Abstract

   Neuropathic pain (NP) is a chronic and intractable disease that is
   widely present in the general population. It causes painful behavior
   and even mood changes such as anxiety and depression by altering the
   metabolism of substances. However, there have been limited metabolomics
   studies conducted in relation to neuropathic pain. Therefore, in this
   study, the effects of NP on metabolites in serum and the dorsal root
   ganglion (DRG) were investigated using a non-targeted metabolomics
   approach detected by gas chromatography–mass spectrometry (GC-MS) and
   liquid chromatography–mass spectrometry (LC-MS) to uncover differential
   metabolites and affected metabolic pathways associated with NP. Sixty
   mice were divided into the following two groups: a chronic constriction
   injury (CCI) of the sciatic nerve group and a sham group (n = 30,
   each). After 7 days of CCI modeling, the metabolite profiles of serum
   and the DRG were analyzed using GC/LC-MS for both the CCI and sham
   groups of mice. Multivariate analysis revealed differential metabolites
   and altered metabolic pathways between the CCI and sham groups. In the
   CCI group, our findings provided insights into the complex
   phospholipid, amino acid and acylcarnitine metabolic perturbations of
   DRG metabolism. In addition, phospholipid metabolic disorders and
   impaired glucose metabolism were observed in the serum. Moreover, the
   metabolic differences in the DRG and serum were correlated with each
   other. The results from this untargeted metabolomics study provide a
   perspective on the metabolic impact of NP on serum and the DRG.

   Keywords: neuropathic pain, metabolomics, DRG, serum

1. Introduction

   Neuropathic pain, which affects approximately 6.9–10% of the general
   population, is characterized by pain resulting from damage to or
   disease of the somatosensory nervous system [[38]1,[39]2]. It can arise
   from a variety of sources such as trauma, inflammation, tumors,
   metabolic disorders, neurodegenerative diseases, and viral infection.
   Depending on the site of injury, neuropathic pain can be divided into
   peripheral and central neuropathic pain [[40]3]. Examples of peripheral
   neuropathic pain include pain following peripheral nerve injury,
   trigeminal neuralgia, and postherpetic neuralgia. Central neuropathic
   pain includes neuropathic pain after spinal cord or brain injury,
   multiple sclerosis, and chronic central post-stroke pain [[41]4]. The
   manifestations of neuropathic pain include hyperalgesia, allodynia, and
   spontaneous pain, which significantly affects the overall well-being
   and quality of life of patients. Despite its prevalence, the underlying
   mechanisms of neuropathic pain remain poorly understood. Consequently,
   further exploration of the mechanisms of neuropathic pain is urgently
   needed to facilitate the development of more effective treatments.

   Metabolomics, a rapidly developing field within biological systems, has
   emerged as a new technique following advances in genomics and
   proteomics. It focuses primarily on the comprehensive analysis of
   low-molecular-weight endogenous metabolites, such as fatty acids,
   carbohydrates, amino acids, lipids, and vitamins in organisms [[42]5].
   This field has been extensively explored across a wide range of
   applications to capture metabolic changes associated with various
   disease states [[43]6]. In addition, researchers can delve into the
   metabolic pathways of biological systems by looking at variations in
   metabolic components or contents. Currently, gas chromatography–mass
   spectrometry (GC-MS), liquid chromatography–mass spectrometry (LC-MS),
   and nuclear magnetic resonance spectroscopy (NMR) are the commonly
   employed technologies for non-targeted metabolomics [[44]7]. Due to the
   complementary nature of LC-MS and GC-MS for metabolic analysis, the
   combination of LC-MS and GC-MS was used to carry out a more
   comprehensive analysis of the metabolic data in our study.

   The dorsal root ganglion (DRG) is an influential relay station for the
   processing and transmission of peripheral pain to the central nervous
   system, and it plays an essential role in the onset and maintenance of
   neuropathic pain [[45]8]. In clinical work, neuropathic pain is mainly
   measured by patient’s subjective description and pain scale, but lacks
   objective indicators. Blood is the most commonly used test substance to
   react to the disease, which is less harmful to the human body. There
   have been numerous studies on DRG and serum organization in neuropathic
   pain, but few have explored metabolic changes in this condition and the
   connection between DRG and serum metabolites. Therefore, the purposes
   of this study are to explore the metabolite changes in the DRG and
   serum, then clarify whether there is a connection between the DRG and
   serum metabolite changes in neuropathic pain mice. The results will
   provide us with a new perspective for future clinical research. In this
   study, we used GC/LC-MS metabolomics techniques to detect the
   metabolites in the DRG and serum and used statistical methods to
   explore the differential metabolites and metabolic pathways in the CCI
   model.

2. Materials and Methods

2.1. Animals

   Sixty male C57BL/6 mice (8 weeks old) were purchased from the
   Laboratory Animal Center of Shanghai Jiao Tong University and divided
   into a CCI group and a sham group (n = 30, each group). All
   experimental protocols were approved by the Experimental Animal
   Management Ethics Committee of Shanghai General Hospital, Shanghai Jiao
   Tong University, School of Medicine (No. 2019AW009). In addition, all
   animal studies were performed in accordance with the ARRIVE guidelines.
   The mice were bred and maintained in a standard animal room with a 12 h
   day and 12 h night cycle, and given free access to food and water.

2.2. CCI Model

   The chronic constriction injury of the sciatic nerve (CCI) model was
   established according to a previous study [[46]9]. In brief, the mice
   were anesthetized by intraluminal injection of 10% chloral hydrate;
   then, the left sciatic nerve of the mice was carefully isolated under
   the mid-femur, and through the use of three segments of 5–0 chromic gut
   sutures to loosely ligate the sciatic nerve with 1 mm intervals. In the
   sham group, the sciatic nerve was exposed in the same way to the CCI
   mice but without ligations.

