Ethical approval for this study was granted according to the requirements of the German Animal Welfare Act on the Use of Experimental Animals and the Animal Care and Use Committees of Saxony-Anhalt (permit number 42502-2-2-947 Uni MD). Male rats (300–400 g; Institute's breeding population of inbred Wistar rats; Harlan-Winkelmann, Borchen, Germany; altogether 162; 128 included, 34 excluded, see below) were housed under controlled laboratory conditions (light cycle of 12 h light/12 h dark; lights on at 6:00 a.m.; temperature, 20±2°C; air humidity, 55–60%) with free access to water and chow. Every effort, including restriction to one single GP dose, was made to minimize the amount of suffering and the number of animals used in the experiments.

The study comprised of the following groups: i) Sham-operated; ii) sham-operated with DMSO applied once at the moment of spontaneous circulation re-establishment; iii) sham-operated with GP applied once at the moment of spontaneous circulation re-establishment; iv) ACA-treated; v) ACA-treated with DMSO applied once at the moment of spontaneous circulation re-establishment; vi) ACA-treated with GP applied once at the moment of spontaneous circulation re-establishment; vii) sham-operated with daily applied DMSO for 7 days; viii) sham-operated with daily applied GP for 7 days; ix) ACA-treated with daily applied DMSO for 7 days; and x) ACA-treated with daily applied GP for 7 days.

Groups i-vi consisted of 18 animals each; 5 for evaluation of mitochondrial parameters 6 h after ACA, 5 for evaluation of mitochondrial parameters 24 h after ACA, 3 for assessment of neurodegeneration 24 h after ACA, 5 for assessment of neurodegeneration 7 days after ACA; groups vii-x consisted of 5 animals each; altogether 128 animals. DMSO groups were installed to discriminate between GP effects and artificial effects of its solubilizer DMSO.

Anesthesia was induced with 5% sevoflurane (Pfizer GmbH, Berlin, Germany) in an oxygen/nitrous oxide mixture (40:60) via facemask followed by endotracheal intubation with a modified laryngoscope and a venous catheter and muscular relaxation with vecuronium (1 mg/kg; Pfizer). Mechanical ventilation was performed with intermittent positive pressure ventilation (IPPV). For drug administration, blood sampling, and continuous blood pressure monitoring both left femoral vessels were cannulated with polyethylene catheters. After 5 min of room air ventilation and baseline control, ACA was induced by an end-expiratory interruption of IPPV on paralyzed rats for 6 min. ACA (defined as a non-pulsatile blood pressure of less than 10 mmHg) was reached within approximately 3 min (Fig. 1).

Resuscitation was performed by the administration of epinephrine (i.v.; 1 µg/kg; Pfizer) and sodium bicarbonate (1 mEq/kg), restarting IPPV with 100% oxygen for 1 h, and manual external chest compression (200/min). Return of spontaneous circulation (ROSC) was defined as a pulsatile mean arterial pressure (MAP) above 60 mmHg. Rats with no ROSC within 2 min were excluded (altogether 34). The 2-min interval is our lab standard to avoid non-homogeneous ACA periods leading to pathophysiological differences and, subsequently, to the need for larger numbers of animals. Vital parameters (ECG, blood pressure, temperature, airway pressure) were monitored continuously during the first 30 min of the post-resuscitation intensive-care phase. At 5, 15 and 30 min after ROSC, arterial blood samples were collected and evaluated for blood gases (pCO₂, pO₂), pH and glucose. After sufficient spontaneous respiration was established (~30 min post ROSC), catheters were removed, cannulated vessels were ligated and incisions were surgically closed. Rats were extubated and returned to their cages. Body temperature was maintained at 37±0.3°C during preparation, insult and the first 30 min post-ROSC. Thereafter, normal body temperature was stably maintained within normal range by placing the rats in an incubator cage for 24 h.

