AR, CR, PRA, JF, and ZR were purchased from Jiangsu Hospital of Integrated Traditional Chinese and Western Medicine. All the crude drugs were morphologically authenticated according to Chinese Pharmacopoeia (2010 Edition). The voucher specimens were deposited at Laboratory of Cellular and Molecular Biology, Jiangsu Province Academy of Traditional Chinese Medicine, Nanjing, China. Reference substances including gallic acid, paeoniflorin, calycosin-7-O-β-D-glucoside, rutin, coumarin, ononin, cinnamic acid, formononetin, astragaloside IV, 6-gingerol and 6-shogaol were purchased from Must Biological Technology Co. Ltd (Chengdu, China). The purity of each reference compound was over 98% by HPLC.

HPLC-grade acetonitrile and formic acid were purchased from ROE Scientific Inc. (United States). Deionized water was purified using a Milli-Q water purification system from Millipore (Bedford, MA, United States).

To prepare AC591 samples, AR, CR, PRA, JF, and ZR were mixed at a ratio of 2:1:1:1:1 and refluxed for 2 h with 10 volumes of water after maceration for 30 min. The water extract was filtered, and the residue refluxed again with 8 volumes of water for 2 h. The combined filtrates were desiccated to produce the AC591 extract at a concentration of 20 g crude drug/mL, and stored at -80°C. The extracts were filtered through a 0.2 μm PTFE syringe filter.

Qualitative analysis was performed by HPLC-MS/MS system consisting of Waters 2695 HPLC instrument and Quattro Premier XE MircoMass triple quadrupole tandem mass spectrometer (Waters Co., Milford, MA, United States). Chromatographic separation was carried out at 35°C on a Waters MS C₁₈ column (250 mm × 4.6 mm, 5.0 μm) using the following gradient profile: 5–15% B (0–0.15 min), 15–20% B (15–35 min), 20–40% B (35–65 min), 40–60% B (65–75 min), 60–100% B (75–90 min). The mobile phase was composed of 0.1% acetic acid (A) and 100% acetonitrile (B). The injection volume was 2 μL. Quantification analysis was carried out using HPLC-DAD method. The concentrations of major constituents were determined by an external standard curve.

Male Wistar rats (200–250 g) for oxaliplatin-induced peripheral neuropathy model and 6-week-old male BALB/c mice (21–23 g) for in vivo tumor growth model were feeded in Experimental Animal Center, Jiangsu Academy of Traditional Chinese Medicine (Nanjing, China). Animals had free access to food and water in their home cages. All experimental procedures were approved by the Animal Care and Use Committee at Jiangsu Academy of Traditional Chinese Medicine. Efforts were made to minimize the number of animals used and their suffering during the study.

Fifty male wistar rats were randomized into five experimental groups (n = 10): vehicle, oxaliplatin, oxaliplatin +AC591 (5 g/kg), oxaliplatin+AC591 (10 g/kg), and oxaliplatin+AC591 (20 g/kg). Oxaliplatin was obtained from Jiangsu Hengrui Medicine Co., Ltd (Jiangsu, China) and solubilized in 5% glucose solution. In vehicle and oxaliplatin groups, 5% glucose solution or oxaliplatin (4 mg/kg) was, respectively, injected intraperitoneally (i.p.) twice a week for 4 weeks (on days 1, 2, 8, 9, 15, 16, 22, and 23). In AC591-treated groups, in addition to oxaliplatin injection (4 mg/kg, i.p., twice a week), AC591 extract was administered orally daily for 4 weeks. The dosage of oxaliplatin was chosen based on a previous report (Ushio et al., 2012). Concurrently, the body weight of the rats was measured, and behavioral tests were performed blindly with respect to drug administration. At the end of the experiment, dorsal root ganglion (DRG) and sciatic nerves were harvested from all rats.

The mechanical allodynia induced by oxaliplatin was assessed by Von Frey hair test before the first administration of drugs (on day 0) and on days 7, 14, 22, and 25. The rats were placed in a clear plastic box (20 cm × 17 cm × 13 cm) with a wire mesh floor and allowed acclimatization for 30 min prior to testing. Von Frey filaments (Stoelting, Wood Dale, IL, United States) ranging from 1 to 26 g bending force were applied to stimulate the mid-plantar skin of each hind paw at 3–4 s intervals. The minimal pressure level [in gram (g)] that initiated a withdrawal response was recorded. A positive response was noted if the tested paw was withdrawn sharply during 5 of the 10 trials.

