Magnetic Resonance Spectroscopy Lipids Peak May Serve as a Potential Biomarker for Back Pain in Intervertebral Disc Degeneration: An Integrative Metabolomics and Proteomics Study Investigating the Role of the Lipid Droplets-Interleukin-17 Inflammatory Axis
Article information
Abstract
Objective
To explore and validate clinical magnetic resonance spectroscopy (MRS) biomarkers associated with patient-reported symptoms in intervertebral disc degeneration, and to further elucidate the pathogenic mechanisms linking these symptoms to MRS biomarkers via an integrative multiomics approach.
Methods
Patients categorized into the predominant lipids peak (pLP) group and the non-pLP group based on MRS spectrum lipids peak. Nucleus pulposus cells underwent lipidomics, proteomics and functional experiments. Outcome measures compared, and Pearson correlation coefficient evaluated relationships between symptoms, interleukin (IL)-17 immune-positive cells, and lipid contents. Multivariate linear analysis was employed to analyze the contributions of various variables to patient-reported symptoms.
Results
The pLP group exhibited significantly higher preoperative visual analogue scale (VAS)-back scores (6.5 vs. 4.7, p<0.01) and Oswestry Disability Index (ODI) scores (63.3% vs. 51.2%, p<0.01) compared to the non-pLP group. The multiomics analysis revealed the pLP group was characterized by lipid droplets accumulation in nucleus pulposus cells, and the activation of interleukin-17 (IL-17) inflammatory pathway. Preoperative VAS-back and ODI scores showed positive correlations with the expressions of IL-17 (r=0.555, p<0.001; r=0.566, p<0.001) and the relative lipid contents (r=0.567, p<0.001; r=0.561, p<0.001). Multivariate linear analysis revealed that percentage of IL-17 positive cells and the relative triglyceride contents were associated with preoperative VAS-back pain (p=0.021, p=0.046).
Conclusion
Patients with the MRS pLP spectrum showed reduced quality of life and upregulation of the lipid droplets-IL-17 inflammatory pathway in nucleus pulposus cells. Inflammatory factors contribute significantly to chronic low back pain development and progression, affecting patient-reported symptoms. The MRS lipids peak may serve as a potential biomarker for diagnosing and monitoring low back pain.
INTRODUCTION
Intervertebral disc degeneration (IDD) significantly contributes to the occurrence of low back pain, which is widely recognized as a primary cause of morbidity and physical disability on a global scale [1]. The deterioration of intervertebral discs can initiate a cascade of symptoms, including radiculopathy, chronic pain, and reduced mobility, profoundly impacting the quality of life for affected individuals [2]. Current treatment modalities for IDD encompass both conservative and surgical strategies; however, they often fall short in addressing the underlying biological issues associated with disc degeneration. As a result, many patients continue to endure persistent pain and disability. Studies suggest that symptomatic IDD accounts for approximately 40% of all cases of low back pain, highlighting the urgent need for effective therapeutic interventions [2,3]. Thus, comprehending the intricate relationship between IDD and chronic low back pain, as well as patient-reported symptoms, could offer significant insights for identifying novel therapeutic targets.
Biomarkers associated with low back pain have garnered significant attention in recent research, owing to their potential to enhance diagnostic accuracy and therapeutic outcomes for patients suffering from chronic low back pain due to IDD [4]. A substantial body of research has examined the roles of inflammatory cytokines, metabolic alterations, and imaging biomarkers in elucidating the pathophysiology of low back pain [5,6]. For example, a significant study investigated the serum concentrations of proinflammatory cytokines, with a focus on interleukin-6 (IL-6), in patients diagnosed with IDD. The results demonstrated a markedly elevated level of serum IL-6 in individuals experiencing low back pain relative to control subjects, indicating a potential systemic inflammatory response linked to IDD [1,4,7]. Furthermore, elevated levels of proinflammatory chemokines such as chemokine (C-C Motif) ligand 5 have been linked to IDD, indicating their potential as systemic biomarkers for diagnosing low back pain [8,9]. Nevertheless, the biomarkers commonly used to detect low back pain often necessitate the invasive extraction of blood or bodily fluids, a process that can cause patient trauma. Additionally, these samples must be subjected to further laboratory analysis, which imposes certain constraints on their clinical utility. Therefore, the integration of noninvasive biomarkers, laboratory-based biomarkers, and radiographic imaging has the potential to revolutionize the management of low back pain by balancing accuracy, patient comfort, and clinical feasibility.
Magnetic resonance spectroscopy (MRS) is a noninvasive method that utilizes the magnetic properties of atomic nuclei to quantitatively analyze the biochemical composition in the human body [10,11]. In clinical practice, MRS has been extensively employed for characterizing metabolic features within various tissues and organs [12]. Using ex vivo MRS, Keshari et al. [13] demonstrated that specific intradiscal metabolites, such as lactate and proteoglycan, can serve as quantifiable metabolic biomarkers for low back pain and IDD. Subsequent in vivo MRS analysis of intervertebral discs categorized into painful and nonpainful groups revealed distinct variations in lactate and proteoglycan levels [14]. In our studies, MRS spectra acquired from the degenerated nucleus pulposus (NP) tissues of IDD patients featured a higher lipids peak at approximately 1.3 ppm, indicating abnormal lipids accumulation [12]. The significance of these findings lies in the potential to develop new biomarkers for back pain based on metabolic profiles. The ability to noninvasively assess disc metabolic profiling in real time may provide clinicians with valuable insights into the underlying mechanisms of low back pain and guide treatment decisions.