The standardized GP powder, obtained by extraction of dried aerial parts with 75% ethanol/25% water, was received from Herbasin Co., Ltd. (Shenyang, China; batch no. 061001E074). The extract composition was certified by the manufacturer. According to the manufacturer, the extract contained a gypenoside content of 99.7%, a heavy metal contamination <10 ppm and no microbiological contamination. In the HPLC-fingerprint performed by the consumer, eight peaks corresponding to gypenosides were detected (15). In particular, Gypenoside LXIII, Gypenoside Rb₃ and Gypenoside VIII could be identified (Fig. 1) (25). The retention times of specific gypenosides were determined by using standard gypenosides in separate experiments. The respective method has been described in detail previously (25). Briefly: Sample preparation: 5 mg of the dried ethanolic extract were dissolved in 1 ml of ethanol, filtered over Millipore^® filtration unit, type 0.45 µm, and injected into the HPLC apparatus. Injection volume: Gynostemma pentaphyllum extract: 20.0 µl. HPLC parameter: Apparatus: MERCK HITACHI D-6000 A Interface, MERCK HITACHI AS-2000 Autosampler, MERCK HITACHI L-6200 A Intelligent Pump. Separation column: LiChroCART^® 250-4 LiChrospher^® 100 RP-18 (5 µm) (Merck). Precolumn: LiChroCART^® 4-4 LiChrospher^® 100 RP-18 (5 µm) (Merck). Solvent: A: dist. Water (Millipore Ultra Clear UV plus^® filtered), B: acetonitrile (Fa. VWR). Gradient: 5–100% B in 60 min, total runtime: 60 min. Flow: 0.8 ml/min. A single quadruple mass spectrometer (LC/MS) with electrospray ionization (ESI) in negative mode was used for detection (26).

GP powder (60 mg) was dissolved in 0.5 ml DMSO and diluted with PBS (phosphate-buffered saline) to a final volume of 10 ml. In pilot studies a volume of 200 µl of this solution injected i.p. was found to be optimal for fast resorption and distribution within the circulation.

The tolerability and efficacy of GP was tested in vitro using both cultures of dispersed astrocytes and hippocampal slices (20). A dosis of 60 µg/ml was found to be optimal in vitro. The applied 200 µl of the stock solution contained 1,200 µg GP. At an average animal weight of about 350 g/a blood volume of about 20 ml that resulted in the intended final concentration of 60 µg/ml.

As vehicle control a corresponding DMSO solution was injected. First dose of GP/DMSO was administered at the time when spontaneous circulation was re-established. In groups vii-x, GP or DMSO were additionally administered every morning for the next 6 days.

After the respective survival times, animals were sacrificed by over-dosed anesthesia (isoflurane 10% in a in a 4.35 l sealed desiccator; Baxter, Unterschleissheim, Germany). Brains were quickly removed and hippocampi were separated on ice, weighed, minced using small scissors, transferred into ice-cold phosphate buffer solution (PBS; pH 7.4; 10% tissue portion) and homogenized at 4°C using a Potter-Elvehjem glass-Teflon homogenizer (10 strokes at 600 rpm). The mitochondrial parameters citrate synthase activity, NADH:cytochrome c oxidoreductase activity, succinate:cytochrome c oxidoreductase activity and cardiolipin were determined by using the homogenates. The data were related either to mg protein of the homogenate or to the activity of the mitochondrial marker enzyme citrate synthase.

Citrate synthase (CS) activity was assayed in homogenates of hippocampi 6 and 24 h after ACA using a standard procedure at 30°C (27). Before running the assay samples were fivefold frozen and defrosted to permeabilize membranes. Homogenates of 40 µg were evaluated. The increase of CoA absorption was monitored at 412 nm with a Cary 100 Bio spectrophotometer (Varian, Darmstadt, Germany).

The parameter was assayed in homogenates of hippocampi 6 and 24 h after ACA. Samples (500 µl) of the respective tissue homogenates were fivefold frozen and defrosted to permeabilize membranes. A volume of 15 µl (40 µg protein) was used to run the standard assay at 30°C. The reduction of cytochrome c was followed by measuring the absorption at 550 nm with a Cary 100 Bio spectrophotometer (Varian). The rotenone-sensitive absorption was used for quantification.