2.3. Behavior Test

   The mechanical withdrawal threshold (MWT) of the mice was detected
   using von Frey filaments (NC12775, North Coast Medical Company,
   Carlsbad, CA, USA); the protocol was described previously [[47]10].
   Before testing, the mice were placed in an acrylic box on a metal mesh
   floor to be quiet for 30 min. Then, the von Frey filaments (0.16–2.0 g)
   were used to stimulate the plantar surface of the left hind paw
   vertically until the filament was bent for 6 s. Brisk withdrawal or paw
   flinching were considered to be positive responses. The MWT of mice was
   detected using the “up-down” method; the filament was started at 0.16 g
   and increased until the mice exhibited positive responses.
   Subsequently, lower-weight filaments were used to evaluate the sensory
   threshold.

   The thermal withdrawal latency (TWL) of the mice was detected via the
   Hargreaves method (Ugo Basile SRL, Gemonio, VA, Italy). The mice were
   placed in a box on a glass platform to be quiet for 30 min. Heat was
   applied to stimulate the left hind paws of the mice; lifting or licking
   of the left hind paw was considered a pain response. The time from the
   onset of the heating stimulus to the appearance of the pain response
   was considered TWL. To avoid tissue damage, a 20 s cut-off time was
   set. Each mouse was tested 3 times with a 5 min interval between
   consecutive stimuli. The average value of 3 tests was taken as the TWL
   of the mice.

2.4. Sample Preparation

   After 7 days of surgery, DRGs from the left L4–L6 segments of 10 mice
   from each group were collected together and used as one sample with a
   weight of approximately 20 mg. The blood of each mouse was collected
   separately by removing the eyeball after anesthesia and leaving it at
   room temperature for 30 min; then, the serum obtained was the
   supernatant of the blood after 1200 rpm centrifugation. The serum of
   six mice in the sham group and eight mice in the CCI group showed
   obvious hemolysis, therefore, the serum of these mice was eliminated.
   The serum of 5 mice (30 μL of each mouse) in each group was mixed to a
   150 μL sample for detection. The protocol of the experiment is shown in
   [48]Figure 1.

Figure 1.

   [49]Figure 1
   [50]Open in a new tab

   The flow chart of experiment.

   Approximately 20 mg of DRG tissue was accurately weighed and
   transferred to a 1.5 mL Eppendorf tube, and 600 μL mixed solution of
   methyl alcohol was added: water (4:1, v/v, containing 4 μg/mL L-2
   chlorphenylalanine); then, the DRG sample was ground at 60 Hz for 2
   min, sonicated in an ice-water bath for 10 min, and stored at −20 °C
   for 30 min; the extract was centrifuged at 12,000 rpm for 10 min.
   Approximately 150 μL of serum was added to a 1.5 mL Eppendorf tube with
   2 μL of L-2-chlorophenylalanine (0.3 mg/mL) dissolved in methanol;
   then, the tube was vortexed to mix the contents; subsequently, 600 μL
   of an ice-cold mixture of methanol and acetonitrile (2/1, v/v) was
   added, and the mixture of serum was extracted via ultrasonic for 10 min
   in an ice-water bath; the extract was centrifuged at 12,000 rpm for 10
   min. All steps were performed at 4 °C.

   GC-MS: After centrifugation, the supernatant (150 μL) was dried in a
   freeze concentration centrifugal dryer. Approximately 80 μL of
   methoxamine hydrochloride pyridine solution (15 mg/mL) was added to the
   evaporated sample. The resultant mixture was vortexed for 2 min and
   incubated at 37 °C for 60 min. Then, the mixture with 50 μL of BSTFA
   and 20 μL of n-hexane was added. This mixture was subjected to
   vortexing for 2 min and then derivatized at 70 °C for 60 min.
   Subsequently, the samples were allowed to stand at ambient temperature
   for 30 min prior to their analysis using GC-MS. LC-MS: The supernatant
   (150 μL) from each sample was collected after centrifugation; then, the
   collected supernatant was filtered through 0.22 μm microfilters before
   being transferred to LC vials. These vials were then stored at −80 °C
   until the LC-MS analysis was performed.

2.5. Analysis Conditions of GC/LC-MS

   GC-MS: An Agilent 7890B gas chromatography system (Agilent Technologies
   Inc., Santa Clara, California, USA) was used to analyze the samples.
   The separation of derivatives was achieved using a DB-5MS fused-silica
   capillary column from Agilent J & W Scientific (Folsom, CA, USA).
   Throughout the analysis, helium served as the carrier gas, with a
   constant flow rate of 1 mL/min through the column being maintained. To
   achieve optimal conditions for analysis, the injector temperature was
   maintained at 260 °C, and the injection volume was 1 μL. The
   chromatographic temperature program commenced with the initial oven
   temperature held at 60 °C for 0.5 min, followed by a gradual increase
   to 125 °C at a rate of 8 °C/min, subsequently to 210 °C at a rate of 5
   °C/min, further to 270 °C at a rate of 10 °C/min, and finally reaching
   305 °C at a rate of 20 °C/min, where it was held for 5 min. For the
   mass spectrometry (MS) analysis, the MS quadrupole temperature was set
   at 150 °C, and the ion source temperature was set at 230 °C. A
   collision energy of 70 eV was applied during the MS analysis. Mass
   spectrometric data were acquired in the full-scan mode, covering a
   range of m/z 50–500.