Lipid droplets (LDs) are increasingly recognized as critical players in the regulation of inflammatory responses within various cell types, particularly in the context of metabolic and cardiovascular diseases [15]. These organelles not only serve as storage sites for neutral lipids but also participate actively in cellular signaling pathways that modulate inflammation [16]. For example, in endothelial cells, the formation of LDs is intricately linked to inflammation induced by lipopolysaccharides [17]. Notably, these LDs, which are composed of highly unsaturated lipids, are associated with pronounced inflammatory responses. This suggests that LDs may act as mediators of inflammation, contributing to the pathophysiology of diseases [15]. Moreover, LDs have been implicated in the immune response, where they play a role in the generation of proinflammatory mediators such as prostaglandins and leukotrienes, and tumor necrosis factor-alpha (TNF-α) [15,18]. LDs are not merely passive storage organelles; they are dynamic entities that play essential roles in mediating proinflammatory responses across various cell types, influencing both local and systemic inflammation. We have identified the accumulation of LDs in NP cells of patients with IDD [12,19]. However, the association between LDs in human NP cells, MRS lipids peak and patient-reported symptoms has not been investigated and remains unclear.
Herein, we aim to elucidate the associations of lipids and patient-reported symptoms, with a particular focus on the role of LDs accumulation and its association with inflammatory pathways. By integrating multiomics approaches, including MRS, metabolomics, and proteomics, we aim to identify specific lipid biomarkers and elucidate their interactions with inflammatory mediators that may contribute to the pathogenesis of low back pain and patient-reported symptoms. This integrative approach not only deepens our understanding of the disease but also has the potential to identify promising therapeutic targets and biomarkers for early diagnosis and prognosis.
MATERIALS AND METHODS
1. Patient and Samples
This study was approved by the Institutional Review Board (IRB) of The First Affiliated Hospital of USTC (IRB No. 2020ky19), and written informed consent was obtained from each patient. This study encompassed patients adhering to the following criteria: (1) chronic low back pain persisting for over 3 months, (2) diagnosed with herniated discs, spinal stenosis, or degenerative lumbar spondylolisthesis; (3) received monosegment interbody fusion surgery; (4) aged between 18 and 80 years. The exclusion criteria were as follows: (1) a history of previous lumbar surgery; (2) diagnosed with spinal fracture, scoliosis greater than 15 degrees, osteoporosis (T scores≤-2.5) or other known pathologies that could potentially cause back pain; (3) with lumbar instability, the motion of the lumbar spinal segments was assessed using flexion and extension radiography. Lumbar instability was defined as the presence of one or more of the following criteria: (1) translational displacement exceeding 8% or 3 mm; (2) angular motion greater than 10° [20,21]; (3) with narrowed intervertebral discs which were ineligible for MRS scanning. Due to the 5-mm voxel size defined for the axial plane (the height direction) of the scan, intervertebral discs with a height of less than 8 mm were excluded to ensure adequate spatial coverage for the MRS acquisition.
The degree of human NP tissue sample degeneration was graded using Pfirrmann grades applied on the basis of MR T2-weighted images: Grades I–II were collectively defined as the early-stage degeneration, and grades III–V were collectively defined as advanced- stage degeneration [19]. Modic changes were categorized into 3 types in accordance with Modic criteria [22]. The anterior disc height was determined by measuring the distance between the foremost points of the superior and inferior endplates. Meanwhile, the posterior disc height was calculated as the distance between the rearmost points of the upper and lower endplates [20]. Disc height was defined as the average of the anterior, central, and posterior heights of the intervertebral space.
NP samples were obtained from patients who received discectomy and fusion. The NP tissues were sliced into fragments and subsequently subjected to enzymatic digestion using 0.25% type II collagenase for a duration of 4 hours at a temperature of 37°C. Following the isolation process, the NP cells were resuspended in Dulbecco’s modified eagle medium (Gibco, USA) enriched with 10% fetal bovine serum, 100 units/mL of penicillin, and 100 mg/mL of streptomycin. These cells were then cultured at 37°C within a humidified incubator, ensuring an atmospheric composition of 95% air and 5% CO2. The culture medium was refreshed every 3 days. Primary NP cells were used for subsequent experiments. Moreover, the cells were employed for immunofluorescence staining in order to ascertain the expression of collagen II, thereby validating the identity of the collected cells as NP cells.
2. Patient-Reported Symptoms
The patient-reported symptoms were evaluated using the visual analogue scale for back (VAS-back) and leg (VAS-leg) pain, and the Oswestry Disability Index (ODI). The questionnaires were completed by the patients independently of the surgeons prior to surgery, at 3-month follow-up, and at 6-month follow-up [23].
3. MRS Analysis
Each patient was positioned supine for magnetic resonance examination by a team of experienced technologists according to standard protocol [10,12,19]. This study used a GE Discovery MR 750W 3.0T machine. All participants were informed about the study and provided their informed consent. In MRS procedure, to specifically target regions of interest in the sagittal, coronal, and axial planes, a single-voxel spectroscopy technique was employed by the scan operator to focus on intervertebral space scan regions that included the disc nucleus while excluding the vertebral body (Fig. 1A). The scan regions were positioned as close as possible to the center of the disc. The voxel size in the coronal plane (right-left direction) was set to 14 mm, whereas the voxel sizes in the sagittal plane (anterior-posterior direction) and axial plane (height direction) were both set to 5 mm. The relative signal-to-noise ratio were set as 100%. Lumbar spinal magnetic resonance imaging (MRI) was performed in accordance with standard protocols. Following patient positioning and acquisition of the localizer image, sagittal T1-weighted and T2-weighted sequences were obtained, along with imaging in the coronal and axial planes. Patients were instructed to remain still during image acquisition to minimize motion artifacts. Upon completion of the scan, images were reconstructed and transmitted to PACS (picture archiving and communication system) for radiological interpretation.
MRS analysis for IDD. (A) Illustration of 3-plane voxel prescription for MRS analysis (left: axial, center: coronal; right: sagittal); (B) Typical MRS analysis was performed on a healthy volunteer, with the x-axis representing the metabolites corresponding to specific chemical shifts (ppm). (C) The pLP group demonstrated the most prominent and highest peak at 1.3 ppm in MRS. (D) The MRS for non-pLP group. MRS, magnetic resonance spectroscopy; IDD, intervertebral disc degeneration; pLP, predominant lipids peak.