Cold hyperalgesia was assessed at 4°C using a cold plate. Cold hyperalgesia was measured as the number of foot withdrawal responses (lifting, shaking, or licking) in 1 min after the application of cold stimuli to the plantar surfaces of both the paws. The testing was repeated 3 times with an interval of approximately 5 min between each test. The cold hyperalgesia was also assessed by acetone test. Rats were placed in a clear plastic box (20 cm × 17 cm × 13 cm) and allowed to habituate for 30 min prior to testing. Acetone (50 μL) was sprayed onto the plantar skin of each hind paw, and the number of withdrawal response was counted for 40 s from the start of the acetone spray. The results were reported as the mean of 3 readings.

The noxious heat thresholds for heat hyperalgesia were measured using an increasing-temperature hot plate (IITC Life Science). After habituation, the rats were placed onto the plate, which was then heated from a starting temperature of approximately 30°C at a rate of 12°C/min until the animal showed nocifensive behavior. The corresponding plate temperature was considered as the noxious heat threshold. All results were reported as the mean of 2 readings from both the hind paws.

On day 28, sciatic nerves and DRG were harvested from rats anesthetized with sodium pentobarbital. The nerves were fixed in 10% paraformaldehyde, washed with saline, dehydrated by incubation in different concentrations of ethanol, and rinsed with xylene. Next, they were paraffin embedded, and the tissue was cut to a thickness of 5 μm, followed by hematoxylin and eosin (H&E) staining. Pathological changes were evaluated under a light microscope (BX51; Olympus Corp., Tokyo, Japan).

DRG tissues of rats from the vehicle, oxaliplatin, oxaliplatin+AC591 (20 g/kg) groups were subjected to total RNA extraction and microarray hybridization. The Rat 12 × 135K Gene Expression Array manufactured by Roche NimbleGen was used. Double-strand cDNA (ds-cDNA) was synthesized from total RNA using an Invitrogen SuperScript ds-cDNA synthesis kit in the presence of 100 pmol oligo dT primers. ds-cDNA was purified and labeled in accordance with the NimbleGen Gene Expression Analysis protocol (NimbleGen Systems, Inc., United States). The purified cDNA was quantified by a NanoDrop ND-1000. For Cy3 labeling of cDNA, the NimbleGen One-Color DNA labeling kit was used according to the manufacturer’s guidelines (NimbleGen Systems, Inc., Madison, WI, United States). Microarrays were hybridized at 42°C for 16 to 20 h with 4 μg of Cy3 labeled ds-cDNA in NimbleGen hybridization buffer/hybridization component A in a hybridization chamber (Hybridization System – NimbleGen Systems, Inc., Madison, WI, United States). Following hybridization, the microarrays were washed with the NimbleGen Wash Buffer Kit. The slides were scanned in an ozone-free environment using the Axon GenePix 4000B microarray scanner (Molecular Devices Corporation) piloted by GenePix Pro 6.0 software (Axon).

The scanned images (TIFF format) were then imported into NimbleScan software (version 2.5) for grid alignment and data analysis. The expression data were normalized through quantile normalization and the Robust Multichip Average (RMA) algorithm included in the NimbleScan software. The Probe level (^∗_norm_RMA.pair) files and Gene level (^∗_RMA.calls) files were generated after normalization. All gene level files were imported into Agilent GeneSpring GX software (version 12.0) for further analysis. Differentially expressed genes with statistical significance were identified through Volcano Plot filtering. Hierarchical clustering was performed with the R software (version 2.15). GO and pathway analyses were performed using the standard enrichment computation method.

CT26 (2 × 10⁶) cells were injected subcutaneously into the back of 6-week-old male BALB/c mice (21–23 g). When the tumors had grown to 200–300 mm³ size (day 10), mice were randomly assigned to five groups (n = 8 each). Vehicle group was treated by intragastrical saline (0.2 mL) for 2 weeks, and oxaliplatin group was processed by intraperitoneal oxaliplatin (10 mg/kg/week, 2 weeks, i.p.). The groups treated with oxaliplatin and AC591 were administrated oxaliplatin (10 mg/kg/week), and AC591 (7, 14, and 28 g/kg, respectively, i.g. daily for 14 days). Tumor size was measured with a caliper rule every 3 days, and the volume was calculated as follows: TV (mm³) = (L × W²)/2, wherein L is the longest and W the shortest radius of the tumor (mm). At the end of the treatment period, the mice were sacrificed, and tumors were weighed.

The clinical trial was reviewed and approved by the ethics committee of Jiangsu Hospital of Integrated Traditional Chinese and Western Medicine (2013LW029). This trial was registered at Chinese Clinical Trials Registry (ChiCTR-TRC-14004765) on 6th June 2014, and conformed to the CONSORT flowchart (Figure 6) and checklist. Written consent was obtained from each patient.

