Geranylgeranylacetone selectively binds to the HSP70 of Helicobacter pylori and alters its coccoid morphology

Geranylgeranylacetone (GGA) is used to treat patients suffering from peptic ulcers and gastritis. We examined the effect of GGA on Helicobacter pylori, which is a causative factor of gastrointestinal diseases. Previously, we have reported that GGA binds specifically to the molecular chaperone HSP70. In this paper, we report that GGA bounds to H. pylori HSP70 (product of the DnaK gene) with 26-times higher affinity than to human HSP70, and induced large conformational changes as observed from surface plasmon resonance and circular dichroism. Binding of GGA suppressed the activity of the H. pylori chaperone. GGA also altered several characteristics of H. pylori cells. GGA-treated cells elicited enhanced interleukin-8 production by gastric cancer cell lines and potentiated susceptibility to complement as compared to untreated cells. GGA also caused morphological alterations in H. pylori as reflected in fewer coccoid-like cells, suggesting that GGA converts H. pylori to an actively dividing, spiral state (vegetative form) from a non-growing, coccoid state. This morphological conversion by GGA resulted in accelerated growth of H. pylori. These results suggest a model in which GGA sensitizes H. pylori to antibiotic treatment by converting the cells to an actively growing state.

. Purified DnaK and HSP70. DnaK and HSP70 were purified as described under "Materials and Methods" and the purity was analyzed by SDS-PAGE (9% gels).   H. pylori cells in the biopsy specimens from patient stomachs showed elongated rod-shaped or coccoid morphologies in variable proportions 5,6 . The predominance of rods in exponentially growing in vitro cultures suggests that this form represents proliferating cells, namely the vegetative form. On the other hand, the coccoid form is recognized as a viable-but-not-culturable state, which is resistant to various environmental stresses, such as starvation. The coccoid form could contaminate environments such as drinking water, thereby leading to oral transmission.
Geranylgeranylacetone (6,10,14,18-tetramethyl-5,9,13,17-nonadecatetraen-2-one; GGA) is an anti-peptic ulcer drug developed and approved in Japan in 1984. GGA is shown to suppress H. pyloriinduced tissue and cell injury 7,8 and inflammatory reaction 9,10 , so it is expected to show beneficial effects on H. pylori-infected tissues. One of pharmacological actions of GGA is induction of molecular chaperone 70-kDa heat shock protein (HSP70) in mammals 11 . The induced HSP70 protects the gastrointestinal tract from cell damage induced by various stresses, such as reactive oxygen, ethanol and non-steroidal anti-inflammatory drugs (NSAIDs) 11,12 . In addition, mammalian HSP70 induced by GGA has been reported to be cytoprotective in a number of pathological lesions other than the gastrointestinal tract, including heart ischemia 13 , hepatectomy 14 , cerebral infarction 15 , and colitis 16 .
The ubiquitous HSP70s and their constitutive cognates (HSC70s) are highly conserved in evolution. HSP70s are present in mammalian cells as two different gene products that are closely related: a stress-inducible form, HSP70 (known as HSP72), and a constitutively expressed form, HSC70 (known as HSP73) 17 . HSP70s consist of two domains, a highly conserved NH2-terminal ATPase domain with a molecular mass of 45 kDa and a COOH terminal domain of 25 kDa 18,19 . The polypeptide-binding site is located in the 18-kDa of the COOH-terminal adjacent to the ATPase domain 20,21 . The ATP-bound form of HSP70 binds and releases peptide rapidly, whereas the ADP form binds peptide slowly and in a more stable complex 22,23 . The C-terminal peptide-binding domain of the HSC70 binds to heat shock factor-1 (HSF-1) and the activation of HSF-1 is suppressed under normal conditions. We have previously reported that HSC70 and HSP70 specifically bind to a GGA-affinity matrix 24 . When concentration of the GGA is increased in the cell, GGA preferentially binds to the C-terminal of HSC70 and dissociates HSF-1. The released HSF-1 is phosphorylated and activated, acquiring the ability to bind to a heat shock element (HSE) in the promoter of the heat-inducible HSP70 gene and resulting in induction of HSP70 expression.
HSPs are highly conserved between mammals and bacteria, so we anticipated that GGA could also bind to the HSP70 of H. pylori (product of the DnaK gene). It is important to establish whether GGA acts on both human and H .pylori cells. A previous report showed that GGA has antibacterial activity against H. pylori, but not Campylobacter jejuni 25 . In the present study, we investigated the properties of binding of GGA to both H. pylori and human HSP70, and also the effect of GGA on H. pylori cells.

