2 by hepatocyte growth factor in HEK293 cells transfected with APP751

Available online at www.sciencedirect.com R

Experimental Cell Research 287 (2003) 387–396

www.elsevier.com/locate/yexcr

Regulation of amyloid precursor protein expression and secretion via activation of ERK1/2 by hepatocyte growth factor in HEK293 cells transfected with APP751 Feng Liu,a Yuan Su,a Baolin Li,a and Binhui Nia,b,* a

Neuroscience Discovery Research, Lilly Research Laboratories, Eli Lilly and Company, Lilly Corporate Center, Indianapolis, IN 46285, USA b Department of Psychiatry, Indiana University Medical School, Indianapolis, IN 46285, USA Received 12 November 2002, revised version received 20 February 2003

Abstract The increased intracellular levels and aberrant processing of the amyloid precursor protein (APP) are associated with ␤-amyloid peptide (A␤) production, cerebrovascular amyloid deposition, and amyloid plaque formation. Here we report that APP level, soluble APP (sAPP) secretion, and A␤ production in HEK293 cells transfected with either wild-type APP751 or APP751 carrying the Swedish mutation are all elevated by hepatocyte growth factor (HGF). We investigated the potential molecular mechanisms underlying the HGF effect. Our data show that HGF stimulated extended activation of extracellular signal-regulated protein kinases (ERK1/2). Pretreatment of cells with inhibitors (UO126 or PD98059) for MEK, the upstream kinase of ERK1/2, abolished ERK1/2 activation evoked by HGF, and abrogated HGF-induced increases in APP levels and sAPP secretion. In addition, transient expression of active MEK1 activated ERK1/2 and increased intracellular APP levels and sAPP secretion. Inhibition of ERK1/2 activity, however, failed to block HGF-stimulated A␤ production. Consistently, transient expression of active MEK1 did not increase A␤ accumulation. Taken together, these results suggest that: (1) HGF regulates the intracellular levels of APP and the secretion of sAPP and A␤; (2) the modulation of APP levels and sAPP secretion induced by HGF is mediated via the MEK1/ERK1/2 signaling pathway; (3) HGF-stimulated A␤ production is independent of ERK activity and, therefore, independent of HGF-evoked elevation of intracellular APP levels. © 2003 Elsevier Science (USA). All rights reserved.

Introduction Amyloid precursor protein (APP)1 is a transmembrane glycoprotein encoded by a single gene located in chromosome 21 and ubiquitously expressed in mammalian tissues [1–3]. Alternative splicing gives rise to multiple isoforms of APP ranging from 365 to 770 amino acids [4]. The three major isoforms are APP695, APP751, and APP770 [3]. APP is * Corresponding author. Fax: ⫹1-317-277-1125. E-mail address: [email protected] (B. Ni). 1 Abbreviations used: AD, Alzheimer’s disease; APP, amyloid precursor protein; A␤, ␤-amyloid peptide; ERK, extracellular signal-regulated protein kinase; MAPK, mitogen-activated protein kinase; sAPP, soluble APP; swAPP751, APP751 bearing the Swedish mutation; wtAPP751, wildtype APP751; HGF, hepatocyte growth factor; PBS, phosphate-buffered saline; PBST, PBS containing 0.1% Tween 20; BSA, bovine serum albumin; mAb, monoclonal antibody; CMV, cytomegalovirus; NGF, nerve growth factor; MEK, MAPK kinase.

subjected to proteolytic processing by a series of proteases known as secretases. The cleavage of APP through ␤-secretase and ␥-secretase generates A␤ and soluble APP␤ (sAPP␤), an amyloidogenic pathway; whereas the ␣-secretase/␥-secretase cleavage leads to the production of P3, a truncated A␤ fragment, and soluble APP␣ (sAPP␣), a nonamyloidogenic pathway [5]. Soluble APP and A␤ are secreted in low levels constitutively by normal cells in culture and are circulated in plasma and cerebrospinal fluid (CSF) of healthy humans and other mammals. Although mechanisms underlying the regulation of APP processing are largely unknown, overexpression and alternative processing of APP play a pivotal role in APP secretion and in A␤ deposition, one of the hallmarks of Alzheimer’s disease. Although the functions of a variety of growth factors in the regulation of APP production have been intensively studied, the effect of hepatocyte growth factor (HGF) on

0014-4827/03/$ – see front matter © 2003 Elsevier Science (USA). All rights reserved. doi:10.1016/S0014-4827(03)00152-6

388

F. Liu et al. / Experimental Cell Research 287 (2003) 387–396

intracellular APP levels and extracellular APP accumulation has not been reported. HGF, also known as scatter factor, is a heparin-binding glycoprotein originally identified as a fibroblast product that induces scattering of contiguous epithelium sheets into isolated cells [6]. Subsequently, HGF was recognized as a multifunctional cytokine secreted by many cell types [7]. Through the ligation with its receptor, a tyrosine kinase encoded by the c-Met proto-oncogene, HGF displays diverse biological effects including mitogenesis, motogenesis, morphogenesis, organogenesis, and cell survival [8,9]. HGF expression is also observed in a variety of cell lines including glioblastoma, glioma, and various astrocytomas [10]. The HGF receptor c-Met has also been identified in human brain tissues [11]. More interestingly, HGF levels are significantly increased in both the astrocytes and the microglial cells surrounding individual senile plaques in AD brains [12]. In this study, we explore the effect of HGF on intracellular APP levels, sAPP secretion, and A␤ production in HEK293 cells transfected with APP, a cell line widely used for studies of APP processing. We also evaluated biochemical signaling pathways potentially responsible for the HGF-initiated biological effects.

Material and methods

supplied with 5% CO2 and 95% air, and used at passages 8 –18. Prior to HGF treatment, 0.5 ⫻ 106 cells per well were plated on 24-well tissue culture plates in 1 ml growth medium and cultured for 24 h. Transient transfection The active MEK1 cDNA plasmid was kindly provided by Dr. Mark Marshall (Eli Lilly & Co.). The transfection reagent FuGENE6 was purchased from Roche (Indianapolis, IN, USA). swAPP751 cells were seeded in 6-well plates 24 h prior to transfection, and cells grew to 50 –70% confluence at the time of transfection. Cells were incubated with DNA/FuGENE6 mixture (at the ratio of 6 ␮l FuGENE6 to 1 ␮g DNA) prepared according to the manufacturer’s recommendation in complete growth medium for 24 h. Cultured supernatants were collected and cells were homogenized in cell lysis buffer containing 10 mM K2HPO4, pH 7.2, 1 mM EDTA, 5 mM EGTA, 10 mM MgCl2, 50 mM ␤-glycerophosphate, 1 mM Na3VO4, 2 mM dithiothreitol, 1% Triton X-100, 1 mM phenylmethylsulfonyl fluoride, 10 ␮g/ml leupeptin, 1 ␮g/ml pepstatin, and 1 ␮g/ml aprotinin. Protein concentrations were determined using the BCA (bicinchoninic acid) protein assay kit (Pierce, Rockford, IL, USA), and 15 ␮g of protein was analyzed by electrophoresis and Western blotting.

