads, e and g)), regional hepatocyte vacuolization and biliary dilatation (g, arrow) are common. By ~5 months of age, p53^QS livers show evidence of widespread parenchymal loss (h, necrotic areas bounded by dashed lines) signifying ongoing liver decompensation in the absence of regeneration. Original magnifications; a, b, c, d, g, h and f, x62.5; e, x125. (D) Kaplan-Meier curve of morbidity and mortality in WT and p53^QSlox/lox:Alb-Cre mice. Median survival of p53^QSlox/lox:Alb-Cre mice; 147 days. **** P < .0001 (Log Rank (Mantel Cox) test). (E) Ethidium-stained gels of Sq-PCR for p53, p21/Cdkn1a, Noxa, senescence markers (p15^INK4b and p16^INK4a) and DAMPs (S100a8 and S100a9) in two 4–6 week old WT mice (lanes 1 and 2), two 4–6 week old p53^QS mice with moderate disease (lanes 3 and 4) and two 4 month old p53QS mice with advanced disease (lanes 4–6). (F) IHC with a phospho-RPS6^(Ser235/6)-specific antibody showing regional phospho-RPS6^(Ser235/6) staining in WT liver and pan-lobular staining in p53^QS livers. Original magnifications, x125. (AEC Chromogen red, hematoxylin counterstain, blue). Scale bars for all images, 50μ.

https://doi.org/10.1371/journal.pgen.1010595.g008

Having found that p53^QS expression generally mimicked the biliary defect, but was not able to induce the robust or widespread hepatocyte loss that accompanied loss of Rps6, we continued to monitor p53^QS mice to determine how chronic stabilization of p53 impacted the liver as mice aged. While p53^QS livers showed evidence of mild hepatic dysplasia and focal necrosis at ~5–6 weeks of age (panel d of Fig 8C), livers became increasingly unstable over time as seen by increasing hepatocyte heterogeneity in terms of size, nuclear morphology and degree of vacuolization. Moreover, small foci of atrophic or dying hepatocytes that had been evident at 4–6 weeks continued to expand to encompass large swathes of the parenchyma suggesting that livers were beginning to fail (panels e, f, g, and h of Fig 8C). Indeed, all p53^QS mice ultimately became moribund, requiring euthanization at or before 6 months of age (median survival: 147 days. Fig 8D). Autopsies performed on three p53^QS mice between 3–5 months of age revealed hepatomegaly and abnormal liver growth with histological analysis showing that much of the parenchyma had been replaced by small hepatoblast-like cells (panels a and b of S17 Fig). IHC with the p53-specific antibody revealed that liver failure in p53^QS mice was being driven by the death of p53^QS-expressing hepatocytes with livers being repopulated by small immature cells that failed to express the p53^QS mutant (panels c and d of S17 Fig). Immunoprofiling of these p53^QS-naïve cells showed that all expressed MYC (panels e and f of S17 Fig), while a subset also expressed EPCAM (panels g and h of S17 Fig) suggesting that p53^QS livers were being repopulated by maturation-arrested cells with immunoprofiles similar to, but distinct from E9.5-E11.5 (EPCAM+) or E12-14 (EPCAM-) hepatoblasts.

The protracted viability of p53^QS-expressing hepatocytes together with the absence of any histological evidence of regeneration in livers of young p53^QS mice suggested the possibility that hepatocytes were being driven into a state of cell-cycle arrest or senescence before livers failed. We therefore performed Sq-PCR analysis of cell-cycle-, senescence- and apoptosis-associated genes p21/Cdkn1a, p15/Ink4b, p16/Ink4a and Noxa in livers of mice at different ages and stages of disease to determine if this was the case and, if so, how changes in expression correlated with disease as livers progressed from dysplasia to failure. Analysis showed that p21/Cdkn1a and Noxa were elevated in p53^QS livers irrespective of age or stage of disease (Fig 8E), a finding that was unexpected given that the p53^QS mutant is defective for transcriptional activation of both of these genes [63]. p15/Ink4b and p16/Ink4a were also induced in p53^QS livers; however unlike p21/Cdkn1a and Noxa, upregulation was only seen in end-stage livers containing immature p53^QS-naïve cells. In light of the fact that senescence can induce innate immunity and the SASP [67,68], we analyzed mRNA expression of the DAMPs S100a8 and S100a9, both of which had been induced in ΔS6 livers, and found that like Cdkn1a and Noxa, both were upregulated in p53^QS livers independent of age or disease state (Fig 8E). Finally, given mTOR’s ability to promote senescence and the SASP [69,70] we performed IHC with the phospho-Rps6^Ser235/236-specific antibody on sections of WT and dysplastic p53^QS livers from 3 month old mice which revealed strong pan-lobular phospho-Rps6 staining in p53^QS liver indicating activation of mTOR/S6K signaling (Fig 8F). Taken together, these results show that while expression of p53^QS mimics loss of Rps6 by inhibiting bile duct development, hepatocytes tolerate p53^QS much better than loss of Rps6 and only die after a protracted period of cell-cycle arrest or senescence. Hepatocytes and biliary cells thus diverge in their response to p53 stabilization or loss of Rps6.

