Fibroblast Growth Factor Receptor Signalling Is Crucial for Liver Homeostasis and Regeneration

Fibroblast growth factor receptor signalling is crucial for liver homeostasis and regeneration

Heike Steiling1, Torsten Wu¨ stefeld2, Philippe Bugnon1, Maria Brauchle3, Reinhard Fa¨ ssler4, Daniel Teupser5, Joachim Thiery5, Jeffrey I Gordon6, Christian Trautwein2 and Sabine Werner*,1

1Institute of Cell Biology, ETH Zu¨rich, Ho¨nggerberg, CH-8093 Zu¨rich, Switzerland; 2Department of Gastroenterology and Hepatology, Medizinische Hochschule Hannover, D-30625 Hannover, Germany; 3Novartis Pharma AG, CH-4002 Basel, Switzerland; 4Max-Planck-Institute of Biochemistry, Am Klopferspitz 18a, D-82152 Martinsried, Germany; 5Institute of Laboratory Medicine, Clinical Chemistry and Molecular Diagnostics, Universita¨tsklinikum Leipzig, D-04103 Leipzig, Germany; 6Department of Molecular Biology and Pharmacology, Washington University School of Medicine, St Louis, MO 63110, USA

Several growth factors have been suggested to play a doderm was shown to be dependent on FGFs (Jung crucial role in liver regeneration, but a functional proof is et al., 1999). However, until now, little is known about still missing. Since ﬁbroblast growth factors are important the function of FGFs in liver homeostasis postnatally, for the initiation of mammalian liver development, we and a role of FGFs in liver repair has not been determined the roles of these mitogens in liver repair by demonstrated. Since FGF7 is a potent mitogen for targeted expression of a dominant-negative ﬁbroblast hepatocytes in vitro and in vivo (Housley et al., 1994; growth factor receptor (FGFR)in hepatocytes of trans- Strain et al., 1994) and since it can protect hepatocytes genic mice. The liver of young animals appeared from tumor necrosis factor-induced apoptosis (Senaldi histologically normal, and liver function was not obviously et al., 1998), we speculated about a role of FGFs in liver impaired. In aged transgenic mice, the frequency of fatty homeostasis and regeneration. To address this question, liver development was strongly increased compared to we chose to express a dominant-negative FGFR mutant control animals. Following partial hepatectomy, trans- in hepatocytes. Such receptor mutants are characterized genic mice showed markedly reduced hepatocyte prolif- by the lack of a functional tyrosine kinase domain. They eration because of an arrest in the late G1 phase of the cell disrupt FGFR signalling by competing for ligand cycle. These data demonstrate a key role of FGFR binding and by forming inactive heterodimers with signalling in repair after liver injury. endogenous FGFRs, thereby blocking signalling Oncogene (2003) 22, 4380–4388. doi:10.1038/sj.onc.1206499 through all FGFRs in response to the same ligand. By contrast, signalling through other growth factor recep- Keywords: cell cycle; hepatectomy; injury; keratinocyte tors is unaffected by the truncated FGFR (Ueno et al., growth factor; regeneration; steatosis 1992). Here, we show that expression of a dominant- negative FGFR2IIIb (dn-FGFRIIIb) mutant in hepatocytes impairs liver homeostasis and results in a severe Introduction defect in liver regeneration.