MRS data were collected during a separate MRI session after completion of clinical imaging. Following precise localization of the MRS voxel, proton MRS scans were performed using a time interval between 2 successive radio frequency pulses (time of repetition) set at 1,000 msec and a duration between the application of a radio frequency pulse and its corresponding echo (time of echo) set at 37 msec. Following the completion of the first MRS scan, the patient remained in the scanner while the MRS voxels were repositioned with only minor adjustments to their placement at the scan site, and the analysis was subsequently repeated. The collected MRS data were not utilized for surgical purposes. A sample of MRS spectrum for healthy intervertebral disc was illustrated in Fig. 1B, in which x-axis represents metabolite, and the y-axis represents abundance. The processing of acquired metabolite spectra, including spectral fitting and metabolite quantification, was conducted using LCModel (ver. 6.2). The data were transferred from the scanners to the spectroscopy tab of the workstation to enable automatic generation of spectral lines. Relative peak area was expressed as area under the lipids peak for IDD normalized against area under the lipids peak for healthy controls. The average value of the lipid peak areas from 2 scans was used for the final analysis.
4. Study Group
The patients were divided into predominant lipids peak (pLP) group and non-pLP based on the predominant peak observed in the MRS spectrum. The pLP group was identified by the MRS spectrum, which exhibited the highest peak at 1.3 ppm. In contrast, patients who did not display this spectral feature were classified into the non-pLP group (Fig. 1C and D). All patients underwent 2 MRS scans; for those who exhibited a pLP peak in one scan and a non-pLP peak in another repeat scan, a third MRS scan was performed. Final group assignment was determined by an experienced technician.
5. Lipidomics
Lipid extraction and analysis via mass spectrometry were performed according to the following procedure: Initially, 200 μL of chilled water and 20 μL of an internal lipid standard mix were introduced into the samples to facilitate lipid extraction. The primary NP cells (1×10⁶ per sample) were homogenized at 4°C using a homogenizer (MP Biomedicals, USA). Following this step, 800 μL of cold methyl tert-butyl ether and 240 μL of methanol were added to the mixture. This was followed by vortex mixing for 30 seconds and ultrasonic treatment at 4°C for 20 minutes. The samples were left to settle for 30 minutes and subsequently centrifuged at 14,000 g for 15 minutes at 4°C to separate the lipid components. The upper organic layer, which contained the extracted lipids, was dried under vacuum conditions. Finally, the dried lipid samples were reconstituted in 200 μL of a solvent mixture composed of isopropanol and acetonitrile (9:1, v/v) in preparation for lipidomic analysis. Subsequently, LC-MS/MS (liquid chromatography-tandem mass spectrometry) analysis was performed using a Q Exactive Plus mass spectrometer (Thermo Scientific, USA) coupled with a UHPLC (ultra–high-performance liquid chromatography) Nexera LC-30A system. Specifically, lipids were fractionated on a Waters ACQUITY PREMIER CSH C18 Column (1.7 μm, 2.1×100 mm) under the following chromatographic conditions: mobile phase A consisted of acetonitrile and water in a ratio of 6:4 (v/v), while mobile phase B comprised acetonitrile and isopropanol in a ratio of 1:9 (v/v). The flow rate was maintained at 300 μL/min, and the column temperature was set to 45°C. Mass spectrometry detection and analysis were conducted via electrospray ionization in both positive and negative ion modes. Full-scan spectra were acquired within m/z ranges of 200–1,800 for positive ions and 250–1,800 for negative ions. Lipid identification, extraction, alignment, and quantification were achieved using LipidSearch software version 4.1 (Thermo Scientific).
The principal component analysis (PCA) and orthogonal partial least squares-discriminant analysis (OPLS-DA) were performed on the OUYI platform by uploading the raw data (https://cloud.oebiotech.com/). In PCA model, principal component (PC) 1, capturing the direction of maximum variance in the data and usually accounting for primary sample differences like intergroup variations, while PC2 is the second principal component, representing the direction of the next highest variance orthogonal to PC1 and generally explaining the second most significant source of variation. In OPLS-DA model, PCo1 represents the variation associated with the grouping, while PC represents systematic variation that is not related to the grouping.
6. Proteomics
The primary NP cells were isolated via flow cytometry into boron-dipyrromethene (BODIPY)-high and BODIPY-low groups, and then followed by proteomic analysis. The data-independent acquisition (DIA) proteomics assay was performed with reference to the previous literature [24,25]. Briefly, after proteins processing, the sample was separated using an HPLC (high-performance liquid chromatography) system (Rigol L3000) with a C18 column at 45°C, monitored at 214 nm, and fractionated into 4 components. Fractions were vacuum-dried and redissolved in 0.1% formic acid. For proteomic analysis, an EASY-nLC 1200 UHPLC system coupled to a Q Exactive HF-X mass spectrometer was employed. In data-dependent acquisition mode, 4 μg of sample with indexed retention time reagent was loaded onto a C18 Nano-Trap column (4.5 cm×75 μm, 3 μm) and eluted via a linear gradient on a 15-cm analytical column (150 μm, 1.9 μm). MS parameters included a 2.1-kV spray voltage, 320°C capillary, full scan (350–1,500 m/z, 120,000 resolution), and higher-energy collisional dissociation fragmentation (top 40 precursors; 27% collision energy). DIA used similar LC conditions with MS1 resolution 60,000 and MS2 30,000.
