Scatter Factor Promote Proliferation of Primary Human Hepatocytes and in Rodent Liver

GASTROENTEROLOGY 2012;142:897–906

BASIC AND TRANSLATIONAL—LIVER Protein Engineered Variants of Hepatocyte Growth Factor/Scatter Factor Promote Proliferation of Primary Human Hepatocytes and in Rodent Liver JACOB ROSS,* ERMANNO GHERARDI,‡,§ NOEMI MALLORQUI–FERNANDEZ,‡,§ MARCO BOCCI,‡,§ ANNA SOBKOWICZ,‡,§ MYRRDIN REES,储 ARTHUR ROWE,# STEPHAN ELLMERICH,¶ ISOBEL MASSIE,* JUNPEI SOEDA,* CLARE SELDEN,* and HUMPHREY HODGSON*

BACKGROUND & AIMS: Hepatocyte growth factor/scatter factor (HGF/SF) stimulates hepatocyte DNA synthesis and protects against apoptosis; in vivo it promotes liver regeneration and reduces fibrosis. However, its therapeutic value is limited by its complex domain structure, high cost of production, instability, and poor tissue penetration due to sequestration by heparin sulfate proteoglycans (HSPGs). METHODS: Using protein engineering techniques, we created a full-length form of HGF/SF (called HP21) and a form of the small, naturally occurring HGF/SF fragment, NK1 (called 1K1), which have reduced affinity for HSPG. We characterized the stability and proliferative and anti-apoptotic effects of these variants in primary human hepatocytes and in rodents. RESULTS: Analytical ultracentrifugation showed that 1K1 and NK1 were more stable than the native, full-length protein. All 4 forms of HGF/SF induced similar levels of DNA synthesis in human hepatocytes; 1K1 and NK1 required heparin, an HSPG analogue, for full agonistic activity. All the proteins reduced levels of Fas ligand–mediated apoptosis, reducing the activity of caspase-3/7 and cleavage of poly(adenosine diphosphate–ribose) polymerase. 1K1 was more active than NK1 in rodents; in healthy mice, 1K1 significantly increased hepatocyte DNA synthesis, and in mice receiving carbon tetrachloride, it reduced fibrosis. In rats, after 70% partial hepatectomy, daily administration of 1K1 for 5 days significantly increased liver mass and the bromodeoxyuridine labeling index compared with mice given NK1. CONCLUSIONS: 1K1, an engineered form of the small, naturally occurring HGF/SF fragment NK1, has reduced affinity for HSPG and exerts proliferative and antiapoptotic effects in cultured hepatocytes. In rodents, 1K1 has antifibrotic effects and promotes liver regeneration. The protein has better stability and is easier to produce than HGF/SF and might be developed as a therapeutic for acute and chronic liver disease. Keywords: Growth Factor; Cirrhosis; Mouse Model; Binding Mutant.

H

epatocyte growth factor (HGF)/scatter factor (SF) is a large multi-domain protein mediating cell growth, survival, and motility, playing important roles in tissue

regeneration and development.1–3 It acts on a variety of cell types, including hepatocytes, through the tyrosine kinase receptor met (UniProt: MET_HUMAN).4,5 HGF/SF (UniProt: HGF_HUMAN) consists of 2 polypeptide chains and 6 domains: the N-terminal domain (n) and 4 kringle domains (k1–k4) within the heavy chain (␣) and a serine protease homology domain within the light chain (␤)6,7(Figure 1A). The 2 chains are cleaved from a single ⬃80-kilodalton polypeptide precursor and are covalently linked by a disulfide bond.8,9 Endogenous HGF/SF is an important factor in liver regeneration and repair; its active form is cleaved from the precursor stored in the liver within minutes of partial hepatectomy,10 leading to phosphorylation of met on hepatocytes; animals deficient in hepatic met have markedly diminished repair after surgery.11 Exogenous HGF/SF initiates hepatocyte proliferation in intact animals, enhances regrowth after surgery, increases survival after injury in the presence and absence of preexisting liver disease, and prevents and reverses hepatic fibrosis.12–17 In vitro, HGF/SF both initiates DNA synthesis and protects against apoptosis of hepatocytes.18,19 HGF/SF has long been considered a potential therapeutic agent. However, the feasibility of HGF/SF-based therapies is limited; its complex structure and intricate level of posttranslational modification mandates use of mammalian expression systems to produce recombinant protein, therefore with limited yield and high costs.20 Additionally, exogenous HGF/SF is rapidly sequestered by heparan sulfate proteoglycans (HSPGs) widely expressed on cell surfaces and the extracellular matrix of the liver.21–24 The native molecule also has a tendency towards aggregation. These features may limit the amount of active growth factor available for sustained levels of receptor binding and activation. Engineered HGF/SF derivatives offer a means of circumventing these biological and manufacturing difficulties. Here we investigated 2 novel variants of HGF/SF (HP21 and 1K1) in which, by protein engineering, we have reAbbreviations used in this paper: BrdU, bromodeoxyuridine; EC50, half-maximal effective concentration; FasL, Fas ligand; HGF/SF, hepatocyte growth factor/scatter factor; HSPG, heparan sulfate proteoglycan; PARP, poly(adenosine diphosphate–ribose) polymerase. © 2012 by the AGA Institute 0016-5085/$36.00 doi:10.1053/j.gastro.2011.12.006

BASIC AND TRANSLATIONAL LIVER

*UCL Hepatology and ¶Centre for Amyloidosis, Royal Free Campus, University College London, London, England, United Kingdom; ‡MRC Centre, MRC, Cambridge, England, United Kingdom and 储Department of Surgery, North Hampshire Hospital, Basingstoke, Hampshire; #School of Biosciences, University of Nottingham, Sutton Bonington, Leicestershire, England, United Kingdom; and §Department of Molecular Medicine, University of Pavia, Pavia, Italy