Results
Purification of human and H. pylori HSP70s. We investigated the sequence homology between human HSP70 and H. pylori DnaK. It has been shown that HSP70 is composed of two domains, ATPase and peptide-binding domain. The ATPase and peptide-binding domains are located in the N-terminal and C-terminal halves of the protein, respectively. There was 47% sequence identity between those two proteins, which was evenly distributed across the two halves. There are some differences in the peptide-binding domains (Supplementary Figure S1). After expression and purification, the human and H. pylori proteins were compared by SDS-PAGE (Fig. 1). The apparent molecular weight of the H. pylori protein (referred to henceforth as DnaK) was slightly larger. These purified proteins were used in the present study.
Affinity of GGA for HSP70 and DnaK. A sensorgram from surface plasmon resonance of the binding of GGA to HSP70 and DnaK is shown in Fig. 2. The kinetic parameters K a , K d , and K D were estimated using BIAcore evaluation software, fitting the binding curves to a simple bimolecular binding algorithm. Based on these results with immobilized HSP70 and soluble GGA, the apparent equilibrium dissociation constant (K D ) was 2.00 × 10 −6 M. In contrast, the K D value of DnaK and GGA was calculated as 7.55 × 10 −8 M. Thus, the affinity of GGA to the bacterial DnaK is about 26 times higher than that to mammalian HSP70.  Table 1.

Effect of GGA on chaperone activity of DnaK and HSC70.
We compared the effect of GGA on the chaperone activities of DnaK and HSP70. Citrate synthase (CS) is very unstable to heat (see Fig. 4 for the thermal aggregation assay). Both DnaK (Fig. 4A) and HSP70 (Fig. 4B) inhibited CS aggregation completely. GGA had only slight effect on chaperone activity of HSP70 (Fig. 4B). HSP70 activity was reduced by only 15% at 5 mM GGA; and reduced by 10 and 5% 2.5 and 0.5 mM, respectively. In contrast, the chaperone activity of DnaK was inhibited almost completely at GGA 5 mM (Fig. 4A). These results are consistent with the influence of the drug on the conformations of the two proteins.
Effect of GGA on growth and morphology of H. pylori. In broth culture medium, GGA accelerated the growth of H. pylori in a dose-dependent manner (Fig. 5). Microscopic observation of the cells from these cultures revealed a mixture of Gram-negative rods and small spherical forms. A Gram stained micrograph of typical coccoid-form cells was shown in Fig. 6-D as a control. The coccoid-form cells were round and deeper pink than the rod-shaped cells. GGA caused a disappearance of coccoid-like form cells in a dose-dependent manner (Fig. 6). Therefore, GGA decreased the occurrence of the viable-but-not-culturable coccoid form, and increased highly-dividing rod-shaped cells in these cultures of H. pylori.
Effect of GGA on serum sensitivity of H. pylori. The changes of cell morphology suggest that cell surface of H. pylori may be altered by GGA treatment. So we examined effect of GGA treatment on  serum sensitivity, namely antibody-independent complement-mediated killing, by using normal rabbit serum as a source of complement. GGA-treated H. pylori cells were more rapidly killed by the serum than untreated cells (Fig. 7), consistent with the GGA-treated cells being more susceptible to complement.