Loss of p53 fails to improve liver disease in ΔS6 mice

Having determined that expression of p53^QS mimicked the biliary, but not the hepatocyte defect in ΔS6 livers, we sought to determine the extent to which each aspect of liver disease depended on p53 by breeding Rps6^lox/lox:Alb-Cre mice to p53-/- mice to generate mice with livers that were deficient for Rps6 and p53 (herein referred to as ΔS6:Δp53 mice). In setting benchmarks that would be used to determine the impact of p53 loss on ΔS6-associated hepatocyte death, we turned to ΔS6:c-Myc mice as our rescue paradigm given that augmentation of c-Myc in ΔS6 livers had preserved hepatocyte viability and corrected other hallmarks of Rps6-insufficiency by restoring normal liver mass and neonatal growth, suppressing the inflammatory signature and normalizing mTOR (Fig 7). Genotyping of progeny from the appropriate matings designed to generate ΔS6:Δp53 mice showed that mice lacking both copies of p53 were again underrepresented in litters irrespective of Rps6 status. Body weight analysis of WT, ΔS6 and ΔS6Δp53 mice between the ages of 4–6 weeks showed that ΔS6:Δp53 mice remained underweight indicating that the loss of p53 did not correct the neonatal growth deficit (Fig 9A). Moreover, although liver weights of half of the ΔS6:Δp53 mice remained within the normal range, the mean liver weight of the group was not significantly different from either ΔS6 mice or WT mice, suggesting that the impact of p53-nullizygosity on ΔS6-associated liver hypoplasia was mixed (Fig 9A). Moreover, because %L/BW is a function of body weight and liver weight, the skewing of ΔS6Δp53 liver weights towards normal coupled with their lower body weights translated into %L/BW values that were normal, precluding us from reaching a definitive conclusion as to whether loss of p53 impacted ΔS6-associated liver hypoplasia. In light of this ambiguity, we turned to LFTs to determine if loss of p53 improved hepatic function in ΔS6 livers. Of the 4 biochemical markers used to assess function, only alkaline phosphatase, a marker of cholestasis, showed a modest improvement in the absence of p53, while total bilirubin levels remained elevated and ALT and AST levels increased further indicating that loss of p53 was exacerbating hepatocyte dysfunction, rather than improving it (S18 Fig). Histological evaluation of ΔS6:Δp53 livers also confirmed that loss of p53 failed to confer the level of hepatoprotection afforded by c-Myc as livers were still being forced to regenerate (Fig 9B). However, nodules were smaller and appeared less distinct than in ΔS6 livers (panels g and h of Fig 9B), in large part due to lack of a robust dr that had accentuated nodule boundaries in ΔS6 livers (panels e and f of Fig 9B). Indeed, Sox9 IHC showed that while the cord-like streaming and duct-forming properties of Sox9-positive cells in ΔS6 livers (panel b of S19A Fig) clearly signified induction of a classical HPC-mediated-dr as a precursor to secondary bile duct development [42], none of the Sox9-positive cells in ΔS6:Δp53 livers were becoming incorporated into ductular structures (panels c and d of S19A Fig). Biliary malformations in the form of cysts in peri-portal areas (panel g of Fig 9B) and a paucity of normal Sox-9-expressing bile ducts indicated that loss of p53 was failing to correct the biliary defect (S19A Fig). Analysis showed that immature hepatoblast-like cells had also begun to accumulate in ΔS6:Δp53 livers by 4–5 months of age, which in some cases, were either proliferating in distinct clusters or had expanded to become the dominant cell-type (S19B Fig). Finally, sq-PCR analysis of p21/Cdkn1a, Noxa and S100a9 mRNAs, all of which had been upregulated in ΔS6 livers and normalized by c-Myc, showed that all remained elevated in ΔS6Δp53 livers (Fig 9C), while immunoblotting showed that loss of p53 also failed to blunt Rps6 phosphorylation (Fig 9D). By showing that loss of p53 fails to significantly improve liver disease in ΔS6 mice, these results indicate that liver disease in ΔS6 is either p53-independent, or that p53 is but one arm of a much broader ribosomal stress response that drives disease in Rps6-deficient livers.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 9. Loss of p53 does not rescue the ΔS6-associated growth defect or hepatic dysfunction.