Reagents Electrophoresis and Western immunoblotting The polyclonal anti-␤-amyloid precursor protein antibody (CT695) was from Zymed Laboratories Inc. (South San Francisco, CA, USA). The monoclonal antibody 8E5 that recognizes APP N terminus was a generous gift from Mr. Edward Johnstone/Dr. Sheila Little (Eli Lilly & Co), and the monoclonal antibody 6E10 that detects sAPP␣ was obtained from Signet (Dedham, MA, USA). Recombinant human HGF was purchased from R&D Systems (Minneapolis, MN, USA) and reconstituted to 50 ␮g/ml with PBS containing 2 mg/ml BSA. UO126 and PD98059 were from Calbiochem (La Jolla, CA, USA). Pan-p44/42 MAPK (ERK1/2) pAb, phospho-p44/42 MAPK pAb, phospho-Akt pAb, and peroxidase-conjugated secondary antibodies were purchased from Cell Signaling (Beverly, MA, USA). ␤Actin Ab was purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Protein molecular weight marker (Precision Protein Standards) was from Bio-Rad (Hercules, CA, USA). All other common reagents and cell culture media were from Invitrogen Life Technologies (Carlsbad, CA, USA) unless specified otherwise. Cells HEK293 cells stably transfected with either wtAPP751 or swAPP751 [13] were cultured in DMEM-F12 medium supplemented with 5% fetal bovine serum, 20 mM Hepes, 300 ␮g/ml G418, and 50 ␮g/ml penicillin–streptomycin. All cell lines were maintained at 37°C in a humidified incubator

The NuPAGE Bis–Tris Gel System (Invitrogen Life Technologies) was used for electrophoresis and Western transfer. For the detection of cell-associated proteins, cells grown and treated in 24-well plates were lysed with 400 ␮l of 1⫻ LDS sample buffer, and boiled for 10 min. For the detection of sAPP released into culture medium, culture supernatants were collected, centrifuged at 14,000 rpm for 10 min (4°C) to remove cell debris, and mixed with 4⫻ LDS sample buffer at a 1:3 ratio to yield a final concentration of 1⫻ sample buffer. Cell lysates or culture supernatants (8 ␮l) were separated on 10 or 12% precast Bis–Tris gels (200 V, 1 h), and transferred to nitrocellulose membranes (0.2 ␮m) at 50 V for 2 h. Nitrocellulose blots were incubated with PBST (PBS with 0.1% Tween 20) containing 5% nonfat milk for 1 h, then reacted with primary antibodies diluted in PBST containing 5% BSA and 0.01% NaN3 for 1 h at room temperature. Alternatively, blots were incubated with primary antibodies at 4°C overnight. Following three washes with PBST (10 min each), blots were incubated with peroxidase-conjugated secondary antibodies (1:5000) diluted in PBST with 5% nonfat milk (1 h), and again washed with PBST (3 ⫻ 10 min). Finally, blots were incubated with the Enhanced Chemiluminescence (ECL) detection system (Amersham Pharmacia Biotech, Piscataway, NJ, USA) for 1 min, and exposed to BioMax ML films (Kodak, Rochester, NY, USA). The protein bands were scanned using a UMAX PowerLook 2100XL scanner,

F. Liu et al. / Experimental Cell Research 287 (2003) 387–396

389

and the net band intensities were quantified using Kodak Digital Science ID Image Analysis software. Measurement of amyloid ␤ peptides in conditioned media Secretion of A␤ peptide was measured using a sandwich enzyme-linked immunosorbent assay (ELISA) as previously described [14] with minor modification. Assays were performed in 96-well Immulon 4 microtiter plates (Dynex Technologies Inc, Chantilly, VA, USA). Affinity-purified monoclonal antibody (mAb) 266.2 (1.5 ␮g/well) was applied as the capture antibody for total A␤, while mAb 21F12 (0.5 ␮g/well) was used as capture antibody for 1– 42 A␤. Biotinylated mAb 3D6 was used as detection antibody. Streptavidin– horseradish peroxidase was purchased from Amersham and the TMB Substrate Kit was from Pierce. Each sample was tested in duplicate in each experiment, which was repeated at least twice. Measurement of cell viability Cell viability was assessed using the 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide thiazolyl blue (MTT; Sigma, St. Louis, MO, USA) assay. Briefly, cells were seeded at 0.5 ⫻ 106 per well in 24-well tissue culture plates and allowed to grow for 24 h. Then cells were rinsed with serum-free medium and subsequently incubated with relevant reagents for 24 h. Conditioned medium was collected and MTT assay was performed as previously described [15]. Statistics Statistical analysis was performed using a paired t test. Data are presented as means ⫾ SE; P ⬍ 0.05 is considered significant.

