Neurotherapeutic effects of Ginkgo biloba extract and its terpene trilactone, ginkgolide B, on sciatic crush injury model: A new evidence

Summary

Ginkgo biloba leaves extract (GBE) was subjected to neuroprotective-guided fractionation to produce eleven fractions with different polarities and constituents. The intermediate polar fraction was shown to be terpene trilactones-enriched fraction (TEGBE). Out of this fraction, pure ginkgolide B (G-B) was further purified and identified based on its spectral data. The effects of GBE and TEGBE were evaluated in comparison to that of G-B in the crush sciatic nerve injury rat model. To evaluate the neuroprotective effects, sixty Wistar male rats were randomly allocated into 6 groups: naive, sham, crush + normal saline, and three treatment groups; crush + GBE, crush + TEGBE, and crush + G-B. Treatments were given one hour following injury, and once daily for 14 days. Neurobehavioral tests, histomorphological examinations, and immunohistochemical analysis of the sciatic nerve and the spinal cord were performed at weeks 3 and 6 post-injury. GBE, TEGBE and G-B were shown to enhance the functional and sensory behavioral parameters and to protect the histological and the ultrastructural elements in the sciatic nerve. Additionally, all treatments prevented spinal cord neurons from further deterioration. It was shown that G-B has the most significant potential effects among all treatments with values that were nearly comparable to those of sham and naive groups.

Keywords:

Neurons – Nerve fibers – Terpenes – Axons – Sciatic nerves – Astrocytes – Myelin sheath

Introduction

Ginkgo biloba L. belongs to the Family Ginkgoaceae and is reported to have a wide range of biological activities due to the synergistic effects of its active components, mainly the flavonols, e.g., quercetin, kaempferol and myricetin, and terpene trilactones (Fig 1).

Typical flavonoids concentrations, in the standardized extracts, range from 24–26%, and the trilactones from 6–8% [1]. Thus, low quality extracts may be subjected to adulteration with some of these flavonoids to meet the minimum requirements of the flavonoids content.

In vitro and in vivo studies showed that the chemopreventive effects of these flavonoids are due to their anti-proliferative, apoptosis, cytotoxic, and angiogenesis actions in cancer models [2–5]. Some researchers have reported the role of flavonoids in preventing the development of Alzheimer’s disease due to their capabilities to attenuate the oxidative stress in the patients [6]. Consequently, flavonoids are considered neuroprotective agents and contribute to the total effects of Ginkgo biloba on cerebrovascular and peripheral nervous system.

Additionally, terpene trilactones, unique components of Ginkgo biloba, have received, by far, the most attention among other compounds in Ginkgo biloba, due to their activity and, thus, importance in the standardization process of GBE. Terpene trilactones are divided structurally into two groups; diterpene trilactones, e.g., ginkgolides A, B and C, and sesquiterpene trilactones, i.e., bilobalides (Fig 1).

Ginkgolides were found to be selective antagonists of platelet activating factor (PAF) receptors [7] with ginkgolide B (G-B) to be the most potent [8]. It was documented that G-B increases the ischemic reperfusion blood flow following cerebral injury [9].

The affinity to PAF receptors was concluded to be due to the cage-like structure of ginkgolides. It is interesting to notice that the presence of an hydroxyl group at C-7 converts ginkgolide B into C, and consequently lowers the activity [10].

Moreover, it was reported that ginkgolide C exhibits the highest affinity to alpha-1 glycine receptors, and that modification of any of the hydroxyl groups leads to the loss of activity [11]. Consequently, the presence of hydroxyl groups in ginkgolides revealed a crucial role in their activity, and any chemical modulation may have a critical impact on their affinity to the target. This is why different ginkgolides, differ only in these hydroxyl groups, have different level of activities.

Bilobalide, on the other hand, is considered to be the most potent neuroprotective compound among other trilactones. Such an effect is due to the strong inhibition of phospholipase A2, the activation of which initiates a cascade of events leading to neuronal death [12]. In addition, bilobalide was reported to antagonize the GABA_A receptor in rat hippocampal tissue, which is also contributing to the neuroprotective effect of bilobalide [13].

Peripheral nerve injury presents a lifelong disability where crush injury is the highest rated type among other injuries [14]. Sciatic nerve crush injury underlies a serious problem with an incidence of 2.8% in multiple-trauma victims and is diagnosed as allodynia, and heat and mechanical hyperalgesia [15]. Among other therapeutic strategies, pharmacotherapy has been shown to be a promising approach to the neurorehabilitation, with the exploration of potential leads from nature being of increasing interest [15].

Towards that, many reports discussed the role of GBE in nerve injury and peripheral neuropathic pain [16–18]. However, the neurotherapeutic effect of GBE on the peripheral nerve crush injury in rats has been rarely studied, particularly on the sciatic nerve.

This study aimed at developing a neuroprotective-guided fractionation protocol to possibly isolate and identify the active leads(s) from GBE. GBE was prepared, then fractionated to afford eleven fractions with different polarities. Fractions containing mainly ginkgolides (TEGBE) were further processed to isolate a representative ginkgolide, i.e., ginkgolide B (G-B). The potential neurotherapeutic effects of GBE, TEGBE and G-B were evaluated on the crushed sciatic nerve. Several tests such as neurobehavioral, histomorphological and immunohistochemical analyses of the sciatic nerve and spinal cord were done.