898

ROSS ET AL

GASTROENTEROLOGY Vol. 142, No. 4

Figure 1. (A–C) Domain structure and (D–F) sedimentation analysis of HGF/SF, HP21, NK1, and 1K1. (A) Domain structure of HGF/SF-HP21 and NK1-1K1: n, N-terminal; k1– k4, kringle; sphd, serine protease homology domain. HP21 is a single point mutant of HGF/SF (R73E). 1K1 is a double mutant of NK1 (K132E:R134E). (B and C) Electrostatic potential of kringle 1 domains of NK1 and 1K1. Atomic coordinates of NK1 and 1K1 are from Protein Data Bank (PDB) accessions 1NK1 and 3MKP, respectively. Blue areas are positively charged, and red areas are negatively charged; figure drawn with Pymol (DeLano Scientific, San Carlos, CA). (D–F) Sedimentation velocity analysis of NK1, 1K1, and HGF/SF. Results of 2 samples for each protein analyzed after dialysis against PBS (day 0) or after a further 72-hour incubation at 37°C in PBS (day 3). For HGF/SF, a further sample is shown (high salt) that, before incubation at 37°C, was adjusted to a concentration of 1.0 mol/L NaCl. Data show plots of c(s) against s*20,w and insets show s*20,w (S), frictional ratios (fr), and molecular mass (m) values for main peak components. The species with S- values between 1.22 and 1.48 in HGF/SF samples represent a small molecular weight degradation product with high A280 absorptivity, enriched due to precipitation of intact HGF/SF following dialysis against PBS.

BASIC AND TRANSLATIONAL LIVER

shaped the binding to HSPG to improve the activity and/or tissue distribution of the growth factor. HGF/SF has 2 binding sites for HSPG: a high-affinity one in the n domain and a lower-affinity one in k1 (Figure 1A). HP21 is a full-length HGF/SF mutant (R73E) in which the high-affinity binding site in the n domain is disrupted through a reverse charge mutation of Arg73.21,25 1K1 (K132E:R134E) is a mutant of NK1, a naturally occurring splice variant of HGF/SF, encoding the n and k1 domains (Figure 1A), in which the low affinity binding site for HSPG of k1 is disrupted through reverse charge mutations of Lys132 and Arg134.26 This changes the electrostatic potential on the surface of k1 (compare Figure 1C [1K1] with Figure 1B [NK1]). Although NK1 and 1K1 only contain 2 of the 6 domains of the parent molecule (n and k1), they retain agonistic activity.27,28 They are convenient agents to develop, because they can be efficiently produced in yeast.29 We show that the engineered proteins HP21 and 1K1 have similar proliferative and anti-apoptotic effects to native HGF/SF protein in primary human hepatocyte cultures and confirm superior solution stability of 1K1. In vivo, we investigated the proliferative effects of the engineered HGF/SF proteins in rodents with intact livers and their regenerative effects after partial hepatectomy. Finally, we investigated the capacity of 1K1 to prevent carbon tetrachloride–induced liver fibrosis.

Materials and Methods Protein Production NK1 and 1K1 were produced in the yeast Pichia pastoris.25,28 HGF/SF was produced in the mouse myeloma line NS0

transfected with a full-length human HGF/SF complementary DNA. Proteins were purified by affinity chromatography on Heparin-Sepharose Fast Flow with a gradient of NaCl (0.15–2.0 mol/L) in 25 mmol/L phosphate buffer, pH 7.4, further purified by cation-exchange chromatography on MonoS (GE Heatlthcare UK Ltd, Little Chalfont, Buckinghamshire, England).30

Sedimentation Velocity Experiments Sedimentation velocity determined solution behavior and stability of NK1, 1K1, and HGF/SF. Proteins, dialyzed at room temperature against phosphate-buffered saline (PBS), were used for sedimentation velocity studies at 20°C using absorption at 280 nm in a Beckman XL-I analytical ultracentrifuge (Beckman Coulter, High Wycombe, England). After survey scans at 3000 rpm, the rotor was accelerated to 45,000 rpm. Two hundred scans at 2-minute/cell intervals were logged to disk using the Beckman ProteomeLab software. Data analysis used the SEDFIT program31 to generate c(s) profiles. Settings used were resolution ⫽ 200 and regularization ⫽ 0.684. The frictional ratio (F) was floated, as were the time-invariant and radialinvariant noise and baseline. Transformation of the c(s) profile into a c(M) profile enabled estimation of molecular weight values for “peaks” seen in the c(s) profiles.

Isolation and Culture of Primary Human Hepatocytes Primary human hepatocytes were isolated from surgical specimens as previously described,32 with informed consent from the patients and approval from the Clinical Ethics Committee at the Royal Free Hospital. Cells were released with gentle agitation in dispersal buffer (Williams’ E medium, Invitrogen, Paisley, Scotland; 10% fetal calf serum, Hyclone, Thermo Fisher Scientific, Loughborough, England; 0.01% deoxyribonuclease I), filtered, and centrifuged at 32g for 6 minutes and then plated after 2 further washes in complete Williams’ E medium (with

10% fetal calf serum) at 190,000 cells/well in collagen-coated 24-well plates (DNA studies) or 40,000 cells/well in 96-well plates (apoptosis/caspase-3/7 studies). Cells were attached overnight at 37°C, followed by washing/medium replacement.

In Vitro DNA Synthesis Assay Cells were stimulated with HGF/SF proteins at several concentrations, both in the presence of 10 ␮g/mL heparin (CP Pharmaceuticals, Wrexham, Wales) (HGF/HP21, 0 –10 nmol/L; NK1/1K1, 0 –30 nmol/L) and without heparin (HGF/HP21, 0 –10 nmol/L; NK1/1K1, 0 –300 nmol/L). At 18 and 26 hours, 1.875 ␮Ci/69.4kBq 3H-thymidine was added; at 42 hours, cells were harvested for isolation of total DNA and protein.