The parameter was assayed in homogenates of hippocampi 6 and 24 h after ACA. Samples (500 µl) of the respective tissue homogenates were fivefold frozen and defrosted to permeabilize membranes. A volume of 15 µl (40 µg protein) was used to run the standard assay at 30°C. The reduction of cytochrome c was followed by measuring the absorption at 550 nm with a Cary 100 Bio spectrophotometer (Varian). The antimycin A-sensitive absorption was used for quantification.

CL was also analyzed in hippocampi 24 h after ACA. Therefore, samples (see above) were frozen on dry ice and stored at −80°C until CL analysis, steps of which were i) extraction of CL, 50 ng of tetra-myristoyl-CL [(C14:0)₄-CL; Avanti Polar Lipids Inc., Alabaster, AL, USA; internal standard] and 4.2 ml chloroform/methanol (2/1, v/v) containing 0.05% butylated hydroxytoluene (BHT; Avanti Polar Lipids Inc.) were added to 100 µl of defrosted tissue homogenates. The lipid and aqueous phases were separated by adding 800 µl of 0.01 M HCl, intensive shaking and subsequent centrifugation. After centrifugation, the lipid phase (lower phase) was collected and dried under nitrogen atmosphere and acidified. Ice-cold methanol (2 ml), chloroform (1 ml) and 1 ml of 0.1 M HCl were added. The solution was intensively mixed. After 5 min of incubation on ice the samples were separated by the addition of chloroform (1 ml) and 0.1 M HCl (1 ml). The chloroform/methanol phase was recovered as CL-containing sample. Afterwards, the samples were dried under nitrogen and dissolved in 0.8 ml chloroform/methanol/water (50/45/5, v/v/v). After mixing and filtering of the mixture over 0.2 µm PTFE membranes the samples were ready for use. ii) HPLC-MS/MS analysis as described in detail earlier (28).

The protein content was determined according to the Bradford method (29) using bovine serum albumin as the standard.

After survival times of 24 h and 7 days, anaesthetized rats were sacrificed by transcardial perfusion with 4% 0.1 M phosphate-buffered paraformaldehyde (PFA; pH 7.4; Millipore, Darmstadt, Germany). The brains were quickly removed, post-fixed in the same fixative at 4°C overnight, cryoprotected in 30% sucrose in 0.4% PFA (pH 7.4) for 2 days, and rapidly frozen at −20°C. Slice preparation and staining procedure were performed as described (24). The following mixtures of primary antibodies (diluted in 10% fetal calf serum and 0.3% Triton-X 100 in PBS) were used: (i) Mouse monoclonal anti-NeuN (neuronal nuclei antibody; Chemicon, Billerica, USA; 1:100) and polyclonal rabbit anti-glial fibrillary acidic protein (GFAP; Progen, Heidelberg, Germany; 1:500), and ii) monoclonal mouse anti-MAP2 (microtubule-associated protein 2; Covance, Münster, Germany; 1:1,000); and rabbit polyclonal anti-ionized calcium binding adaptor molecule 1, (IBA1; Abcam, Cambridge, UK; 1:1,000). The mixture of the secondary antibodies consisted of goat anti-mouse Alexa 488 (green; Invitrogen, Carlsbad, USA; 1:500) and donkey anti-rabbit Cy3 (red; Dianova, Hamburg, Germany; 1:500) and was diluted in 1% normal goat serum and 0.3% Triton-X in PBS.

Using an AxioImager.M1 fluorescence microscope (Zeiss, Jena, Germany; with a Plan-Neofluar fluorescein/rhodamine objectives) slices were evaluated. For MAP2 analysis, five alternating slices/animal were scanned image field by image field (objective ×40/0.75) to compose an image of the complete hippocampus (1,388×900 pixel; AxioVision software ‘Panorama’, Zeiss). After that a standard evaluation window (500×300 pixel), including the complete hippocampal CA1 region, was selected manually and the respective staining intensity was quantified using ImageJ software (http:/rsbweb.nih.gov/ij/).

