Gene Expression Profiles Perturbed by Injury to the Mouse Intervertebral Disc

Abstract

Objectives.

Back pain subsequent to intervertebral disc (IVD) injury is a common clinical problem. Previous work examining early molecular changes post injury mainly used a candidate marker approach. In this study, gene expression in the injured and intact mouse tail IVDs was determined with a nonbiased whole transcriptome approach.

Design.

Mouse tail IVD injury was induced by a needle puncture. Whole murine transcriptome was determined by RNASeq. Transcriptomes of injured IVDs were compared with those of intact controls by bioinformatic methods.

Results.

Among the 18,078 murine genes examined, 592 genes were differentially expressed (P.adj <0.01). Novel genes upregulated in injured compared with intact IVDs included Chl1, Lum, etc. Ontology study of upregulated genes revealed that leukocyte migration was the most enriched biological process, and network analysis showed that Tnfa had the most protein-protein interactions. Novel downregulated genes in the injured IVDs included 4833412C05Rik, Myoc, etc. The most enriched downregulated pathways were related to cytoskeletal organization.

Conclusion.

Novel genes highly regulated post disc injury were identified with an unbiased approach; they may serve as biomarkers of injury and response to treatments in future experiments. Enriched biological pathways and molecules with high numbers of connections may be targets for treatments post injury.

INTRODUCTION

Numerous individuals experiencing back pain may recollect a strain or injury, often seek treatments days, weeks, or even years after the initial incident. Examining the early events that trigger inflammation/repair and determining the optimal timing for treatments in humans is challenging due to delayed clinic visits and restricted access to biological samples. Utilizing animal models of disc degeneration proves valuable in addressing this knowledge gap.

The mouse intervertebral disc (IVD) injury model is extensively employed, as it has been well-documented for its consistent molecular, histological, and biomechanical indicators of disease progression.^1–3 The advantages of the tail injury method include easy access to the tail discs and low morbidity.² Selected sets of inflammatory markers and extracellular matrix genes have been well described post injury; these markers have been selected based on experience and literature.^{2, 4–6} However, genes previously undescribed or unsuspected that could be more sensitive or more reliable markers were not used. Therefore, a nonbiased approach could uncover novel markers, and reveal mechanistic insights.

RNA-Seq (sequencing) is a sequencing technique that uses next-generation sequencing to reveal the presence and quantity of RNA in a biological sample, representing an aggregated snapshot of the cells’ dynamic pool of RNAs known as transcriptome,⁷ which is the sum total of all the messenger RNA molecules expressed from the genes. Human and bovine IVD tissue transcriptomes have been examined, which shed light on cell types in this unique tissue and its degenerative process.^{8, 9} In the mouse model, whole transcriptome analysis of notochord-derived cells during embryonic formation of the mouse nucleus pulposus (NP) has been described.¹⁰ The RNASeq method was also used to identify genes regulated by the intrinsic circadian clock, which allows cells to cope with the drastic biomechanical and chemical changes that occur throughout the day.^{11, 12} Here, we used RNASeq to examine transcriptome of the mouse tail IVDs 1 week following injury, to gain insight into the early events post injury. The findings will serve as comparison in future studies of genetically modified mice, and may identify novel molecular targets and pathways for treatment of disc injury. The current work aims to define the early molecular changes with an unbiased approach by measuring expression of all known genes following an injury to the mouse tail IVD.

MATERIALS AND METHODS

Mice.

Approval for all animal experimental procedures was obtained from the Institutional Animal Care and Use Committee of the University of Pennsylvania, Philadelphia, PA. This study involved seven young adult male mice, aged 10–11 weeks, on the DBA background (DBA/1LacJ, the Jackson Laboratory, Bar Harbor, ME, USA). Four mice underwent the tail IVD injury experiment, while three served as intact controls. The mice were kept in a pathogen-free environment with environmental enrichment (nestlets by Ancare, Bellmore, NY, USA), accommodating up to 5 animals per cage. They were provided with PicoLab diet no. 5053 (LabDiet, Fort Worth, TX, USA) without restriction and had access to acidified bottled water. The housing conditions included a room temperature ranging from 21.1–24.4°C (equivalent to 70–76°F) with 30–70% humidity, following a 12:12-hour light: dark cycle.

Tail injury surgery.

Surgery was performed.² Specifically, each mouse underwent subcutaneous anesthesia using Ketamine (90mg/kg) and Xylazine (10mg/kg). While under anesthesia, the skin was sterilized with betadine. The coccygeal (Co) intervertebral discs (IVDs) were located, and a 26G needle was inserted into the IVD space until the needle tip reached approximately 2/3 of the disc thickness. In this study, the Co3/4, 4/5, 5/6, and Co6/7 IVDs in each mouse were subjected to injury in the “injured” group, while the corresponding intact IVDs were collected from the “intact” control group (Fig. 1). Mice were assessed 4 hours after the surgery and the following day, with daily monitoring until the study endpoint. Despite the absence of apparent signs of pain or distress, mice received preemptive buprenorphine at the time of sedation and 4 hours later. Animal sacrifice was performed by exposing them to CO2 one week after the tail disc injury.