In Vitro Caspase-3/7, Poly(Adenosine Diphosphate–Ribose) Polymerase Cleavage, and Lactate Dehydrogenase Release Toxicity Assays HGF/SF proteins were used at a range of concentrations to determine the optimum dose for protection against Fas ligand–induced apoptosis (FasL, 20 ng/mL) (R&D Systems, Abingdon, England). Cells were stimulated with HGF/SF proteins (HGF/HP21, 10 nmol/L; NK1/1K1, 30 nmol/L) before or at the same time as addition of FasL (prevention) or after FasL (rescue). Controls received vehicle alone (PBS ⫹ 0.01% bovine serum albumin). Four hours after addition of FasL, cells were lysed and apoptosis induction markers caspase-3 and caspase-7 measured using the Caspase-Glo3/7 Assay Kit (Promega, Southampton, England); after 6 hours, poly(adenosine diphosphate– ribose) polymerase (PARP) cleavage was assessed using the Cleaved PARP ELISA Kit (Invitrogen) as per the manufacturer’s instructions. In other experiments, primary human hepatocytes were exposed to 100 nmol/L 1KI for 24 hours, and the supernatants were harvested for estimation of lactate dehydrogenase activity using the CytoTox 96 toxicity assay (Promega).

Phosphorylation of c-met, Akt, and Erk1/2 by Sodium Dodecyl Sulfate/Polyacrylamide Gel Electrophoresis and Western Blotting Primary human hepatocyte cultures were stimulated with HGF/SF, 1K1, or NK1 for 5 minutes at 37°C, lysed, and frozen. Samples were analyzed for total and phosphorylated Met, Akt, and Erk1/2 by Western analysis. Antibodies were from New England Biolabs (UK) Ltd (Hitchin, England) and Cell Signaling (Danvers, MA) unless otherwise stated: rabbit anti– phospho-Met (Tyr1234/1235) (D26) XP (c/n3077); mouse antiMet (L41G3) (c/n3148); rabbit anti–phospho-Akt (Ser473) (c/ n9271); rabbit anti-Akt (c/n9272); mouse anti–phospho-Erk (Sigma, M8159); and rabbit anti-Erk1/2 (Promega, V114). Horseradish peroxidase– conjugated secondary antibodies, goat antirabbit immunoglobulin G (Dako, Ely, England) and rabbit antimouse immunoglobulin G (Dako) were used for 1 hour at room temperature, and blots were developed with SuperSignal West Pico Chemiluminescent Substrate (Thermo Scientific, Loughborough, United Kingdom).

DNA Synthesis in Intact Mouse Liver All animals received humane care in accordance with the Guide for the Care and Use of Laboratory Animals,33 and experiments were performed under a project license from the UK Home Office. In experiment 1, female C57BL/6 mice (18 –21 g) were injected intravenously (IV) via the tail vein with 1K1, HP21

THERAPEUTIC POTENTIAL OF ENGINEERED HGF PROTEINS

899

(800 ␮g/kg), or saline alone, twice a day for 2.5 days (total 5 injections, total dose 4 mg/kg). In experiment 2, female Balb/c mice (18 –21 g) were injected with 1K1 (800 ␮g/kg) or saline alone, either via the IV or intraperitoneal (IP) route, as per experiment 1. In all cases, an 80mg/kg flash label of bromodeoxyuridine (BrdU; Sigma, Poole, England) was given IP with the final injection, and animals were killed 6 hours later with the livers fixed in formalin for histology/BrdU immunohistochemistry. In further experiments, 18- to 21-g Balb/c mice (n ⫽ 4 in each group) were injected IV with 950, 4000, or 8000 ␮g/kg 1K1 or saline vehicle in a 200␮L volume and after 24 hours were killed with collection of liver, spleen, muscle, lung, and kidney specimens for histologic examination. Blood was collected by cardiac puncture centrifuged at 2000g, plasma separated, and stored at ⫺20°C for estimation of serum amyloid A component using the Phase Mouse SAA Elisa Kit (Tridelta Development Ltd, Co, Kildare, Ireland).

Liver Growth After 70% Hepatectomy in the Rat Male Sprague–Dawley rats (200 –225 g) were anesthetized under isoflurane inhalation and subcutaneous buprenorphine injection (60 ␮g/kg) (Alstoe, York, England). Laparotomy was performed, followed either by 70% hepatectomy removing left and median lobes34 or by sham manipulation of the liver. In some animals, BrdU (80 mg/kg) was administered IP 22 hours later to “flash label” cells in S-phase, and animals were killed 2 hours later. In other animals, from the day of surgery, IV injections of 1 mg/kg NK1 or 1K1 or saline alone were administered daily for 4 days. On day 5, these animals received BrdU as previously described and were killed 2 hours later. Remnant liver was excised, blotted, and weighed, and samples were formalin fixed for histology/BrdU immunohistochemistry and snap frozen for total protein content analysis.

Prevention of CCl4-Induced Liver Fibrosis in the Mouse Male Balb/c mice (18 –22 g) were treated with escalating IP doses of CCl4 in olive oil 3 times a week for 4 weeks: 0.125 (week 1), 0.25 (week 2), and 0.5 mL/kg CCl4 (weeks 3 and 4). On alternate days, twice a week, 1K1 (500 ␮g/kg) or saline was administered IP. Animals were killed after 4 weeks, serum was sampled for alanine aminotransferase (ALT) levels, and liver tissue was fixed in formalin for paraffin embedding for Sirius Red staining.

BrdU Immunohistochemistry Mouse anti-BrdU (Dako) at 1:100 dilution was used as described elsewhere.35 The percentage of BrdU-positive nuclei was measured by double-blinded counts of a total of ⬃3000 nuclei across 10 fields of view per animal (n ⫽ 6; 20⫻ magnification).

Sirius Red Histologic Collagen Stain Four-micrometer sections of mouse liver were deparaffinized in xylene and rehydrated in ethanol (100%, 70%) and then tap water. Sections were treated with Sirius Red (Sigma) 0.1% in saturated aqueous picric acid for 8 minutes. Fibrotic area was calculated by quantification of red-stained areas (excluding blood vessels) on 5 double-blinded random images per animal (n ⫽ 8; 4⫻ magnification) from different lobes of the liver using ImageJ software (National Institutes of Health, Bethesda, MD).

