h-Prune as a novel binding protein of NS5A that regulates ERK1/2 activation

Hepatitis C virus (HCV) non-structural 5A (NS5A) protein is associated with a wide variety of host signaling pathways by binding to C-terminal polyproline (PxxP) motifs of various proteins. In this study, we used yeast two-hybrid analysis and a GST pull-down assay to screen a novel NS5A interacting protein and elucidate the binding site and cellular signaling by focusing on recombinant human epidermal growth factor (rhEGF)-mediated ERK1/2 activation. Screening a liver cDNA library revealed that h-prune, a member of the DHH (Asp-His-His) protein superfamily, directly interacted with HCV NS5A C-terminus. In particular, a mutation of five proline amino acids to alanine in this region revealed that these two proteins produced strong interaction through this domain. It is known that h-prune possesses a highly conserved DHH motif, which has exopolyphosphatase activity that accelerates hydrolysis of inorganic polyphosphate. A time-chasing analysis after rhEGF treatment demonstrated that h-prune overexpression almost restored NS5A-mediated attenuation of ERK1/2 phosphorylation, but h-prune itself did not alter this signaling. Although the detailed mechanisms need to be clarified, this study demonstrates that h-prune interacts directly with the PxxP motif of the HCV NS5A C-terminus and that this binding alters the rhEGF-mediated ERK1/2 signaling cascade in liver cells.


Introduction
Hepatitis C virus (HCV) is a small enveloped virus in the Flaviviridae family, which is most closely related to the pestiviruses, and which has a positive-sense single-stranded RNA genome. HCV RNA encodes a single large polyprotein of 3000 amino acids (Macdonald et al. 2003). Processing of this precursor polyprotein generates three structural proteins (core, E1, and E2) and seven nonstructural proteins (p7, NS2, NS3, NS4A, NS4B, NS5A, and NS5B). Among them, NS5A is a serine phosphoprotein (Yamauchi et al. 2015). Unlike other HCV proteins, the function of NS5A is not well understood.
C-terminal polyproline (PxxP) domain of NS5A is highly conserved throughout a range of HCV genotypes (Tan et al. 1999) and affects a wide variety of host cellular signaling proteins by binding to Src-homology (SH) 3 domains of the adaptor proteins, thus finally influencing immune invasion and carcinogenesis (Street et al. 2004). Also, NS5A inhibits phosphorylation of extracellular signal-regulated kinases (ERK) by activating epidermal growth factor (EGF) (Macdonald et al. 2003) and perturbs tumor suppressor protein p53 (Lan et al. 2002). Especially, NS5A regulates mitogenic and replication-related signaling pathways in HCV-infected hepatocytes, thereby regulating host cell growth.
h-Prune is a human ortholog of Drosophila prune, which is originally identified as a gene involved in Drosophila eye development. The Drosophila prune eye color mutant lethally interacts with the killer-of-prune allele on the abnormal wing disk (awd) locus, which is the Drosophila homolog gene of the mammalian tumor metastasis gene nm23 encoding a nucleotide diphosphate kinase. In addition, prune encodes GTPase-activating proteins (GAPs), which control Ras-like proteins (Hackstein 1992). The N-terminus of the h-prune protein contains DHH domain named after the characteristic Asp-His-His motif. The DHH domain is highly conserved within species and has phosphodiesterase and exopolyphosphatase activities (Tammenkoski et al. 2008). The h-prune C-terminus has a role in the interaction with glycogen synthase kinase-3b and nm23-H1, which are known as suppressors of cancer metastasis (Reymond et al. 1999;Lacombe et al. 2000;Kobayashi et al. 2006). These two proteins bind simultaneously to different parts of h-prune, thereby regulating its cellular signaling activity. Moreover, overexpression of h-prune regulates cellular motility and metastasis in several solid tumors, such as prostate cancer (Carotenuto et al. 2015), breast cancer (Virgilio et al. 2012), and colorectal cancer (Muller et al. 2010). Physical interaction of this protein with nm23-H1 inhibits its anti-metastatic function in vivo (D'Angelo et al. 2004;Garzia et al. 2006). Taken together, these findings suggest that h-prune may play a negative role in the anti-metastatic activity of nm23-H1.
Inorganic polyphosphate (PolyP) is a biopolymer that has been detected in a variety of organisms from microorganisms to humans. Nuclear PolyP participates in the regulation of gene expression and the stress response. High-polymeric PolyP must be fragmented into shorter polymers to function in response to exopolyphosphatase or endopolyphosphatase. h-Prune is a member of the DHH protein superfamily and functions as a nucleotide phosphodiesterase and an exopolyphosphatase, thus supplying phosphate for DNA synthesis, ATP synthesis, and energy metabolism.
We report the identification of the NS5A-interacting protein, h-prune, and the effect of this protein in the NS5Amediated ERK1/2 signaling pathway. The results clarify NS5A function in hepatocarcinoma.