For quantification of pyknotic (24 h post-intervention) and normal NeuN-positive (7 days post-intervention) cells of the CA1 region, the AxioVision z-stack software (Zeiss) was used to get a composition image from 5–8 single images (objective ×40/0.75) taken at different focal distances of the slice. All clearly recognizable pyknotic/hyperchromatic cells and NeuN-positive cells were counted unbiased.

GFAP and IBA1 immunostainings of the hippocampal CA1 region were only descriptively analyzed.

For each antibody, microscopic settings and the exposure time of the fluorescence channels were set on the basis of control slices and kept equal for the corresponding preparation.

After 7 days of survival, rats underwent the object recognition test in an open field (clear Plexiglas, 50 cm wide ×25 cm high) combined with a video tracking system (VPC-FH1; Sanyo Electric, Moriguchi, Japan). The experiment started with an adaptation trial without object in the open field, followed by two sessions, a sampling and a test trial, each 6 min long. In the sampling session, rats explored one object (A1, glass bottle 12 cm tall, 5 cm wide). The test session was conducted 120 min later by allowing rats to explore object A (now A2) together with a novel object (B, metal box, 5 cm tall ×5 cm wide). Thereby object A was placed at the same place as in the first session. Each session was video recorded for later analysis. Three independent experimenters, blinded to group treatment, scored each behavioral test. Exploration was defined as sniffing or touching the object with the nose. Longer exploration duration of object B vs. object A2 can be interpreted as evidence for an intact recognition memory (30). Moreover, global habituation (comparison of the total time spent in exploring object A1 during the sampling session to the summarized total time spent in exploring objects A2 and B in the test phase) was determined. Data are presented as mean ± SD.

All quantitative data are presented as the mean ± SD per animal. The means (the exact n is given in the respective figure legends) were analyzed with the non-parametric Kruskal-Wallis test and the Dunn's multiple comparison post-hoc test using Graph Pad Prism 6 (GraphPad Software Inc., La Jolla, CA, USA). For intra-group differences of sham- and ACA-animals one-way ANOVA was performed. Differences of groups with equal treatment regimens but different survival times were analyzed with the Wilcoxon-Mann-Whitney-test. In either case P≤0.05 was considered to indicate a statistically significant difference.

The preparation of animals including anesthesia via facemask, trachea intubation, cannulation of the left femoral vessels and determination of baseline parameters required 22±5 min. Fig. 2 illustrates the recording of ECG, arterial blood pressure and ventilation. In sham-operated animals, DMSO/GP intervention did not influence ECG and blood pressure pattern. In ACA animals, arterial blood pressure dropped to null line after 170±28 sec of asphyxiation without statistical difference between the three ACA groups. Shortly after stop of ventilation, the ECG signal amplitude became suppressed with remarkable fluctuation. It proceeded to asystole within 180±50 sec. The recording was aggravated by typical technical transients. When 6 min of asphyxiation were completed, resuscitation was initiated. Thereby, ROSC was achieved within 30±24 sec. Independent of DMSO/GP intervention, resume of ECG activity started with an initial pattern resemblant to burst-suppression.

The time course of the basic vital parameters is displayed in Fig. 3. ACA induced a transient increase of MAP (hypertensive in tendency 1 min after ACA but, not significant, P=0.07 returning to normal values afterwards; Fig. 3A), a significant increase of heart rate (Fig. 3B), of arterial carbon dioxide tension (pCO₂; Fig. 3C) and of blood glucose values (Fig. 3E), whereas pH levels were significantly reduced (Fig. 3F). In the presence of GP, the ACA-mediated increase of pCO₂ (Fig. 3C) and blood glucose concentration (Fig. 3E) was prevented. DMSO was able to normalize both ACA-mediated increases too (Fig. 3C and E).

Preparation of the animals did not affect these vital parameters; all parameters were within the physiological ranges (baseline). The arterial pO₂ of ACA-animals was increased after 30 min as consequence of ventilation with 100% oxygen (Fig. 3D). The body temperature (tympanal and rectal) was not affected within the 30 min-monitoring period after resuscitation (data not shown).