Materials and methods

Equipment and chemicals

UV spectra were measured in methanol and recorded using an UV-Visible dual beam-spectrophotometer (Spectroscan 50), while IR spectra were recorded on a JASCO FT/IR-4200 spectrophotometer. ¹H and ¹³C NMR spectra of G-B were obtained using a Bruker Avance II 600 MHz spectrometer. Both the ¹H and ¹³C spectra were recorded in deuterated methanol (MeOH-d₄)_, and the chemical shift values were expressed in parts per million (ppm) relative to the internal standard, tetramethylsilane. Carbon multiplicities were determined using DEPT angles at 90°, 45°, and 135°. Two-dimensional NMR data were obtained using the standard pulse sequence of the Bruker 600 for correlation spectroscopy (COSY), heteronuclear single quantum coherence (HSQC) and heteronuclear multiple bond correlation (HMBC). HREIMS data were determined using a double-focusing magnetic sector mass spectrometer (GS-MS DFS/Thermo). High pressure liquid chromatography (HPLC) analyses were performed using Waters machine equipped with a reversed-phase column (ODS, 5 μm, 4.6 x 150 mm, Waters), and Photodiode Array Detector (PDA, Waters). Additionally, preparative HPLC separations were done on Waters HPLC machine equipped with XBridge^TM Prep column (ODS, 5 μm, 10 x 150 mm, Waters). The operating module was equipped with Empower software. Solvents used for chromatographic fractionation and thin layer chromatographic (TLC) analyses were of a general-purpose reagent (GPR) grade, and those for HPLC and spectral analyses were of HPLC and analytical grades, respectively.

Plant material

Coarsely powdered Ginkgo biloba leaves were purchased from Vitaspace, USA. Pure authentic ginkgolides A, B, C and bilobalide were obtained commercially (Merck, Germany) to serve as reference standards.

Extraction

Three kilograms of the powdered Ginkgo biloba leaves were percolated in 10 l of 80% methanol for 24 hours. The extract was collected and the percolation process was repeated two more times using fresh methanol each time. The combined methanol extracts were evaporated in vacuo till dryness to afford 70 g of brownish syrupy residue, abbreviated as GBE.

Neuroprotective-guided fractionation

Bioactivity-guided fractionation protocol was followed, where 50 g of GBE were fractionated using vacuum liquid chromatography [19]. Initially, the extract was pre-adsorbed on a 50 g C18 silica gel. The loaded silica was applied on a 400 g C18 silica gel column (8 x 30 cm ODS, 40–63 μm, 230–400 mesh, ASTM, Merck). The column was initially eluted with 100% water. The elution was run in a gradient mode, where the eluent strength increased by 10% increments of methanol in water, and ended up with 100% methanol. One-liter fractions of each polarity were collected, evaporated and weighed. This afforded 11 fractions, DGA-I-5A – 5K, with different polarities. TLC analyses of the obtained fractions were carried out using pre-coated glass plates (5 x10 cm, 250 μm) with UV₂₅₄ indicator (Anal Tech). The plates were developed in toluene: acetone; 8:2, then, visualized via spraying with acetic anhydride spray reagent, heating for 10 minutes, and then, exposing them to both long (λ_max = 366 nm) and short (λ_max = 254 nm) wavelengths of the UV light (CAMAG).

HPLC analysis

A modified HPLC method from a previously reported one [20] was proposed and applied to analyze, isolate and purify G-B. Fractions DGA-I-5F and 5G, eluted by 50% and 60% methanol in water, respectively, and shown to contain G-B (TLC analysis, Fig 2A and 2B), were subjected to HPLC analysis for further confirmation (Fig 2C).

On the other hand, preparative HPLC was used to isolate G-B in quantities large enough to evaluate its pharmacological effects. Authentic G-B sample (10 mg dissolved in 1 ml of MeOH) was initially analyzed to find the reference sample’s retention time and the wavelength corresponding to the maximum absorbance. Then, 8 mg of each of the above two fractions were dissolved in 2 ml of MeOH, filtered and subjected to analysis. Authentic G-B and the two fractions were injected in a volume of 20 μl each, and eluted using methanol: tetrahydrofuran: water; 1: 3: 15, as the mobile phase, at a flow rate of 1.5 μl/min for a total elution time of 25 minutes. Detection was done using a PDA detector set at λ_max = 219 nm. The produced chromatograms from authentic G-B were compared to those obtained from the other two fractions.

Isolation and purification of G-B

Further purification processes were performed to isolate G-B using preparative HPLC. The solvent system used in the analytical HPLC method (MeOH: THF: H₂O; 1: 3: 15) was applied in the preparative HPLC protocol. The reference compound, G-B, was initially tested, following a dilution of 10 mg in 500 μl MeOH, to confirm the running instrument parameters, to use the obtained chromatogram as a reference for collecting the pure isolated sample. Reference G-B was injected at a volume of 100 μl, and eluted at a flow rate of 5 ml/min for a 25-minute duration. The chromatogram of G-B was recorded. Fractions DGA-I-5F and 5G were diluted appropriately in MeOH (1 g in 10 ml), and filtered as before. The filtered fractions were injected at a volume of 500 μl and a flow rate of 5 ml/min for a total run time of 30 minutes. Different peaks were separately collected, as detected by the PDA detector (λ_max = 219 nm). This procedure was repeated several times, and similar fractions were collected and finally pooled together. The pooled fractions were dried, and tested by TLC to confirm the presence of pure G-B, abbreviated as DGA-I-81D.

DGA-I-81D: amorphous powder (0.263 yield %); UV (MeOH) λ_max (log ε) 218 (3.45) nm; IR (neat) v_max 3446 and 1774 cm^-1; ¹H NMR (CD₃OD, 600 MHz) see Table 1; ¹³C NMR (CD₃OD, 150 MHz) see Table 1; HREIMS (70 eV) m/z 424.1365 [M]⁺; R_f = 0.19 (toluene: acetone; 8:2).

Terpene trilactones-enriched fraction (TEGBE)

Fractions DGA-I-5H and 5I, eluted by 70% and 80% methanol in water, respectively, in the fractionation step above, were subjected to TLC analysis to verify its content. These fractions were added to the leftover subfractions after eluting G-B from fractions DGA-I-5F and 5G. The pooled together fractions were dried, weighed, and abbreviated TEGBE. This fraction was then subjected to HPLC analysis to confirm the identity of the terpene trilactones.

Animals

Male Wistar rats, weighing 300–350 g and aging between 2–3 months old, were obtained from the animal facility of the Health Science Center, Kuwait University. The animals were kept under conditions of constant temperature (23±2°C) and humidity with 12/12-hr light/dark cycle. The rats were housed in pairs with food and water ad libitum. Ethical approval was obtained for all procedures from the animal ethics committee at the Health Sciences Center, Kuwait University, Kuwait, and in accordance with the guidelines of laboratory animal welfare and the National Institutes of Health guide for the care and use of Laboratory Animals (NIH Publications No. 8023, revised 1978). All efforts were made to minimize animal suffering and to reduce the number of animals used in the study.