BASIC AND TRANSLATIONAL LIVER

April 2012

900

ROSS ET AL

Measurement of Total Protein and DNA From Whole Liver Total liver protein and total liver DNA were measured as described previously.35

Results Stability: Solution Behavior of NK1, 1K1, and HGF/SF We used analytical ultracentrifugation to assess protein monodispersity and stability in solution. When NK1 and 1K1 proteins were dialyzed against a buffer of physiological ionic strength, the proteins were recovered in high yield and appeared monodisperse (⬎90% for NK1 and ⬎95% for 1K1 in the mean peak component) in sedimentation velocity (Figure 1D–F), with molecular mass values very close to those predicted from sequence (21,135 and 21,109 daltons for NK1 and 1K1, respectively) (Figure 1D and E). Incubation of NK1 and 1K1 for 3 days in PBS at 37°C did not cause aggregation (Figure 1D and E). In contrast, dialysis of recombinant HGF/SF against PBS led to precipitation of ⬃50% of the protein and a further ⬃20% was lost in the lowspeed (3000 rpm) phase of the experiment, confirming the presence of large oligomers/aggregates. HGF/SF

GASTROENTEROLOGY Vol. 142, No. 4

remaining in solution behaved as an HGF/SF dimer (125,442 daltons against a value of 72,313 daltons obtained from velocity of HGF/SF in 1.0 mol/L NaCl and against a molecular mass of 79,661 daltons calculated from sequence) (Figure 1F). Furthermore, on incubation in PBS for 3 days at 37°C, sample polydispersity of HGF/SF increased with the peak component, yielding a molecular mass of 242,692 daltons (Figure 1F). These results confirm the instability and tendency to aggregate of HGF/SF in physiological buffers but clearly show that under identical conditions, the truncated variant NK1 and its engineered derivative 1K1 are monodisperse and stable.

In Vitro DNA Synthesis In freshly isolated human hepatocytes (Figure 2 and Supplementary Figure 1), all the growth factors induced DNA synthesis significantly greater than that of the basal level. HGF/SF induced maximally a 21.8-fold increase greater than basal (mean of three experiments on different livers) (Figure 2; P ⬍ .01); for HP21, the maximal increase was 11.5-fold (Supplementary Figure 1), whereas 1K1 and NK1 induced more modest increases of 4.5fold and 3.7-fold, respectively (Figure 2; P ⬍ .01). The

BASIC AND TRANSLATIONAL LIVER

Figure 2. DNA synthesis in primary human hepatocytes in response to HGF/SF, NK1, and 1K1, with (right) and without (left) 10 ␮g/mL heparin. Data from ⱖ 3 livers per curve, taken from a total of 5 liver preparations. Values are expressed as mean (n ⫽ 3) ⫾ range. Statistics are from a 2-tailed t test, and asterisks indicate significance compared with basal, corresponding to the least significant data point for each concentration (*P ⬍ .05, **P ⬍ .01; ***P ⬍ .001). Results are expressed as disintegrations per minute (DPM) 3H in DNA per microgram protein.

April 2012

THERAPEUTIC POTENTIAL OF ENGINEERED HGF PROTEINS

Anti-apoptotic Effects To assess anti-apoptotic effects on primary human hepatocytes, caspase-3/7 activity was measured in cultures treated with the HGF/SF proteins 1–2 hours before, at the same time as (prevention models), or 1 hour after (rescue model) FasL. In the absence of added growth factors, the average induction of caspase-3/7 activity by FasL was 2.23-fold greater than the basal (Figure 3). In the preven-

tion models, all 4 proteins significantly reduced FasLmediated caspase-3/7 activity (Figure 3A–C; P ⬍ .001). For 1K1 and NK1, this reduction was nearly to that of basal (ranging from 1.14-fold to 1.36-fold over basal), whereas for HGF/SF and HP21 it was to a level statistically indistinguishable from basal (HGF/SF in all prevention models and HP21 when given 1–2 hours before; P ⬎ .05 in all cases). In the rescue model, all 4 proteins significantly reduced caspase-3/7 activity, but not to basal levels (Figure 3D: HGF/SF, 1.44-fold greater than basal, P ⬍ .05; HP21, 1.48-fold greater than basal, P ⬍ .05; NK1, 1.46-fold greater than basal, P ⬍ .01; 1K1, 1.43-fold greater than basal, P ⬍ .01). In contrast to the effects on DNA synthesis, addition of heparin did not further increase the antiapoptotic activity of 1K1 and NK1 (data not shown). The anti-apoptotic effect of all 4 proteins was confirmed by inhibition of PARP cleavage in response to FasL (Figure 3E and F) at 6 hours, reflecting a later event in the apoptotic pathway.

DNA Synthesis In Vivo Intact mouse liver. The effects of the HGF/SF proteins as inducers of DNA synthesis in intact mouse liver were investigated (Figure 4). In healthy C57BL/6 mice, IV-administered 1K1 induced a 3.0-fold increase in hepatocyte DNA synthesis over vehicle alone (Figure 4A; P ⬍ .01), whereas HP21 only induced a 1.5-fold induction (not significant). In healthy Balb/c mice, 1K1 administered via the IP and IV routes induced 8.5-fold and 11.5-fold increases over vehicle (each P ⬍ .01), respectively (Figure 4B–D). Effects after 70% hepatectomy. In animals undergoing laparotomy and liver manipulation without hepatic resection, after 24 hours there was a 4-fold increase in the proportion of hepatocytes labeled with BrdU (0.25% ⫾ 0.06% vs 1.07% ⫾ 0.41%; P ⬍ .05) in animals receiving 1K1 compared with control (vehicle only) injections. The proportion of hepatocytes flash labeled in 1K1-treated animals at 24 hours after 70% partial hepatectomy, the peak of the regenerative response, was 30.5% ⫾ 6.7% compared with control treated animals (26.1% ⫾ 7.6%). This difference was not statistically significant. However, when 1K1 was administered daily for 4 days after 70% partial hepatectomy, liver mass and the proliferative rate of hepato-