Fibroblast growth factors (FGFs) comprise a group of 22 structurally related proteins, which stimulate a wide Results variety of target cells (reviewed by Ornitz and Itoh, 2001). They have a high affinity to heparin and heparan Overexpression of a dominant-negative FGFR mutant in sulfate proteoglycans and require heparan sulfate to the gut and the liver of transgenic mice activate one of the four transmembrane tyrosine kinase receptors (FGFR1–FGFR4), which mediate their bio- We expressed a dominant-negative FGFR2IIIb mutant logical effects. Additional diversity is generated by under the control of the rat fatty acid binding protein alternative splicing of fibroblast growth factor receptor (fabpl) gene promoter in transgenic mice. This promoter (FGFRs) 1–3, which affects their ligand-binding specifi- targets the expression of transgenes to hepatocytes and cities (reviewed by Johnson and Williams, 1993). intestinal epithelial cells (Sweetser et al., 1988). A FGFs and their receptors are important regulators of truncated form of human FGFR2IIIb (dn-FGFR2IIIb) development and repair of various tissues and organs that lacks both tyrosine kinase domains was used (reviewed by Ornitz and Itoh, 2001). For example, (Figure 1a). We chose this mutant since (i) FGFR2IIIb initiation of mammalian liver development from en- is strongly expressed by hepatocytes, in particular during liver regeneration (Hu et al., 1995) and since *Correspondence: S Werner; (ii) the truncated FGFR2IIIb had been successfully used E-mail: [email protected] to block FGFR signalling in the skin and the lung Received 2 January 2003; accepted 17 February 2003 (Peters et al., 1994; Werner et al., 1994). FGF in liver regeneration H Steiling et al 4381

Figure 1 Expression of dn-FGFR2IIIb in the liver of transgenic mice. (a): The dn-hFGFR2IIIb cDNA lacking the tyrosine kinase domains (TK1 and TK2) was inserted downstream of the fabpl gene promoter and upstream of the human growth hormone (hGH) gene. Functional elements of dn-hFGFR2IIIb include immunoglobulin-like domains I, II and IIIb, an acidic box (AB), and the transmembrane domain (TM). The IIIb exon, which is speciﬁc for the FGFR2IIIb variant of FGFR2, is indicated with a thick black line. (b): Total cellular RNA (20 mg) from livers of wild-type and transgenic mice was analysed for the presence of dn-FGFR2IIIb mRNA using a human FGFR2IIIb probe. A measure of 1 mg of each RNA sample was loaded on a 1% agarose gel and stained with ethidium bromide (shown below the autoradiogram). 3, 43, 109: hemizygous transgenic mice of lines 3, 43, and 109. (c): Total cellular RNA (20 mg) was analysed for the presence of mFGFR2IIIb mRNA by RNase protection assay. Owing to the three stretches of identical nucleotides in the murine and human FGFR2 mRNAs, this fragment also detects the RNA derived from the human transgene (three lower bands). tRNA (50 mg) was used as a negative control. A 1000 c.p.m. of the hybridization probe was loaded in the lane labelled ‘probe’ and used as a size marker. Hybridization of the same set of RNAs with a GAPDH riboprobe was used to conﬁrm equal loading. Wt: wild-type mice; 43: hemizygous transgenic mice of line 43. (d): Membrane proteins (200 mg) from liver were analysed by immunoblotting for the presence of the truncated FGFR2IIIb protein (arrow). (e): Total cellular RNA (20 mg) from various tissues of transgenic mice (tg; line 43) and wild-type littermates were analysed for the presence of dn-FGFR2IIIb mRNA using the probe described in Figure 1b. For ‘tRNA’ and ‘probe’, see (b). Each RNA (1 mg) sample was loaded on a 1% agarose gel and stained with ethidium bromide (shown below the autoradiogram)

Three transgenic founder mice with different integra- levels (Figure 1b). Using a murine FGFR2IIIb tion sites of the transgene (data not shown) were probe that cross-hybridizes with the human transgene obtained and used to establish transgenic mouse lines. RNA, we found a 15-fold excess of the transgene These animals express the transgene at different mRNA compared to endogenous FGFR2 mRNA in