Surgical procedures on animals

A total of 60 rats were randomly assigned to 6 groups; naive (no surgery or sciatic nerve injury) (n = 6); sham (sham-injury surgical control group) (n = 12); crush (saline-treated, crushed sciatic nerve) (n = 12); crush + GBE (50 mg/kg GBE-treated crushed sciatic nerve rats) (n = 6); crush + TEGBE (50 mg/kg TEGBE-treated crushed sciatic nerve rats) (n = 12); and crush + G-B (15 mg/kg G-B-treated crushed sciatic nerve rats) (n = 12). All treated groups received i.p. injections of the treatment an hour post-surgery and once daily for 14 days. Doses, route of administration and duration of treatments were established previously in many studies [21]. In this study, the sciatic nerve crush injury was performed as previously described [22]. All approved parameters were followed to minimize animal-animal variation due to injury and to induce a standard direct trauma as previously described [23]. Briefly, animals were anesthetized with an intraperitoneal injection of a mixture of ketamine and xylazine. The sciatic nerve was crushed at a distance of 10 mm from the sciatic notch for 60 seconds using micro mosquito forceps (12.5 cm, straight, World Precision Instruments, Inc.). The nerve was examined to ensure that the epineurial sheath was intact but translucent (axotomy). Then it was replaced under the muscle, and the skin incision was sutured. Sham surgery was done for the rats in the sham group, where the right sciatic nerve was exposed as described before, and skin was sutured, without crushing the nerve [24].

Assessment of motor and sensory functional recovery

The rats in all experimental groups were evaluated for motor and sensory neurobehavioral functions 2 weeks preoperatively and 6 weeks post-injury as described before [25]. All tests were repeated three times (with 3 min—20 min interval) for each rat. The mean of the 3 measurements was used as the data point for that rat for calculating the group mean for further statistical analysis. Investigators were blinded to all treatments in all experiments.

Several tests of reflexive sciatic nerve function (motor tests), such as foot position, toe spread, extensor postural thrust (EPT), hopping and rotarod tests were conducted as described in previous studies [22]. Rotarod performance was measured using the rotarod test instrument (47750—Rat Rotarod NG, Italy) set at the acceleration mode. The time and the number of rounds for each rat were recorded once it felt into its lane. The rotarod time latency that the animal falls is indicated on the Y-axis.

Neurobehavioral sensory tests (mechanical and thermal hyperalgesia, and tail flick tests) were evaluated for all experimental animals as previously described [26]. Mechanical hyperalgesia was measured using Analgesia Meter (Ugo-Basile, Linton instrumentation, Italy). This test was used to calculate the percentage of sensory deficits and paw pressure latencies following sciatic nerve injury. The maximum applied force was 250 g to avoid skin damage. Thermal nociception was evaluated using hot/cold plate (50°C; Ugo-Basile, Hot/Cold Plate, Linton instrumentation, Italy). The time between the placement of the animals on the plate and the onset of paw licking, jumping off the plate, and shaking as the paw withdrawal latency was noted. A standard cutoff latency of 35 sec was employed to avoid animal injury. Tail flick test was applied to assess the central nociception using Analgesia Meter Apparatus (Ugo-Basile, Linton instrumentation, Italy).

Processing of sciatic nerve tissue for light and electron microscopy

For morphometric analysis and TEM study, half of the animals were randomly selected and sacrificed at week 3 (day 21), and week 6 (day 42) for histopathological and morphometric evaluation as previously described [25]. Specimens of sciatic nerves were cut and processed following the previously published protocol [27]. Briefly, rats were euthanized with carbon dioxide, the right sciatic nerve and lumbar spinal cord were dissected and fixed by immersion overnight at 4° C, in a mixture of 2% glutaraldehyde and 2% paraformaldehyde in 0.1 M phosphate buffer, pH 7.4. On the following day, sciatic nerve specimens distal to the crush sites (or corresponding location in control rats) were washed in phosphate buffer (pH 7.4) twice, post-fixed in 1% osmium tetroxide, dehydrated through a graded alcohol series, immersed into propylene oxide and embedded in Epon resin [25, 27]. From each tissue block, semi-thin (1 μm) cross-sections were cut using RMC MT-7 ultra-microtome (Research and Manufacturing Co, Tucson, AZ, USA), and stained with 1% toluidine blue for light microscopic histopathological examination and morphometric analysis [25, 27]. A light microscope (Olympus BH 40, Tokyo, Japan) equipped with Olympus DP71 digital camera and an image manager system (Olympus, DP-Controller) was used for conducting the histomorphometric analysis on each semi-thin sections of the sciatic nerve. For TEM, ultrathin sections were cut from the same tissue blocks and processed for EM analysis.

Morphometric / stereological analysis

It was performed according to the principles described previously [24]. Sections were randomly photographed under a 100x oil-immersion objective via DP-Controller-Olympus software. The number of nerve fibers/field in each photomicrograph were counted. The two fundamental parameters, axon diameter (d) and nerve fiber diameter (D), were measured. Myelin thickness [m = (D-d)/2], g-ratio (d/D) and myelin thickness/axon diameter ratio were calculated manually. The thickness measurements acquired from all sampled axons were then averaged to obtain the mean myelin thickness. The g-ratio, on the other hand, was calculated by dividing the axon diameter (d) by the fiber diameter (D) and the coefficient of variation (CV) was obtained as standard deviation/mean x 100. These calculations were performed using Microsoft Excel.

Western blotting of the sciatic nerve and spinal cord proteins

Western blot was performed as described before [24]. Briefly, the sciatic nerve tissues from 3 rats/group were homogenized in RIPA buffer. The proteins measurement of the cell lysate was done using Epoch Microplate Spectrophotometer. The proteins were transferred onto PVDF membranes. The membranes were blocked with 5% milk in TBS-Tween for 1 h and incubated with the primary antibody Anti- MBP (D-18: sc-13912, molecular weight:14–22; Santa Cruz Biotechnology, Inc., Heidelberg, Germany) overnight at 4°C. The membranes were then incubated with the HRP-conjugated secondary antibody for 2 h at room temperature, washed and developed using ECL kit (RPN2109). Blots were scanned, and the band density was quantified using a densitometer (Biorad GS-800). The same protocol was applied on spinal cord tissues obtained from all experimental groups; n = 3/group, except for GBE-treated group.