   LC-MS: The metabolic profiling analysis was conducted using an ACQUITY
   UPLC I-Class system from Waters Corporation (Milford, CT, USA) in both
   electrospray ionization (ESI) positive and ESI negative ion modes. In
   the analysis, mobile phase A was composed of water and mobile phase B
   was a mixture of acetonitrile and methanol (2/3 v/v), with both
   containing 0.1% formic acid. Linear gradient: 1% B at 0 min, 30% B at 1
   min, 60% B at 2.5 min, 90% B at 6.5 min, 100% B at 8.5 min, held at
   100% B until 10.7 min, then returned to 1% B at 10.8 min, and finally
   maintained at 1% B until 13 min. During the analysis, the flow rate was
   set at 0.4 mL/min, and the column temperature was maintained at 45 °C.
   Data acquisition was carried out using a combination of the full scan
   mode, covering m/z ranges from 50 to 1000, and the MSE mode. The MSE
   mode involved acquiring two independent scans with different collision
   energies (CE) alternately during the analysis. The mass spectrometry
   parameters employed during the analysis were set as follows: A
   low-energy scan with a collision energy (CE) of 4 eV and a high-energy
   scan with a CE ramp ranging from 20 eV to 45 eV were applied to induce
   fragmentation of the ions. Argon gas was used as the collision-induced
   dissociation gas. For data acquisition, a scan time of 0.2 s was used,
   followed by an interscan delay of 0.02 s. The capillary voltage was set
   to 2.5 kV, and the cone voltage was maintained at 40 V. The source
   temperature was set to 115 °C, while the desolvation gas temperature
   was maintained at 450 °C. The desolvation gas flow rate was 900 L/h.

2.6. Data Preprocessing and Statistical Analysis of GC/LC-MS

   The GC/MS data were processed using MS-DIAL software, which facilitated
   various analytical steps, including peak detection, peak
   identification, peak alignment, wave filtering, and missing value
   interpolation. By referencing the LUG database, metabolite
   characterization was achieved, and a data matrix was generated as a
   result. To ensure data reliability and accuracy, all peak signal
   intensities within each sample were segmented and normalized based on
   the standards, taking into account the relative standard deviation
   (RSD) greater than 0.3 after screening. The subsequent step of peak
   merging and redundancy removal was carried out, leading to the final
   formation of the data matrix.

   The original LC-MS data underwent processing in Progenesis QI V2.3
   software (Nonlinear Dynamics, Newcastle, UK). Subsequently, a
   qualitative analysis was performed by identifying the compounds using
   the Human Metabolome Database (HMDB), lipid maps, and self-built
   databases, based on the precise mass-to-charge ratio (m/z), secondary
   fragments, and isotopic distribution. Compounds with scores below 36
   (out of 80) points were considered inaccurate and excluded.
   Subsequently, a data matrix was generated by combining the positive and
   negative ion data.

   Data collected from LC-MS and GC-MS were used for the multivariate
   statistical analysis. To distinguish between the differing metabolites
   in the groups, orthogonal partial least-squares-discriminant analysis
   (OPLS-DA) was employed. The model’s validity was assessed using the
   model parameters R2X(cum), R2Y(cum), and Q2(cum). Additionally, SPSS
   25.0 software was utilized for the univariate statistical analysis,
   including Student’s t-test and fold change (FC) analysis. To rank the
   overall contribution of each variable to group discrimination, variable
   importance of projection (VIP) values derived from the OPLS-DA model
   were utilized. Differential metabolites were selected based on VIP
   values greater than 1.0 and p-values less than 0.05, and their
   representation was displayed through a volcano map and a cluster map.
   Furthermore, the metabolic pathway enrichment analysis of these
   differential metabolites was conducted using the Kyoto Encyclopedia of
   Genes and Genomes (KEGG) pathway database.

3. Results

3.1. Behavioral Changes in the Mice

   The baseline values of MWT and TWL on the day before surgery showed no
   significant difference between the sham and CCI groups. After 7 days of
   surgery, the mice in the CCI group exhibited heightened sensitivity to
   mechanical and thermal stimuli compared to those in the sham group
   ([51]Figure 2). These results confirm the successful establishment of
   the CCI model.

Figure 2.

   [52]Figure 2
   [53]Open in a new tab

   The behavioral changes in the mice. Compared to the sham group, the MWT
   (A) and TWL (B) of the CCI mice significantly decreased after surgery
   (n = 6, * p < 0.05, *** p < 0.001).

3.2. Typical Metabolic Spectrums of the DRG and Serum

   The total ion chromatograms of the DRG and serum samples were acquired
   by GC-MS, and the base peak chromatograms (BPIs) of the DRG and serum
   samples were acquired by LC-MS in the negative and positive ion modes
   ([54]Figure S1). A strong signal response was observed in all samples,
   with a large number of metabolites detected over the entire run time.

3.3. The Differences between the CCI and Sham Groups in the DRG and Serum

   To analyze the data distinction between the two groups, the OPLS-DA
   model was constructed to analyze the GC-MS and LC-MS data for the DRG
   and serum. In the OPLS-DA model score plots, there was significant
   separation both in the GC-MS and LC-MS data on the DRG and serum
   between the CCI and sham group ([55]Figure 3). Both in the DRG and
   serum tissues, the model parameters R2X(cum), R2Y(cum), and Q2(cum)
   ([56]Table 1) of the two models were all greater than 0.5 and close to
   1, indicating that the fitting accuracy of the model was excellent and
   that the two groups had significant differences. Moreover, we found
   that the differences within each group were tiny, indicating that the
   metabolites of the mouse DRG and serum were different in the two
   groups.

Figure 3.

   [57]Figure 3
   [58]Open in a new tab

   OPLS-DA score plot of LC and GC in the DRG and serum. (A) OPLS-DA score
   plot of the DRG-GC. (B) OPLS-DA score plot of the DRG-LC. (C) OPLS-DA
   score plot of the serum-GC. (D) OPLS-DA score plot of the serum-LC.
   DRG-GC and serum-GC: the GC-MS data of the DRG or serum. DRG-LC,
   serum-LC, the LC-MS data of the DRG or serum.

Table 1.