Oncogene FGF in liver regeneration H Steiling et al 4382 hemizygous mice of line 43 (Figure 1c). A 10–30-fold cellular RNA was pooled from 3 to 10 animals at the age excess of the truncated receptor compared to the of 8–12 weeks or 1 year, respectively, and used for endogenous receptor is required to completely block RNase protection assays. signalling through the endogenous receptor in the We found FGF1 mRNA but not FGF7, FGF10 and presence of physiological concentrations of the ligand FGF22 mRNAs in the livers of young and old animals (Ueno et al., 1992; Werner et al., 1993). Transgenic (Figure 2a, b and data not shown), and we never progeny of the other two founder mice showed detected FGF3 mRNA in the liver of young mice moderate (line 109, ratio 8 : 1) or low (line 3, ratio (Figure 2a). However, this FGF was expressed in the 3 : 1) levels of transgene expression (data not shown). liver of many old control and transgenic mice The levels of dn-FGFR2IIIb in the various pedigrees (Figure 2b). Expression of endogenous FGFR2IIIb of mice correlated with the levels of its protein product. was slightly higher in many transgenic animals com- Immunoblots detected a protein of the expected size pared to control mice, but this was not consistently (70 kDa) in membrane fractions prepared from the observed (Figure 1c and 2a). Expression of FGFR4, pooled livers of transgenic mice (n ¼ 3; Figure 1d). This FGFR2IIIc, FGFR1IIIc and FGFR3 was not affected band was not detected when the antibody was pretreated by either age or transgene (Figure 2a, b, and data not with the immunization peptide (data not shown). shown). Thus, overexpression of a dn-FGFR2IIIb Endogenous FGFR2, which is expressed at much lower mutant does not lead to a compensatory induction of levels, was not detectable under these conditions. any endogenous FGFR. We also analysed the expres- In all three mouse lines, we found highest levels of dn- sion of the endogenous liver fatty acid binding protein FGFR2IIIb mRNA in the liver and in the small gene, and found no differences between wild-type and intestine, with lower levels of expression in the colon transgenic mice (Figure 2a, b). Thus, there is no (shown in Figure 1e for line 43). All other tissues depletion of transcription factors by the fabpl gene surveyed had undetectable amounts of dn-FGFR2IIIb promoter-driven transgene. mRNA (Figure 1e). In the liver, we found expression in hepatocytes as determined by in situ hybridization (data Old transgenic mice develop fatty livers not shown). All transgenic mice appeared generally healthy and Expression of FGFs and FGFRs in the liver of wild-type fertile. However, animals from line 43, which expressed and transgenic mice high levels of the transgene, were smaller and had a reduced body weight (9% reduction compared to the We subsequently determined whether the transgene weight of age-matched females (n ¼ 20 transgenic, 14 affects expression of the FGFR2IIIb ligands FGF1, normal mice); 13% reduction in males (n ¼ 12 trans- FGF3, FGF7 and FGF10 or endogenous FGFRs. Total genic, 13 wild-type mice, age range ¼ 6–12 weeks).

Figure 2 Expression of FGFRs, FGFR2IIIb ligands and FABPl in livers of young (a) and old (b) dn-FGFR2IIIb mice and nontransgenic littermates. Total cellular RNA (20 mg) was analysed by RNase protection assay for the expression of dn-hFGFR2IIIb, endogenous mFGFR2IIIb, FGFR4, FGF1, FGF3, FABPl and GAPDH mRNAs. For probe and tRNA, see legend to Figure 1. Wt: wild-type mice; tg: transgenic mice