Qualitative and quantitative analysis of spinal cord immunostained neurons

A parallel study to sciatic nerve analysis, lumbar spinal cords were also dissected and fixed by immersion overnight at 4°C, in a mixture of 4% paraformaldehyde and 0.1% glutaraldehyde in 0.2M phosphate buffer, pH 7.4. Three spinal cord tissues (L3-L6) per group (except for naive group) were processed, cut, then sectioned through multiple steps as prescribed before [28]. The protocol for histological and immunohistological staining were done following previously published work [22,29]. The primary antibodies used were Anti-NeuN (Clone A60, Mouse Monoclonal Antibody, Cat# MAB377, Millipore. Billerica, Massachusetts, USA), Anti-GFAP antibody (GA-5: sc-58766, Mouse Monoclonal, Santa Cruz Biotechnology) or Anti-GAP-43 (rabbit polyclonal IgG, ab16053, ABCAM, Cambridge, UK). The sections were observed, and the stained neurons were quantified under a light microscope. The number of Neu-N labeled neurons, GFAP immunostained astrocytes, and the intensity of GAP-43 in the ventral and dorsal grey horns of the spinal cord were determined using Cell Sens Dimension software, and analyzed as described previously [26].

Statistical analysis

All data were analyzed by one-way ANOVA followed by Bonferroni’s and LSD post hoc test to determine the differences in the individual baseline values using SPSS^® statistical program (version 22, SPSS Inc., Chicago, IL, USA). Results were considered significant when p < 0.05. Data were represented as the mean ± standard deviation (SD). Graphs and images were generated by Microsoft^® Excel and Publisher [24].

Results and discussion

Extraction of Ginkgo biloba leaves

Three kilograms of coarsely powdered Ginkgo biloba leaves were extracted in 80% methanol and evaporated to afford 70 g of brownish syrupy residue abbreviated GBE.

Neuroprotective-guided fractionation of GBE

Most of the bioactive natural products were isolated using bioactivity-guided fractionation [19]. In this technique, the extract of a plant is fractionated into several fractions with different polarities based on their solubility in aqueous and organic solvents. Simultaneously, the biological activities of the fractions are tested to determine the active fraction(s). Then, the bioactive fraction is further purified into subfractions using chromatographic methods. Similarly, the purified subfractions are subjected to biological activity evaluation. This procedure allows for tracking any alteration in the bioactivity due to the purification process, which may lead to a total loss of bioactivity. It is useful, as well, to select and make changes in the purification scheme to purify the active principle(s) without significant changes in its activity.

Isolation, purification and identification of DGA-I-81D

Part of the obtained extract (50 g) was subjected to a vacuum liquid chromatography, where it was fractionated using increasing concentrations of MeOH in H₂O, starting with 100% H₂O and ending up with 100% MeOH, to afford eleven fractions with different polarities. Highly polar fractions, abbreviated DGA-I-5A, 5B, 5C, 5D and 5E, presumably contain flavonoids, were not dealt with in this study. Thin layer chromatographic analysis of the intermediate polar fractions, DGA-I-5F, 5G, 5H and 5I, suggested that they contain terpene trilactones, including ginkgolide B (Fig 2A and 2B). This analysis also showed that fractions DGA-I-5F and 5G, weighing 1.11 and 1.32 g, respectively, contain mainly ginkgolide B (R_f = 0.19) and small quantities of other ginkgolides. This assumption was further confirmed by the HPLC analysis that showed a weakly UV-active compound (λ_max = 219 nm), which is typical for the lactone carbonyl chromophore of the ginkgolides. Upon further purification by preparative HPLC, these two fractions afforded a pure compound, DGA-I-81D, as amorphous powder (Fig 2C). Several other methods are documented to isolate different ginkgolides using different solvent systems and chromatographic techniques [20,30].

The molecular formula of DGA-I-81D was determined as C₂₀H₂₄O₁₀ on the basis of its molecular ion peak at m/z 424.1365 [M]⁺ and NMR data. On the other hand, IR spectrum showed absorption bands for hydroxyl group(s) and lactone carbonyl group. ¹³C NMR and DEPT 135° spectra (S1 and S2 Figs) showed eighteen resonances distributed as two quartets (q, CH₃), one triplets (t, CH₂), seven doublets (d, CH), and eight singlets (s, C). Six carbons resonated in the aliphatic region, and nine in the oxygenated carbons region (δ_C 69–112 ppm). The other three carbons resonated in the carbonyl groups region. The possibility of having identical carbons was confirmed by the presence of cross peaks in the HSQC (S3 Fig) spectra between δ_C 29.6 quartet and δ_H 1.13 singlet that was integrated for nine protons. This confirmed the presence of three identical methyl groups: C-18, 19 and 20. The remaining methyl group (C-16) resonated at δ_C 8.2, and showed cross contours with three protons resonated as a doublet at δ_H 1.24, in the HSQC spectra (Table 1). From COSY spectrum (S4 Fig), this methyl group, coupled to a proton, resonated at as a quartet at δ_H 3.03. This proton was assigned to H-14.

Additionally, ¹H NMR spectrum (S5 Fig) showed the presence of a clear AB system. This system resonated as a pair of doublets (J = 7.8 Hz) at δ_H 4.20 and 4.59, and they were assigned as H-1 and H-2, respectively. This assignment was further confirmed by the coupling contours in COSY spectrum between these two protons. Consequently, C-1 and C-2 were unambiguously assigned at δ_C 75.6 and 93.5, respectively from HSQC spectra. Other COSY contours aided the unambiguous assignments of H-6 (δ_H 5.41), H-7 (δ_H 2.11 and 2.27) and H-8 (δ_H 1.92). On the other hand, C-2 (δ_C 93.5) showed three-bond cross peaks in HMBC spectra (S6 Fig), with a carbon resonated as a singlet at δ_C 178.5. This carbon was concluded to be C-15, which was further confirmed due to the presence of another three-bond cross peaks with H-16 (δ_H 1.24, d, 6.6), and two-bond cross peaks with H-14 (δ_H 3.03, q, 7.2). Similarly, other carbonyl groups were assigned at δ_C 175.4 (C-11) and δ_C 172.8 (C-13). The assignments of other quaternary carbons were confirmed by HMBC correlations as shown in Table 1. These spectral data constituted a solid body of evidence that unambiguously confirmed the identity of this compound as ginkgolide B. The spectral data were found to be distinguishable from those previously reported [31].