(A) Graphs of body weights, liver weights and % liver/body weights of 32–40 day old WT, ΔS6 and ΔS6Δp53 mice. P values: *P ≤ .05, **P ≤ .01. NS; not significant (2-tailed unpaired Student’s t-test). (B) Photomicrographs of H&E stained sections of livers from 4–5 week old WT, ΔS6, Δp53 and ΔS6Δp53 mice showing normal architecture and the presence of bile ducts in WT (a, b) and Δp53 (c, d) livers and well defined nodules and a prominent dr in the ΔS6 liver (e, f). Note moderate hepatocyte vacuolation in Δp53 livers (d). In the ΔS6Δp53 liver (g, h), nodules are present but are less defined due to the absence of a prominent dr which accentuates nodular boundaries in ΔS6 livers. Normal bile ducts are absent and biliary malformations are evident (arrow heads). Original magnifications; a, c, e and g (x 31.25; scale bars 100μ), b, d, f and h (x 250; scale bars 25μ). (C) Ethidium stained gels of sq-PCR performed on cDNA prepared from individual WT, ΔS6, Δp53 and ΔS6Δp53 livers. Note that loss of p53 fails to normalize the expression of any of the mRNAs induced in ΔS6 livers including p21/Cdkn1a. (D) IHC of livers of WT (a), ΔS6 (b), ΔS6:c-Myc (c) and ΔS6Δp53 (d) mice with a phospho-RPS6^(Ser235/6)-specific antibody showing that mTOR activation in ΔS6 livers is suppressed by overexpression of c-Myc, but not by loss of p53. Original magnifications; a, c, d (x62.5; scale bars, 100μ), b (x125; scale bar 50μ). (AEC chromagen, red; hematoxylin counterstain, blue).

https://doi.org/10.1371/journal.pgen.1010595.g009

Ethics statement

All experiments were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and with approval of the University of Texas Southwestern Medical Center Institutional Animal Care and Use Committee (IACUC).

Mouse strains

Conditional Rps6^lox/lox mice, Albumin-c-Myc mice and conditional Trp53^LSL-Q25S26 mice were generously provided by George Thomas (Catalan Institute of Oncology, Barcelona, Spain), Eric Sandgren (University of Wisconsin, Madison) and Laura Attardi (Stanford University) respectively. Albumin-Cre (Speer6-ps1^{Tg(Alb-cre)21Mgn}, Stock # 003574), Pten^lox/lox (Pten^tm1/Hwu, Stock # 004597) and p53-/- (Trp53^tm1Tyj, Stock # 002101) mice were purchased from The Jackson Laboratory (Bar Harbor, ME). ApoE-rtTA:TRE2-Cre transgenic mice were generated as described in S8 Fig. Mixed hybrid strain (C57Bl6/129/SJL) male and female mice were used for all studies unless otherwise stated. Mice were housed in a conventional colony under standard 12 hour light/dark cycles and allowed access to water and chow ad libitum.

Genotyping of mice and analysis and quantitation of recombination

Oligonucleotide sequences and conditions for PCR-genotyping of mouse strains and assessment of Cre-mediated recombination of the loxp-flanked Rps6 and Pten alleles are shown in S5 Table. Ethidium stained gels showing recombination were imaged using Image J software [164]. To determine the % Rps6 recombination in livers, the intensity signal values for the imaged Rps6^lox (upper) band and the recombined ΔS6^del (lower) band were added together to produce an arbitrary value of 100%. The extent of recombination was determined by representing the signal value of the ΔS6^del (lower) band as a percentage of the total.

Analysis of liver function

Peripheral blood was collected from the retro-orbital sinus using a heparinized glass capillary tube. Biochemical analysis of liver function (ALT, AST, T-Bil and Alk-Phos) was performed by UT Southwestern Medical Center’s Metabolic Phenotyping Core using a Vitros 250 bioanalyzer. Plasma was isolated from ≥3 individual mice/group.