Plasmid construction
Full-length or partial NS5A constructs (NS5A-F, aa 1-447; NS5A-A, aa 1-150, NS5A-B, aa 1-300, NS5A-C, aa 300-447) were produced as reported previously (Ahn et al. 2004). Briefly, the polymerase chain reaction (PCR) was performed with the primer pairs indicated in Table 1, and cloned into the pHybLex/Zeo vector (Thermo Fisher Scientific, Rockford, IL, USA) using the liner sites at both ends (Fig. 1A). In addition, two PCR reactions were performed with a pair of PCR primers (Bioneer, Daejeon, Korea; NS5A-mut1/NS5A(447-R) and NS5A (1-F)/NS5A-mut2; Table 1) using pHybLex/Zeo-NS5A-F as a template to construct the pHybLex/Zeo-NS5A full-length (F)-mutant, in which five proline amino acids of the PxxPxR (class II) in NS5A were all mutated to alanine. After then, a bloc PCR was performed with the NS5A (1-F) and NS5A (447-R) primers using the two PCR fragments as a template. The resulting PCR fragment was treated with Not I and Sal I and ligated into the Not I/Sal I site of pHybLex/Zeo.

Yeast two-hybrid analysis
A yeast two-hybrid screening was performed as reported previously (Ahn et al. 2004) to search for host proteins that interact with HCV NS5A, following the protocol recommended by the manufacturer (Hybrid Hunter TM ; Thermo Fisher Scientific) with a minor modification. The entire or partial NS5A cDNA-coding region (genotype 1b, generously donated by Dr. Soichiro Miura, National Defense Medical College, Tokorozawa, Saitama, Japan) was fused in-frame with the Lex DNA-binding domain into the pHybLex/Zeo vector as bait for screening a liver cDNA library. In addition, full-length h-prune (liver cDNA library) was cloned into the yeast pYESTrp2 vector to construct the B42 activation domain (B42AD)/h-prune fusion protein as a prey plasmid (Table 1). Positive (pHybLex/Zeo-Fos and pYESTrp2-Jun) and negative (pHybLex/Zeo-lamin and pYESTrp2-Jun) control plasmid combinations were added during the analysis for validation. The pYESTrp2/h-prune and pHybLex/NS5A derivatives were cotransfected into the pEGY48/pSH18-34 yeast reporter strain using the standard lithium acetate method. Transformed cells were spread on synthetic dropout (SD)/ zeocin (Zeo) agar plates [0.67 % (w/v) yeast nitrogen base without amino acids, 20 % (w/v) glucose, 2 % (w/v) agar powder, 10 9 dropout solution without uracil, tryptophan, histidine supplement, and 300 lg/mL Zeo] for 5 days. Yeast cells on SD minimal plates were transferred to filter paper to assess b-galactosidase activity in a colony-lift filter assay. The filter was placed in liquid nitrogen for 30 s and incubated at room temperature in Z-buffer (60 mM Na 2 HPO 4 , 40 mM NaH 2 PO 4 , 10 mM KCl, and 1 mM MgSO 4 ), containing 0.82 mM 5-bromo-4-chloro-3-indolylb-D-galactopyranoside. Then, the filter was placed at 30°C and monitored for a color change to indicate expressed b-galactosidase activity.