Results Hepatocyte growth factor increases cellular APP levels and enhances sAPP and A␤ secretion We first investigated the effect of HGF on intracellular APP accumulation in 293 cells stably transfected either with wtAPP751 or with swAPP751. Whole-cell lysates were analyzed by Western blotting using an antibody recognizing the APP C terminus. HGF treatment (50 ng/ml, 24 h) significantly augmented APP levels in both wtAPP751 and swAPP751 cells (Fig. 1A). To ensure equal sample loading and to confirm the increase in APP is a specific effect rather than a general proliferative activity of HGF, the same APP blots were reprobed with anti-␤-actin antibody. The immunoblots shown in Fig. 1B indicate that the actin levels remain constant after HGF treatment, suggesting that HGF specifically enhances APP accumulation in wtAPP751 and

Fig. 1. Hepatocyte growth factor promotes the synthesis of amyloid precursor protein in cultured cells transfected with either wtAPP751 or swAPP751. (A) 293 cells stably transfected with either wtAPP751 or swAPP751 were incubated with either vehicle control (10 ␮g/ml BSA) or HGF (50 ng/ml) in serum-free medium for 24 h. Cells were lysed and homogenized in LDS sample buffer, then analyzed by NuPAGE gel electrophoresis followed by Western immunoblotting using anti-␤-amyloid precursor protein C-terminal antibody (CT695). HGF significantly increased APP expression in both wtAPP751 and swAPP751 cells. (B) The same blots shown in (A) were rehybridized with actin antibody to verify equal sample loading and the specificity of the HGF effect on APP expression. (C) Net intensities of both mature and immature APP bands (mAPP ⫹ imAPP) were measured and combined, and APP expression in the presence of HGF is presented as a percentage of control. Data from three independent experiments were pooled for the statistical analyses. Shown are means ⫾ SE. *P ⬍ 0.05. The net increases in APP synthesis (HGF treated vs control) are similar in the wtAPP751 and the swAPP751 cells.

swAPP751 cells. For statistical analysis, the band densities of both mature and immature APP were measured and combined. Data from three independent experiments reveal that HGF stimulated similar increases in APP levels in both wtAPP (183 ⫾ 8 % of control) and swAPP (179 ⫾ 27% of control) cells (Fig. 1C). To determine whether HGF specifically stimulates APP expression or simply enhances the transcriptional activity of the CMV promoter built into the APP construct, we investigated the effect of HGF on the overexpression of a newly characterized protein, named RIFLE (unpublished data). RIFLE with a HA tag at the C terminus was cloned into the same vector as that of APP, and was transfected into the culture cells. Pooled stable RIFLE transfectants were incubated with either vehicle control or HGF (50 ng/ml, 24 h) in serum-free medium, and the expression of RIFLE was detected with anti-HA antibody. As shown in Fig. 2, HGF failed to alter the expression of RIFLE (Fig. 2A), while it consistently stimulated the expression of APP. These data suggest that HGF specifically elevates APP expression via a specific cellular mechanism rather than the CMV promoter of the construct.

390

F. Liu et al. / Experimental Cell Research 287 (2003) 387–396

cell-associated APP level in a time-dependent manner. The increases in APP by HGF were first detectable at 3 h and continued up to the end of the test (24 h). That was consistent with the time course of HGF-stimulated sAPP release (Fig. 4B). However, HGF-induced A␤ secretion was observed 1 h following the same treatment (Fig. 4C) when increased APP expression was not detectable. HGF-stimulated A␤ secretion continued over the 24-h incubation time. These data suggest that the HGF-enhanced A␤ release may not be associated with the increases in the overall intracellular APP levels.

Fig. 2. HGF does not alter the synthesis of RIFLE. Cells stably transfected with RIFLE-HA were incubated with either vehicle control (10 ␮g/ml BSA) or HGF (50 ng/ml) in serum-free medium (24 h). RIFLE protein expression was detected by Western immunoblotting using an antibody against HA tag (A). As controls, wtAPP751 and swAPP751 cells were also treated with HGF and APP expression was determined (B). Finally, all samples were probed with actin antibody as protein loading controls (C).

It is known that increased APP expression may lead to elevated APP secretion and A␤ production; therefore, we next evaluated the effect of HGF on APP and A␤ secretion in wtAPP751 and swAPP751 cells. The relative concentrations of total sAPP, the N-terminal fragments of APP after ␤ cleavage as well as ␣ cleavage, and sAPP␣, the Nterminal fragments of APP after ␣ cleavage only, in conditioned media were revealed by Western immunoblotting using antibodies 8E5 and 6E10, respectively. HGF produced compelling increases in both sAPP (213 ⫾ 9% of control) and sAPP␣ (228 ⫾ 26% of control) in wtAPP751 cells, and comparable changes in swAPP751 cells (259 ⫾ 27 and 210 ⫾ 30%, respectively) (Figs. 3A–D). Moreover, HGF substantially increased the production of total A␤ peptides from 0.19 ⫾ 0.05 to 0.43 ⫾ 0.04 ng/ml in the wtAPP751 cells, whereas no A␤1– 42 peptides were detectable. In swAPP751 cells, HGF significantly increased total A␤ from 10.27 ⫾ 1.62 to 14.32 ⫾ 2.10 ng/ml and elevated A␤1– 42 secretion from 0.34 ⫾ 0.08 (n ⫽ 12, P ⬍ 0.0005) to 0.48 ⫾ 0.11 (n ⫽ 12, P ⬍ 0.0001) ng/ml (Fig. 3E). Because HGF may potentially display diverse biological effects including mitogenesis [8], we were concerned that the effect of HGF on the secretion of A␤ and APP might simply be due to the induction of cell proliferation. However, this possibility was ruled out by cell viability measurements (MTT assay). The cell numbers were not altered by the incubation with HGF (50 ng/ml, 24 h; see the first bar of the graph in Fig. 7B) under the same condition shown above, suggesting that HGF provokes a specific mechanism in the production of APP and A␤ in these cells. Since both wtAPP751 and swAPP751 cells respond to HGF in a similar fashion (unpublished observation) and sAPP and A␤ are readily detectable in the swAPP751 cells, we decided to use the swAPP751 cells in all following studies. As shown in Fig. 4A, HGF treatment increased the

Extracellular signal-regulated protein kinases (ERK1/2) mediate APP expression and secretion induced by HGF HGF potentially activates multiple signal transduction pathways such as MEK1/ERK1/2 and PI3-kinase/Akt cas-

Fig. 3. HGF enhances the secretion of sAPP and A␤. wtAPP751 or swAPP751 cells were incubated with either vehicle control (10 ␮g/ml BSA) or HGF (50 ng/ml) in serum-free medium for 24 h. Conditioned media were analyzed by Western immunoblotting using antibody 8E5, which recognizes all sAPP isoforms (A,B), and antibody 6E10, which recognizes sAPP␣ (C,D). HGF increased the accumulation of sAPP and sAPP␣ in the conditioned media of both cell lines. The immunoblots shown are representative of three experiments. The statistical analyses are presented in (B) and (D). Data are means ⫾ SE, n ⫽ 3. *P ⬍ 0.05. (E) production of A␤total and A␤1– 42 in the absence or presence of HGF as measured by ELISA. HGF significantly increased the concentrations of both A␤total (mean ⫾ SE, n ⫽ 12, P ⬍ 0.0005) and A␤1– 42 (n ⫽ 12, P ⬍ 0.0001) in conditioned media of swAPP751 cells and increased the release of total A␤ from wtAPP751 cells (n ⫽ 4, P ⬍ 0.05). A␤1– 42 was not measurable in wtAPP751 culture media.