Histopathology and immunohistochemistry (IHC)

Small pieces (~1cm x 1cm) of freshly harvested liver were immersion fixed in 10% neutral-buffered formalin for 24–48 hours and processed to paraffin for sectioning at 5 microns for routine H & E staining or IHC as previously described [165]. Primary and secondary antibodies and reagents and conditions used for IHC are listed in S6 Table. Antigen retrieval (Retrievagen A pH 6.0, BD Pharmingen) was performed in a microwave at 95°C for 10 minutes for phospho-S6, total S6 and pan-CK, 20 minutes for p53 and 50 minutes for c-Myc IHC. All primary antibody incubations were performed overnight at 4°C except for c-Myc which was performed for 48 hours at 4°C. All secondary antibody incubations were performed at room temperature for 1–2 hours and slides were incubated with streptavidin horseradish peroxidase (HRP) conjugate at room temperature for 20–60 minutes. 3-Amino-9-Ethylcarbazole (AEC) chromagen was prepared as recommended by the manufacturer. After color development, sections were either immediately cover-slipped (no counterstain) or counterstained with Hematoxylin QS.

Analysis of apoptosis

Liver cell apoptosis was determined using the DeadEnd Colorimetric TUNEL System (Promega, Catalog #G7360) applied to de-paraffinized liver sections as recommended by the manufacturer. For quantitation of the number of TUNEL-positive hepatocytes, at least 30-fields per liver section were counted. Livers from at least three mice per genotype and age were evaluated.

Bile duct and biliary epithelial cell (BEC) quantitation

To quantify the number of patent bile ducts per portal vein (bds/PV) in H&E stained sections, 10 portal veins (sampled from three or more animals for each genotype and age) were scored. To quantify the number of Sox9+ biliary epithelial cells (BECs), livers from three mutant animals and three controls were examined and positive cells scored from at least ten portal tracts per animal. Portal veins were identified by the presence of five or more biliary cells in the perivascular region and associated bile ducts were counted. P values were calculated by Student’s t-test.

RNA isolation, Northern blotting and probe labeling

Total liver RNA was prepared using RNA STAT 60 (Fisher Scientific, Catalog # CS-111) as previously described [165]. For Northern blotting, 12μg of total RNA was denatured at 50°C with glyoxal before separating on a 1.5% agarose-formaldehyde gel using a vertical gel apparatus. Fractionated RNA was transferred to Hybond-N⁺ membranes (GE Healthcare) overnight and hybridized to [α-³²P]dCTP radiolabeled cDNA probes prepared using a Random Primed DNA labeling kit (Sigma, 11004760001). The following day membranes were washed and exposed to Blue Lite Film (ISC BioExpress, F-9024) film with intensifying screens at -80°C for 8–24 hours, after which they were stripped and re-probed with a radiolabeled oligonucleotide probe specific for 18S rRNA (5′ GCCGTGCGTACTTAGACATGCATG 3′ corresponding to nucleotides 50–73 of the rat ribosomal RNA gene) end-labeled with [γ-³²P]ATP using T4 polynucleotide kinase (Fisher, Pittsburgh, PA) or a cyclophilin A-specific cDNA probe for assessment of RNA loading. The 840bp cDNA probe for Rps6 was obtained by NotI/SalI digestion of plasmid pCMV-Sport6 containing mouse Rps6 cDNA (ATCC Product # 10538869). The 4.8kb cDNA probe for mouse Myc was isolated and purified by digesting plasmid pSV-c-Myc-1 (ATCC # 41029) with BamHI and XbaI. The 721bp mouse Ppia/cyclophilin A-specific probe was obtained from Ambion (AM 7375).

mRNA expression analysis by semi-quantitative (sq) RT-PCR

1μl of cDNA synthesized from DNase 1-treated total liver RNA was used as a template for sq PCR amplification using TAq Polymerase (Qiagen) and primers and conditions listed in S7 Table. Confirmation of differential expression was obtained by running half of the PCR product on an agarose gel and quantifying relative band intensities using Image J software.

Analysis of rRNA processing

Pre-rRNA processing was analyzed by Northern blotting using 2 synthetic radiolabeled oligonucleotide probes: ITS1 (5′ CTCTCACCTCACTCCAGACACCTCGCTCCA 3′), which is homologous to nucleotides 5977–6006 and ITS2 (5′ ACCCACCGCAGCGGGTGACGCGATTGATCG 3′) which is homologous to nucleotides 7036–7065 of the mouse 45S pre-ribosomal RNA (NCBI Ref. Seq. NR_046233.2). Blots were stripped and re-probed with a radiolabeled oligonucleotide probe specific for 18S rRNA.