Oncogene FGF in liver regeneration H Steiling et al 4383 We performed a detailed histological analysis of the To determine the type of FGF that might signal via transgenic animals and control littermates or age- FGFR2IIIb during liver regeneration, we performed matched controls of the same sex (8–10 weeks: n ¼ 5 RNase protection assays with liver RNAs prepared wild-type, 14transgenic mice, 6–8 months: n ¼ 7 wild- from wild-type and transgenic mice (line 43), before and type, 13 transgenic mice, 10–14months: n ¼ 18 wild- at different time points after PH. Consistent with type, 45 transgenic mice). No obvious abnormalities published data (Marsden et al., 1992), expression of were observed in the intestine (data not shown) and in fgf1 was slightly induced after injury (1.5–2-fold; the liver of young transgenic animals (Figure 3a–d). Figure 4e; n ¼ 2–4animals per time point and genotype). However, mice from line 43 above the age of 12 months In contrast, fgf3 and fgf7 were not expressed at any frequently developed fatty livers as determined by stage of repair (data not shown). hematoxylin/eosin staining of paraffin sections (Figure Liver regeneration requires a priming step that leads 3e, f) and by staining of frozen sections with Sudan III, a to induction of primary response genes (reviewed by fat-soluble red dye (Figure 3g, h). The results were Fausto, 2000). To determine if this initial step is already confirmed with Oil Red O and Sudan Black, which also impaired in our transgenic mice, we analysed the mRNA stain fat (data not shown). In all, 81% of transgenic levels of c-jun and c-myc (n ¼ 3 wild-type mice and n=4 animals belonging to line 43 (n ¼ 26) developed fatty transgenic mice per time point; Figure 4e). Interestingly, livers. In contrast, only 32% of the mice that expressed expression of these genes was not reduced in dn- moderate levels of the transgene (line 109, n ¼ 19) had FGFR2IIIb transgenic mice but rather was increased fatty livers, a frequency comparable to wild-type (c-jun) or prolonged (c-myc), indicating that the animals (n ¼ 18, 38%). impaired regeneration is not a result of reduced immediate early gene expression. Liver function is not significantly impaired in Cyclin D1 mRNA levels were increased (4–8-fold) in dn-FGFR2IIIb transgenic mice transgenic mice at t ¼ 0(n ¼ 15 control, 15 transgenic mice). The time course of change after PH resembled Serologic analysis revealed that the activities of alanine that in normal littermate controls, but at each time and aspartate aminotransferases were not altered in point, the concentration of transcripts was higher in transgenic mice, whereas pseudocholinesterase activity mice expressing the transgene (Figure 4e). Immunoblot was higher in mice of the high expressor line in both analysis of cyclin protein levels revealed a comparable young (wild-type mice: 98 mmol/l s, n ¼ 11; transgenic kinetics of cyclin E expression in transgenic and wild- mice: 137 mmol/l s, n ¼ 17) and old animals (wild-type: type animals, although the levels of this protein were 136 mmol/l s, n ¼ 7; transgenic: 179 mmol/l s, n ¼ 7). increased in dn-FGFR2IIIb transgenic mice at t ¼ 24 Plasma glucose, total triglyceride and cholesterol and 48 h (Figure 4f). In contrast, cyclin A levels were levels, the lipoprotein profile as well as concentrations significantly reduced in the transgenic mice compared to of a-, b-, and g-globulins and of albumin were controls and the increase in cyclin A expression after indistinguishable between transgenic and control mice. hepatectomy was delayed. While in wild-type animals In agreement with the lack of obvious serological cyclin A expression peaked after 48 h, a first significant abnormalities, we could not detect signs of inflammation increase was only found 72 h after surgery in transgenic or fibrosis in the livers of our transgenic animals as animals (Figure 4f). Together, these results indicate that determined by histologic analysis and by measuring the impaired liver regeneration in dn-FGFR2IIIb mice is levels of mRNAs encoding interleukin-1b and tumor because of a defect in overcoming the G1 restriction necrosis factor a, type I and III collagens, and after hepatectomy. fibronectin (data not shown).

Severely reduced hepatocyte proliferation in transgenic mice after partial hepatectomy (PH) Discussion We next used our transgenic mice to investigate the role In this study, we generated and characterized transgenic of this receptor during liver regeneration. First, we mice expressing a dn-FGFR2IIIb mutant in hepatocytes determined the proliferation rate of hepatocytes after and intestinal epithelial cells. We observed a high PH. Animals received BrdU intraperitoneally 2 h before frequency of steatosis in old transgenic mice, which being killed. Liver sections were stained with an express high levels of the transgene. In contrast, in mice antibody directed against BrdU (Figure 4a–d). DNA with moderate or low levels of transgene expression, the synthesis reached its maximum 48 h posthepatectomy. frequency of fatty liver development was comparable to Representative results are shown for each group of mice that of wild-type mice. Although we cannot completely at this 48 h time point (Figure 4a–c). In all, 30% of rule out the possibility that the fatty liver phenotype is hepatocytes of wild-type mice (n ¼ 4; Figure 4a, d), but because of integration of the transgene into an only 5% of hepatocytes of the high expressor line (line endogenous gene that is important for liver function, 43; n ¼ 5) had incorporated BrdU (Figure 4b, d). The this seems unlikely, since the same phenotype was seen effect was dependent on the expression levels of the in both hemizygous and homozygous animals and was transgene, since an intermediate effect was seen in line more pronounced in the latter. Furthermore, the 109 (n ¼ 4; Figure 4c, d). phenotype occurred in the tissue where the transgene