   Model parameters for the DRG.
                  DRG       Serum
   Parameters GC-MS LC-MS GC-MS LC-MS
   R2X(cum)   0.720 0.730 0.718 0.781
   R2Y(cum)   1.000 1.000 1.000 1.000
   Q2(cum)    0.887 0.932 0.848 0.879
   [59]Open in a new tab

   R2X(cum), explanation rate in the X-axis. R2Y(cum), explanation rate in
   the Y-axis. Q2(cum), cumulative prediction rate.

3.4. The Differential Metabolites between the CCI and Sham Groups in the DRG
and Serum

   The different metabolites in the CCI and sham groups were analyzed by
   t-test, and volcano plots of the mice’s DRG and serum metabolites were
   produced to visually display the changed metabolites ([60]Figure 4).
   There were 336 metabolites of the DRG and 350 metabolites of the serum
   in the two groups showing differences (p < 0.05). In addition, the VIP
   value obtained from the OPLS-DA was used to extract the significant
   differential metabolites. Notably, metabolites with VIP > 1.0, p <
   0.05, and FC > 1 exhibited a significant increase, while metabolites
   with VIP > 1.0, p < 0.05, and FC < 1 showed a significant decrease.
   There were 51 significant differential metabolites identified in the
   mice DRG sample between the CCI and sham groups, including organic
   acids, lipids, and nucleosides ([61]Table 2). Similarly, there were 44
   significant differential metabolites also observed in the serum sample
   between the two groups, mainly including organic acids, lipids,
   carbohydrates and nucleosides ([62]Table 3). In addition, the
   differential metabolites of the DRG and serum are visually displayed in
   heatmaps ([63]Figure 5). The above results indicate that there were a
   large number of differential metabolites in the DRG and serum in the
   CCI group and sham group.

Figure 4.

   [64]Figure 4
   [65]Open in a new tab

   Metabolites of the DRG and serum detected by GC-MS and LC-MS. CCI vs.
   the sham. (A) Differential metabolites between two groups, based on
   GC-MS data from DRG. (B) Differential metabolites between two groups,
   based on LC-MS data from DRG. (C) Differential metabolites between two
   groups, based on GC-MS data from serum. (D) Differential metabolites
   between two groups, based on LC-MS data from serum. Blue dots: compared
   to sham group, down-regulated metabolites in CCI group, p < 0.05 and
   fold change < 1. Red dots: compared to sham group, up-regulated
   metabolites in CCI group, p < 0.05 and fold change > 1.

Table 2.

   List of DRG metabolites with significant changes in the CCI and sham
   groups.
   Metabolites VIP p-Value log2 (FC) FC Data Class
   Organic acids and derivatives N-Acetyl-L-Aspartic Acid (NAA) 2.032
   0.015 −0.679 0.625 GC
   2s-Amino-Butanoic Acid 1.643 0.047 −0.478 0.718 GC
   N-Acetyl-L-Glutamic Acid 1.452 0.045 −0.373 0.772 GC
   L-Gluconic Acid 1.586 0.047 −0.459 0.727 GC
   2-Hydroxyglutaric Acid 1.662 0.021 −0.467 0.724 GC
   N-Acetyl-1-aspartylglutamic acid (NAAG) 7.209 0.022 −0.502 0.706 LC
   L-Aspartic acid 4.407 0.046 −0.447 0.734 LC
   (1R)-Hydroxy-(2R)-glutathionyl-1,2-dihydronaphthalene 2.331 0.007
   −0.524 0.695 LC
   Trans-3-hydroxy-L-proline 1.152 0.002 −0.529 0.693 LC
   N1-Acetylspermidine 1.236 0.040 −0.518 0.698 LC
   Isocitric acid 4.202 0.006 −0.412 0.752 LC
   Lipids and lipid-like molecules Oleic Acid 4.172 0.002 2.623 6.160 GC
   14-Methylpentadecanoylcarnitine 8.562 0.044 −0.346 0.787 LC
   16-methylnonadecanoylcarnitine 2.403 0.027 −0.547 0.685 LC
   2-Hydroxyhexadecanoylcarnitine 9.042 0.008 −0.361 0.778 LC
   3-Hydroxyhexadecadienoylcarnitine 1.165 0.001 −0.835 0.561 LC
   3-Hydroxyarachidonoylcarnitine 1.430 0.016 −0.488 0.713 LC
   8-Hydroxyoctadecanoylcarnitine 6.045 0.007 −0.426 0.744 LC
   Arachidonoylcarnitine 5.844 0.043 −0.354 0.782 LC
   hydroxyoctadecenoylcarnitine 2.967 0.003 −0.473 0.721 LC
   Propionylcarnitine 3.042 0.013 −0.839 0.559 LC
   Stearoylcarnitine 9.281 0.010 −0.416 0.749 LC
   trans-Hexadec-2-enoyl carnitine 4.306 0.027 −0.436 0.739 LC
   MGDG(16:0/18:2(9Z,12Z)) 5.015 0.009 −2.552 0.171 LC
   Myristoleoylcarnitine 3.411 0.007 −0.546 0.685 LC
   L-Acetylcarnitine 10.425 0.038 −0.643 0.640 LC
   LysoPA(22:5(4Z,7Z,10Z,13Z,16Z)/0:0) 1.847 0.015 −3.526 0.087 LC
   LysoPC(22:1(13Z)/0:0) 1.746 0.012 −1.101 0.466 LC
   LysoPC(24:1(15Z)/0:0) 2.291 0.027 −4.447 0.046 LC
   O-Linoleoylcarnitine 6.888 0.002 −0.633 0.645 LC
   PA(13:0/22:2(13Z,16Z)) 1.508 0.030 −8.663 0.002 LC
   PA(15:0/22:1(13Z)) 1.221 0.003 −0.955 0.516 LC
   PA(P-16:0/13:0) 7.741 0.027 −2.308 0.202 LC
   PC(22:6(4Z,7Z,10Z,13Z,16Z,19Z)/17:1(9Z)) 6.316 0.015 −9.492 0.001 LC
   PE(15:0/18:1(9Z)) 2.907 0.010 −10.981 0.000 LC
   PE(20:2(11Z,14Z)/20:5(5Z,8Z,11Z,14Z,17Z)) 2.908 0.037 −12.171 0.000 LC
   PE(22:5(4Z,7Z,10Z,13Z,16Z)/P-16:0) 1.239 0.046 −23.919 0.000 LC
   PE(22:6(4Z,7Z,10Z,13Z,16Z,19Z)/22:5(7Z,10Z,13Z,16Z,19Z)) 2.854 0.024
   1.446 2.725 LC
   PE-Cer(d14:1(4E)/21:0) 4.508 0.044 −4.718 0.038 LC
   PE-Cer(d14:2(4E,6E)/23:0) 1.866 0.011 −2.636 0.161 LC
   PE-NMe(20:3(5Z,8Z,11Z)/22:5(4Z,7Z,10Z,13Z,16Z)) 1.501 0.013 −10.238
   0.001 LC
   PE-NMe2(16:1(9Z)/20:5(5Z,8Z,11Z,14Z,17Z)) 1.821 0.030 −5.815 0.018 LC
   PE-NMe2(22:5(4Z,7Z,10Z,13Z,16Z)/18:4(6Z,9Z,12Z,15Z)) 2.677 0.047 −1.459
   0.364 LC
   PE-NMe2(22:1(13Z)/14:1(9Z)) 1.060 0.018 −0.304 0.810 LC
   PC(O-10:1(9E)/0:0) 1.308 0.009 0.727 1.656 LC
   PS(22:6(4Z,7Z,10Z,13Z,16Z,19Z)/0:0) 2.492 0.001 −0.257 0.837 LC
   Medica 16 6.398 0.009 −7.468 0.006 LC
   Phosphoribosyl formamidocarboxamide 2.393 0.026 −1.570 0.337 LC
   Nucleosides FAD 1.949 0.043 −0.217 0.860 LC
   and PE(18:2(9Z,11E)+=O(13)/22:0) 1.546 0.031 −24.503 0.000 LC
   others PG(18:1(9Z)/20:4(6Z,8E,10E,14Z)-2OH(5S,12R)) 1.768 0.005 −2.866
   0.137 LC
   [66]Open in a new tab