We used the activity of CS as marker for the amount of mitochondria in order to study changes in mitochondrial mass. In sham animals, 6 h after treatment (early response) increased values of CS were found in animals exposed to DMSO as the vehicle control (Fig. 4A). However, in combination with GP no effect was observed (Fig. 4A). Twenty four hours after treatment (late response), no effect of DMSO could be detected in sham animals (Fig. 4B). In ACA treated animals, DMSO did not cause changes in CS activity at all (Fig. 4A, B). GP caused increase in CS activity only in ACA animals at both time points (Fig. 4A, B).

In order to explore the effect of ACA on the expression of respiratory chain complexes in mitochondria the activities of NADH-cytochrome c oxidoreductase (complex I + III) and succinate:cytochrome c oxidoreductase (complex II + III) were analyzed. For clear interpretation of the data, the complex activities were related to protein content of homogenate samples (Fig. 4C, D, G, H) or to CS activities representing the mitochondrial amount in the homogenate (Fig. 4E, F, I, J). When complex I + III activities were related to mg protein of homogenate samples similar results of GP as for citrate synthase activities were obtained (Fig. 4C, D). The specific complex I + III activities (related to CS activity) displayed no singular DMSO or GP effect. 6 h after treatment, we detected increased specific activities of complex I + III of all ACA-treated animals in comparison to the respective sham operated groups (Fig. 4E). Twenty four h after treatment, the specific complex I + III activities of all sham animals were in tendency elevated possibly indicating a stress response to anesthesia (Fig. 4F).

The analysis of complex II + III activities revealed that DMSO caused decrease in complex activity in sham operated and in ACA-treated animals 6 h post ACA as a side effect (Fig. 4G, I). This effect declined with survival time (Fig. 4H, J). ACA alone caused increase in complex II + III activity 6 h after treatment (Fig. 4G, I) that was not detected in the presence of GP. Moreover, GP neutralized the side effect of DMSO (Fig. 4G, I).

To evaluate the CL tissue content the amount of CL was related to mg protein of homogenate. The corresponding data are presented in Fig. 5A. For the determination of the specific mitochondrial content of CL the data were related to the activity of the mitochondrial marker enzyme CS. These data are presented in Fig. 5B. ACA caused significant increase in both the tissue content of CL and in the mitochondrial CL content. The stimulation of CL synthesis by ACA was prevented by the administration of DMSO and GP, respectively. We did not find any change in the composition of molecular CL species under the conditions of investigation (data not shown).

In sham animals, the NeuN-immunostaining showed intact hippocampal CA1 neurons (Fig. 6A). The MAP2-immunostained fiber net (Fig. 6B) was well developed. The patterns of GFAP-positive astroglia (Fig. 6A) or IB1-positive microglia (Fig. 6B) conformed to norms. In DMSO-treated sham animals, a couple of pyknotic cells was seen in the CA1 pyramidal cell layer (Fig. 6C, arrows). Pyknotic cells are characterized by size-reduction and/or condensation usually associated with hyperchromatosis, that has been demonstrated by NeuN immunostained nuclei. It indicated cell death induction by DMSO, which was confirmed by MAP2 staining, offering an ongoing loosening up of the nerve fiber net (Fig. 6D). The pattern of astroglia (Fig. 6C) and microglia (Fig. 6D) showed signs of activation. The respective staining patterns of GP-treated sham animals were identical to those of the untreated group (data not shown).

In the hippocampal CA1 pyramidal cell layer of ACA treated animals (Fig. 6E), a massive increase of pyknotic cells (arrows) as well as first cell losses (arrowheads) were found. The respective MAP2-positive fiber nets revealed distinct signs of disruption (Fig. 6F). Yet, astroglia activation was not evidenced by GFAP immunostaining (Fig. 6E). Activation of microglia was, however, clearly demonstrable; the cell bodies became more rounded and the originally ramified branches became shorter and stockier (arrow in Fig. 6F). The respective staining patterns of DMSO-treated ACA-animals were identical to those of the untreated ACA group (data not shown). In case of GP administration, neuroprotective tendencies were found: lower number of pyknotic neurons (arrows in Fig. 6G) and less disruption of the MAP2-positive fiber net (Fig. 6H). Semi-quantification of pyknotic NeuN-positive cells is given in Fig. 6I.