Oncogene FGF in liver regeneration H Steiling et al 4384

Figure 3 Histological abnormalities in the livers of transgenic mice. (a)–(f): Parafﬁn sections (6 mm) of the liver from a 7-week-old wild-type mouse (wt.; a and c), a transgenic mouse of the same age (tg; line 43, b, d), a 12-month-old wild-type animal (e) and a transgenic animal of the same age (line 43, f) were stained with hematoxylin/eosin (HE). Paraformaldehyde-ﬁxed cryosections (6 mm) of livers from old wild-type (11-month old, g) and transgenic mice of line 43 (11-month old, h) were stained with Sudan III and counterstained with hematoxylin. Fat droplets appear red. Bars indicate 100 mmin(a, b), 50 mmin(c–f), and 25 mmin(g, h)

was expressed. Finally, all other abnormalities were The increased frequency of fatty liver development in observed in mice that express either high or low levels of aged mice that express high levels of dn-FGFR2IIIb the transgene. suggests that FGFs can at least partially prevent the

Oncogene FGF in liver regeneration H Steiling et al 4385

Figure 4 Liver regeneration is impaired in dn-FGFR2IIIb mice. The livers of 6–8-week-old wild-type (wt) and transgenic mice were removed at different time points after PH. (a)–(c): Mice were injected with BrdU and killed 2 h after injection. Liver sections of wild- type mice (a), hemizygous mice of line 43 (b) and hemizygous mice of line 109 (c) were stained for BrdU-labelled cells to monitor DNA synthesis. (d): BrdU-positive cells in three to six representative areas were counted for each genotype at each time point. Average values and s.d.s are indicated. (e): RNA was isolated from the liver of control and dn-FGFR2IIIb mice before and at different time points after PH. Samples of 10 mg were analysed by RNase protection assay for the expression of fgf1,c-myc,c-jun , cyclin D1 or gapdh.(f): Protein (10 mg) from nuclear extracts of mouse liver were analysed by immunoblotting for the expression of cyclin E and cyclin A before and at different time points after PH development of this pathological fat deposition. Pre- (Nonogaki et al., 1995), indicating that FGFs can vious studies demonstrated that FGF7 increases fatty directly regulate liver fatty acid metabolism and thus acid mobilization and hepatic triglyceride secretion prevent hepatic steatosis. Alternatively, FGFs could