F. Liu et al. / Experimental Cell Research 287 (2003) 387–396

391

Fig. 4. Time course of HGF-stimulated APP expression, secretion, and A␤ production. (A) swAPP751 cells were incubated with either vehicle control (10 ␮g/ml BSA) for 24 h or HGF (50 ng/ml) for the indicated time. The culture medium was switched to serum-free medium in all wells at the beginning of the experiments (Time 0). HGF was added to cells at different time points and all samples were collected at the end of the experiments (Time 24 h) so that the total incubation time in serum-free medium for all cells was the same (24 h). Whole-cell lysates were subjected to Western analysis using antibody CT695. The increase in APP expression was detected 3 h after HGF and continued increases were evident through the 24-h treatment. Data shown are representative of three experiments. (B) swAPP751 cells were incubated with HGF (50 ng/ml) for the indicated time in serum-free medium. There was a vehicle control (10 ␮g/ml BSA) for each time point. Conditioned medium was subjected to Western analysis using antibodies 8E5 to detect sAPP and 6E10 to detect sAPP␣. Similar to the APP expression induced by HGF, the increases in both sAPP and sAPP␣ were detected 3 h after HGF treatment and evidenced throughout the 24-h treatment. Data shown are representative of three experiments. (C) Portions of the conditioned media used in (B) were analyzed for A␤ production by ELISA. Unlike APP expression and secretion, A␤ secretion was increased as early as 1 h after HGF stimulation (approx twofold increase). Data are representative of three experiments.

cades leading to gene expression and a variety of cellular functions [16,17]. Hence, to unveil the mechanism of HGFmediated APP and A␤ production, we set out to determine pathways involved in the HGF signaling in the swAPP751 cells. As shown in Fig. 5, HGF (50 ng/ml) rapidly stimulated ERK1/2 activation as detected with anti-phosphop44/42 MAPK (Thr202/Tyr204; Fig. 5A). While the total ERK protein level was constant (Fig. 5B), the phosphorylated ERK1/2 changed in a time-dependent manner, which peaked 10 –15 min after HGF exposure, followed by a slow and steady decline, but remained above baseline beyond 6 h and finally returned to baseline at 24 h (Fig. 5A). However, the phosphorylation of Akt at Ser473 was unchanged in the presence of HGF (Fig. 5C), implying that the Akt activities may not be affected by HGF in these cells. To assess the possible role of ERK in the HGF-induced APP expression, we first examined the effectiveness of the specific ERK1/2 kinase MEK1 inhibitor UO126 on the blockade of ERK1/2 activation and on HGF-stimulated APP synthesis. UO126 treatment (30 min) significantly attenuated HGF-induced ERK1/2 activation, as measured by phospho-ERK, in a dose-dependent manner, while the total ERK protein level was not affected. UO126 reduced HGFevoked ERK phosphorylation starting at a concentration as low as 0.25 ␮M and completely blocked the ERK activation at 1 ␮M and higher (Fig. 6A). Meanwhile, UO126 at 1 ␮M

Fig. 5. HGF activates p44/42 ERK1/2 MAP kinase, but not Akt kinase, in a time-dependent manner. swAPP751 cells were incubated with either vehicle (10 ␮g/ml BSA) or 50 ng/ml of HGF in serum-free medium for the indicated period as described in the legend to Fig. 3A. Cell homogenates were analyzed by Western immunoblotting. Parallel blots were hybridized with anti-phospho-p44/42 ERK (ERK1/2) antibody (A), anti-p44/42 antibody (B), or anti-phospho-Aktser473 antibody (@). HGF stimulated rapid and sustained ERK activation, which peaked at least 15 min after HGF treatment and lasted more than 6 h with a steady decline. HGF had no influence on Akt phosphorylation. Western blots shown are representative of three experiments.

392

F. Liu et al. / Experimental Cell Research 287 (2003) 387–396

alternative signaling cascade is involved in the HGF-induced A␤ secretion. This result is consistent with our observation that the A␤ production in response to HGF stimulation is temporally independent of increased cellular APP expression by HGF. Both rapid and the prolonged ERK activation are required for HGF-stimulated APP expression and secretion

Fig. 6. HGF-induced ERK phosphorylation and APP synthesis are blocked by UO126. (A) swAPP751 cells were pretreated with either vehicle (0.1% DMSO; lanes 1, 2) or a serial-diluted UO126 (lanes 3– 8) for 30 min, followed by incubation with either BSA (10 ␮g/ml; lane 1) or HGF (50 ng/ml; lanes 2– 8) in serum-free medium for 15 min. Cell homogenates were analyzed by Western immunoblotting using anti-phospho-p44/42 ERK (top) and anti-p44/42 ERK antibodies. UO126 dose-dependently reduced HGF-stimulated ERK1/2 activation without affecting the ERK protein level. (B) swAPP751 cells were pretreated with either vehicle (0.1% DMSO; lanes 1, 2) or a serial-diluted UO126 (lanes 3–7) in serum-free medium for 30 min, followed by incubation with either BSA (10 ␮g/ml; lane 1) or HGF (50 ng/ml; lanes 2–7) for 24 h. Cell homogenates were analyzed by Western immunoblotting using anti-␤-amyloid precursor protein C-terminal antibody (CT695). UO126, at a concentration as low as 1 ␮M, blocked the increases in APP expression mediated by HGF.

and higher attenuated HGF-induced APP expression (Fig. 6B), which is in a complete agreement with the effect of UO126 on HGF-induced ERK activation. We next extended our study to determine whether ERK1/2 MAP kinase activation is involved in HGF-induced APP synthesis, sAPP secretion, and A␤ production. Prior to HGF stimulation, swAPP751 cells were treated with either UO126 (10 ␮M, 30 min) or PD98059 (50 ␮M, 30 min), another well-known MEK1 inhibitor, followed by the assessment of APP expression and sAPP/A␤ secretion. UO126 and PD98059 treatment alone slightly reduced the basal APP level and they both abolished HGF-evoked APP expression and secretion (Fig. 7A). Fig. 6B demonstrates that neither UO126 nor PD98059 significantly affected cell viability under our experimental condition (serum-free medium, 24-h incubation). Since there were no changes in cellular ␣-actinin, which was used to monitor cell proliferation and protein loading (Fig. 7A, bottom blot), our data suggest that the MEK/ERK pathway is responsible for the HGF-stimulated APP synthesis and sAPP release. However, inhibition of the ERK pathway failed to block HGF-induced secretion of total A␤ and A␤1– 42 (Fig. 7C), suggesting an