   VIP, variable important in projection; p-value, significance of
   difference; FC, fold change.

Table 3.

   List of serum metabolites with significant changes in the CCI and sham
   groups.
   Metabolites VIP p-Value log2 (FC) FC Data Class
   Organic acids and derivatives (S)-3,4-Dihydroxybutyric Acid 1.433 0.006
   0.589 1.505 GC
   L-Glutamine 3.172 0.029 2.380 5.204 GC
   Methoxyacetic Acid 1.654 0.022 −0.850 0.555 GC
   N-Formylglycine 1.362 0.023 −0.573 0.672 GC
   L-Carnitine 6.030 0.002 0.481 1.396 LC
   Leukotriene C5 4.168 0.019 −0.484 0.715 LC
   Proline betaine 1.449 0.014 0.517 1.431 LC
   Protocatechuic acid 3-O-sulfate 1.048 0.001 −0.911 0.532 LC
   Symmetric dimethylarginine 1.136 0.017 0.617 1.533 LC
   Carbohydrates 2-Deoxy-D-Ribose 2.222 0.045 1.521 2.870 GC
   2-Ketoglucose 1.385 0.047 −0.638 0.642 GC
   D-Ribose 1.101 0.031 −0.389 0.764 GC
   Erythronic Acid 1.405 0.032 0.611 1.527 GC
   Glucose 1.478 0.026 0.663 1.583 GC
   N-Acetylmannosamine 1.274 0.025 0.496 1.410 GC
   D-Fructose 5.918 0.010 0.339 1.265 LC
   Invert sugar 2.296 0.037 0.410 1.328 LC
   Lipids and lipid-like 3-Beta-D-Galactosyl-Sn-Glycerol 1.400 0.012 0.582
   1.497 GC
   (21-Methyl-8Z-pentatriacontene 1.526 0.022 0.426 1.344 LC
   molecules GM4(d18:1/16:0) 11.617 0.023 −0.527 0.694 LC
   hexadecanedioic acid mono-L-carnitine ester 2.327 0.001 2.928 7.612 LC
   LysoPC(0:0/16:0) 20.970 0.024 −0.415 0.750 LC
   LysoPC(17:0/0:0) 1.008 0.009 −0.377 0.770 LC
   LysoPC(18:0/0:0) 5.459 0.016 −0.301 0.812 LC
   LysoPC(18:2(9Z,12Z)/0:0) 2.495 0.037 −0.377 0.770 LC
   LysoPC(22:0/0:0) 2.696 0.006 2.161 4.473 LC
   Medica 16 1.381 0.042 7.823 226.438 LC
   Montiporic acid D 1.224 0.036 −0.544 0.686 LC
   N6-lauroyl lysine 1.290 0.007 0.278 1.213 LC
   PA(22:1(13Z)/19:2(10Z,13Z)) 1.278 0.048 1.090 2.129 LC
   PA(P-16:0/13:0) 2.816 0.020 2.471 5.543 LC
   PC(0:0/20:4;O) 1.087 0.027 −0.581 0.668 LC
   PC(18:3(9Z,12Z,15Z)/20:4(8Z,11Z,14Z,17Z)) 4.044 0.026 13.300 ### LC
   PC(20:4(8Z,11Z,14Z,17Z)/20:4(8Z,11Z,14Z,17Z)) 3.861 0.032 9.718 842.057
   LC
   PC(O-16:0/0:0) 1.592 0.034 −0.329 0.796 LC
   PE(18:1(9Z)/20:5(5Z,8Z,11Z,14Z,17Z)) 5.844 0.002 8.141 282.319 LC
   PE-Cer(d14:1(4E)/21:0) 1.536 0.039 26.618 ### LC
   PE-NMe(20:3(5Z,8Z,11Z)/22:5(4Z,7Z,10Z,13Z,16Z)) 1.784 0.011 11.105
   2202.369 LC
   PE-NMe(20:4(8Z,11Z,14Z,17Z)/18:0) 5.221 0.018 −1.535 0.345 LC
   PI(20:4(5Z,8Z,11Z,14Z)/18:0) 5.069 0.011 9.040 526.430 LC
   SM(d16:1/17:0) 4.910 0.007 2.992 7.956 LC
   Nucleosides and others Guanosine 3.322 0.040 2.412 5.324 GC
   PE(18:2(9Z,11E)+ = O(13)/22:0) 1.728 0.038 6.857 115.920 LC
   PG(18:1(9Z)/20:4(6Z,8E,10E,14Z)-2OH(5S,12R)) 1.628 0.001 3.281 9.717 LC
   [67]Open in a new tab