Table 1. EC50 Values and Corresponding Values for Stimulation Over Basal, for Each Engineered HGF Protein, Calculated From DNA Synthesis Dose Responses in Primary Human Hepatocytes With heparin (10 ␮g/mL)

Without heparin

HGF HP21 NK1 1K1

EC50 (nmol/L)

Fold stimulation over basal at EC50 (disintegrations per minute/␮g protein)

EC50 (nmol/L)

Fold stimulation over basal at EC50 (disintegrations per minute/␮g protein)

0.018 ⫾ 0.012 0.012 ⫾ 0.003 — 2.0 ⫾ 0.6

13.0 ⫾ 9.4 6.6 ⫾ 1.2 — 2.7 ⫾ 0.8

0.48 ⫾ 0.18 0.15 ⫾ 0.09 1.1 ⫾ 0.4 2.2 ⫾ 1.1

7.1 ⫾ 4.7 7.4 ⫾ 3.6 7.6 ⫾ 5.4 7.3 ⫾ 1.8

NOTE. There was no distinct dose response observed for NK1; therefore, no EC50 value was obtained. Values are means from at least 3 liver preparations each ⫾ range.

BASIC AND TRANSLATIONAL LIVER

maximal activity of 1K1 and NK1 was at 10 –30 nmol/L ⬃100-fold higher than that of HGF/SF and HP21 (both at 0.3 nmol/L). We confirmed (Supplementary Figure 2) that the small HGF/SF engineered proteins 1K1 and NK1 phosphorylated the c-met receptor and AKT in a dose-dependent fashion. Because the cell isolation process to generate hepatocytes for in vitro study has the potential to destroy or diminish the functionality of the HSPG normally present on hepatocyte surfaces, we investigated the effect of free molecular heparin as a structural and functional mimic of HSPG function. In the presence of 10 ␮g/mL heparin, maximal induction of DNA was reduced nearly 2-fold for HGF/SF from 21.8-fold to 12.6-fold (Figure 2) but as expected not for the HP21 mutant (11.0-fold compared with 11.6-fold; Supplementary Figure 1). Strikingly, in the presence of heparin, maximal induction of DNA synthesis by 1K1 and NK1 markedly increased to 17.3-fold and 11.9-fold, respectively (Figure 2; P ⬍ .001), to a similar magnitude to that found with the full-length proteins. For native HGF/SF, the concentration at which maximal DNA synthesis occurred was 10-fold higher in the presence of heparin compared with that without heparin, whereas for the other growth factors the presence of heparin did not significantly alter the concentration at which maximal effect was achieved. The half-maximal (EC50) values for the HGF/SF proteins (concentrations at which induction of DNA synthesis is half that of the maximum) are shown in Table 1. Without heparin, the EC50 value for 1K1 was ⬃100-fold higher than that for the full-length HGF/SF and HP21 proteins (2.0 compared with 0.018 and 0.012 nmol/L, respectively). However, with heparin, the EC50 value for HGF/SF and HP21 increased by 27-fold and 13-fold to 0.48 and 0.15 nmol/L, respectively, whereas that of 1K1 remained unchanged.

901

902

ROSS ET AL

GASTROENTEROLOGY Vol. 142, No. 4

Figure 3. Reduction in FasL-induced apoptosis in response to HGF/SF proteins in primary human hepatocytes. (A–C) Prevention models: caspase-3/7 activity, with HGF/SF proteins given 2 hours or 1 hour before, or at the same time as FasL insult, respectively. (D) Rescue model: HGF/SF proteins given 1 hour after FasL. (E and F) Prevention models: PARP cleavage in cells treated with HGF/SF proteins 2 hours before or at the same time as FasL insult, respectively. Data from one representative experiment, values are mean (n ⫽ 6) ⫾ SD. Statistics are from a 2-tailed t test (*P ⬍ .05, **P ⬍ .01; ***P ⬍ .001). LU, luminescence units.

BASIC AND TRANSLATIONAL LIVER

cytes, measured at the time the animals were killed on day 5, were significantly greater in the animals receiving 1K1. 1K1 increased the liver mass/body mass ratio (g tissue/g body weight) by 12.5% over vehicle (Figure 5D; P ⬍ .05), and at 4 days there was 36% greater hepatocyte DNA synthesis (over the 6 hours before the animals were killed) compared with rats receiving vehicle alone (Figure 5A–C; P ⬍ .01). Total protein content was increased by 13.7% (P ⬍ .05) in 1K1-treated over saline-treated animals, and a similar 11.5% increase in total DNA content was observed, although it was not statistically significant (Figure 5 E and F). The increases in liver mass and hepatocyte proliferation rate over control animals observed with NK1 administration were not significant.

Prevention of CCl4-Induced Fibrosis in the Mouse To investigate protective effects during chronic liver damage, 1K1 was administered during 4 weeks of CCl4-induced liver fibrosis in Balb/c mice. After the animals were killed, 1K1 had significantly reduced liver fibrosis by 19% compared with vehicle alone (Figure 6A–C; P ⬍ .05), assessed from the fibrotic area in Sirius Red–

stained sections. ALT levels at the time the animals were killed were reduced by 30% compared with vehicle (Figure 6D; P ⬍ .05).