In vitro protein binding assay
We performed in vitro pull-down assays to confirm the existence of a protein-protein interaction between NS5A and h-prune. First, full-length NS5A and h-prune were cloned with the pGEX4T-1 (GE Healthcare Life Sciences, Wauwatosa, WI, USA) and pMAL-cR1 vectors (New England Biolabs, Ipswich, MA, USA), respectively (Ahn et al. 2004). The resulting plasmids were transformed into Escherichia coli (E. coli) strain DH5-a [(genotype: F -/80dlacZDM15D (lacZYA-argF) U169 deoR recA1 endA1 hsdR17 (r k -, m k ? ) phoA supE44kthi-1 gyrA96 relA1)]. GST/NS5A and Mal/ h-prune fusion protein expression was induced 5 h after adding 1 mM isopropyl-b-D-thiogalactopyranoside and the expression levels of each fusion protein were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and Western blotting. In detail, cells expressing GST/NS5A or the Mal/h-prune fusion proteins were harvested, resuspended in PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na 2 HPO 4 , and 1.8 mM KH 2 PO 4 ) containing 1 mM of the protease inhibitor phenylmethylsulfonyl fluoride and lysed by sonication in RIPA buffer (Cell Signaling Technology, Danvers, MA, USA). The supernatant was clarified by centrifugation at 4°C, the lysates expressing NS5A and the h-prune protein were mixed together, pulldowned using Glutathione Sepharose 4B (GE Healthcare Life Sciences), and reacted for 12 h at 4°C. The Glutathione Sepharose 4B was washed out with PBS containing 1 % Triton X-100 and phosphatase inhibitors to remove unbound non-specific proteins, and beads were collected.

Determination of exopolyphosphatase activity
Exopolyphosphatase activity was determined by the rate of P i formation. h-Prune was cloned into the pET-28a(?) expression vector with a His tag and expressed using BL21 (DE3) (F -ompT hsdS(r B m B -) gal dcm k(DE3) pLysS (Cam r )) competent E. coli cells to obtain the protein for this assay. Expressed proteins were clarified by centrifugation. After removing the cell debris, the h-prune protein was concentrated with His-binding resin by overnight activation with Ni 2? at 4°C, and 20 lg of purified protein was mixed with 100 lM inorganic polyphosphate 45 (polyP 45 ) in a reaction buffer containing 50 mM Tris-HCl (pH 7.5), 5 mM MgCl 2 , and 100 mM NaCl (Lorenz et al. 1997) and reacted for 30 min or 1 h at 37°C (Eldarov et al. 2013). PolyP hydrolysates were subjected to PAGE on a 22 % polyacrylamide gel with 7 M urea, and the gels were stained with 0.05 % toluidine blue in water containing 25 % methanol and 1 % glycerol and then washed with distilled water (Mullan et al. 2002).

Cell lines
The Hep3B, Huh7, HepG2, and Chang hepatocarcinoma cell lines were cultured in high glucose Dulbecco's A. PCR primers used in the construction of pHybLex/Zeo-NS5A-F, -F-A, -F-B, and -F-C modified Eagle medium (DMEM) containing 10 % fetal bovine serum (FBS), 100 IU/mL penicillin, and 100 lg/ mL streptomycin. A stable cell line expressing NS5A was constructed using 500 lg/mL G418. The mammalian cell lines were incubated at 37°C in a 5 % CO 2 incubator.

RNA isolation
The hepatocarcinoma cell lines were washed once with icecold PBS. After aspirating the PBS, the cells were treated with 1 mL TRIzol (Thermo Fisher Scientific) and the plate was scraped briefly. After pipetting, the cell lysates were transferred to 1.5 mL Eppendorf tubes, 250 lL chloroform was added, the tube was shaken vigorously for about 15 s, and incubated at room temperature for 5 min. After centrifugation at 12,0009g for 15 min at 4°C, the aqueous phase containing RNA was transferred to a new tube, and the RNA was precipitated by mixing with an equal volume of isopropyl alcohol. The tubes were incubated for 12 h at -20°C. The supernatant was centrifuged at 12,000 rpm for 20 min at 4°C. After aspirating the supernatant, the remaining pellet was washed once with 75 % cold ethanol in diethylpyrocarbonate (DEPC)-treated H 2 O, the ethanol was aspirated, and the samples were dried. The pellets were re-dissolved in DEPC-treated water, added to the same volume of phenol, inverted slowly, and centrifuged at 12,000 rpm for 2 min. The supernatants were transferred to a new tube, and two-fold volumes of ethanol and 1 M sodium acetate were added. The supernatants were centrifuged at 12,000 rpm at 4°C and re-dissolved in DEPC water.