   VIP, variable important in projection; p-value, significance of
   difference; FC, fold change; ### >10,000.

Figure 5.

   [68]Figure 5
   [69]Open in a new tab

   Heatmap of differential metabolites between the CCI and sham groups in
   the DRG and serum. (A) Heatmap of differential metabolites in the DRG.
   (B) Heatmap of differential metabolites in serum. The colors indicate
   the expression abundance of the metabolites.

3.5. The Differential Metabolic Pathways between the CCI and Sham Groups in
the DRG and Serum

   Metabolic pathway enrichment analysis of differential metabolites was
   performed based on the KEGG database. Data on differential metabolites
   were imported into the KEGG database to investigate the weights of the
   metabolic pathways. The enrichment map and bubble chart of the
   metabolic pathways showed the top 20 differential metabolic pathways of
   the DRG ([70]Figure 6A,B) and serum ([71]Figure 6C,D). In the DRG,
   there were seven significantly different metabolic pathways between the
   sham and CCI groups (raw p < 0.05, impact > 0.05), mainly including the
   alanine, aspartate, and glutamate metabolisms; arginine biosynthesis;
   central carbon metabolism; and autophagy and choline metabolism, and
   details on the metabolic pathway analysis are presented in [72]Table 4.
   In the serum, there were 18 significantly different metabolic pathways
   between the sham and CCI groups (raw p < 0.05; impact > 0.05), mainly
   including the pentose phosphate pathway, choline metabolism,
   glycerophospholipid metabolism, central carbon metabolism, and so
   forth, and details on the metabolic pathway analysis are presented in
   [73]Table 5.

Figure 6.

   [74]Figure 6
   [75]Open in a new tab

   Differential metabolic pathways analysis between the CCI and sham
   groups. The differential metabolites were subjected to metabolic
   pathway enrichment analysis, with p < 0.05, VIP > 1; top 20 related
   metabolic pathways were analyzed. (A,C) Enrichment map: the red dummy
   line indicates a p-value of 0.01, and the blue dummy line indicates a
   p-value of 0.05. Signal pathways higher than the blue dummy line
   represent significant differences. (B,D) Bubble chart: the size of the
   dots represents the number of metabolites. RichFactor, the number of
   differential metabolites/total number of metabolites in this pathway.

Table 4.

   The details of the metabolic pathways of the DRG.
   Pathway Name Hit Match Status p-Value FDR Rich Factor
   Alanine, aspartate, and glutamate metabolism N-Acetyl-L-Aspartic Acid,
   N-Acetyl-1-aspartylglutamic acid, L-Aspartic acid 3/38 0.00023 0.00796
   0.10714
   Arginine biosynthesis N-Acetyl-L-Glutamic Acid, L-Aspartic acid 2/23
   0.00462 0.07847 0.08696
   Central carbon metabolism in cancer L-Aspartic acid, Isocitric acid
   2/37 0.01171 0.11442 0.05405
   Autophagy—other PE (polytypes) 1/3 0.01355 0.11442 0.33333
   Kaposi sarcoma-associated herpesvirus infection PE (polytypes) 1/5
   0.02248 0.11442 0.20000
   Neuroactive ligand–receptor interaction N-Acetyl-1-aspartylglutamic
   acid, L-Aspartic acid 2/53 0.02323 0.11442 0.03774
   Glycerophospholipid metabolism PE, LysoPC (polytypes) 2/56 0.02575
   0.11442 0.03571
   Autophagy—animal PE (polytypes) 1/6 0.02692 0.11442 0.16667
   Cysteine and methionine metabolism 2s-Amino-Butanoic Acid, L-Aspartic
   acid 2/66 0.03494 0.13200 0.03030
   Choline metabolism in cancer LysoPC (polytypes) 1/11 0.04884 0.16605
   0.09091
   [76]Open in a new tab

   Row p, the p value calculated from the enrichment analysis; FDR, false
   discovery rate.

Table 5.