Oncogene FGF in liver regeneration H Steiling et al 4386 protect the liver from harmful insults during postnatal tion (Arora et al., 2000; Behrens et al., 2002), indicating life that damage the liver and thus lead to metabolic that the early induction of primary response genes is abnormalities. It will be interesting to determine if functionally important. Our finding that expression of defects in the FGF/FGFR system are also associated c-jun and c-myc is not reduced after PH and even with fatty liver development in humans. increased or prolonged suggests that G0 exit can occur In spite of these abnormalities in aged mice, liver in dn-FGFR2IIIb transgenic mice. This is consistent function was not obviously affected in dn-FGFRIIIb with the results from other studies that revealed an transgenic mice. Furthermore, preliminary studies from important role of the proinflammatory cytokines tumor our laboratory revealed that a normal acute phase necrosis factor-a and interleukin-6, but not of growth response can be initiated by injection of interleukin-6 factors in the priming step (Cressman et al., 1996; (data not shown). Thus, FGFR signalling appears to be Yamada et al., 1997; Fausto et al., 2000). In the second dispensable for normal liver function, a finding which is phase of liver repair, growth factors are required for consistent with the lack of obvious abnormalities in the progression through the cell cycle (Loyer et al., 1996; liver of embryonic and newborn FGFR2IIIb null mice Fausto, 2000). Descriptive in vivo expression data and (de Moerlooze et al., 2000; C Dickson, personal results from functional in vitro studies suggested that communication). FGFs have been shown to play a transforming growth factor-a and hepatocyte crucial role in the initial stage of hepatogenesis (Jung growth factor (HGF) are the key regulators of liver et al., 1999). The normal liver development observed in repair (Loyer et al., 1996; Fausto, 2000), but this has dn-FGFR2IIIb transgenic mice is not contradictory to not yet been functionally proven in vivo. The results these findings, since the fabpl gene promoter is not presented in this study reveal a crucial role of FGFs active at this early stage of development (Sweetser et al., in this process. This effect is likely to be direct, since 1988). FGFs can directly stimulate hepatocyte proliferation The most interesting finding of our study was the (Housley et al., 1994; Strain et al., 1994). Furthermore, severe impairment in liver regeneration of dn- the effect is not because of reduced expression of HGF FGFR2IIIb transgenic mice. The type of FGF that is in our transgenic mice, since expression of this growth crucial for liver repair has not yet been identified. FGF7 factor was even higher in the injured liver of dn- is a candidate because of its potent mitogenic effect for FGFR2IIIb mice compared to control animals (data not hepatocytes (Housley et al., 1994; Strain et al., 1994). shown). Although we and others could not detect any FGF7 Expression of cyclin D1 was neither reduced nor transcripts within the first 3 days after PH (Kan et al., delayed after PH in dn-FGFR2IIIb transgenic mice, 1992 and data not shown), this type of FGF is produced and the mRNA levels of this cyclin were even enhanced by the spleen (Suzuki et al., 1993) and could thus reach before and after injury. By contrast, expression of the injured liver via the portal circulation. However, it is cyclin A, which is required during the S phase, was more likely that FGF1 is the responsible factor. strongly delayed in our transgenic animals. Although Previous expression studies suggested that FGF1 is induction of this cyclin did finally occur, it was involved in liver regeneration and hepatic differentiation significantly reduced in dn-FGFR2IIIb transgenic via the stem cell compartment, and levels of FGF1 mice. Most importantly, the increase in cyclin A after mRNA are increased after PH (Marsden et al., 1992 and 72 h was not accompanied by DNA replication as this study). However, a direct role of FGF1 in liver shown by the BrdU incorporation studies. Taken regeneration has not yet been demonstrated. together, our findings indicate that inhibition of FGFR In contrast to FGF7, which only binds FGFR2IIIb, signalling causes a specific G1/S checkpoint arrest FGF1 binds to all FGFR variants with high affinity. after PH. Therefore, it could also exert a mitogenic effect via FGFR4, another FGFR expressed by hepatocytes (Korhonen et al., 1992; Kan et al., 1999). Although mice deficient in FGFR4have smaller gallbladders and Materials and methods show enhanced bile acid production and secretion, liver regeneration after PH was not affected in these animals Generation and identification of transgenic mice (Yu et al., 2000), demonstrating that FGFR4is not the The truncated human FGFR2IIIb cDNA (Werner et al., 1994) FGFR that is crucial for liver regeneration. Based on was inserted into an expression cassette that includes nucleo- these observations, we suggest that FGFR2IIIb is the tides +3 to +2150 of the human growth hormone gene receptor that mediates this effect. downstream of nucleotides À596 to +21 of the rat fabpl gene Results from previous studies suggest that liver (Sweetser et al., 1988). Standard procedures were followed in regeneration occurs in two phases: priming that leads order to generate transgenic B6D2F2 mice. Mice were regularly tested and found to be negative for bacterial to exit from G0 and cell cycle progression (reviewed by Fausto, 2000). Priming causes activation of transcrip- pathogens and viruses, including mouse hepatitis virus and Helicobacter. tion factors such as NF-kB and STAT3, followed by Genomic DNA was analysed by Southern blot analysis, induction of primary response genes (Fausto, 2000). using the truncated FGFR2IIIb fragment as a probe, or by Interestingly, reduction in c-myc expression by antisense PCR (amplification of rFABPl-hFGFR2IIIb fragments: 50- oligonucleotides as well as targeted inactivation of the primer: 50-TGATCCTG-GCCATAAAGA-30 and 30-primer: c-jun gene in the liver caused impaired hepatic regenera- 50-TCTAATGTGGTATCCTCA-30.