Northern blot analysis
Isolated RNAs were mixed with 10 9 formaldehyde gelloading buffer (50 % glycerol, 10 mM EDTA, 0.25 % bromophenol blue, and 0.25 % xylene cyanol FF), electrophoresed on a formaldehyde gel containing 1.2 % agarose, and the separated RNA was transferred to a Hybond N ? membrane (Thermo Fisher Scientific) and cross-linked using UV light. Saline sodium citrate (SSC) (0.15 M NaCl in 15 mM sodium citrate, pH 7.0) was used as the transfer buffer. DNA probes with sequences (TAGTGCAATATATATGGATTTGGAG) were end-labeled with [c-32 P] CTP. The membranes were pre-hybridized with Rapid-Hyb buffer (GE Healthcare Life Sciences) for 2 h at 65°C and then hybridized to radiolabeled probes for 2 h at the same temperature. The hybridized blots were washed in 6 9 SSC for 20-30 min at room temperature and were washed 2 times in 2 9 SSC with 0.1 % SDS for 20-30 min at 42°C, one time in 0.1 9 SSC with 0.1 % SDS for 20-30 min at 42°C, and then exposed to X-ray film (FUJIFILM Medical Systems, Stamford, CT, USA) for 24 h at -70°C. Finally, the signals were detected using an automatic X-ray film processor.

Immunoprecipitation
NS5A and h-prune were cloned with pCMV/Tag1 with a FLAG tag (Dr. Cho Rok Jung, Biomedical Translational Research Center, KRIBB, Daejeon, Korea) and pCMV/ Tag1 with a Myc tag, respectively, to assay protein binding and each cloned plasmid was expressed in the Huh7 hepatocarcinoma cell line. In brief, Huh7 cells were transfected with pCMV-Tag1/FLAG-NS5A and pCMV-Tag1/Myc-h-prune using a calcium phosphate transfection kit (Thermo Fisher Scientific). After transfection, NS5A and h-prune were expressed for 24 h. The cells were lysed by sonication in RIPA buffer, the supernatants were clarified by centrifugation at 4°C, and cell debris was removed. The supernatants were incubated with anti-FLAG antibody (1 lg/mL) (Sigma, St. Louis, MO, USA) for 12 h at 4°C, 20 lL (1:2 slurry) of protein agarose A beads were added (Sigma), and incubated for another 4 h at 4°C. The protein agarose A beads were washed out five times with lysis buffer to remove unbound non-specific protein. All reactions were performed under ice-cold conditions.

Western blot analysis
Bound proteins were released by boiling in SDS sample buffer containing 62.5 mM Tris/HCl pH 6.8, 2 % SDS, 0.002 % bromophenol blue, 0.7135 M b-mercaptoethanol, and 10 % glycerol, resolved by SDS-PAGE, and transferred to a PVDF membranes (Pall, Port Washington, NY, USA). After incubation with horseradish peroxidase-conjugated antibodies, the blots were visualized using an enhanced chemiluminescence system (ECL Western blotting detection reagent; GE Healthcare). The GST/NS5A and Mal/h-prune fusion proteins were detected with the GST polyclonal antibody (Santa Cruz Biotechnology, Dallas, TX, USA) and the MBP antibody (New England Biolabs).

Determination of Erk 1/2 phosphorylation
The pCMV-Tag1/h-prune was transfected into Huh7 cells, the Huh7 cells were overexpressed with FLAG-NS5A using Lipofectamine reagent (Thermo Fisher Scientific), and incubated for 24 h. The transfected cells were starved with 0.1 % FBS/DMEM for 12 h, stimulated with 20 ng/ mL recombinant human EGF (rhEGF) (Cell Signaling Technology), and chased. The cells were lysed by sonication with lysis buffer and a Western blot analysis was performed. Erk1/2 phosphorylation was detected with a phospho ERK1/2 antibody (Cell Signaling Technology). bactin, NS5A, and h-prune expression were detected with FLAG and Myc antibodies (Sigma), respectively.