Proteins were identified using Proteome Discoverer 2.2 against the Human Uniprot database, with 10-ppm precursor and 0.02 Da product ion tolerances. Fixed (carbamidomethyl) and variable (methionine oxidation) modifications were applied, allowing 2 missed cleavages. Peptide-spectrum matchs required >99% confidence, and proteins were filtered at <1% false discovery rate with ≥1 unique peptide. DIA data were analyzed in Spectronaut 14.0 using a spectral library for ion-pair extraction and quantification, with iRT-based retention time correction and q-value <0.01. Proteins with a differential expression fold change exceeding 1.2 (both up- and downregulated) and a p-value (from t-test) less than 0.05 were identified as differentially expressed proteins. Bioinformatics analysis of the proteomics data mainly includes PCA and Kyoto Encyclopedia of Genes and Genomes pathway analysis.
7. ELISA Assay
Measurements of IL-6, IL-1β, TNF-α, and prostaglandin E2 (PGE2) were conducted using commercially available enzymelinked immunosorbent assay kits in accordance with the manufacturer’s protocol (Solarbio, China). The enzyme reaction was quantified at an optical density of 405 nm. Standard curves were generated using a 4-parameter logistic regression model based on the absorbance values.
8. Flow Cytometry Isolation
BODIPY 493/503 staining (1:10,000 dilution; Beyotime, China) was utilized for primary NP cells staining according to the manufacturer’s instructions. Specifically, following a 30-minute incubation with the BODIPY working solution in the dark, cells were washed 3 times with 1×hanks’ balanced salt solution. The NP cells were collected from 3 IDD patients, and further isolated as BODIPY-higher and BODIPY-low group. Subsequently, NP cells were resuspended and isolated via fluorescence-activated cell sorting for further experimentation.
9. Triglyceride Assay
The intracellular triglyceride (TG) level was determined using a TG assay kit (Solarbio, China) in accordance with the manufacturer’s instructions. A total of 1×105 cells were collected for quantification of the total TG content. The relative TG content in the pLP group was subsequently normalized to that in the non-pLP group.
10. Transmission Electron Microscopy Observation
Primary NP cells were fixed in 2.5% glutaraldehyde overnight and then treated with 1% osmium tetroxide at 4°C for 1 hour. They underwent dehydration with increasing ethanol concentrations from 30% to 90%, with each step lasting 30 minutes. The cells were further dehydrated 3 times with pure ethanol, each for 30 minutes. Samples were embedded in propylene oxide medium, and ultrathin sections of 70 nm were cut and placed on copper grids. These sections were stained with lead citrate for 10 minutes and uranyl acetate for 30 minutes, followed by washes with deionized water. Images were captured using a JEM-1400 transmission electron microscope.
11. Immunohistochemical Staining
Totally 26 NP tissue samples from the pLP group and 31 NP tissue samples from the non-pLP group were collected. The NP tissues were fixed in a 4% paraformaldehyde solution. Subsequently, the NP tissues were dehydrated and embedded in paraffin, followed by slicing into 5μm sections. After deparaffinization and rehydration, the sections were subjected to microwave boiling in 0.01-mol/L citrate buffer (pH 6.0) for 15 minutes. Then, blocking with goat albumin was performed for 30 minutes before incubating the samples overnight at 4°C with the primary antibody. The primary antibodies used are as follows: anti-IL17 (Servicebio, GB11110; 1:200 dilution), anti-IL-6 (Servicebio, GB11117; 1:400 dilution), anti-IL-1β (Servicebio, GB11113; 1:400 dilution), anti-TNF-α (Servicebio, GB11188; 1:200 dilution) and anti-PGE2 (Bioss, bs-2639R; 1:200 dilution). Following that, a secondary antibody was applied to the samples for an hour at room temperature, followed by hematoxylin staining.
For the immunohistochemical (IHC) staining analysis, the entire tissue section was initially scanned at ×100 magnification to assess staining homogeneity and to identify viable, cellrich regions, while excluding areas with significant artifacts, tears, or acellular debris. Subsequently, a systematic approach was employed to select 5 nonoverlapping, representative fields of view per section for quantitative analysis. This was achieved by superimposing a virtual grid over the NP region and selecting fields from both the center and periphery (upper left, upper right, center, lower left, lower right) to account for potential regional variations. Cell counting was then performed at a high magnification of ×400. In each of the 5 selected fields, a total of 2 hundred nucleated cells were counted. The positive cell rate was determined for each field, and the results from all fields were averaged to yield a final positive rate for each sample.
12. Statistical Analysis
The statistical analyses were performed using Prism 8 (Graph-Pad, USA). All statistical data were analyzed using the Shapiro-Wilk method to estimate the normal distribution of data. An independent t-test was utilized to compare age, body mass index (BMI), patient-reported symptoms measurements, and the percentage of immune-positive cells between pLP and non-pLP groups. Categorical variables were presented as numbers, gender, diagnosis, disc degeneration, and Modic changes in the groups were evaluated using the chi-square test or Fisher exact test. Pearson correlation coefficients were used to examine the correlations between patient-reported symptoms and the percentages of immune-positive cells, as well as the relative lipid areas. Multiple linear regression analysis was used to examine the contribution of related factors to the VAS-back pain and ODI score. A p-value <0.05 was considered statistically significant.
RESULTS
1. Patient Characteristics
A total of 57 patients were enrolled in this study, with an average age of 66.1 years (spanning from 49 to 78 years). Within this cohort, 45 were identified with lumbar disc herniation and/or degenerative lumbar spinal stenosis, and 12 exhibited degenerative spondylolisthesis. Twenty-six patients were classified into the pLP group (12 females and 14 males), with an average age of 66.5 years (ranging from 49 to 77 years). The remaining 31 patients formed the non-pLP group (13 females and 18 males), with an average age of 66.9 years (ranging from 55 to 78 years). No significant difference was detected between the 2 groups with respect to age, gender, BMI, diagnostic classifications, comorbid conditions, the degree of disc degeneration, disc height and surgical levels. In general, the Modic changes were more frequently observed in the pLP group than in the non-pLP group (p<0.01) (Table 1).