Oncogene FGF in liver regeneration H Steiling et al 4387 Histological analysis provided by B Munz), mouse FGF3 (kindly provided by G Martin), mouse c-myc (nt 1629–1928; X01023), mouse c-jun Freshly isolated livers were separated into individual lobes, (kindly provided by P Angel) and mouse glyceraldehydepho- fixed in 4% paraformaldehyde and embedded in paraffin or sphate dehydrogenase (GAPDH; nt 566–685; NM_008084). tissue-freezing medium. Paraffin sections (6 mm) were stained Quantification of mRNA levels was performed with the NIH with hematoxylin/eosin; frozen sections (6 mm) were stained image program or by phosphoimaging. with Sudan III and counterstained with hematoxylin. All animal experiments were carried out with permission of the local authorities of Zu¨ rich or Hannover. Two-thirds hepatectomy The 8-week-old male mice were anesthetized by i.p. injection of Western blot analysis a combination of rompun–ketamine.After a small subxyphoid Membrane proteins from liver or nuclear extracts were incision, two-thirds hepatectomy was performed as described separated by SDS–PAGE, and transferred onto nitrocellulose (Lu¨ dde et al., 2001). At different time points after surgery, filters. The following antibodies were used: a polyclonal rabbit mice were killed, the remaining livers were removed, rinsed in antibody directed against a peptide from the extracellular ice-cold PBS, and part of the livers were immediately frozen in domain from FGFR1, which crossreacts with FGFR2 (Werner liquid nitrogen or tissue-freezing medium. The remaining livers et al., 1993), anti-Cyclin A (Santa Cruz Biotechnology, Santa were used to prepare nuclear extracts (Trautwein et al., 1998). Cruz, CA, USA) and anti-Cyclin E (Santa Cruz). Detection was performed with the enhanced chemoluminescence detec- Labelling cells in S phase tion system (Amersham, Braunschweig, Germany). BrdU (30 mg/g body weight) was injected i.p. 2 h before the animals were killed. Liver tissue was frozen in tissue-freezing Serology medium. Cryosections (5 mm) were fixed in acetone/methanol Mouse plasma was tested for alanine and aspartate amino- and stained according to the cell proliferation kit manual transferase and pseudocholinesterase activities, as well as (Amersham Pharmacia Biotech Inc., NJ, USA). glucose, triglyceride, cholesterol, total protein and albumin concentrations using a Hitachi PP-Modular clinical chemistry analyzer (Roche Diagnostics). The relative amounts of Acknowledgements albumin, a1-, a2-, b- and g-globulin fractions were determined by agarose gel electrophoresis. We thank Dr P Mo¨ ller, Ulm, Germany and Dr R Wanke, Munich, Germany, for help and advice with the liver histology, and Dr Clive Dickson, London, UK, for critically reading the RNase protection assay manuscript. This work was supported by grants from the RPA was carried out as described (Werner et al., 1993). Human Frontier Science Program (to SW), the Swiss National Template cDNAs: mouse FABPl [nt 53–232, Y14660], human Science Foundation (Grant No. 31-61358.00 to SW) and FGFR2IIIb (Werner et al., 1994), mouse FGFR2IIIb and the Deutsche Forschungsgemeinschaft (SFB 566, project B8 mouse FGF1 (Werner et al., 1993), mouse FGFR4(kindly to CT).