Effect of GGA on induction of interleukin-8 (IL-8) in gastric cancer cell lines by H. pylori.
We examined effect of GGA pretreatment of H. pylori cells on their induction of IL-8 in gastric carcinoma cell lines, MKN28 and MKN45. IL-8 is the most important chemokine produced by epithelial cells for the inflammatory response induction and pathogenesis of H. pylori 26 . GGA pretreatment decreased the ability of H. pylori to induce IL-8 in both cancer cell lines (Fig. 8).

Discussion
GGA is a gastromucoprotective drug with a low incidence of side effects. GGA is employed to treat gastritis/gastric ulcers in Japan and other areas of Asia. GGA is commonly employed as an HSP inducing agent. HSP promotes protein folding/membrane passage and protects cells against various types of stressors including infection and inflammation. Organs/cells in which HSP has been induced show a potent activity to resist toxic substances 11 . There are many reports of in vivo experiments, including a human clinical study 27 , that GGA induces HSPs responsible for cell protection, for example, inhibiting NSAID-related acute gastric/small intestinal mucosal injury.
We have previously reported that mammalian HSP70/HSC70 was a GGA-specific binding protein and proposed induction mechanisms of HSP70 by GGA 24 . HSPs are highly conserved protein among organisms, including prokaryote and eukaryote. The sequence homology between human HSP70 and H. pylori DnaK is 47%. So we anticipated GGA binding to H. pylori DnaK. In this study, we investigated binding of GGA to H. pylori DnaK and the influence of GGA on H. pylori viable cells. GGA bound to H. pylori DnaK with higher affinity than to human HSP70. The same concentration of GGA selectively  GGA shares cell protective effects, especially gastric mucosa cells. There are a few reports that GGA protects from H. pylori-induced gastric injury. The reduction of H. pylori-induced gastric mucosa injury by GGA was shown by using a rat in vivo mode l7 . GGA attenuates growth suppression of human umbilical vein epithelial cells induced by H. pylori cell extract, and the authors speculate that GGA could recover host angiogenesis, which is important for the healing of gastric ulcer 8 . So GGA is a promising agent that is effective against H. pylori-induced pathologies. Effects of GGA on H. pylori cells and H. pylori-induced immune responses have been found in several reports and were observed in our present study. Ishii reported that GGA showed antibacterial activity to H. pylori, however, it highly depended on supplements to media 25 . The antibacterial activity was observed on Brucella agar containing albumin, charcoal, or egg yolk emulsion, but it was not observed on the agar containing horse defibrinated blood or β -cyclodextrin. Conversely, our present study indicated that GGA accelerated growth rate of H. pylori in BHI-FBS broth (Fig. 5) and Brucella broth containing FBS (data not shown). An explanation of the growth enhancement was that GGA converted H. pylori to vegetative phase from the coccoid form (Fig. 6). The coccoid form of H. pylori is recognized as a viable-but-not-culturable state that is more resistant to environmental stresses than the vegetative state with active proliferation. Treatment with antibiotics, including β -lactams, such as amoxicillin, causes conversion of H. pylori to the coccoid form [28][29][30] . Generally, antibiotics are effective on actively proliferating cells and, in fact, the coccoid form of H. pylori is less susceptible to antibiotic treatment 30,31 . This incomplete eradication could explain relapse of H. pylori-induced gastric disorder after the antibiotic treatment 32 . GGA is expected promote eradication of H. pylori through conversion to the vegetative form.
H. pylori is susceptible to antibody-independent complement killing 33 . We also observed this serum sensitivity of H. pylori and, furthermore, GGA-treated cells were more susceptible. It seems that the morphological change from the coccoid to vegetative form reflects an alternation of cell surface properties, resulting in greater sensitivity to complement killing.
Effects of GGA on a H. pylori cell-induced inflammatory response, IL-8 production, have been reported 9,10 . In these reports, live H. pylori cells and GGA were simultaneously incubated with human gastric carcinoma cell lines, such as KATOIII and MKN28. It is unclear if GGA affected the gastric or H. pylori cells. We performed this experiment under conditions in which H. pylori cells were pre-treated with GGA and then cocultured with human gastric cells under GGA-free conditions. We therefore suggest that suppression of IL-8 induction occurred by direct action of GGA on the bacteria. Induction of NF-κ B activation and proinflammatory cytokine production has been shown to be mediated by the type IV secretion system of bacterial cells 34,35 . The type IV secretion system consists of proteins encoded by genes located on cag pathogenicity islands 36,37 . GGA could suppress expression or function of this system. Poursina et al. reported that cagE mRNA, which encodes a key factor of type IV secretion system CagE, was expressed in coccoid cells, however, the expression levels in coccoid cells were lower than in the spiral form cells 38 . This issue needs to be clarified by future work.
In conclusion, GGA acts on both human cells and H. pylori cells for protection from gastric disorder. For human cells, HSP70 induced by GGA protects the gastric cells from injury induced by H. pylori. For the H. pylori cells, GGA causes inhibition of DnaK function, increase of complement susceptibility, suppression of IL-8 induction, and conversion to vegetative form from coccoid form. The sum of these actions on H. pylori should have a positive effect on the symptoms caused by the H. pylori infection. Of greatest interest, the conversion to the vegetative form should increase the susceptibility of H. pylori to many antibiotics.