Univariate analysis of demographics and magnetic resonance findings in patients of pLP and non-pLP group
2. Patient-Reported Symptoms Evaluation
The Fig. 2A and B show typical MR imaging and MRS spectrum in pLP and non-pLP group. The pLP group exhibited a significantly higher preoperative VAS-back score (6.5±1.6 vs. 4.7±1.4, p<0.01) compared to the non-pLP group, while no significant difference was observed in VAS-leg scores (5.4±1.6 vs. 5.3±1.8, p=0.781) between the 2 groups. Furthermore, the pLP group demonstrated a higher preoperative ODI score (63.3%±8.7% vs. 51.2%±9.0%, p<0.01) compared to the non-pLP group. However, at the 3- and 6-month postoperative follow-ups, no statistically significant differences were noted in ODI, VAS-back, or VAS-leg scores between the 2 groups (Fig. 2C–E).
Patient-reported symptoms before operation and follow-up. (A and B) Typical MRS in pLP and non-pLP group; (C) Preoperative patient-reported symptoms. (D) Patient-reported symptoms at 3-month follow-up. (E) Patient-reported symptoms at 6-month follow-up. MRS, magnetic resonance spectroscopy; pLP, predominant lipids peak; ns, not significant. ***p<0.001.
3. Lipidomics Analysis Revealed LDs Accumulation in NP Cells
Because the baseline lipid peaks in healthy controls were not clearly defined, we conducted MRS on a subset of healthy controls (Fig. 3A). The MRS spectra with a higher lipid peak is appreciably noted in pLP group (Fig. 3B). The pLP group exhibited the most pronounced lipid peak at 1.3 ppm, indicating potential abnormal lipid accumulation. To validate the MRS findings, lipidomics analysis was performed. PCA and OPLS-DA effectively differentiated between the 2 groups (Fig. 3C and D). A total of 176 metabolites were identified as significantly different between the groups, and lipid classification revealed distinct lipid subtypes (Fig. 3F). The heatmap of the top differentially expressed metabolites showed that the level of TG was markedly elevated in the pLP group compared to the non-pLP group (Fig. 3G). In validation cohort, the relative TG content was noted significantly higher in the pLP group than in the non-pLP group (Fig. 3H). Transmission electron microscopy (TEM) observations further confirmed increased intracellular LD accumulation in the pLP group (Fig. 3I).
Lipidomics analysis for pLP and non-pLP group. (A) MRS analysis for healthy controls. n=15. (B) Relative lipids area was significantly higher in pLP group and non-pLP group than healthy controls. (C and D) Principal component analysis (PCA) and orthogonal partial least squares-discriminant analysis (OPLS-DA) for Lipidomics. The results indicated a clear separation between the 2 groups. (E and F) Volcano plots and lipids classification are utilized to visualize significantly differentially expressed metabolites. (G) Heatmap for significantly differentially expressed metabolites. (H) The relative TG levels in NP cells were determined for 26 patients in the pLP group and 31 patients in the non-pLP group. (I) TEM observation of LDs in the pLP group; the red arrow indicates LDs. pLP, predominant lipids peak; MRS, magnetic resonance spectroscopy; TG, triglyceride; NP, nucleus pulposus; TEM, transmission electron microscopy; LD, lipid droplet; GL, glycerolipids; LPC, lysophosphatidylcholine; LPE, lysophosphatidylethanolamine; PC, phosphatidylcholine; PE, phosphatidylethanolamine; PI, phosphatidylinositol; SL, sphingolipids; CerG1, monohexosylceramide; CerG2, Dihexosylceramide; SP, sphingophospholipids; phSM, phytosphingomyelin; SM, sphingomyelin; So, sphingosine. Compared to non-pLP group, ***p<0.001.
4. Proteomic Analysis Reveals Activation of IL-17 Inflammatory Pathway
To elucidate the molecular characteristics of LD-enriched NP cells, we employed flow cytometry to isolate NP cells into BODIPY-high and BODIPY-low groups, followed by proteomic analysis (Fig. 4A). The PCA analysis showed large variations in both proteins (Fig. 4B). A total of 217 proteins exceeded the threshold, in which 109 proteins exhibited a decrease while 108 proteins showed an increase in their expression levels (Fig. 4C). Subsequently, we examined the potential biological functions of these proteins. As illustrated in Fig. 4D, it is noteworthy that pathways associated with inflammation, especially the interleukin-17 (IL-17) pathway, were found to be enriched. Furthermore, the IHC staining confirmed that IL-17 was more commonly expressed in the pLP group than non-pLP group (Fig. 4E and F).
Proteomics analysis for the LDs-enriched NP cells. (A) Workflow of BODIPY staining and flow cytometry sorting for BODIPY-low and -high NP cells. (B) Principal component analysis (PCA) analysis for proteomics. (C) Volcano plots show significantly differentially expressed proteins. (D) KEGG analysis of the different proteins were related to “IL-17 pathway.” (E and F) Immunohistochemical staining and corresponding statistical qualification of IL-17 expression in human NP tissues. Scale bar= 50 μm. BODIPY, boron-dipyrromethene; LD, lipid droplet; NP, nucleus pulposus; TEM, transmission electron microscopy; PC, principal component; KEGG, Kyoto Encyclopedia of Genes and Genomes; FC, fold change; IL, interleukin; pLP, predominant lipids peak. Compared to non-pLP group, ***p<0.001.
The activation of the IL-17 pathway leads to the production of other proinflammatory cytokines, including IL-6, IL-1β, TNF-α and PGE2, which collectively contribute to the inflammatory response and back pain in IDD. Our IHC staining also confirmed that the positive rate cell of TNF-α, IL-1β, IL-6 and PGE2 were significantly higher in pLP group than that of in non-pLP group (Fig. 5A and B). In both the pLP group and the BODIPY-high group, blocking the activation of IL-17 with inhibitor Secukinumab (anti-IL-17A) resulted in decreased expression levels of TNF-α, IL-1β, IL-6 and PGE2 (Fig. 5C and D). These findings collectively suggest that the activation of the IL-17 pathway in pLP group is associated with inflammatory responses and discogenic pain.