Ginkgolide B is a member of closely related compounds called diterpenoid trilactones. Their exceptional carbon skeletons are made of hexacyclic trilactones with an etheric bridge that links rings A/E to ring C, creating ring D, which makes these compounds unique to Ginkgo biloba.

On the other hand, other intermediate polar fractions, DGA-I-5H and 5I, were shown to be rich in all ginkgolides. Finally, the remaining two fractions, DGA-I-5J and 5K, were expected to contain the highly nonpolar constituents and were not evaluated in this study.

Terpene trilactones-enriched fraction

TLC analyses of the intermediate polar fractions DGA-I-5H and 5I, eluted by 70% and 80% MeOH in H₂O, respectively, revealed that these samples contain other terpene trilactones, in addition to some ginkgolide B (Fig 2B). These fractions were pooled together with the leftover from fractions DGA-I-5F and 5G, after isolating G-B, and evaporated till dryness to give 17.3 g of terpene trilactones-enriched fraction (TEGBE). This fraction was evaluated, along with the pure G-B, for their nerve regeneration and recovery activities.

Neurotherapeutic effects evaluation

After the induction of sciatic nerve crush injury, the hind paws of the sham-operated rats showed a normal clinical appearance throughout week 1 post-surgery, and similar to the naive animals, indicating that operative wound did not interfere with the gait and posture of the rats. In the crush and crush-treated groups, GBE, TEGBE and G-B, hind paws stayed profoundly flexed, and the animals were unable to stand throughout week 1 post-surgery. In contrast, all treated groups showed an improved clinical picture and weight-bearing characteristics during the successive weeks, and until the end of the third week. Different from GBE and TEGBE-treated groups, the G-B-treated animals recovered earlier. Additionally, neurobehavioral assessments showed in the treated groups, noticeably in G-B-treated, the foot positioning recovery started from week 4 post-injury until the end of the study as compared to the crush group (Fig 3A). Likewise, toe spread analysis revealed a faster recovery in all treated groups, particularly TEGBE-treated group which was significantly (P<0.05) improved as compared to crush animals (Fig 3B). Expectedly, G-B and GBE-treated animals regained their normal toe spread scores faster than crush animals. Treated groups showed a remarkable recovery in hopping ability as compared to crush group, P<0.05 in G-B and TEGBE-treated animals in week 4 post injury (Fig 3C). In the extensor postural thrust test (EPT) (Fig 3D), G-B-treated animals displayed faster recovery than TEGBE and GBE-treated and demonstrated significant (p< 0.05) decrease in the functional motor deficits by the end of week 5 as compared to crush animals. Rotarod test analysis revealed that all treated groups recovered faster as compared to crush group, starting from week 4, and profoundly in GBE-treated group at week 5 (P<0.05) followed by G-B and TEGBE at week 6 (Fig 3E). In contrast, crush animals showed severe motor coordination disabilities as compared to sham animals in day 1 post-injury and until week 6 (p< 0.0001), indicating a profound injury to motor neurons. Similar effects by GBE were previously reported [32] on a transient sciatic nerve injury, but not in a sciatic crush injury model.

Sensory tests, on the other hand, including paw pressure test (PPT, mechanical hyperalgesia), tail flick test and hot plate test (thermal hyperalgesia) revealed that treated animals exhibited improvement in their thermal, mechanical and central latencies (Fig 3). Previous investigations [21,33] indicated that GBE is an effective remedy, as compared to the control group, based on sensory analysis of neuropathic parameters. However, these findings have been documented on GBE only and not on the trilactones-enriched extract (TEGBE) or pure G-B. In the present study, PPT findings revealed that GBE-treated animals recovered faster than TEGBE and G-B, however, they showed no significance (P>0.05) by the end of week 6 (Fig 4A). On the other hand, the mean baseline of thermal hyperalgesia (withdrawal reflex latency) of crush animals was significantly greater as compared to sham animals (p<0.05) (Fig 4B). However, treatment with GBE, TEGBE or G-B prevented the thermal hyperalgesia. There was no significant (P>0.05) difference in paw withdrawal latency in hot plate test between all treated animals throughout the second week post-surgery. The mean tail flick latency of sham (12.26±0.85 sec) and naive (12.58±0.55 sec) rats showed no significant difference on week 1 post-surgery until the day of sacrifice (p>0.05) (Fig 4C). GBE, TEGBE and G-B treatments significantly (P<0.05) prevented the spinally mediated thermal hyperalgesia as compared to the crush rats by week 3 post-surgery.

However, by end of week 6, tail flick withdrawal reflex was totally regained in all treated groups as compared to sham and naive groups (P>0.05). Functional and sensory deficits in crush group in this study were consistent with our previous data, indicating the successful reproduction of the sciatic crush injury model [22]. Further, enhancements of functional and sensory recovery in the GBE-treated sciatic crush injury animals were in agreement with previously reported work on GBE treatment in facial crush injury model [34]. Consistent with the current findings, a published study [35] reported that GBE was effective in protecting ganglion cells in optic nerves following crush injury.

Despite that all treatments (GBE, TEGBE and G-B) showed positive effects on neurobehavioral tests when compared to the control crush group, G-B and TEGBE modulated the motor and sensory tests more effectively than did GBE. This indicates that terpene trilactones, i.e., ginkgolides, might have a more critical impact on the recovery of the sciatic nerve crush injury than the flavonoids in GBE do. The current study also examined the role of G-B in sciatic nerve recovery following crush injury, as there were no previous reports on either TEGBE or G-B. Thus, the current research intended to study the neuroprotective implications of a pure ginkgolide, i.e., G-B, and other ginkgolides, in one treatment (TEGBE), following nerve crush injury.