For animal experiments, the values are represented as mean ± SD. Statistical significance between the groups was analyzed by Student t-test. The differences were considered statistically significance at P < 0.05. For the human study, the rates of neurotoxicity were calculated for each grade and compared between trial and control group. The difference in the grades of neurotoxicity and tumor response between two groups was compared using Chi-square test and P < 0.05 defined as statistically significant.

The spectral profile of AC591 is presented in Figure 1, and the enumerated peaks from the chemical fingerprint are listed in Supplementary Table 1. A total of 36 individual phytochemical peaks were used to define the overall chemical fingerprint pattern, constituting >85% of the total ion current. Among the identified compounds, 25 were characterized using markers, exact mass comparison, and MS/MS fragmentation. Moreover, most of the molecular ions in the fingerprint arise from AR (41.67%) followed by PRA (27.78%), CR (16.67%), ZR (11.11%), and JF (2.78%). The 36 components, uniquely identified to 1 out of the 5 main herbs, could result in a signature of the composition, extraction method, and the herbal ratios. To standardize AC591, up to six marker compounds for main herbal ingredient were quantitatively analyzed: paeoniflorin (PRA), gallic acid (PRA), albiflorin (PRA), ononin (AR), cinnamic acid (CR), and 6-ginger (ZR). Using linear regression analysis on the commercial markers as standards, the amount of AC591 extract phytochemical (mg/g) was determined. The reproducibility, recovery, intra-day and inter-day analyses, and contents of the compounds are summarized in Table 1. The combination of chemical fingerprinting and quantitation of a subset of phytochemical markers provide a global quality control of AC591.

The effects of AC591 on body weight loss in the oxaliplatin-induced neuropathy model was evaluated. The results are shown in Figure 2A. It was found that the rats started to lose weight after the first injection of oxaliplatin, which reached statistical significance after the fourth dosage as compared to the vehicle-treated rats. None of the rats in the vehicle control group exhibited mortality during the experiment. While in the oxaliplatin-treated group (4 mg/kg), two rats perished after eight injections and eight other rats turned slim at the end of the treatment. Unexpectedly, AC591 (5, 10, and 20 g/kg) treatment did not attenuate oxaliplatin-induced body weight loss.

Oxaliplatin was reported to lower the rat’s threshold to mechanical stimuli which do not normally invoke pain (Ushio et al., 2012). Using the Von Frey apparatus, the withdrawal threshold to the non-noxious mechanical stimulus was estimated. As shown in Figure 2B, oxaliplatin caused a lowering of response threshold to mechanical stimuli. On day 25, the withdrawal threshold of the touched paw was significantly lower in rats treated with oxaliplatin (n = 10, 3.14 ± 0.89 g) than that in vehicle-treated rats (12.4 ± 2.94 g; P < 0.01). AC591 could significantly imped the alterations of the pain threshold evoked by oxaliplatin. After AC591 treatment, the withdrawal threshold to the non-noxious mechanical stimulus increased from 3.14 ± 0.89 g (oxaliplatin) to 9.75 ± 1.09 g (oxaliplatin + 20 g/kg AC591). Rats treated with AC591 in the absence of oxaliplatin did not show differences in the pain threshold as compared to the vehicle group (data not shown).

We studied cold hypoesthesia in rats injected with 4 mg/kg of oxaliplatin on a cold plate set at 4 ± 0.2°C. By 4 weeks, oxaliplatin markedly reduced the latency to escape (n = 10, 128 ± 15 s) vs. the vehicle alone (204 ± 23 s; P < 0.001) or AC591 alone (221 ± 19 s; P < 0.001). AC591 at a dose of 20 g/kg markedly abrogated the alterations induced by oxaliplatin (197 ± 25 s with oxaliplatin+AC591 vs. 128 ± 15 s with oxaliplatin alone; P < 0.001) (Figure 2C). In the acetone test, no significant changes in withdrawal responses were observed in animals receiving the vehicle (Figure 2D). In contrast, the number of withdrawal responses significantly increased in oxaliplatin-treated rats when compared with the vehicle group on days 4, 7, 14, and 23 (P < 0.05). AC591 prevented the cold hypoesthesia and exerted a more pronounced effect (P < 0.01) at a dose of 20 g/kg. There was no statistically significant difference in the hot response thresholds between the groups of rats at any time point after oxaliplatin treatment, consistent with the clinical observations (Figure 2E).