Identification of the h-prune interacting domain in NS5A
We previously detected hepatocellular proteins interacting with HCV NS5A using a yeast two-hybrid system (Ahn et al. 2004). Through further study with a human liver cDNA prey-library for yeast two-hybrid screening, we presently identified, a hepatocellular protein, h-prune, that interacted with the HCV NS5A ( Fig. 1A-a). h-Prune is a member of the DHH protein superfamily that functions as a nucleotide phosphodiesterase and an exopolyphosphatase, thus being involved in the progression of cancer metastasis through negative regulation of the metastasis suppressor nucleoside diphosphate kinase A (Dooley et al. 1994).). To define the interacting portion of h-prune and NS5A, we created four different NS5A derivatives as baits; full-length NS5A (NS5A-f, aa 1-447) and its three different derivatives, denoted as NS5A-a (aa 1-150), -b (aa 1-300), and -c (aa 300-447) ( Fig. 1A-a). As shown in (Fig. 1A-b), the full-length h-prune construct strongly interacted with the NS5A full-length (NS5A-F) construct and the NS5A-Cterminal fragment (NS5A-c). However, the NS5A-a and -b fragments did not bind to h-prune. This was an interesting finding because NS5A-a also contains the proline-rich PxxP (class I) motif ( Fig. 1A-b), suggesting that the specific PxxP motif (class II) in NS5A may be associated with h-prune binding. Moreover, a pull-down assay demonstrated that Mal/h-prune fusion protein interacted directly with the GST/NS5A protein (Fig. 1B).

The class II proline-rich SH3-binding motif of HCV NS5A is the h-prune binding domain
The NS5A proline-rich motif is highly conserved among HCV species and plays an important role in various NS5Amediated signaling functions. Therefore, we examined if a mutation in this motif alters the interaction between these two proteins. The proline amino acids of the class II PxxP motif in full-length NS5A were mutated to alanine using a PCR method (Ahn et al. 2004) ( Fig. 2A) and the interactions were verified by a colony filter-lift assay. As shown in (Fig. 2B), the NS5A mutant (NS5A-mut) did not bind to h-prune contrary to the case of normal NS5A, indicating that the PxxP region of the NS5A-c fragment is responsible for mutual interactions of the two proteins.

h-Prune consensus sequences in different species and their exopolyphosphatase activity
From previous results, we found a new protein interacting with a specific proline-rich motif of NS5A. As mentioned above, h-prune is a well-conserved protein in many species. As previously reported by another group (Aravind and Koonin, 1998), a NCBI blast homology search (http://blast. ncbi.nlm.nih.gov/Blast.cgi) showed that h-prune has similarities with exopolyphosphatases from Schizosaccharomyces pombe and D. melanogaster. In particular, four representative motifs (designated motif I, II, III, and IV) Appl Biol Chem (2016) 59(4):543-551 547 and their five invariant aspartate residues, which are characteristic of the DHH phosphoesterase family, were conserved (Fig. 3A), asterisks with black box and gray square box). As described above, these conserved signature regions were involved in h-prune exopolyphosphatase activity. Exopolyphosphatases cleave the inorganic polyphosphatase by Pi. Therefore, we tested h-prune-mediated exopolyphosphatase activity by measuring the timechasing hydrolysis of PolyP using purified h-prune. As indicated on the PolyP PAGE-gel stained with toluidine blue to show the metachromatic shift in phosphate (Mullan et al. 2002), the 20 lg h-prune protein reacted with 100 lM PolyP (type 45) for 1 h increased degradation of PolyP compared to that of the control, showing a decrease in PolyP chain length (Fig. 3B).

h-Prune in liver cells directly interacts with NS5A
Since the present results showed that h-prune binds to HCV NS5A. We next examined which liver cell line expressed h-prune to demonstrate the role of h-prune in liver cells. The yeast two-hybrid assay of HCV NS5A was performed with a liver cDNA library and expression was confirmed by Northern blot analysis in the Hep3B, Huh7, HepG2, and Chang liver cell lines. h-Prune was expressed in all liver cell lines, showing the highest expression in Huh-7 cells (Fig. 4A). Next, we performed co-immunoprecipitation to determine if the two proteins directly interacted. Therefore, FLAG-tagged NS5A and Myc-tagged h-prune were constructed, and Western blots revealed that h-prune and NS5A bound together in Huh-7 cells (Fig. 4B), indicating that h-prune may have some role in liver cells through interaction with NS5A.