Light and transmission electron microscopic (TEM) examinations

Histological examinations of the toluidine blue-stained sciatic nerve specimens of sham and naive groups revealed a normal anatomical appearance of sciatic nerves, with regular distribution of small and large diameter nerve fibers, as well as normal proportion between myelin sheath thickness and fiber diameter at both 3 weeks and 6 weeks post-surgery (Fig 5A–5D). These observations of naive and sham groups were confirmed by electron microscopy, which showed healthy myelin sheaths thickness and typical arrangement of collagen fibers at both time intervals (Fig 6A1–6A4 and 6B1–6B4).

Additionally, the crush group sections displayed Wallerian degeneration features (Fig 5E) at week 3 post-surgery. Collagen fibers in endoneurium showed large separations and discontinuations within the nerve bundles. Also, unmyelinated and irregular axons were abundant as well as the disintegration of axonal cytoskeleton elements in week 3. By the end of week 6 post-surgery (Fig 5E), crush sections showed predominantly newly-regenerated thin myelinated nerve fibers with widespread myelin sheath degeneration. The morphological and ultrastructural analysis of the crush group illustrated a diminished nerve regeneration process, a decrease in macrophage recruitment, reduced activation of phagocytic Schwann cells, and a decline in Wallerian degeneration, even after 6 weeks post-injury. These findings were similar to those previously described [29].

Crush sections at week 3 post-surgery, in electron microscopy, displayed irregularly shaped and highly condensed abnormal myelin sheaths and axonal degradations (Figs 6 and S7). At week 6 after crush injury, however, crush sections exhibited newly regenerated nerve fibers, organized in small fascicles, which were characterized by smaller size and thinner myelin sheaths as compared to sham sections. There was some existence of degenerative fibers among newly regenerated ones, indicating that regeneration was still in progress at week 6 (Fig 6E and 6F). These findings are consistent with other studies [25] supporting the current findings that even after 6 weeks post-injury the process of maturation of regenerated nerve fibers is still incomplete and very slow, when compared to groups treated with GBE, TEGBE and G-B.

In contrast, sciatic nerve specimens of animals treated with G-B, TEGBE and GBE showed a remarkable axonal and myelin regeneration with more newly regenerated nerve fibers as compared to crush sections at week 3 post-injury (Fig 5G, 5I and 5K). The ultrastructural changes of the sciatic nerve for animals treated with G-B improved largely as compared to TEGBE and GBE treated animals. At week 6, the tissue specimens of the treated animals with G-B, TEGBE and GBE (Fig 5H, 5J and 5L) showed almost normal morphological appearance of sciatic nerves as compared to sham and naive groups (Fig 5B and 5D). Further, the number and density of regenerated axons in G-B-treated rats was distinctly lower than other treated groups (TEGBE and GBE) (Fig 5J, 5H and 5L). Evidently, TEM sections of sciatic nerves from animals treated with G-B, TEGBE and GBE showed that nerve fibers were recovered following sciatic crush lesion at week 3, and improved as time progressed (Figs 6G, 6I and 6K and S7). At week 3, thin and regular myelin sheaths were present predominantly with the absence of disintegrated myelin figures as in crush group. Moreover, the extracellular matrix of TEM sections obtained from treated groups displayed normal appearance of collagen fibers as compared to crush group (Figs 6E and S7). At week 6 following crush injury, treatments with G-B, TEGBE and GBE successfully recovered the Schmidt-Lantermann clefts associated with a typical and healthy-looking myelin sheath (Figs 6H, 6J and 6L and S7). Findings from light and electron microscopic examinations showed that morphological alterations by G-B, TEGBE and GBE had an important role in the functional recovery of the sciatic nerve. The current data support other recent in vitro and in vivo studies that showed GBE has a critical modulation on axonal regeneration and Schwann cell activation [36,37]. Furthermore, the positive effects of GBE on sciatic nerve regeneration and neovascularization in current findings were in agreement with many reports [17,18].

Morphometric analysis

Quantitative stereological analysis of sciatic nerve after 3 and 6 weeks post-surgery are presented in Fig 7A–7E.

The results from morphological parameters (axon numbers, areas, and diameters, nerve fiber diameter and myelin thickness) per unit area of the sciatic nerve cross sections obtained from all groups, supported the qualitative data from microscopic examinations. Mean axon numbers per unit nerve area (Fig 7A) were significantly (P<0.0001) lower in G-B-treated group as compared to crush group at week 3 post-injury, but no difference was shown as compared to other treated groups. However, at week 6, G-B-treated animals showed lower mean axon numbers than other treated groups, but equivalent to sham group. Also, the mean axon area per unit nerve area was significantly (P<0.0001) higher in G-B-treated group than crush group, but no significant difference (P>0.05) as compared to sham, TEGBE and GBE-treated groups at week 3 post-injury (Fig 7B). At week 6 post-surgery, the average area of axons per unit nerve area in sections obtained from G-B-treated rats was equivalent to sham and naive (p>0.05) and significantly (p<0.05) was higher than crush group, and unlike TEGBE, and GBE-treated groups that showed no significant difference in their myelinated axon area (p>0.05) as compared to crush group.

The axon diameter sizes in all treated groups were equivalent (P>0.05) to sham, but significantly (P<0.01) higher than those in crush group at 3 weeks post-surgery (Fig 7C). At 6 weeks, the axon diameter size in G-B-treated group was significantly higher than that in the crush group (P<0.05) as compared to other treated groups, which were not more significant than crush group (P>0.05). Further, G-B-treated animals were not significantly different than the sham group (P>0.05).

Similar to axon diameter, nerve fiber diameter was significantly (P<0.0001) high in naive and sham groups as compared to the crush group at both time points (Fig 7D). All treated groups also displayed significantly (P<0.0001) higher nerve fiber diameter per unit area as compared to the crush group, however, it was significantly (P<0.0001) lower as compared to the sham and naive groups at week 3 post-surgery. At week 6 post-injury, nerve fiber diameters obtained from G-B and GBE-treated groups were significantly (P<0.01) higher than those in the crush group, however, TEGBE-treated rats showed no significant (P>0.05) difference when compared to the crush group at week 6 (Fig 7D).