Hepatic and Systemic Tolerability of 1K1 The possibility that 1K1 might induce liver cell necrosis was explored by exposing primary human hepatocytes in vitro to 1K1 for 24 hours. There was no increase in release of lactate dehydrogenase into the medium greater than that from incubation with vehicle alone. Further, mice were injected IV with 950 ␮g/kg 1K1 (the dose used to show the in vivo mitogenic effect on whole liver; Figure 4) or with 4000 ␮g/kg 1K1, 8000 ␮g/kg 1K1, or vehicle alone. Histologic assessment of liver, lung, muscle, and kidney showed no abnormality at the time the animals were killed. There was no acute phase response to 1K1 evidenced by unchanged levels of serum amyloid A component in the blood (mean ⫾ SD: vehicle, 8.05 ⫾ 1.9 ␮g/mL; 950 ␮g/kg, 8.39 ⫾ 2.0 ␮g/mL; 4000 ␮g/kg, 6.04 ⫾ 0.62 ␮g/mL; 8000 ␮g/kg, 8.74 ⫾ 1 ␮g/mL). Others report normal serum amyloid A levels in plasma of Balb/c mice to be less than 12 ␮g/mL. Thus, no adverse responses to 1K1 were detected.

April 2012

THERAPEUTIC POTENTIAL OF ENGINEERED HGF PROTEINS

903

Discussion Protein engineering has the potential to improve natural proteins for therapeutic applications and thus impact on the prospect of harnessing the activity of HGF/SF for tissue and organ regeneration. Here we showed that the HP21 and 1K1 engineered variants of HGF/SF retain agonistic activity in vitro and have therapeutic benefit in vivo: specifically, in primary human hepatocytes, induction of DNA synthesis and prevention of apoptosis, and in rodents, induction of DNA synthesis in intact liver, enhancement of liver growth following 70% hepatectomy, and reduction of liver fibrosis in mice treated with CCl4. A strength of these observations is that the in vitro effects on DNA synthesis and apoptosis were assessed on human hepatocytes. These results were complemented by studies in vivo in the mouse and rat. Human HGF/SF binds mouse met with full affinity, and mouse models of disease have provided the preclinical setting for testing the activity of HGF/SF and derivatives for therapy.36 The results of our study therefore offer relevant information toward the prospective use of these proteins for the treatment of human liver disease. The in vitro comparison of the effects of native HGF/ SF, the natural fragment NK1, and the 2 engineered proteins shows significant differences in potency and the dependency of the truncated proteins (NK1 and 1K1) on the presence of heparin, which will mimic the HSPG present in vivo. HSPGs regulate the activity of HGF/SF at multiple levels. HSPGs constitute a storage form of inactive (single-chain) HGF/SF in the tissue whence the active (2-chain) species can be rapidly generated on liver damage.10 Binding of HGF/SF to HSPG can enhance directly the activity of the growth factor.37 In freshly isolated

human hepatocytes, heparin negatively regulated the DNA synthesis–promoting effect of native HGF/SF but had little effect on this activity in HP21 full-length mutant, the interaction of which with HSPG has been diminished by point mutation. In contrast, the stimulatory effects on DNA synthesis of the truncated variants 1K1 and NK1 were significantly enhanced by heparin, in line with the results of earlier studies.26 These findings likely reflect changes in cell surface HSPG density on freshly isolated hepatocytes consequent on collagenase-mediated liver digestion during primary hepatocyte isolation. The truncated variants 1K1 and NK1 can be readily produced in yeast cells, unlike HGF/SF, and we show here that both proteins are monodisperse and stable in physiological buffers, unlike HGF/SF (Figure 1), and activate the classic downstream pathways of the parent molecule. Of the variants we assessed, 1K1 emerges as a strong candidate for further development as a therapeutic agent. We showed here that this small engineered protein has growth-promoting functions similar to those of HGF/SF. Importantly, in vivo, free heparin was not required for the action of 1K1 due to the presence of fully functional HSPGs on cell surfaces and in the extracellular matrix, evidenced by the capacity of the 1K1 fragment to induce DNA synthesis when injected into normal rodents with intact livers, as well as to enhance liver growth after massive liver resection. It is possible that exogenous heparin might further enhance these actions in vivo. Among the potential HGF/SF variants assessed for their ability to induce DNA synthesis in vivo, the abilities of the fulllength native molecule have been repeatedly shown; HP21 failed to induce DNA synthesis, at least at the dose we used, in intact liver (Figure 4), and NK1 failed to significantly enhance liver regeneration in rodents. Thus, 1K1

BASIC AND TRANSLATIONAL LIVER

Figure 4. DNA synthesis in intact, healthy mouse liver. (A) DNA synthesis in C57BL/6 mice in response to HP21 and 1K1, administered IV. Mean values (n ⫽ 3) ⫾ range; statistics are from a 2-tailed t test (**P ⬍ .01). (B) DNA synthesis in Balb/c mice in response to 1K1, administered IP and IV. Mean values (vehicle IP, n ⫽ 2; vehicle IV, n ⫽ 2; 1K1 IP, n ⫽ 2; 1K1 IV, n ⫽ 3) ⫾ range. (C and D) Liver sections immunostained for BrdU in Balb/c mice receiving vehicle and 1K1 (IP).

904

ROSS ET AL

GASTROENTEROLOGY Vol. 142, No. 4

Figure 5. Enhanced liver regeneration in rats by engineered HGF/SF proteins 4 days after 70% hepatectomy. (A–C) DNA synthesis in liver on day 4 sections stained for BrdU. Animals treated with (A) saline vehicle or (B) 1K1 daily after surgery. (C) Percent BrdU-positive hepatocytes on day 4. (D) Liver size on day 4 (liver mass:body mass). (E) Total liver protein on day 4 (total protein:body mass). (F) Total liver DNA on day 4 (total DNA: body mass). Values are mean (n ⫽ 6) ⫾ SD. Statistics are from 2-tailed t test (**P ⬍ .01; *P ⬍ .05).