Discussion
HCV is a small and enveloped Hepacivirus in the family Flaviviridae that causes hepatocellular carcinoma as well as chronic liver disease and cirrhosis. Thus, HCV infection is the leading indicator for liver transplantation (Kim 2002). The HCV genome encodes a polyprotein precursor of approximately 3,000 residues (Bartenschlager and Lohmann 2000), consisting of 10 distinct virion proteins, including four structural proteins (core, E1, E2, and p7) and six non-structural proteins (NS2, NS3, NS5A, NS4B, NS5A, and NS5B) (Bartenschlager et al. 1994). NS5A is generated as a mature protein by NS3/NS4A serine protease. NS5A is localized to cytoplasmic and perinuclear regions of the cell and exists as a serine/ threonine phosphoprotein (Tanji et al. 1995). A characteristic NS5A signature is the PxxP proline-rich sequence, which is highly conserved in most HCV isolates and interacts with a variety of host signaling proteins, thus regulating immune evasion and carcinogenesis. This motif of NS5A is a binding site for various proteins, particularly those containing SH3 domains within a diverse group of signal-transducing proteins (Pawson 1995).
A domain analysis revealed binding of the NS5A protein to h-prune at the C-terminal proline-rich region. This is an interesting finding because this specific proline-rich NS5A motif has many biological effects on a number of signal transduction pathways. A functional analysis indicated that NS5A noticeably attenuated rhEGF-induced ERK phosphorylation in hepatocarcinoma cells. But, h-prune restored completely NS5A-mediated ERK signal attenuation. ERK is characterized as a virion-associated kinase that can modulate human immunodeficiency virus infectivity (Jacque et al. 1998). Several viruses interfere with the ERK pathway to support viral replication. For example, infection with herpes simplex virus 1 or dengue virus serotype 2 suppresses ERK activity (Chang et al. 2012;Chuluunbaatar et al. 2012), whereas activation of the ERK signaling cascade is required for efficient replication of enterovirus 71 and vaccinia virus to facilitate host cell survival (Liu Fig  Several reports have suggested that HCV triggers the release of various cytokines including alpha interferon (IFN-a) and interleukin-6, which are important in the initial antiviral immune defense, thereby restricting viral replication. However, reports on the effect of the ERK signal pathway are controversial in view of the pathway's roles in viral infection and IFN-a signaling. For example, the HCV protein induces activation of the ERK and p38 mitogenactivated protein kinase (MAPK) signaling pathways in transgenic mice (Tsutsumi et al. 2003), whereas HCV NS5A inhibits transcription factor-activating protein-1 function by perturbing the ERK pathway (Macdonald et al. 2003). Viruses use various strategies for infection, and ERK signaling may be affected at various stages, which needs to be clarified.
The h-prune found by us to be a NS5A-interacting protein promotes metastasis by binding and inactivating the anti-metastatic protein nm23-H1, but its role in liver cells has not been reported. Presently, while h-prune overexpression itself did not affect rhEGF-mediated ERK1/2 activity, co-expression of h-prune and NS5A rescued the decreased ERK1/2 signal by NS5A overexpression. Independent of cell mortality regulation through phosphodiesterase activity of h-prune, the protein helps survival of liver cells infected with HCV NS5A.
Inorganic polyphosphate is used as an energy source in mammalian systems and is involved in tumor metastasis Zollo 2009), apoptosis (Hernandez-Ruiz et al. 2006), and cell proliferation (Wang et al. 2003). For example, PolyP activates mammalian target of rapamycin (mTOR) signaling and S. cerevisiae PPX1, the human hprune ortholog, inhibits phosphorylation of 4E-BP1 by mTOR (Wang et al. 2003). Therefore, h-prune may also downregulate 4E-BP1 as in PPX1. For this reason, we may need to examine how the interaction between h-prune and HCV NS5A is coordinated with other signaling pathways, such as mTOR. Fig. 5 h-Prune restores NS5Amediated attenuation of Erk1/2 signaling. (A) Western blots for phospho-ERK1/2. Huh7 cells transfected with FLAG-NS5A and Myc-h-prune were treated with 20 ng/mL rhEGF and immunoblotted with antiphospho ERK1/2 antibody. (B) Bars denote the density ratio of phospho-ERK1/2 relative to b-actin. All ratio values were normalized to the WT ratio at time 0 and defined as 1 (*P \ 0.05, **P \ 0.01, ***P \ 0.001 vs. WT 0 time; #P \ 0.05, ##P \ 0.01, ns (not significant) versus mock at each time point; mean ± SE, n = 4 experiments)