The behavioral tests demonstrated that chronic oxaliplatin treatment induced sensitivity to mechanical and cold stimuli. Next, whether oxaliplatin exhibited neurotoxic changes in the structure of peripheral neurons was determined histologically. Figures 3A,B illustrated the histopathological changes on DRG and sciatic nerve specimens obtained after the final drug dosage. HE staining of lumbar spinal cord (L4–L5) DRG presented decreased nucleolar size and significant alteration in nuclear and cell body diameter in rats treated with oxaliplatin alone. Changes in DRG morphology were also previously documented in rats after treatment with multiple doses of oxaliplatin and other platinum drugs (Cavaletti et al., 2001). AC591 at doses of 10 and 20 g/kg significantly prevented the decrease in the nucleolar size of neurons highlighted in oxaliplatin-treated animals (Figure 3A). Examination of sciatic nerve sections by light microscopy did not exhibit remarkable histological abnormalities during oxaliplatin treatment, which may be explained by the fact that the main pathological changes in oxaliplatin-induced neuropathy occur predominantly in DRG but not sciatic nerves (Figure 3B) (De Grandis, 2007).

We applied a microarray assay to detect 14,976 genes in total RNA samples of DRG specimens obtained from normal rats (n = 3), and PN rats treated with either AC591 (n = 3) or vehicle (n = 3). All samples passed quality control based on the inspection of the normalized intensity values, principle component analysis, and hybridization controls. Compared with normal rats, oxaliplatin-treated animals demonstrated ≥1.5-fold change in 782 genes and ≥2-fold change in 231 genes. Out of the 782 genes, 407 were annotated, and among them, 60 genes were related to immune or inflammatory response. In comparison with oxaliplatin, AC591 treatment corrected 391 (49.94%) out of 782 genes, and 181 genes among them were found to be significantly reversed (with >1.5-fold change in expression) (Figure 4A). A principle component analysis (PCA) plot using these 181 probe sets illustrated the variations in gene expression patterns between the various rat groups, and the AC591-treated ones were proximate to the vehicle controls (Figure 4B). A heat map showed 83 probe sets that were more abundant in the AC591-treated and vehicle control rats, while the other 98 oxaliplatin-induced probe sets were suppressed by AC591 (Figure 4C).

We next organized the AC591-affected genes into functional groups in order to obtain a better insight into the biological consequences of the gene expression changes. GO analysis using the DAVID program indicated that the highest proportion of the 391 genes were immune response (GO ID 6955) and defense response (GO ID 6952). As a result of AC591 administration, expression of a large number of genes related to wounding (17 genes, P = 6.8e^-3) and inflammation (11 genes, P = 8.4e^-3) was reduced to a level closer to that of the vehicle control. Other overrepresented GO pathways are shown in Table 2. KEGG pathway analysis was also used to discover the key signaling pathways and correlations between differentially expressed genes (Supplementary Table 2). Inflammation mediated by chemokine and cytokine signaling pathway (P = 0.0006) as well as Wnt signaling pathway (P = 0.0005) were identified as the most significant signaling mechanisms in enrichment analysis. The chemokine signaling pathway encompassed 13 differentially expressed genes: CCL4, CCL5, CCL6, CXCL10, CXCL11, CXCL13, CXCL16, CXCL9, HCK, NCF1, PIK3CD, STAT1, and VAV1.

In the present study, we also tested whether AC591 interfered with the tumor-reducing effects of oxaliplatin. Male BALB/c mice bearing CT26 tumors were randomly selected for treatment with saline, oxaliplatin (10 mg/kg/week, 2 weeks, i.p.) alone, and oxaliplatin+AC591 (7, 14, and 28 g/kg/d, i.g.). Figure 5A shows that oxaliplatin administration led to a substantial decrease in the body weight from day 6 after inoculation. Co-administration of AC591 exhibited no obvious effects on the weight loss induced by oxaliplatin. The mean tumor volume was similar in each of the five groups at the commencement of the experiment. On days 10 and 13, oxaliplatin significantly inhibited the increase of tumor volumes as compared to that of the vehicle in tumor cells implanted mice (P < 0.05, Figures 5B,C). AC591 did not attenuate, but rather resulted in the synergistic inhibition of the tumor growth as compared to oxaliplatin alone (P < 0.05, Figures 5B,C). After 2-week caliper measurement, the tumor was excised and weighed at 8.4 ± 2.1 g and 7.5 ± 1.0 g for vehicle control group and oxaliplatin-treated group, respectively. For oxaliplatin +AC591-treated group, tumor size averaged 5.2 ± 1.4 g, as significantly lower than that of vehicle (P < 0.001) or oxaliplatin-treated group (P < 0.05) (Figure 5D). These preliminary results suggested that AC591 did not attenuate, but rather enhance the anti-tumor activity of oxaliplatin.