While transient ERK activation has been implicated to mediate certain growth factor- and hormone-induced rapid secretion of sAPP in some cell systems [18,19], sustained ERK activation has been suggested to be the major determinant for critical gene expression and cell function [20,21]. We tested the chronological requirement of ERK activation in HGF-stimulated APP expression and secretion. MEK1 inhibitor UO126 was added to the culture medium of swAPP751 cells at various time points before and after the HGF treatment and APP expression as well as sAPP secretion was then assessed 24 h following HGF addition (Fig. 8). UO126 abolished the HGF-induced expression and secretion of APP (lane 2) when added to cells prior to (lane 3) or at the same time as (lane 4) the HGF addition, indicating the importance of the early ERK activation in the APP expression and secretion stimulated by HGF. Additions of UO126 at later stages of ERK activation, e.g., 1, 3, and 6 h following HGF treatment, only partially blocked the HGF effect (lanes 5, 6, 7) and the effect was completely lost when UO126 was added 20 h after HGF treatment (lane 8). The fact that UO126 was able to attenuate APP synthesis and secretion 6 h after HGF challenge signifies the requirement of prolonged ERK activation for the maximum induction of APP by HGF. We next examined the effect of HGF on cellular levels of the transcription factor Egr-1, one of the immediate downstream targets of ERK. Egr-1 mRNA transcription and protein synthesis have been reported to be upregulated by epidermal growth factor and platelet-derived growth factor in human malignant glioma cells [22], suggesting its important role in growth factor-initiated gene transcription in those cells. In this study, we found that treatment of swAPP751 cells with HGF caused rapid induction of Egr-1, as detected by Western immunoblot analysis, in a highly time-dependent manner (Fig. 9A). The initial synthesis of Egr-1 protein induced by HGF (50 ng/ml) occurred between 30 and 60 min, which followed the maximal ERK1/2 activation (see Fig. 5A). The Egr-1 protein level reached the peak at 3 h, which corresponded to the time when APP synthesis began (see Fig. 4A). Egr-1 was subsequently diminished at 6 h and became undetectable at 24 h. Pretreatment of the cells with UO126 abolished HGF-induced ERK activation and completely prevented Egr-1 synthesis induced by HGF (50 ng/ml, 3 h) without affecting ERK protein level (Fig. 9B). Our data suggest that HGF-evoked Egr-1 expression is mediated via ERK activation. It is con-

F. Liu et al. / Experimental Cell Research 287 (2003) 387–396

393

Fig. 7. Both UO126 and PD98059 block HGF-induced APP expression and secretion, but not A␤ production. (A) swAPP751 cells were pretreated with vehicle (0.1% DMSO; lane 1, 2), UO126 (10 ␮M; lanes 3, 4), or PD98059 (50 ␮M; lanes 5, 6) for 30 min, followed by incubation with either 10 ␮g/ml BSA (lanes 1, 3, 5) or 50 ng/ml HGF (lanes 2, 4, 6) in serum-free medium for 24 h. Cell homogenates and conditioned media were analyzed by Western immunoblotting. Cell-associated APP was determined by antibody CT695, whereas sAPP and sAPP␣ in conditioned media were detected by antibodies 8E5 and 6E10, respectively. Both UO126 and PD98059 blocked HGF-induced APP expression and secretion. Actin was used for protein loading control. (B) swAPP751 cells were treated as in (A) and then subjected to viability assessments using MTT assay. Cell viabilities were not significantly altered by HGF, UO126, or PD98059. The data are presented as percentages of control (mean ⫾ SE, n ⫽ 3). (@) swAPP751 cells were treated as in (A) and the conditioned media were analyzed for total A␤ and A␤1– 42 concentration by ELISA. Bars 1 and 4: cells were pretreated with 0.1% DMSO; Bars 2 and 5: cells were pretreated with 10 ␮M UO126; Bars 3 and 6: cells were pretreated with 50 ␮M PD98059. Data shown are the fold increases of total A␤ and A␤1– 42 after HGF treatment (mean ⫾ SE, n ⫽ 3). Neither UO126 nor PD98059 blocked HGF-induced A␤ (total A␤ and A␤1– 42) secretion.

ceivable that Egr-1 may regulate the APP synthesis induced by HGF. However, more studies are needed to confirm that Egr-1 is directly involved in HGF-stimulated APP expression. Finally we further evaluated the involvement of MEK/ ERK, and possibly Egr-1, in APP synthesis and secretion by activating the ERK pathway using a constitutively active MEK1 construct. As shown in Fig. 10, transient transfection of constitutively active MEK1 cDNA construct in swAPP751 cells (24 h) induced sustained ERK activation (Fig. 10A), Egr-1 expression (Fig. 10B), APP expression (Fig. 10C), and the secretion of sAPP (Figs. 10E and F). These data imply that ERK activation is sufficient for increasing APP expression and secretion in cultured cells transfected with swAPP751.

Discussion In this report, we characterize the biological function of HGF in the regulation of APP expression/secretion and A␤ production, and investigate the mechanism underlying the HGF effect on APP processing in the widely used culture system. We show that HGF enhances APP synthesis and sAPP secretion in cultured cells transfected with either

wtAPP751 or swAPP751. HGF incites activation of MAPK ERK and transcription factor Egr-1. Blockade of the MEK1/ ERK signaling cascade by specific pharmacological inhibitors abolishes the HGF effect on Egr-1 induction, APP synthesis, and sAPP secretion. Transfection of a constitutively active MEK1 mutant induces sustained activation of ERK1/2, resulting in continued Egr-1 expression and modest increases in intracellular APP and secretion of sAPP, suggesting that ERK activation is not only required but also sufficient for the HGF effect. It is worth noting that cells treated with a MEK1 inhibitor alone caused a small reduction in basal level APP expression, suggesting that the endogenous MEK1/ERK activity is important for the maintenance of basal APP synthesis. We also show that HGF rapidly increases A␤ peptide secretion, via a MEK/ERKindependent mechanism. These findings may shed light on our understanding of the potential mechanism underlying the APP regulation and A␤ production by HGF, which is greatly elevated in the surrounding cells of senile plaques in AD brains [19]. HEK293 cells have been widely used for the study of APP processing, relevant to the pathogenesis of Alzheimer’s disease [23,24]. Cells transfected with either wtAPP or swAPP displayed similar responses to HGF stimulation (e.g., ERK activation, APP synthesis and secretion), sug-