The calculated myelin thickness for the crush and all treated groups at both time points following sciatic nerve injury showed a significant (p<0.0001) decrease as compared to the naive and sham groups (Fig 7E). However, myelin thickness obtained from G-B and GBE-treated groups showed a significant increase (p<0.0001) as compared to the crush group at both time points. Seemingly, TEGBE-treated group displayed no significance as compared to crush group at both time points.

Only one study reported the positive effects of GBE on nerve regeneration and functional recovery in an animal model of crush sciatic injury [38]. However, this study lacks the histological and immunohistochemical analyses of the spinal cord following injury. Also, no thorough investigation on neurotherapeutic effects of GBE isolated constituents, such as G-B, was reported in that study.

As such, further confirmation of the morphometric findings was obtained from the axon and nerve fiber diameter distribution at weeks 3 and 6 post injury (Fig 8A–8D).

The crush group displayed a remarkable shift to the left in the distribution of both axon and nerve fiber diameters as compared to the sham animals at week 3 and week 6. Although all the treated groups showed an improvement pattern of the axon and nerve fiber diameters distributions, the G-B-treated group displayed more normalization of axon and nerve fiber diameters distributions as compared to TEGBE and GBE-treated groups at week 3 and week 6. So, according to the qualitative and quantitative analysis, crushed sciatic nerves in animals treated with G-B were recovered in a quick manner as compared to other treated groups. These findings strongly support that G-B effects on sciatic nerve regeneration might be through the involvement PAF mechanism. Additionally, this study showed that the TEM ultrastructural findings were coherent with those obtained from the stereological analysis on the sciatic nerve following the administration of GBE, TEGBE, and G-B treatments.

As supportive data to the stereological analysis, western blotting of myelin basic protein (MBP) in sciatic nerves obtained from naive, sham, crush, G-B and TEGBE-treated groups showed that treatment with G-B and TEGBE (Fig 7F) either protected or increased the MBP as compared to the crush group. A significant (P<0.0001) decrease in the MBP content in the crush as compared to all other groups was shown. In contrast, sciatic nerves from G-B-treated group exhibited significantly (P<0.05) high MBP content as compared to TEGBE-treated group. Western blotting of MBP was also performed to confirm the morphological analysis in sciatic nerve injury treated with G-B as compared to the crush, sham, and naive groups.

Nissl staining

A qualitative analysis of cresyl-violet-stained spinal cord sections clearly showed that Nissl substances in the crushed sections were absent in most of the dorsal (sensory) (Fig 9C and 9D) and ventral (motor) neurons along with the chromatolysed neurons.

Moreover, the crush cross-sections displayed abundant irregular neurons and eccentric nucleus of motor neurons in the ventral horn at week 3 following injury. Further, sections obtained at week 6 showed remarkable Nissl-devoid neurons as compared to sham group. In contrast, sham group showed normal distribution of Nissl substance in all neurons with clear nuclei in both dorsal and ventral horns at week 3 and week 6 post-injury.

The spinal cord sections obtained from treated groups with G-B, TEGBE and GBE at week 3 following sciatic injury displayed more recovered Nissl stained neurons than the crush group, and more abundant trapezoidal shape of motor neurons in ventral horn. However, dorsal neurons in G-B-treated group (Fig 9) exhibited a noticeably higher number than other treated groups (Fig 9E and 9I) at week 3 post-injury. At week 6 post-injury, sections obtained from the treated groups displayed much more Nissl stained neurons as compared to crush spinal cord sections (Fig 9F, 9H and 9J). These current findings were in agreement with previous investigations using Nissl staining analysis on G-B [39] and GBE [40] treated models of spinal cord injury.

Immunohistochemical analysis

The findings of Neuronal Nuclei (Neu-N) immunostaining neurons (Fig 10A) in both laminas (dorsal and ventral) revealed a clear difference between crush and sham groups, at weeks 3 and 6 following sciatic nerve injury.

Further, treatments with G-B, TEGBE and GBE showed remarkably higher numbers in Neu-N immunoreactive neurons than those in the crush group. Neu-N immunostaining and quantitative analysis indicate the survival of neurons in the spinal cord laminae in the treated groups as compared to the crush group. Fig 10B and 10C show highly significant (P<0.0001) decrease in neuronal counting in the crush group as compared to the sham group in both laminas. Expectedly, G-B, TEGBE and GBE-treated groups showed significant (P<0.05) higher number in Neu-N immunoreactive neurons in dorsal and ventral horns at both time points as compared to the crush group. Meanwhile, all treated groups showed no significant difference in Neu-N immunoreactivity as compared to the sham group.

The neuronal protective characteristics of GBE in the spinal cord revealed in the current study was in agreement with previous findings [40] that showed the positive influence of GBE on dorsal sensory neurons following peripheral nerve injury. G-B was suggested to be the most effective treatment as a neuronal protective when compared to TEGBE and GBE treatments in the ipsilateral spinal cord region based on this analysis (Fig 10B and 10C). It was suggested that the neuroprotective effects of G-B may be due to its potent PAF antagonistic activity in the CNS, and this may contribute to the current findings of the Neu-N analysis [41]. Further, this study demonstrated, for the first time, the positive effects of G-B, TEGBE and GBE on the quantitative analysis of Neu-N immunostained spinal cord neurons post sciatic crush injury.

Immunohistochemical analysis (IHC) of glial fibrillary acid protein (GFAP) indicated the state of astrocytes and glial cells [42]. Staining the spinal cord sections with GFAP (Fig 11A1–11A10) revealed a remarkable increase in astrocytes activity in the crush group among other groups.