Inhibition of the IL-17 pathway blocks the inflammatory pathway. (A and B) Immunohistochemical staining and corresponding statistical qualification of IL-6, IL-1β, TNF-α, and PGE2 positive cells. Scale bar=50 μm. (C) Inhibition of the IL-17 pathway blocks the expression of IL-6, IL-1β, TNF-α, and PGE2 in BODIPY-high NP cells. (D) Inhibition of the IL-17 pathway blocks the expression of IL-6, IL-1β, TNF-α, and PGE2 in pLP group NP cells. pLP, predominant lipids peak; IL, interleukin; TNF-α, tumor necrosis factor-alpha; PGE2, prostaglandin E2; BODIPY, boron-dipyrromethene; NP, nucleus pulposus. Compared to non-pLP or Bodipy-high group, ***p<0.001.
5. Correlation Between Patient-Reported Symptoms and the Level of IL-17 Expression and Lipids Contents
As illustrated in Fig. 6A and B, the preoperative VAS-back pain exhibited moderate positive correlations with the expressions of IL-17 (r=0.561, p<0.001) and relative TG contents (r=0.567, p<0.001). The preoperative ODI exhibited moderate positive correlations with the expressions of IL-17 (r=0.566, p<0.001) and relative TG contents (r=0.555, p<0.001). Conversely, there were no significant correlations between preoperative VAS-leg pain and the level of IL-17 expression and relative lipid contents (both p>0.05).
Association between the patient-reported symptoms and LDs-IL-17 pathway. (A) Association between the preoperative patient-reported symptoms and relative TG contents. (B) Association between the preoperative patient-reported symptoms and the expression of IL-17. LD, lipid droplet; IL, interleukin; TG, triglyceride; VAS, visual analogue scale.
Finally, age, sex, BMI, Modic changes, disc degeneration, IL-17 positive cells and relative TG contents were included in multivariate linear regression analysis, which showed that the IL-17 positive cells (p=0.021) and relative TG contents (p=0.046) were significant factors for the preoperative VAS-back pain (Table 2). Furthermore, the analysis also demonstrated that the IL-17 positive cells (p=0.047) and relative TG contents (p=0.037) were significantly factors for the preoperative ODI scores (Table 3).
Multivariate linear regression analysis of potential contributing factors and VAS-back scores
Multivariate linear regression analysis of potential contributing factors and ODI scores
DISCUSSION
The present study compared patient-reported symptoms between the pLP group and non-pLP group. By employing an integrated approach of MRS-based metabolomics, lipidomics and proteomics, we identified the LDs-IL17-inflammatory axis and analyzed its association with clinical symptoms. Our findings revealed that individuals exhibiting a higher lipids peak in the MRS spectrum, particularly around 1.3 ppm, experienced a poorer quality of life. Additionally, LDs accumulation and lipotoxic conditions were observed in the NP cells of patients with pLP spectrum. Furthermore, elevated expression of the IL-17 signaling pathway and inflammatory cytokines such as IL-6, IL-1β, TNF-α, and PGE2 was noted in these NP cells. LDs accumulation may serve as a potential risk factor for the activation of the IL-17 inflammatory pathway, which can significantly impact the pathophysiology of NP cells, ultimately leading to a poorer quality of life. The clinical implications of this study are summarized as follows: (1) it elucidates the underlying pathological mechanisms responsible for pLP observations in MRS; (2) it uncovers the relationship between LDs and inflammatory signaling pathways. These findings provide a theoretical foundation for utilizing MRS as a noninvasive, real-time biomarker in clinical practice, thereby enhancing the selection of effective treatment strategies for IDD and the assessment of therapeutic efficacy.
Alterations in the metabolic profile within an organ are intricately linked to its physiological function, thereby serving as a critical indicator for disease monitoring [25-27]. These metabolic changes can provide valuable insights into the organ’s health status and can be used to detect early signs of dysfunction or disease progression [19]. By elucidating the specific metabolic pathways that are affected, we can gain a deeper understanding of the underlying causes of various conditions and develop more targeted treatment strategies for diseases such as IDD [12,19]. Moreover, longitudinal monitoring of these metabolic alterations can help in assessing the effectiveness of therapeutic interventions and predicting disease progression, thus playing a pivotal role in personalized medicine and patient care [10]. MRS is a noninvasive technique that leverages the magnetic properties of atomic nuclei, enabling quantitative analysis of the biochemical composition within living human tissues. Its primary advantage lies in its noninvasive nature and real-time capabilities [13]. MRS is based on nuclear magnetic resonance, using a magnetic field and radiofrequency pulses to excite nuclei such as protons, which resonate at frequencies influenced by their molecular environment. The resulting signals are transformed into a spectrum with chemical shifts (in ppm) on the x-axis and signal amplitude on the y-axis. MRS facilitates the noninvasive quantification of metabolites in specific tissues, thereby enhancing our understanding of the pathophysiological mechanisms underlying conditions like low back pain [28]. In clinical applications, MRS has been extensively utilized to characterize in vivo metabolic profiles in NP tissues, liver, and other tissue types [16,29]. In our previous studies, we identified a subset of patients exhibiting a pLP spectrum in MRS analysis; however, the clinical significance of this finding remains to be fully elucidated.