394

F. Liu et al. / Experimental Cell Research 287 (2003) 387–396

Fig. 8. Both early and sustained ERK1/2 activation are required for the maximal effect of HGF on APP expression and secretion. swAPP751 cells were incubated with either 10 ␮g/ml BSA (lane 1) or 50 ng/ml HGF (lane 2– 8) in serum-free medium for 24 h. UO126 (5 ␮M) was added to cells 0.5 h prior to HGF addition (lane 3), or at the same time as HGF addition (lane 4), or 1, 3, 6, and 20 h following HGF addition. Cell homogenates were analyzed by Western immunoblotting using antibody CT695 (A) and anti-p44/42 ERK antibody for loading control (B). The conditioned media were analyzed using antibody 6E5 to determine sAPP␣ (@). Early additions of UO126 abolished HGF-induced APP expression and secretion (lanes 3, 4), and delayed addition of UO126 attenuated the HGF effect (lanes 5–7) depending on the time point. These results suggest the importance of both initial and prolonged ERK activation in HGF-mediated APP expression and secretion.

gesting that the mutations do not fundamentally alter the cellular responses to HGF. We used swAPP cells in the majority of our studies because the cells expressing swAPP produce substantially higher levels of A␤ than cells expressing wtAPP [25], making it less difficult for us to trace A␤ production, especially A␤1– 42, and to examine the potential mechanisms underlying APP expression/sAPP secretion and A␤ production by HGF. Given the distinct functionalities of neuronal and somatic cells, the signaling pathways that regulate their molecular and cellular processes might be different. Therefore, it is plausible to extend our study into other cell types, especially neuronal cells, to generalize our finding. In the studies of APP processing regulation, much of the attention has been focused on alterations in sAPP release via the stimulation of protein kinase C (PKC) and receptors linked to phospholipase C [26 –28]. A report by Mills et al. has demonstrated that nerve growth factor (NGF) and phorbol ester (PMA) induce rapid secretion of sAPP in PC12 and HEK293 cells stably transfected with APP695 bearing the Swedish mutation, with the maximum sAPP induction occurring 15 min following NGF or PMA stimulation [18]. Certain hormones also stimulate similar prompt sAPP release [19,29]. These studies also suggest that the phosphorylation/activation of ERK is necessary for the rapid induction of sAPP secretion [18,19,29]. Our data show that HGF

stimulates a more induction of sAPP secretion, which becomes detectable on Western blots at 3 h and reaches a maximum between 6 and 24 h following HGF exposure (Fig. 3B). The discrepancy between our finding and those of others is likely due to the differences in the duration of the ERK activation by the different stimuli, which is critical for cell signaling decisions [30]. We observed that HGF stimulates a prolonged ERK activation, which is required for the enhancement of APP expression and sAPP secretion. Several growth factors, such as NGF, stimulate APP promoter activity and induce APP gene expression via activation of the Ras/Raf/MAPK signaling pathway [31,32] since transfection of dominant inhibitory ras, raf, or MAPK mutants prevents the response of the APP promoter to the neurotrophins [32]. However, NGF-induced secretion of sAPP is unaffected by the dominant negative ras, implying that the sAPP release elicited by NGF is independent of the stimulation of APP expression by NGF in PC12 cells [32], although it is unclear whether NGF-activated signaling pathways leading to APP promoter activation and sAPP

Fig. 9. HGF induced ERK-dependent transcription factor Egr-1 expression. (A) swAPP751 cells were incubated with either vehicle control (10 ␮g/ml of BSA) or 50 ng/ml HGF in serum-free medium for the indicated periods as described in the legend to Fig. 3A. Cell homogenates were then analyzed by Western immunoblotting using anti Egr-1 (C-19). HGF-stimulated time-dependent Egr-1 protein synthesis, which was detected at as early as 30 min, peaked at 3 h, and diminished 24 h following HGF treatments. (B) swAPP751 cells were pretreated with either 0.1% DMSO (lanes 1, 2) or UO126 (10 ␮M; lanes 3, 4) for 30 min, followed by incubation with either 10 ␮g/ml BSA (lanes 1, 3) or 50 ng/ml HGF (lanes 2, 4) in serum-free medium for 3 h. Cell homogenates were analyzed by Western immunoblotting using Egr-1 antibody C-19 (top), phospho-ERK1/2 antibody (middle), or pan-ERK antibody (bottom). UO126 abolished HGF-induced ERK1/2 phosphorylation and abrogated HGF-induced Egr-1 expression, while the total ERK protein remained constant.

F. Liu et al. / Experimental Cell Research 287 (2003) 387–396

Fig. 10. Transient overexpression of constitutively active MEK1 mimics the HGF effect on APP expression and secretion. swAPP751 cells were transiently transfected with either vector (pcDNA3.1) or constitutively active MEK1 cDNA (24 h). Cells were lysed and protein concentrations were measured. Fifteen-microgram protein samples were analyzed by Western immunoblotting using various antibodies. (A) Anti-phosphop44/42 ERKs; (B) anti-Egr-1 (C-19); (@) anti-C-terminus APP (CT695); (D) anti-p44/42 ERKs. The conditioned media were collected and analyzed using 8E15 (E) and 6E10 (F). The expression of active MEK1 evoked ERK activation, Egr-1 expression, APP synthesis, and sAPP/sAPP␣ secretion. Data are representative of three independent experiments.