Preparation of proteins and immunoblotting

Preparation of total protein lysates was performed as previously described [47]. Nuclear and cytoplasmic proteins were fractionated as follows; ~ 250mg of freshly isolated or previously frozen (-80°C) liver was homogenized in 2.5 mls of ice-cold homogenization buffer (20mM Tris-Cl (pH7.4), 2mM MgCl₂, 0.25M sucrose, 10mM EDTA, 10mM EGTA, 0.2mM Na₃VO₄, 1mM NaO₇P₂.10H₂0, 10mM NaF containing protease inhibitors (cOmplete Protease Inhibitor cocktail (Millipore Sigma, Catalog #11697498001), 0.4μM Microcystin LR (Millpore Sigma, Catalog # 475815), 0.2μM Okadaic Acid (Enzo Life Sciences, Catalog # ALX-350-003-C100) and 30μg/ml PMSF Millipore Sigma, Catalog # 10837091001) using a Dounce tissue grinder and centrifuged at 1000 x g for 5 mins at 4°C. The supernatant and pellet were then processed as follows. The supernatant was removed and re-centrifuged at 100,000 x g for 1 hour at 4°C, after which the pellet was discarded and the supernatant (cytosolic fraction) aliquoted, snap frozen in liquid N₂ and stored at -80°C. The pellet generated by the 1000 x g centrifugation was washed and gently resuspended in 3 mls of homogenization buffer and re-centrifuged at 1000 x g for 5 minutes at 4°C. After discarding the supernatant, the pellet was gently re-suspended in 200μl of ice-cold nuclear extract buffer (20mM HEPES (free acid) NaOH (pH7.6), 2.5% glycerol, 0.42M NaCl, 1.5mM MgCl₂, 1mM EDTA, 1mM EGTA, 0.2mM Na₃VO₄, 1mM NaO₇P₂.10H₂0, 10mM NaF containing protease inhibitors (cOmplete Roche Protease Inhibitor Cocktail), 0.5μM Microcystin LR, 0.25μM Okadaic acid and 30mg/ml PMSF) and incubated at 4°C with gentle rocking for 45 minutes. Following centrifugation at 100,000 x g for 30 minutes at 4°C, the supernatant containing nuclear proteins was removed and aliquoted, snap-frozen in liquid N₂ and stored at -80°C. Immunoblotting was conducted as previously described [165]. 100 micrograms of protein (or 50 micrograms in the case of abundant housekeeping proteins such as GADPH) was loaded per lane. Antibodies used for immunoblotting are listed in S8 Table.

Imaging and quantitation

Ethidium bromide-stained gels, Northern blot and immunoblot films were imaged using Image J software. mRNA and protein levels were quantified by determining fold-changes after normalizing signal intensities to cyclophilin A for sq-PCR cDNA analysis, 18S rRNA or cyclophilin A for Northern Blot analysis and GAPDH for immunoblot analysis.

Rapamycin treatment

5–6 week old WT and ΔS6 mice received 6 intrapertioneal (i.p) injections of rapamycin (Calbiochem, Catalog # 553210) (2.5mg/kg body weight) or vehicle (DMSO/saline) on day 1, 3, 5, 7, 10 and 12. On day 13, mice were sacrificed and livers harvested for histology, IHC processing and protein lysate preparation. 4 mice (2 WT and 2 ΔS6 mice) received vehicle and 6 mice (3 WT and 3 ΔS6) mice received rapamycin.

Microarray analysis and bioinformatics

Microarray analysis of WT and ΔS6 livers was performed using cDNA generated from pooled total liver RNA isolated from 3 individual WT and 3 individual ΔS6 male mice at 5 weeks of age. cDNA was hybridized to Affymetrix Mouse 430 v2.0 GeneChips and data normalized using the RMA procedure [166], followed by quantile normalization [167]. 235 differentially expressed genes (up or downregulated ≥8 fold) were identified in ΔS6 livers. Gene association network enrichment was performed using Ingenuity Pathway Analysis (IPA) (Ingenuity Systems, Inc. Redwood City, CA). The top 4 networks activated in ΔS6 liver (Cdkn1a, Jun/Spp1, NF-κB and MAPK; S7 Fig) were obtained from the 235 differentially expressed genes with an enrichment score of 38 to 36. Colored shapes (red: over-expression, green: under-expression) represent genes within the differential gene list while open shapes represent genes are not on the significant gene list but are associated with the network. Arrows represent positive regulation of gene expression, with solid arrows indicating direct regulation, and broken arrows indicating indirect regulation.