In untreated sham-operated animals, NeuN (Fig. 7A) as well as MAP2 (Fig. 7B) immunostaining revealed intact hippocampal CA1 neurons with a dense fiber network. A consistent standard pattern of GFAP-positive astroglia (Fig. 7C) and IB1-positive microglia (Fig. 7D) was also evident. In both, onetime (shown in Fig. 7E-H) or multiple (not shown) DMSO-treated sham-animals, the NeuN-stained CA1 pyramidal cell layer showed a spotty cell loss (arrows in Fig. 7E). These areas offered a narrow accumulation of IBA1-positive microglia cells (arrows in Fig. 7H). MAP2 (Fig. 7F) and GFAP (Fig. 7G) immunostainings revealed, however, regular patterns. In GP treated sham-animals, there were no signs of abnormal neuronal viability or cell pattern, irrespective of whether GP was applied once or multiple (Fig. 7I-L, demonstrated for onetime applied GP). That indicates that GP was able to counteract cell stress mediated by the vehicle DMSO.

In ACA-animals, the hippocampal NeuN-stained CA1 pyramidal cells were massively degenerated (Fig. 7M). In parallel, MAP2 immunostained nerve fibers showed a massive failure (Fig. 7N). Activation of GFAP-positive astroglia (Fig. 7O) and IBA1-positive microglia (Fig. 7P) was clearly evident. That was demonstrated by enhanced immunoreactivity of the resident glia cells. Moreover, microglia migration as well as conversion from ramified to amoeboid cell morphology took place (insert Fig. 7D vs. insert 7P). DMSO intervention, irrespective of its application regime, did not further impair these patterns (Fig. 7Q-T; demonstrated for onetime applied DMSO). GP, independently of its application regime, was able to counteract these degenerative changes to some degree; the band of NeuN-positive CA1 neurons seemed to be partly retained (Fig. 7U, demonstrated for onetime applied GP) and some of MAP2-positive fibers survived (Fig. 7V). Activation of astroglia (Fig. 7W) as well as microglia (Fig. 7X) was as high as in ACA and ACA-DMSO animals. Obviously, the strong neurodegenerative potential of ACA limited the GP-mediated neuroprotection. Semi-quantification is given for NeuN-positive CA1 neurons in Fig. 7Y and for MAP2-stained CA1 fibers in Fig. 7Z.

Sham-operated rats, independent of non-/DMSO-/GP-treatment, were able to discriminate between a ‘known’ and a novel object; the exploration time of object B at simultaneous presence of object A (A2) was roughly twice as long (Fig. 8A). Moreover, the totalized exploration time for A2 plus B in the test trial was significantly higher than for object A alone (A1) in the sampling trial (Fig. 8B).

The recognition ability of ACA-stressed animals without DMSO- or GP-intervention was, actually unexpected, unchanged when compared with sham-animals. The exploration time difference (B vs. A2) was even more pronounced (Fig. 8A). Now, this pattern was moderately affected by DMSO and dramatically by GP. DMSO-treated animals seemed to be better in the recognition of the ‘known’ object A (A1 vs. A2 significantly different; P<0.05). This effect of better recognition, even more distinctive (P<0.001), was also found in GP-treated rats. Additionally, the sampling behavior of the GP-treated ACA-stressed animals was extremely pronounced. They explored object A1 significantly longer than animals from all other groups (Fig. 8A). And the GP-treated animals were the only group with significant shorter (P<0.05) exploration time for the novel object B when compared with the exploration time of object A at the first session.

As demonstrated in Fig. 8B, the GP-effect in ACA-animals was strong enough to completely reverse the significant difference of the global habituation pattern. In the test trial (object A2 + B), the exploration time of GP-treated ACA-animals was significantly lower (P<0.05) then the time they spend to explore object A1 in the first session. DMSO treatment of ACA-animals reduced the difference between exploration times when both sessions were compared, but it was not able to reverse the difference of the global habituation pattern.