   The details of the metabolic pathways of serum.
   Annotation Hit Match Status Row p FDR Rich Factor
   Pentose phosphate pathway 2-Deoxy-D-Ribose, Glucose, D-Ribose 3/36
   0.00050 0.02744 0.08333
   Choline metabolism in cancer LysoPC, PC (polytypes) 2/11 0.00104
   0.02849 0.18182
   Glycerophospholipid metabolism LysoPC, PC, PE (polytypes) 3/56 0.00183
   0.03360 0.05357
   Mineral absorption Glucose, L-Glutamine 2/29 0.00729 0.08020 0.06897
   Non-alcoholic fatty liver disease Glucose 1/2 0.00905 0.08171 0.50000
   Central carbon metabolism in cancer Glucose, L-Glutamine 2/37 0.01171
   0.08171 0.05405
   Diabetic cardiomyopathy Glucose, L-Carnitine 2/39 0.01296 0.08171
   0.05128
   Autophagy—other PE (polytypes) 1/3 0.01355 0.08171 0.33333
   Insulin signaling pathway Glucose 1/4 0.01803 0.08262 0.25000
   FoxO signaling pathway Glucose 1/5 0.02248 0.08833 0.20000
   Kaposi sarcoma-associated herpesvirus infection Glucose 1/5 0.02248
   0.08833 0.20000
   Autophagy—animal Glucose 1/6 0.02692 0.09255 0.16667
   Type II diabetes mellitus Glucose 1/6 0.02692 0.09255 0.16667
   Glutamatergic synapse Glucose 1/8 0.03575 0.11565 0.12500
   GABAergic synapse Glucose 1/9 0.04013 0.11616 0.11111
   AGE-RAGE signaling pathway in diabetic complications Glucose 1/9
   0.04013 0.11616 0.11111
   Prolactin signaling pathway Glucose 1/11 0.04884 0.13431 0.09091
   [77]Open in a new tab

   Row p, the p value calculated from the enrichment analysis; FDR, false
   discovery rate.

4. Discussion

   In this study, we explored the abnormal metabolism of the DRG and serum
   in CCI models using an untargeted metabolomics approach. We found 51
   metabolites in the DRG and 44 metabolites in serum that appeared to be
   significantly altered in the CCI group. These differential metabolites
   mainly include organic acids, carbohydrates, lipids and lipid-like
   molecules, nucleosides, and others.

   In the DRG metabolites, we reported reductions in metabolites
   associated with phospholipid biosynthesis. Phospholipids are the main
   constituents of the phospholipid bilayer of the cell membrane and
   maintain the normal activity and function of the cell. At the same
   time, phospholipids are also vital in mitochondrial function and myelin
   maintenance [[78]11,[79]12,[80]13]. PC is a phospholipid with elevated
   content in peripheral myelin [[81]14]. The myelin sheath is an
   important component of neurons, which plays a dual role in promoting
   nerve impulse conduction and preventing axonal injury [[82]15].
   Phospholipids contains phosphatidyl choline (PC), phosphatidyl
   ethanolamine (PE), phosphatidyl serine (PS), phosphatidyl inositol
   (PI), phosphatidyl glycerol (PG), and so on. In this study, we found
   that some PC, PE and PS contents were significantly downregulated in
   the CCI models. Additionally, PE-NME, a precursor of phosphatidic
   choline synthesis, and phosphatidic acid (PA), a precursor to
   phospholipid synthesis, were also significantly downregulated
   ([83]Figure 7A). However, the opposite result was observed in the
   serum, where the expression of sectional PC, PE, and PI was
   significantly upregulated, and the precursor substances for
   phospholipid synthesis, PA and PE-NME, were significantly upregulated
   ([84]Figure 7A). Moreover, among the differential metabolites, there
   were four identical metabolites that showed opposite trends in DRG and
   serum ([85]Figure 7B), which may be related to the metabolites in the
   DRG entering the blood, suggesting that some blood metabolites can
   reflect metabolic changes in the DRG. The above results indicate that
   NP may affect mitochondrial function or myelin maintenance by
   inhibiting phospholipid synthesis in the DRG, and studies have shown
   that structural abnormalities in either the mitochondria or myelin
   sheaths can cause pain behavior [[86]16,[87]17]. Therefore, NP may
   cause neuronal dysfunction and hyperalgesia by inhibiting phospholipid
   synthesis in the DRG. The opposite expression trends of differential
   metabolites in serum and the DRG suggest that there exists a close
   correlation between metabolic disorders of the blood and the DRG.

Figure 7.

   [88]Figure 7
   [89]Open in a new tab

   Differential expression of metabolites in the DRG and serum. (A,B)
   Differential expression of phospholipids in the DRG and serum. (C)
   Amino acids and oleic acid expressions in the DRG. (D) Acylcarnitine
   expression in the DRG. (E) Carbohydrates expression of serum. The
   colors represent the p-value, with red representing upregulation and
   blue representing a decrease. D, DRG. S, serum.

   Among the differential metabolites identified in the DRG, four amino
   acids, N-Acetyl-L-Aspartic Acid (NAA), N-Acetyl-L-Glutamic Acid,
   N-Acetyl-1-aspartylglutamic acid (NAAG), and L-Aspartic acid, were
   significantly decreased ([90]Figure 7C). NAA, a marker of neuronal
   density and viability, is mainly synthesized in neurons [[91]18]. A
   decreased NAA concentration suggests the possible neuronal damage or
   dysfunction, and several studies have identified decreased NAA as a
   clinical indicator of neurological dysfunction [[92]19,[93]20].
   Additionally, some studies have shown that the NAA reduction was also
   observed in the brain of patients with neuropathic pain
   [[94]19,[95]21], and the NAA level was positively correlated with the
   severity of neuropathic pain symptoms. In this study, we observed a
   decreased NAA expression in the DRG of CCI mice via metabolomics,
   suggesting that DRG neurons in NP mice may be damaged or dysfunctional.
   Studies have implied that NAA is synthesized from ASP in mitochondria,
   and then transported to the Schwann cells to synthesize myelin
   [[96]22]. However, the decrease in NAA in the DRG suggests a possible
   mechanism for demyelination in NP. NAAG is the third most common
   neurotransmitter in the nervous system behind glutamate and GABA
   [[97]23]. It inhibits the release of glutamate through the activation
   of metabotropic glutamate mGlu3 receptors at the presynaptic membrane
   and stimulates the release of neuroprotective growth factors through
   the activation of mGlu3 receptors on glial cells [[98]24]. Therefore,
   NAAG exerts neuroprotective effects by inhibiting excitotoxicity.
   Furthermore, it has been demonstrated that the inhibition of NAAG
   degradation can be used for the treatment of several diseases, such as
   animal models of traumatic brain injury and neuropathic pain [[99]24].
   The above indicates that NAGG has a protective effect on neurons. In
   this study, we found a significant decrease in NAAG in the DRG of the
   CCI mice, suggesting that it may reduce the inhibition of
   excitotoxicity to promote the progression of NP. Similarly, KEGG
   enrichment analysis showed significant changes in the alanine,
   aspartate, and glutamate metabolic pathways involved by NAA, NAAG, and
   L-aspartate ([100]Table 4). Studies have shown that the depletion of
   amino acids in the tissues suggests cell death or neurodegeneration
   [[101]25,[102]26,[103]27]. We observed a significant decrease in amino
   acids in the DRG of CCI model, which suggested that cell death or
   neural degeneration may occur in DRGs, but further studies are needed
   to confirm this.