BASIC AND TRANSLATIONAL LIVER

has the benefits of stability in solution, small size, aiding tissue penetration, and mutations in the negatively regulating HSPG binding site, making it perhaps the most suitable of the HGF/SF variants for growth promoting activity and efficacy in vivo. Although this study did not address a full toxicological analysis, there was no evidence of direct hepatocyte or liver toxicity, nor was systemic administration of doses substantially higher than that required for the beneficial effects associated with any evidence of induction of an acute phase response, as a correlate of systemic toxicity. 1K1 also showed in vitro a marked protective effect against FasL-induced apoptosis of primary hepatocytes, as does native HGF/SF,15 with the important ability to be effective both before and after administration of the apoptotic agent. When the HGF/SF proteins were given before/at the same time as FasL, caspase-3/7 activity was decreased to basal levels under HGF/SF and HP21 and to levels close to basal under NK1 and 1K1 (Figure 3). A second marker of apoptosis, PARP cleavage, confirmed these results. Thus, 1K1 has the capacity to promote proliferation and survival of hepatocytes in vitro, increase liver DNA and mass acutely in vivo, and enhance acute regenerative

repair. Furthermore, in a model of chronic disease, 1K1 reduced fibrosis and was also associated with evidence of less hepatocellular necrosis after 4 weeks of continuous treatment, evidenced by a reduction in circulating serum transaminase levels, reflecting previous observations with full-length HGF/SF.17 Further work will define the effects of greater doses/longer duration of administration and the effects on reversal of fibrosis. There are a series of clinical reports from China on the effect of the native recombinant HGF/SF molecule in patients with acute and chronic liver failure. Indeed, more than 5000 patients have been treated in clinical trials.38 A recent meta-analysis commented on the relatively poor quality of these trials but highlighted nonetheless that improvement in survival was almost universally reported.38 As an agent for development and investigation in human disease, the experiments reported here indicate that engineered analogues of HGF/SF, and particularly 1K1, offer substantial promise as therapeutic agents not only in promoting repair and regeneration but in preventing fibrosis. Compared with the native molecule, the greater ease of production using industrially applicable expression systems, and the smaller molecular size aiding tissue penetration, provide strong arguments for phase 1 clinical studies with 1K1.

April 2012

THERAPEUTIC POTENTIAL OF ENGINEERED HGF PROTEINS

905

Supplementary Material Note: To access the supplementary material accompanying this article, visit the online version of Gastroenterology at www.gastrojournal.org, and at doi: 10.1053/j.gastro.2011.12.006. References 1. Stoker M, Gherardi M, Perryman M. Scatter factor is a fibroblastderived modulator of epithelial cell mobility. Nature 1987;327: 239 –242. 2. Huh CG, Factor VM, Sánchez A, et al. Hepatocyte growth factor/ c-met signaling pathway is required for efficient liver regeneration and repair. Proc Natl Acad Sci U S A 2004;101:4477– 4482. 3. Zhang YW, Vande Woude GF. HGF/SF-met signaling in the control of branching morphogenesis and invasion. J Cell Biochem 2003; 88:408 – 417. 4. Bottaro D, Rubin J, Faletto D, et al. Identification of the hepatocyte growth factor receptor as the c-met proto-oncogene product. Science 1991;251:802– 804. 5. Bladt F, Riethmacher D, Isenmann S, et al. Essential role for the c-met receptor in the migration of myogenic precursor cells in the limb bud. Nature 1995;376:768 –771. 6. Gherardi E, Sandin S, Petoukhov MV, et al. Structural basis of hepatocyte growth factor/scatter factor and MET signalling. Proc Natl Acad Sci U S A 2006;103:4046 – 4051. 7. Gherardi E, Hartmann G, Hepple J, et al. Domain structure of hepatocyte growth factor/scatter factor (HGF/SF). Ciba Found Symp 1997;212:84 –93; discussion 93–104. 8. Miyazawa K, Shimomura T, Kitamura N. Activation of hepatocyte growth factor in the injured tissues is mediated by hepatocyte growth factor activator. J Biol Chem 1996;271:3615–3618. 9. Miyazawa K, Tsubouchi H, Naka D, et al. Molecular cloning and sequence analysis of cDNA for human hepatocyte growth factor. Biochem Biophys Res Commun 1989;163:967–973. 10. Mars WM, Liu ML, Kitson RP, et al. Immediate early detection of urokinase receptor after partial hepatectomy and its implications for initiation of liver regeneration. Hepatology 1995;21:1695– 1701.

11. Borowiak M, Garratt AN, Wustefeld T, et al. Met provides essential signals for liver regeneration. Proc Natl Acad Sci U S A 2004;101: 10608 –10613. 12. Patijn GA, Lieber A, Schowalter DB, et al. Hepatocyte growth factor induces hepatocyte proliferation in vivo and allows for efficient retroviral-mediated gene transfer in mice. Hepatology 1998;28: 707–716. 13. Xue F, Takahara T, Yata Y, et al. Hepatocyte growth factor gene therapy accelerates regeneration in cirrhotic mouse livers after hepatectomy. Gut 2003;52:694 –700. 14. Kosai K, Matsumoto K, Funakoshi H, et al. Hepatocyte growth factor prevents endotoxin-induced lethal hepatic failure in mice. Hepatology 1999;30:151–159. 15. Kosai K, Matsumoto K, Nagata S, et al. Abrogation of Fas-induced fulminant hepatic failure in mice by hepatocyte growth factor. Biochem Biophys Res Commun 1998;244:683– 690. 16. Masunaga H, Fujise N, Shiota A, et al. Preventive effects of the deleted form of hepatocyte growth factor against various liver injuries. Eur J Pharmacol 1998;342:267–279. 17. Matsuda Y, Matsumoto K, Yamada A, et al. Preventive and therapeutic effects in rats of hepatocyte growth factor infusion on liver fibrosis/cirrhosis. Hepatology 1997;26:81– 89. 18. Gohda E, Yamasaki T, Tsubouchi H, et al. Biological and immunological properties of human hepatocyte growth factor from plasma of patients with fulminant hepatic failure. Biochim Biophys Acta 1990;1053:21–26. 19. Schulze-Bergkamen H, Brenner D, Krueger A, et al. Hepatocyte growth factor induces Mcl-1 in primary human hepatocytes and inhibits CD95-mediated apoptosis via Akt. Hepatology 2004;39: 645– 654. 20. Park JS, Kim H, Park J, et al. Overproduction of recombinant human hepatocyte growth factor in Chinese hamster ovary cells. Protein Expr Purif 2010;70:231–235. 21. Hartmann G, Prospero T, Brinkmann V, et al. Engineered mutants of HGF/SF with reduced binding to heparan sulphate proteoglycans, decreased clearance and enhanced activity in vivo. Curr Biol 1998;8:125–134. 22. Catlow KR, Deakin JA, Wei Z, et al. Interactions of hepatocyte growth factor/scatter factor with various glycosaminoglycans re-