Data analysis of GFAP (Fig 11B and 11C) revealed a significant (P<0.0001) increase in GFAP immunostained astrocytes in the crush as compared to the sham group in both laminas at both time points following sciatic nerve injury. The reactive astrocytes and gliosis, in the crush group, and the increased GFAP-immunoreactive distribution was similar to astrocyte hypertrophy observed in previous studies [43]. However, in the sham group, the astrocytes’ reactivity appeared to be completely quiescent, as consistent with a previous study [44]. Unlike, the crush group, the treated groups with GB and TEGBE showed a significant (P<0.0.001) decrease in GFAP immunostained neurons in the dorsal horn (Fig 11C). G-B-treated group showed no significant change (p>0.05) in the GFAP immunoreactive astrocytes intensity as compared to the sham rats in the dorsal and ventral horns at week 3 and week 6 post-injury. On the other hand, TEGBE and GBE-treated groups showed mixed results in the calculated mean values. The inhibition of glial hypertrophy in the spinal cord, by G-B, following sciatic nerve injury might be attributed to its neuroprotective mechanisms due to PAF antagonistic activity, which prevented the GFAP immunoreactivity in astrocytes [41].

Moreover, western blotting data for GFAP immunoreactivity (Fig 11D) supported the significant (P<0.0001) increase in astrocytes activity in the crush group as compared to the other groups. However, G-B and TEGBE-treated groups showed a substantial decrease in GFAP immunoblotting, but still significantly (P<0.001) higher than those of the naive and sham animals at weeks 3 and 6 post-injury. The protective effect of GBE on astrocytes of rat hippocampus was shown previously [45].

Further, immunohistological findings, reported before [46], showed the neuroprotection of GBE in a model of vascular dementia, which was accompanied by a reduction in the astrogliosis and a decrease in GFAP immunoreactivity. However, there were no reported studies on the effects of GBE on GFAP immunoreactivity in the spinal cord post crush sciatic injury.

Growth-associated proteins-43 (GAP-43) is a membrane protein involved in neuronal development and plasticity. It is also considered an essential factor for proper neuronal regeneration, and an integral component of the axotomy response of primary sensory neurons [47]. In addition to IHC tests, GAP-43 immunoreactivity photomicrographs (Fig 12A1–12A10) showed ubiquitously high expression of axonal sprouting in the crush group, unlike the treated groups, in both ventral and dorsal horns. GAP-43 data analysis (Fig 12B and 12C), showed significant (P<0.001) increase in GAP-43 intensity in the crush group as compared to sham group at both time points.

A previous study [48] supported the current data that GAP-43 immunoreactivity increased in a sciatic crush injury model indicating an active attempt by the injured spinal cord neurons to repair or protect themselves from the crush injury insult. This was supported by western blotting data analysis of GAP-43 immunoreactivity (Fig 12D) which showed a highly significant (P<0.0001) increase in GAP-43 expression in the crush group as compared to other groups. In contrast, G-B treatment successfully maintained the levels of GAP-43 in both the dorsal and ventral horns as compared to the sham group (p>0.05) at week 3 and week 6 following crush injury. Likewise, TEGBE and GBE-treated groups showed mixed results, however, a significant (P<0.01) decrease in GAP-43 intensity at week 6 in both dorsal and ventral horns was noticeable as compared to the crush group.

Furthermore, it was revealed that TEGBE and G-B treatments induced a substantial decrease in the GAP-43 immunoblotting but still significantly (P<0.05) higher as compared to naive animals at weeks 3 and 6 post-injury. The decrease in GAP-43 immunoreactivity in the current findings by GBE, TEGBE and G-B was in full agreement with previous studies, thus indicating the central prevention of axonal sprouting following sciatic crush injury [49]. Further, the GAP-43 decrease may reveal that the spinal neurons are well protected by the treatments and survived the crush injury insult. The intensity of the GAP-43 immunoreactivity was highly ameliorated by G-B more than GBE and TEGBE treatments, indicating a more potent central neuroprotective effects for G-B.

To date, there were no reported studies on the GAP-43 immunohistological analysis conducted following the administration of G-B, TEGBE, or GBE after sciatic nerve crush injury.

The current investigations of IHC were consistent with previous studies examined the neuroprotective properties of GBE on spinal cord neurons both in vitro and in vivo, [50,51] and neurogenerative diseases, such as Alzheimer’s and Parkinson’s diseases [52].

Although GBE and TEGBE are shown to be effective in the recovery of the sciatic nerve following crush injury, G-B was shown to be the most effective treatment. Such evidence might be due to the peculiar pharmacokinetic properties of G-B, as it was reported to cross the blood-brain barrier in an active form and at adequate concentrations when compared to other ginkgolides [53]. Pharmacokinetic studies showed that ginkgolide C concentration in plasma is negligible. Additionally, the effects of the terpene trilactones were shown to be due to only ginkgolides A and B and bilobalide [54]. Furthermore, another study [55] reported that not all flavonoids were centrally absorbed.

Therefore, the reasons that GBE was less effective in protecting the spinal cord neurons as compared to the terpene trilactones-enriched extract (TEGBE) and the isolated pure G-B may be due to its poor absorption into the CNS.

In conclusion, this study presented a strong evidence for GBE and its ginkgolides, in particular G-B, on the early recovery of the crushed nerve, neuronal protection, and modulation of the histopathological alterations of the spinal cord milieu in the sciatic crush nerve rat model.

Nonetheless, further experiments should be performed to clarify the mechanism of action of GBE, G-B, as well as other ginkgolides. Future investigations might be required to study the role of GBE and G-B on Schwann cells and neurotrophic factors. Also, experiments should be planned to compare G-B or other ginkgolides, or even GBE, with an approved effective standard therapy on crush sciatic nerve injury model, such as vitamin B complex.

Supporting information

S1 Fig [pdf]
C NMR (150 MHz, CDOD) spectrum of DGA-I-81D.

S2 Fig [pdf]
DEPT 135° spectrum of DGA-I-81D.

S3 Fig [pdf]
HSQC spectrum of DGA-I-81D.

S4 Fig [pdf]
COSY spectrum of DGA-I-81D.

S5 Fig [pdf]
H NMR (600 MHz, CDOD) spectrum of DGA-I-81D.

S6 Fig [pdf]
HMBC spectrum of DGA-I-81D.

S7 Fig [5000x]
Electron microscopy.