References

Arora V, Knapp DC, Smith BL, Statdﬁeld ML, Stein DA, Loyer P, Ilyin G, Cariou S, Glaise D, Corlu A and Guguen- Reddy MT, Weller DD and Iversen PL. (2000). Pharmacol. Guillouzo C. (1996). Prog Cell Cycle Res. 2, 37–47. Exp. Ther., 292, 921–928. Lu¨ dde T, Kubicka S, Plumpe J, Liedtke C, Manns MP and Behrens A, Sibilia M. David JP, Mohle-Steinlein U, Tronche Trautwein C. (2001). Oncogene, 20, 5264–5278. F, Schutz G and Wagner EF. (2002). EMBO J., 21, 1782– Marsden ER, Hu Z, Fujio K, Nakatsukasa H, Thorgeirsson 1790. SS and Evarts RP. (1992). Lab. Invest., 67, 427–433. Cressman DE, Greenbaum LE, DeAngelis RA, Ciliberto G, Nonogaki K, Pan XM, Moser AH, Staprans I, Feingold KR Furth EE, Poli V and Taub R. (1996). Science, 274, 1379– and Grunfeld C. (1995). Endocrinology, 136, 4278–4284. 1383. Ornitz DM and Itoh N. (2001). Genome. Biol., 2, 3005 de Moerlooze L, Spencer-Dene B, Revest J, Hajihosseini M, (reviews). Rosewell I and Dickson C. (2000). Development, 127, 483–492. Peters K, Werner S, Liao X, Wert S, Whitsett J and Williams Fausto N. (2000). J. Hepatol., 32, 19–31. L. (1994). EMBO J., 13, 3296–3301. Housley RM, Morris CF, Boyle W, Ring B, Biltz R, Tarpley Senaldi G, Shaklee CL, Simon B, Rowan CG, Lacey DL and JE, Aukerman SL, Devine PL, Whitehead RH and Pierce Hartung T. (1998). Hepatology, 27, 1584–1591. GF. (1994). J. Clin. Invest., 94, 1764–1777. Strain AJ, McGuinness G, Rubin JS and Aaronson SA. Hu Z, Evarts RP, Fujio K, Marsden ER and Thorgeirsson SS. (1994). Exp. Cell Res., 210, 253–259. (1995). Cell Growth Differ., 6, 1019–1025. Suzuki M, Itoh T, Osada H, Rubin JS, Aaronson SA, Suzuki Johnson DE and Williams LT. (1993). Adv. Cancer Res., 60, T, Koga N, Saito T and Mitsui Y. (1993). FEBS Lett., 328, 1–41. 17–20. Jung J, Zheng M, Goldfarb M and Zaret KS. (1999). Science, Sweetser DA, Birkenmeier EH, Hoppe PC, McKeel DW and 284, 1998–2003. Gordon JI. (1988). Genes Dev., 2, 1318–1332. Kan M, Wu X, Wang F and McKeehan WL. (1999). J. Biol. Trautwein C, Rakemann T, Malek NP, Plumpe J, Tiegs G and Chem., 274, 15947–15952. Manns MP. (1998). J. Clin. Invest., 101, 1960–1969. Korhonen J, Partanen J and Alitalo K. (1992). Int. J. Dev. Ueno H, Gunn M, Dell K, Tseng Jr A and Williams LT. Biol., 36, 323–329. (1992). J. Biol. Chem., 267, 1470–1476.

Oncogene FGF in liver regeneration H Steiling et al 4388 Werner S, Smola H, Liao X, Longaker MT, Krieg T, Yamada Y, Webber EM, Kirillova I, Peschon JJ and Fausto Hofschneider PH and Williams LT. (1994). Science, 266, N. (1997). Proc. Natl. Acad. Sci. USA, 94, 1441–1446. 819–822. Yu C, Wang F, Kan M, Jin C, Jones RB, Weinstein M, Werner S, Weinberg W, Liao X, Peters KG, Blessing M, Deng CX and McKeehan WL. (2000). J. Biol. Chem., 275, Yuspa SH, Weiner RL and Williams LT. (1993). EMBO J., 15482–15489. 12, 2635–2643.

Oncogene