Recently, MRS has emerged as a valuable tool in the study of low back pain, providing insights into the biochemical changes associated with IDD and various spinal conditions [28]. In algorithm for IDD, carbohydrate/collagen (CA) and proteoglycan (PG) regions are used as markers of structural integrity. Conversely, alanine (AL), lactate (LA), and propionate (PA) regions, which act as acidic pain markers, are expected to increase in the presence of discogenic pain [11,14]. In their study, Zuo et al. [30] showed that the water/PG peak area ratio was positively associated with increasing Pfirrmann grade in both low back patients and controls. In low back pain patients, discs demonstrating positive provocative discography exhibited a significantly higher water/PG peak area ratio than provocative discography-negative discs. Moreover, an elevated water/PG peak area ratio was significantly linked to poorer outcomes in ODI and 36-item Short Form health survey scores. Gornet et al. [11,14] performed a prospective study on patients affected by low back pain that underwent MRS and subsequently received PD. MRS generated spectral features, with CA and PG serving as structural markers, and AL, LA, and PA as acidic pain markers, and the MRS-based evaluation can effectively identify painful lumbar discs. However, these studies attempted to explain the causes of IDD-related low back pain from the structural basis of the intervertebral disc, but overlooked the significant role of metabolic environmental stress and inflammatory factors. As our previous research has shown, the intervertebral disc is a closed metabolic environment characterized by lipotoxicity and MRS higher lipids peak [12]. The above-mentioned research cannot be explained from the perspective of “metabolic environment-MRS manification-symptoms.” As a useful supplementary, the present study, providing a new perspective for MRS to explore the causes of low back pain.
In our study, we observed a subset of patients exhibiting significantly higher lipids peak at approximately 1.3 ppm, indicative of lipids accumulation in NP tissues. However, our findings are partially inconsistent with prior studies indicating increased levels of certain disc chemicals, such as LA and PGs, which serve as spectroscopically quantifiable biomarkers for low back pain and IDD [11]. The discrepancies between previous studies and our work can be partially attributed to variations in the patient samples. In prior research, patients with spondylolisthesis, scoliosis, and disc herniation were excluded. By contrast, our study incorporates these patient groups.
The lipid family comprises a wide variety of molecular species, including TGs, sphingolipids, phosphoglycerides, and sterols [31]. To improve the interpretability of the MRS analysis, we performed targeted lipidomics sequencing to analyze specific alterations in a defined subset of lipids. The integration of lipidomics data with experimental findings demonstrates an accumulation of abnormal LDs within the NP cells of patients in the pPL group. The accumulation of LDs is often associated with impaired lipid metabolism, which can lead to a cascade of pathological events in a variety of human diseases [15]. For instance, in the context of metabolic diseases such as obesity and type 2 diabetes, LDs accumulation in adipose tissue and other organs can exacerbate insulin resistance and inflammation, contributing to the progression of these conditions [32,33]. In the aging brain, LDs accumulation in microglia is linked to a dysfunctional and proinflammatory state [16]. These LDs accumulating microglia are defective in phagocytosis and produce high levels of reactive oxygen species and proinflammatory cytokines, contributing to neurodegenerative diseases. Similarly, in Alzheimer’s disease, microglia associated with amyloid plaques accumulate LDs, which are influenced by genetic factors such as the TREM2 mutation, affecting their inflammatory response [34]. This accumulation is thought to contribute to neuroinflammation and neuronal cell death, exacerbating the progression of these diseases. In the current study, we identified a significant accumulation of LDs in the NP cells of patients within the pLP group. This finding may suggest an inflammatory state within the intervertebral disc and appears to be associated with more severe clinical manifestations in these patients.
In the present study, we identified the activation of the IL-17 pathway within LDs-enriched NP cells though proteomics approach. IL-17, a proinflammatory cytokine, plays a pivotal role in facilitating inflammatory processes across a spectrum of diseases [35]. It is predominantly secreted by Th17 cells, a subset of T-helper cells, and has been associated with the pathogenesis of several autoimmune and inflammatory disorders. IL-17 mediates its effects by binding to the IL-17 receptor, which triggers downstream signaling cascades resulting in the production of additional proinflammatory cytokines and chemokines, thus perpetuating the inflammatory response [36]. For instance, IL-17 has been shown to enhance the production of IL-6 and IL-1β in keratinocytes, contributing to skin inflammation through the ROS-NLRP3-caspase-1 pathway [37]. Additionally, IL-17’s role in promoting IL-6 and IL-1β secretion is evident in the context of osteoarthritis, where it contributes to the inflammatory milieu and disease progression [36,38,39]. Moreover, IL-17 interaction with other cytokines like TNF-α can lead to synergistic effects, further amplifying the inflammatory response [40]. This is particularly significant in autoimmune diseases such as rheumatoid arthritis, where the combination of IL-17 and TNF-α has been shown to potentiate the secretion of proinflammatory cytokines, thereby exacerbating the disease [41]. Furthermore, IL-17’s ability to stimulate the production of PGE2, a lipid compound involved in inflammation and low back pain, underscores its role in disease progression [35,36].