secretion converge at MEK or ERK in PC12 cells. HGFmediated increases in sAPP might be a direct consequence of elevated cytoplasmic APP concentration, since increased intracellular APP could provide additional substrate for secretases to generate more sAPP. Consistent with this notion, the induction of APP expression and that of sAPP secretion are temporally correlated and both events require activation of MEK/ERK pathway (Figs. 6A, 9). However, we cannot exclude the possibility that the stimulation of APP synthesis is parallel to the induction of sAPP secretion. It is tempting to assume that increased cellular APP leads to A␤ production. However, in our study, A␤ secretion is increased less than 1 h following HGF treatment when no increases in APP expression are yet detected (Fig. 3). Moreover, unlike APP synthesis/secretion, the A␤ secretion stimulated by HGF is not altered by MEK/ERK inhibition (Fig. 6C), nor by activation of ERK signaling pathway via transfection of active MEK1 (data not shown). Collectively. these data indicate that HGF differentially regulates APP expression and A␤ production. One possibility is that HGF may mediate alterations in protein trafficking resulting in increases in A␤ release from the intracellular A␤ pool. Multiple transcription factors have been reported to be involved in HGF signaling leading to critical gene expression and physiological/pathological consequences. For ex-

395

ample, HGF induces transcription factor STAT3 synthesis, which contributes to acute phase protein synthesis in hepatocytes during wound healing [33]. HGF is also known to stimulate the expression of the transcription factor Ets-1 in hepatoma cells resulting in cell invasion [34] and to activate c-Fos to promote cell proliferation in epithelial cells [35]. However, effects of HGF on the expression of Egr-1, a zinc finger transcription factor, are unknown. Egr-1 is an immediate downstream target of ERK [36] and is required for ERK-mediated p35 gene expression elicited by NGF [20]. Egr-1 is inducible by epidermal growth factor and plateletderived growth factor in human malignant glioma cells [22]. Our data indicate that HGF stimulates a robust, but transient Egr-1 expression (Fig. 8), which is a precedent of HGFinduced APP expression (Fig. 3). Consistent with HGFinduced APP expression, the Egr-1 synthesis stimulated by HGF is also dependent on the increases in ERK activity, suggesting the possible involvement of Egr-1 in APP synthesis in the presence of HGF. Further experimentation is, however, warranted to clarify the direct role of Egr-1 in HGF-evoked APP synthesis. In conclusion, we demonstrate for the first time that HGF increases the expression and secretion of APP and elevates the secretion of A␤ peptides in cultured cells stably transfected with normal APP or APP bearing a Swedish mutation. Our studies define MEK1/ERK as an essential signaling pathway mediating HGF-induced APP synthesis/release and also show that HGF-stimulated A␤ secretion is independent of ERK activation and cellular APP accumulation caused by HGF.

Acknowledgments We thank Dr. Mark Marshall for providing the MEK1 construct, Dr. Sheila Little and Mr. Edward Johnstone for providing the APP antibody 8E5, and Dr. Bruce Gitter for critical review of this manuscript. We also thank John Ryder and Yan Zhou for technical assistance.

References [1] J. Kang, H.G. Lemaire, A. Unterbeck, J.M. Salbaum, C.L. Master, K.H. Grzeschik, G. Multhaup, K. Beyreuther, B. Muller-Hill, The precursor of Alzheimer’s disease amyloid A4 protein resembles a cell surface receptor, Nature 325 (1987) 733–736. [2] D. Goldgaber, M.I. Lerman, O.W. McBride, U. Saffiotti, D.C. Gajdusek, Characterization and chromosomal localization of a cDNA encoding brain amyloid of Alzheimer’s disease, Science 235 (1987) 877– 880. [3] R.E. Tanzi, J.F. Gusella, P.C. Watkins, G.A. Bruns, P. St. GeorgeHyslop, M.L. Van Keuren, D. Patterson, S. Pagan, D.M. Kurnit, R.L. Neve, Amyloid ␤ protein gene: cDNA, mRNA distribution and gene linkage near the Alzheimer’s locus, Science 235 (1987) 880 – 884. [4] K.S. Kosik, Alzheimer’s disease: a cell biological perspective, Science 256 (1992) 780 –783. [5] D.J. Selkoe, M. Yamazaki, M. Citron, M.B. Podlisy, E.H. Koo, D.B. Teplow, C. Haass, The role of APP processing and trafficking path-

396

[6]

[7] [8] [9]

[10]

[11]

[12]

[13]

[14]

[15]

[16]

[17]

[18]

[19]

[20]

[21]

F. Liu et al. / Experimental Cell Research 287 (2003) 387–396 ways in the formation of amyloid ␤-protein, Ann. NY Acad. Sci. 777 (1996) 57– 64. M. Stoker, E. Gherardi, M. Perryman, J. Gray, Scatter factor is a fibroblast-derived modulator of epithelial cell mobility, Nature 327 (1987) 239 –242. E.M. Rosen, I.D. Goldberg, Scatter factor and angiogenesis, Adv. Cancer Res. 67 (1995) 257–279. K. Matsumoto, T. Nakamura, Emerging multipotent aspects of hepatocyte growth factor, J. Biochem. (Tokyo) 119 (1996) 591– 600. L. Zhang, T. Himi, I. Morita, S. Murota, Hepatocyte growth factor protects cultured rat cerebellar granule neurons from apoptosis via the phosphatidylinositol-3 kinase/Akt pathway, J. Neurosci. Res. 59 (2000) 489 – 496. P. Kunkel, S. Muller, P. Schirmacher, D. Stavrou, R. Fillbrandt, M. Westphal, K. Lamszus, Expression and localization of scatter factor/ hepatocyte growth factor in human astrocytomas, Neuro-oncology 3 (2001) 82– 88. T. Yamada, H. Tsubouchi, Y. Daikuhara, M. Prat, P.M. Comoglio, P.L. McGeer, E.G. McGeer, Immunohistochemistry with antibodies to hepatocyte growth factor and its receptor protein (c-Met) in human brain tissues, Brain. Res. 637 (1994) 308 –312. H. Fenton, P.W. Finch, J.S. Rubin, J.M. Rosenberg, W.G. Taylor, V. Kuo-Leblanc, M. Rodriguez-Wolf, H.M. Schipper, E.G. Stopa, Hepatocyte growth factor (HGF/SF) in Alzheimer’s disease, Brain Res. 779 (1998) 262–270. T. Oltersdorf, P.J. Ward, T. Henriksson, E.C. Beattie, R. Neve, I. Lieberburg, L.C. Fritz, The Alzheimer amyloid precursor protein, J. Biol. Chem. 265 (1990) 4492– 4497. P. Seubert, C. Vigo-Pelfrey, F. Esch, M. Lee, H. Dovey, D. Davis, S. Sinha, M. Schlossmacher, J. Whaley, C. Swindlehurst, R. McCormack, R. Wolfert, D. Selkoe, I. Lieberburg, D. Schenk, Isolation and quantification of soluble Alzheimer’s beta-peptide from biological fluids, Nature 359 (1992) 268 –269. T. Mosmann, Rapid colorimetric assay for cellular growth and survival: application to proliferation and cytotoxicity assays, J. Immunol. Methods 65 (1983) 55– 63. G. Dong, Z. Chen, Z.Y. Li, N.T. Yeh, C.C. Bancroft, C. Van Waes, Hepatocyte growth factor/scatter factor-induced activation of MEK and PI3K signal pathways contributes to expression of proangiogenic cytokines interleukin-8 and vascular endothelial growth factor in head and neck squamous cell carcinoma, Cancer Res. 61 (2001) 5911– 5918. C.F. Yu, B. Roshan, Z.X. Liu, L.G. Cantley, ERK regulates the hepatocyte growth factor-mediated interaction of Gab1 and the phosphatidylinositol 3-kinase, J. Biol. Chem. 276 (2001) 32552–32558. J. Mills, D.L. Charest, F. Lam, K. Beyreuther, N. Ida, S.L. Pelech, P.B. Reiner, Regulation of amyloid precursor protein catabolism involves the mitogen-activated protein kinase signal transduction pathway, J. Neurosci. 17 (1997) 9415–9422. D. Manthey, S. Heck, S. Engert, C. Behl, Estrogen induces a rapid secretion of amyloid ␤ precursor protein via the mitogen-activated protein kinase pathway, Eur. J. Biochem. 268 (2001) 4285– 4291. T. Harada, T. Morooka, S. Ogawa, E. Nishida, ERK induces p35, a neuron-specific activator of Cdk5, through induction of Egr1, Nat. Cell Biol. 3 (2001) 453– 459. C.C. Liang, H.C. Chen, Sustained activation of extracellular signalregulated kinase stimulated by hepatocyte growth factor leads to integrin alpha 2 expression that is involved in cell scattering, J. Biol. Chem. 276 (2001) 21146 –21152.