   In the DRG metabolites, we observed a significant increase in oleic
   acid ([104]Figure 7C). Oleic acid, as a known endogenous neurotrophic
   factor, combined with albumin and suggests neuroprotective effects by
   inhibiting nociceptive reflexes in SCI, resulting in neuroprotective
   effects [[105]28]. One study has shown that extracellular albumin and
   oleic acid, when entering neurons, can promote dendritic growth
   [[106]29,[107]30,[108]31] and also promote sphingomyelin synthesis
   [[109]32,[110]33]. Furthermore, oleic acid inhibits the production of
   proinflammatory mediators in microglia cultures [[111]34] and can
   attenuate the pain response in patients with arthritis [[112]35].
   Therefore, oleic acid plays an important role in repairing nerve damage
   and attenuating pain responses. In this study, we found a significant
   increase in oleic acid in the CCI model, which could be a self-healing
   process after nerve injury, but its role in the DRG needs further
   investigation.

   Furthermore, we observed a significant reduction in acylcarnitine
   levels in the DRG of the CCI group ([113]Figure 7D). Acylcarnitine is
   located in the inner mitochondrial membrane, and its role is to
   transport the acyl of fatty acids from the cytoplasm to the
   mitochondria, so as to perform β-oxidation to produce energy; it is an
   important part of the energy metabolism [[114]36,[115]37]. In our
   study, we found a significant decrease in acylcarnitine levels in the
   CCI group, suggesting that NP may cause an energy metabolism disorder
   of the fatty acids in the DRG that affects the supply of energy to the
   cell.

   In the serum metabolites, we found that there were significant
   differences in most of the carbohydrates, such as glucose, between the
   sham group and the CCI group ([116]Figure 7E), and metabolic pathway
   analysis showed that the pentose phosphate pathway was the most
   significant differential metabolic pathway between the sham and CCI
   groups ([117]Figure 7C). In addition to providing energy, the pentose
   phosphate pathway also provides raw materials for the synthesis of
   various substances. It has been shown that the brain glucose metabolism
   is increased in rats with chronic pain [[118]38,[119]39], and a
   clinical investigation showed a significant association between
   patients with chronic pain and diabetes [[120]40]. In our study, we
   observed a significant increase in serum glucose in the CCI mice,
   suggesting that the increase in glucose may be caused by increased
   energy metabolism or the eating changes in the mice with pain behavior.
   In addition, the increased serum glucose further aggravated the pain
   response in turn. It is known that patients with diabetes suffer from
   peripheral neuropathic pain, which is caused by peripheral neuropathy
   caused by increased blood glucose [[121]41]. Studies have shown that
   diabetes leads to peripheral neuropathy by reducing phospholipid
   synthesis to induce myelin abnormalities in nerves [[122]42,[123]43].
   In this study, serum glucose and other carbohydrates of serum in the
   CCI group showed significant increases in the CCI group, suggesting
   that NP may lead to abnormal energy metabolism and elevated blood
   glucose in the serum; elevated blood glucose may further inhibit
   phospholipid synthesis and finally exacerbate the pain response.

   There are some limitations in our study: only male mice were used in
   this experiment to study the metabolite changes in DRG and serum in
   neuropathic pain, and the effect of gender on pain and metabolites was
   not explored. Further experiments are needed to test this. In this
   study, we simply observed the results of metabolite changes in the DRG
   and serum of NP mice, but failed to further explore the reasons for
   metabolite changes caused by NP, which is a limitation of this
   experiment. It has been shown that dietary strategies influence pain
   behavior in patients with chronic pain [[124]44,[125]45]. And lifestyle
   interventions may alter the gut microbiome and affect the pain response
   [[126]46]. Therefore, we guess that in addition to the direct nerve
   damage, the stress associated with the nerve injury and the changes in
   the animal’s activity, energy metabolism, and diet in turn affect the
   metabolites. These points require further study.

5. Conclusions

   Through untargeted metabolomics tested by GC-MS and LC-MS, our study
   found that peripheral nerve injury caused dramatic changes in the DRG
   and serum metabolites and the major metabolic pathway changes. Taken
   together, peripheral nerve injury results in significant alterations in
   phospholipid, amino acid, oleic acid, and acylcarnitine in the DRG and
   alterations in glucose in serum. Furthermore, some metabolites showed
   opposite trends in the DRG and serum, which implicated a close
   relationship between DRG and serum metabolites, and serum metabolites
   could reflect the metabolic changes in the DRG. In addition, the
   reasons for metabolite changes could be as follows: direct nerve
   damage, the stress associated with the nerve injury and the changes in
   the animals’ activity, and energy metabolism and diet affected by pain
   behavior. Moreover, the changed metabolites can reflect cellular
   structural disruption or abnormal energy metabolism, as well as
   postinjury self-protection. There are some limitations to this
   experiment, and structural and functional changes in tissues still need
   to be validated by further histomorphology and molecular biology
   studies.

Acknowledgments