Our study revealed a potential correlation between the activation of IL-17 pathways and LDs formation in IDD. The activation of IL-17 pathways and LDs formation are interconnected processes with significant implications for lipid metabolism and inflammation. Excessive LDs accumulation, as observed in metabolic disorders such as obesity and non-alcoholic fatty liver disease, induces cellular stress, leading to the release of damage-associated molecular patterns and proinflammatory mediators [42]. These signals activate innate immune cells, including macrophages and dendritic cells, which produce cytokines like IL-1β, IL-6, and IL-23, thereby promoting the differentiation and activation of Th17 cells [43]. Notably, saturated fatty acids stored within LDs can activate Toll-like receptors on immune cells, further amplifying IL-17 production through nuclear factor kappa-lightchain-enhancer of activated B and mitogen-activated protein kinase signaling pathways [44]. Additionally, LDs generate lipid peroxidation byproducts, such as oxidized phospholipids, which act as ligands for inflammasomes, fostering IL-1β driven Th17 polarization [43]. Experimental models have demonstrated that reducing LDs content via genetic or pharmacological inhibition of lipogenesis attenuates IL-17 signaling and associated tissue damage [45]. Conversely, environments enriched in LDs under conditions such as atherosclerosis exacerbate IL-17-mediated pathology, underscoring a bidirectional crosstalk between LDs and IL-17 signaling [43,46]. This LDs-IL-17 axis may creates a vicious cycle: metabolic lipids overload fuels inflammation, while IL-17 further promotes lipogenesis and LDs formation. Our findings suggest that the activation of IL-17 pathways and LDs formation may play a significant role in chronic low back pain.Our study revealed a potential correlation between the activation of IL-17 pathways and LDs formation in IDD. The activation of IL-17 pathways and LDs formation are interconnected processes with significant implications for lipid metabolism and inflammation. Excessive LDs accumulation, as observed in metabolic disorders such as obesity and non-alcoholic fatty liver disease, induces cellular stress, leading to the release of damage-associated molecular patterns and proinflammatory mediators [42]. These signals activate innate immune cells, including macrophages and dendritic cells, which produce cytokines like IL-1β, IL-6, and IL-23, thereby promoting the differentiation and activation of Th17 cells [43]. Notably, saturated fatty acids stored within LDs can activate Toll-like receptors on immune cells, further amplifying IL-17 production through nuclear factor kappa-lightchain-enhancer of activated B and mitogen-activated protein kinase signaling pathways [44]. Additionally, LDs generate lipid peroxidation byproducts, such as oxidized phospholipids, which act as ligands for inflammasomes, fostering IL-1β driven Th17 polarization [43]. Experimental models have demonstrated that reducing LDs content via genetic or pharmacological inhibition of lipogenesis attenuates IL-17 signaling and associated tissue damage [45]. Conversely, environments enriched in LDs under conditions such as atherosclerosis exacerbate IL-17-mediated pathology, underscoring a bidirectional crosstalk between LDs and IL-17 signaling [43,46]. This LDs-IL-17 axis may creates a vicious cycle: metabolic lipids overload fuels inflammation, while IL-17 further promotes lipogenesis and LDs formation. Our findings suggest that the activation of IL-17 pathways and LDs formation may play a significant role in chronic low back pain.
As a downstream effector of the IL-17 signaling pathway, the expression levels of several key inflammatory mediators, including IL-6, IL-1β, TNF-α, and PGE2, were significantly upregulated. The pathogenesis of discogenic pain is closely associated with the upregulation of these proinflammatory cytokines, which promotes sustained inflammation, facilitating the infiltration of nerve fibers and blood vessels into the normally avascular intervertebral disc through neurotrophic factors like nerve growth factor and vascular endothelial growth factor 8. This aberrant neurovascular ingrowth, coupled with the secretion of pain-related neuropeptides, directly sensitizes nociceptors, contributing to chronic discogenic pain [4]. Notably, previous studies have been demonstrate that IL-6, IL-1β, and TNF-α levels in herniated disc tissues correlate positively with pain severity, as measured by VAS scores [47,48]. PGE2, synthesized via cyclooxygenase-2 activation under cytokine stimulation, further exacerbates pain by enhancing neuronal excitability and peripheral sensitization [23]. Our study demonstrated that blocking the expression of IL-17 resulted in decreased expression levels of IL-6, IL-1β, TNF-α, and PGE2. These findings collectively elucidate the mechanisms underlying the exacerbation of low back pain in patients with pLP.
The clinical significance of these findings is substantial, as they provide a theoretical foundation for using MRS as a noninvasive, real-time biomarker in clinical practice. This innovative approach has the potential to monitor disease progression in real time and analyze its correlation with low back pain, thereby offering valuable insights for clinical applications. Consequently, this advancement in diagnostic technology enhances the selection of effective treatment strategies for IDD and improves the evaluation of therapeutic efficacy. By offering precise and timely insights into biochemical changes within NP tissues, MRS analysis with a focus on lipid peaks enables clinicians to make more informed decisions, thereby improving patient outcomes and potentially reducing the necessity for invasive diagnostic procedures. In summary, the integration of MRS into clinical practice holds the promise of revolutionizing the management of IDD by providing a noninvasive tool that offers both diagnostic and prognostic information, ultimately leading to better patient care and more efficient healthcare resource utilization.
Several limitations should be acknowledged. Firstly, the sample size for MRS and lipidomics is relatively limited, and the research was conducted at our single center. This may restrict the generalizability of the findings to the broader target population. However, most importantly, this study integrates multiomics data with experimental validation to provide mechanistic insights into the pLP spectrum, LDs accumulation, inflammation, and patient symptoms observed. Secondly, our multiomics approach was primarily focused on proteins and lipids. Integrating untargeted metabolomics analysis into the multiomics profile could potentially enhance our understanding of the pathogenic mechanisms. Thirdly, the grouping method of pLP and non-pLP group has a certain degree of subjectivity for patients at the borderline morphological spectrum. To address this limitation, we supplemented our MRS results with lipidomic analysis, TEM observation for LDs, and TG content measurement.
CONCLUSION
In summary, patients with a pLP spectrum exhibited a poorer quality of life, more pronounced LDs accumulation, as well as upregulated IL-17 pathway and downstream inflammatory factors IL-6, IL-1β, TNF-α, and PGE2 in NP cells. LDs-related inflammation may play a significant role in the onset and progression of low back pain in these patients, thereby influencing patient-reported symptoms.
Notes
Conflict of Interest
The authors have nothing to disclose.
Funding/Support
This project has received funding from the Fundamental Research Funds for the Central Universities (YD9110002020).
Author Contribution
Conceptualization: XC, JH; Formal analysis: YD, YL; Investigation: JJ, CL; Methodology: ST, YD; Project administration: WZ; Writing – original draft: XC; Writing – review & editing: YL, JH.