Materials and Methods
GGA was obtained from Eisai Co., Ltd. (Tokyo, Japan). Human HSP70 cDNA (kindly provided from Dr. Richard Morimoto, Northwestern University) was used in this study. Human HSP70 cDNA was amplified by PCR (iCycler BioRad, CA, USA) using HSP70-specific primer containing the restriction enzyme site which forward primer is encoding Xho I site 5′ -CTCGAGATGGCCAAAGCCGCGGCAGTC-3′ and reverse primer encoding Xba I site 5′ -TCTAGACTAATCTACCTCTCAATGGT-3′ from human HSP70 cDNA. The product obtained by PCR were inserted into Xho I/Xba I sites of the pCold I vector (TaKaRa Bio, Japan). The pCold I-human HSP70 constructs were confirmed by DNA sequencing (PRISM 3100, ABI, CA, USA).
H. pylori genomic DNA derived from ATCC43504 was used in this study. DnaK cDNA was amplified by PCR (iCycler BioRad, CA, USA) using DnaK-specific primer containing the restriction enzyme site which forward primer is encoding Xho I site 5′ -CTCGAGATGGGAAAAGTTATTGGAATTGAT-3′ and reverse primer encoding Xba I site 5′ -TCTAGAGATTAAAACCGCTCGCTTGA-3′ from H.pylori genomic DNA. The product obtained by PCR were inserted into Xho I/Xba I sites of the pCold I vector (TaKaRa Bio, Japan). The pCold I-H. pylori DnaK constructs were confirmed by DNA sequencing (PRISM 3100, ABI).
The Surface plasmon resonance (SPR) assay. Surface plasmon resonance (SPR) assay were performed as previously described 39 . All SPR measurements were performed on a BIAcore 2000 instrument (GE Healthcare, Buckinghamshire, UK) at 25 o C, and 25 mM HEPES-KOH (pH 7.4) buffer containing 0.005% Tween 20, 5 mM MgCl 2 , and 150 mM KCl was used as the running buffer at the flow rate of 10 μ l/min. HSP70 or DnaK (1 μ M) were dissolved in 100 mM sodium acetate buffer (pH 4.0) and immobilized on CM5 sensor chips (GE Healthcare, Buckinghamshire, UK) by amino coupling. HSP70 (~15,000 RU) was immobilized on the chip, and GGA solutions (8.6, 17.2 and 172 nM) were loaded on the chip using running buffer. DnaK (~15,000 RU) was immobilized on the chip, and GGA solutions (3, 6, 60, and 120 nM) were loaded on the chip using running buffer. Regeneration of the sensor chip surface was achieved by 1-min pulse (10 μ l) of 0.5 M NaCl in running buffer. The dissociation constant (K D ) was obtained by non-linear curve fitting based on a steady-state affinity using BIAevaluation 3.0 software.