BASIC AND TRANSLATIONAL LIVER

Figure 6. Preventative effects of 1K1 in CCl4-induced liver fibrosis in Balb/c mice. Mice were treated for 4 weeks with CCl4 while receiving 1K1 or saline (vehicle) alone. Sirius Red collagenstained liver section at 4 weeks in mouse receiving (A) vehicle or (B) 1K1. (C) Quantification of redstained fibrotic area (excluding blood vessels) in CCl4-treated mice. (D) Serum ALT level in CCl4-treated mice at 4 weeks. Values are mean (n ⫽ 8) ⫾ SEM. Statistics are from a 2-tailed t test (*P ⬍ .05).

906

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33. 34.

ROSS ET AL veal an important interplay between the presence of iduronate and sulfate density. J Biol Chem 2008;283:5235–5248. Rubin JS, Day RM, Breckenridge D, et al. Dissociation of heparan sulfate and receptor binding domains of hepatocyte growth factor reveals that heparan sulfate-c-met interaction facilitates signaling. J Biol Chem 2001;276:32977–32983. Zioncheck TF, Richardson L, Liu J, et al. Sulfated oligosaccharides promote hepatocyte growth factor association and govern its mitogenic activity. J Biol Chem 1995;270:16871–16878. Lietha D, Chirgadze DY, Mulloy B, et al. Crystal structures of NK1-heparin complexes reveal the basis for NK1 activity and enable engineering of potent agonists of the MET receptor. EMBO J 2001;20:5543–5555. Schwall RH, Chang LY, Godowski PJ, et al. Heparin induces dimerization and confers proliferative activity onto the hepatocyte growth factor antagonists NK1 and NK2. J Cell Biol 1996;133: 709 –718. Matsumoto K, Takehara T, Inoue H, et al. Deletion of kringle domains or the N-terminal hairpin structure in hepatocyte growth factor results in marked decreases in related biological activities. Biochem Biophys Res Commun 1991;181:691– 699. Chirgadze DY, Hepple JP, Zhou H, et al. Crystal structure of the NK1 fragment of HGF/SF suggests a novel mode for growth factor dimerization and receptor binding. Nat Struct Biol 1999;6:72–79. Mack M, Wannemacher M, Hobl B, et al. Comparison of two expression platforms in respect to protein yield and quality: Pichia pastoris versus Pichia angusta. Protein Expr Purif 2009;66:165–171. Gherardi E, Gray J, Stoker M, et al. Purification of scatter factor, a fibroblast-derived basic protein that modulates epithelial interactions and movement. Proc Natl Acad Sci U S A 1989;86:5844 –5848. Schuck P. Size distribution analysis of macromolecules by sedimentation velocity ultracentrifugation and Lamm equation modeling. Biophys J 2000;78:1606 –1619. Saich R, Selden C, Rees M, et al. Characterization of pro-apoptotic effect of liver failure plasma on primary human hepatocytes and its modulation by molecular adsorbent recirculation system therapy. Artif Organs 2007;31:732–742. National Academy of Sciences. Guide for care and use of laboratory animals. 2011;11–124. Higgins GM, Anderson RM. Experimental pathology of the liver: I. Restoration of the liver of the white rat following partial surgical removal. Arch Pathol 1931;12:186 –202.

GASTROENTEROLOGY Vol. 142, No. 4 35. Malik R, Mellor N, Selden C, et al. Triiodothyronine enhances the regenerative capacity of the liver following partial hepatectomy. Hepatology 2003;37:79 – 86. 36. Nakamura T, Sakai K, Matsumoto K. Hepatocyte growth factor twenty years on. J Gastroenterol Hepatol 2011;26(Suppl 1):188 – 202. 37. Delehedde M, Lyon M, Vidyasagar R, et al. Hepatocyte growth factor/scatter factor binds to small heparin-derived oligosaccharides and stimulates the proliferation of human HaCaT keratinocytes. J Biol Chem 2002;277:12456 –12462. 38. Cui YL, Meng MB, Tang H, et al. Recombinant human hepatocyte growth factor for liver failure. Contemp Clin Trials 2008;29:696 – 704.

Received September 7, 2011. Accepted December 1, 2011. Reprint requests Address requests for reprints to: Humphrey Hodgson, DM, FMedSci, UCL Hepatology, Royal Free Campus, University College London, London, NW3 2PF, England, United Kingdom; e-mail: [email protected]; and Ermanno Gherardi, MD, PhD, MRC, Centre, MRC, Hills Road, Cambridge, CB2 2QH, England, United Kingdom; e-mail: [email protected]. Acknowledgments The authors thank Sherri Chalmers, Pat Blake, Dave Brown, and Mark Neal for their laboratory administrative assistance. Conflicts of interest The 1K1 (and similar) variants of HGF/SF are the subject of an MRC patent (US 7,179,786), and Ermanno Gherardi is one of 4 named inventors (Ermanno Gherardi, Daniel Lietha, Thomas L. Blundell, and Dimitry Y. Chirgadze). All the rights from the patent are assigned to the MRC and pursued by MRC Technology. The remaining authors disclose no conflicts. Funding Supported by a grant from the Wellcome Trust (grant no. 2291: 081377). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

BASIC AND TRANSLATIONAL LIVER