[22] K. Kaufmann, G. Thiel, Epidermal growth factor and platelet-derived growth factor induce expression of Egr-1, a zinc finger transcription factor, in human malignant glioma cells, J. Neurol. Sci. 189 (2001) 83–91. [23] J.S. Jacobsen, M.A. Spruyt, A.M. Brown, S.R. Sahasrabubhe, A.J. Blume, M.P. Vitek, H.A. Muenkel, J. Sonnenberg-Reines, The release of Alzheimer’s disease beta amyloid peptide is reduced by phorbol treatment, J. Biol. Chem. 269 (1994) 8376 – 8382. [24] P. Marambaud, N. Chevallier, H. Barelli, S. Wilk, F. Checler, Proteasome contributes to the alpha-secretase pathway of amyloid precursor protein in human cells, J. Neurochem. 68 (1997) 798 –703. [25] M. Citron, T. Oltersdorf, C. Haass, L. McConlogue, A.Y. Hung, P. Seubert, C. Vigo-Pelfrey, I. Lieberburg, D.J. Selkoe, Mutation of the beta-amyloid precursor protein in familial Alzheimer’s disease increases beta-protein production, Nature 360 (1992) 672– 674. [26] R.K. Lee, R.J. Wurtman, A.J. Cox, R.M. Nitsch, Regulation of APP synthesis and secretion by neuroimmunophilin ligands and cyclooxygenase inhibitors, Proc. Natl. Acad. Sci. USA 92 (1995) 8083– 8087. [27] B.E. Slack, J. Breu, L. Muchnicki, R.J. Wurtman, Rapid stimulation of amyloid precursor protein release by epidermal growth factor: role of PKC, Biochem. J. 327 (1997) 245–246. [28] B.A. Wolf, A.M. Wertkin, Y.C. Jolly, R.P. Yasuda, B.B. Wolfe, R.J. Konrad, D. Manning, S. Ravi, J.R. Williamson, V.M.-Y. Lee, Muscarinic regulation of Alzheimer’s disease amyloid precursor protein secretion and amyloid beta-protein production in human neuronal NT2N cells, J. Biol. Chem. 270 (1995) 4916 – 4922. [29] S. Goodenough, S. Enger, C. Behl, Testosterone stimulates rapid secretory amyloid precursor protein release from rat hypothalamic cells via the activation of the mitogen-activated protein kinase pathway, Neurosci. Lett. 296 (2000) 49 –52. [30] C.J. Marshall, Specificity of receptor tyrosine kinase signaling: transient versus sustained extracellular signal-regulated kinase activation, Cell 80 (1995) 179 –185. [31] Cosgaya, J., Latasa, M., Pascual, M.J., A. Nerve growth factor and ras regulate beta-amyloid precursor protein gene expression in PC12 cells, J. Neurochem. 67, 1996 98 –104. [32] A. Villa, M.J. Latasa, A. Pascual, Nerve growth factor modulates the expression and secretion of ␤-amyloid precursor protein through different mechanisms in PC12 cells, J. Neurochem. 77 (2001) 1077– 1084. [33] F. Schaper, E. Siewer, M.J. Gomez-Lechon, P. Gatsios, M. Sachs, W. Birchmeier, P.C. Heinrich, J. Castell, Hepatocyte growth factor/scatter factor (HGF/SF) signals via the STAT3/APRF transcription factor in human hepatoma cells and hepatocytes, FEBS Lett. 405 (1997) 99 –103. [34] Y. Jiang, W. Xu, J. Lu, F. He, X. Yang, Invasiveness of hepatocellular carcinoma cell lines: contribution of hepatocyte growth factor, c-met, and transcription factor Ets-1, Biochem. Biophys. Res. Commun. 286 (2001) 1123–1130. [35] K. Takeuchi, S. Shibamoto, K. Nagamine, I. Shigemori, S. Omura, N. Kitamura, F. Ito, Signaling pathways leading to transcription and translation cooperatively regulate the transient increase in expression of c-Fos protein, J. Biol. Chem. 276 (2001) 26077–26083. [36] C. Hodge, J. Liao, M. Stofega, K. Guan, C. Carter-Su, J. Schwartz, Growth hormone stimulates phosphorylation and activation of Elk-1 and expression of c-Fos, egr-1, and junB through activation of extracellular signal-regulated kinase 1 and 2, J. Biol. Chem. 273 (1998) 31327–31336.