Far-UV Circular Dichroism (CD).
The CD measurements were performed by a J-720 spectropolarimeter (Jasco, Tokyo, Japan) as previously described 40 . The CD spectrum (190-240 nm) of HSP70 (1.1 μ M) or DnaK (1.1 μ M) in 50 mM HEPES-NaOH buffer (pH 7.4) in the presence or absence of GGA (11.0 μ M) were recorded at 25 o C using a cuvette with a 0.5-mm path length. The observed specific ellipticity after normalization against a blank was converted to the mean residue ellipticity [θ ] (degrees cm 2 dmol −1 ). The secondary structure of HSP70 or DnaK was calculated using an analytical program for secondary structure of proteins (SSE-338W, Jasco, Tokyo, Japan).

Measurement of protein aggregation and chaperone function of HSPs.
The thermal aggregation of citrate synthase (CS) (Roche Diagnostics, Mannheim, Germany) was monitored at 50 o C as previously described 24 . Briefly, the concentration of CS used was 0.25 μ M in 50 mM Hepes buffer, pH 7.4, in the presence or absence of the HSP70 (50 μ M) or DnaK (50 μ M), and in the presence of each HSP70 and GGA (0.5 to 5 mM). Light scattering CS was monitored for 20 min at an optical wavelength of 500 nm by Ultrospec 3000 UV-vis spectrophotometer (GE Healthcare) equipped with a temperature control unit using semi-micro-cuvettes (0.5 ml) with a path-length of 10 mm.

Bacterial cells and cultivation.
H. pylori strain SS1 was donated from Dr. Adrian Lee (The University of New South Wales, Sydney, Australia). SS1 was cultured in brain heat infusion broth containing 5% (vol/ vol) heat-inactivated fetal bovine serum (BHI-FBS) at 37 o C in 5% CO 2 , or Helicobacter agar plate (Nissui Pharmaceuticals, Tokyo, Japan) at 37 o C under microareophic condition using AnaeroPack-MicroAero (Mitsubishi Gas Chemical, Tokyo, Japan). Cell growth was determined by turbidity (measured by absorbance at 600 nm). Cell numbers were determined by colony formation using tryptic soy agar plate containing 5% (v/v) sheep blood. Coccoid form cells were prepared according to Yamaguchi et al. 41 . Briefly, cells on Helicobacter agar plates cultured in microaerophic condition were subsequently cultured at 37 o C for 7 d under anaerobic condition using AnaeroPack-Anaero.
Gram staining. Gram staining was performed by Hucker's modified method using Gram Solutions (Wako Pure Chemical, Osaka, Japan). Serum sensitivity assays. Cultured H. pylori SS1 cells were adjusted at a cell density of 10 7 cells/ml. GGA was added to the cell suspension at a concentration of 0, 1, or 5 mM, and then incubated for 48 h. The cells were collected by centrifugation, and then suspended with PBS at a cell density of approximately 10 7 cells/ml. The aliquots (200 μ l) of the cell suspension were mixed with 200 μ l of pooled normal rabbit serum (Cedarlane, Ontario, Canada), and then incubated for various times. After incubation, the treated cell suspension was immediately diluted with PBS at 10 to 10,000-fold and inoculated to tryptic soy agar containing 5% sheep blood. The plates were incubated for 72 h, and the resulting colonies were counted.

IL-8 inducing activity of H. pylori cells.
The human gastric carcinoma cell lines MKN28 and MKN45 were obtained from the Japanese Collection of Research Biosources (JCRB; Ibaraki, Japan). MKN28 and MKN45 were routinely cultured in Dulbecco's modified minimum essential medium supplemented with 10% (vol/vol) heat-inactivated fetal bovine serum (DMEM-FBS). H. pylori cells were precultured in BHI-FBS for 72 h, and then added GGA at a concentration of 0, 1, or 5 mM and further cultured for 48 h. The cells were collected by centrifugation and then washed with PBS twice. The treated cells were suspended with DMEM-FBS, and inoculated to MKN28 or MKN45 cells with approximately 70% confluency at a multiplicity of infection 10 or 100. After 24 h incubation, the amounts of IL-8 in the