Figure 1. The mouse coccygeal (Co) intervertebral disc (IVD) injury model.

A: the Co3/4, 4/5, 5/6 and 6/7 IVDs were injured with a 26 Gauge needle (arrow) under fluoroscopic guidance. B: schematic drawing of tail IVD injuries. C: Alcian Blue or Picro-Sirius Red stained coccygeal IVDs; red arrows: direction of injury; bar: 200μm.

RNA isolation.

From each mouse tail, Co3/4, 4/5,5/6 and Co6/7 IVDs were shaved off the cartilaginous endplate with a scalpel under a dissecting microscope (VistaVision, VWR International, Radnor, PA). The IVD tissues were from all 4 coccygeal IVDs of each animal were isolate, and immersed in RNALater (Ambion, Foster City, CA) overnight and preserved at −80°C until extraction. Total cellular RNA was isolated by the Trizol method. Specifically, RNALater was eliminated, and the tissues were rapidly frozen with liquid Nitrogen before being transferred into Trizol (Invitrogen, Carlsbad, CA). Subsequently, the tissues were homogenized using a homogenizer equipped with disposable OmniTip probes designed for hard tissue (Omni International, Kennesaw, GA). RNA was then precipitated with 70% ethanol and further purified using a RNeasy Micro Kit (Qiagen).

RNASeq.

RNA Library Preparation and Sequencing was performed by Azenta US (South Plainfield, NJ). Specifically, RNA samples were quantified using Qubit 2.0 Fluorometer (ThermoFisher Scientific, Waltham, MA, USA) and RNA integrity was checked with 4200 TapeStation (Agilent Technologies, Palo Alto, CA, USA). This step was carried out by Azenta, the company that performed the RNASeq. Briefly, small amounts of samples are separated in the micro-fabricated chip channels according to their molecular weight and then detected by laser detection. The result is an electropherogram in which the amount of changed fluorescence correlates with the amount of RNA of a given size. The software calculates the ratio of two ribosomal bands. The RNA integrity number (RIN) measurement is based on a machine learning algorithm that uses a capillary electrophoresis pathway and not just on the ratio of ribosomal subunits, although it is highly dependent on the ratio. RIN provides a numerical score (range 1–10) for RNA quality. A higher RIN value indicates a higher degree of RNA integrity. Once the RIN values were deemed acceptable, rRNA depletion was performed to further purify mRNA before the RNASeq procedure. Sequencing library was prepared using QIAGEN FastSelect rRNA HMR Kit (Qiagen, Hilden, Germany). RNA sequencing library preparation uses NEBNext Ultra II RNA Library Preparation Kit for Illumina by following the manufacturer’s recommendations (NEB, Ipswich, MA, USA). Briefly, enriched RNAs were fragmented for 15 minutes at 94 °C. First strand and second strand cDNA were subsequently synthesized. cDNA fragments were end-repaired and adenylated at 3’ends, and universal adapters were ligated to cDNA fragments, followed by index addition and library enrichment with limited cycle PCR. Sequencing libraries were validated using the Agilent Tapestation 4200 (Agilent Technologies, Palo Alto, CA, USA), and quantified using Qubit 2.0 Fluorometer (ThermoFisher Scientific, Waltham, MA, USA) as well as by quantitative PCR (KAPA Biosystems, Wilmington, MA, USA). The sequencing libraries were multiplexed and clustered on one flowcell. After clustering, the flowcell was loaded on the Illumina HiSeq instrument according to the manufacturer’s instructions. The samples were sequenced using a 2×150 Pair-End (PE) configuration Raw sequence data (.bcl files) generated from Illumina HiSeq were converted into fastq files and de-multiplexed using Illumina bcl2fastq program version 2.20. One mismatch was allowed for index sequence identification.

Data Analysis.

After demultiplexing, sequence data were checked for overall quality and yield. Then, sequence reads were trimmed to remove possible adapter sequences and nucleotides with poor quality using Trimmomatic v.0.36. The trimmed reads were mapped to the reference genomes using the STAR aligner v.2.5.2b. The STAR aligner is a splice aware aligner that detects splice junctions and incorporates them to help align the entire read sequences. BAM files were generated as a result of this step. Unique gene hit counts were calculated by using feature counts from the Subread package v.1.5.2. Only unique reads within exon regions were counted. After extraction of gene hit counts, the gene hit counts table was used for downstream differential expression analysis. Using DESeq2, a comparison of gene expression between the groups of samples was performed. The Wald test was used to generate p-values and Log2 fold changes. Genes with adjusted p-values (p.adj) < 0.01 were called as differentially expressed genes for each comparison.

Generation of Heatmap.

18,078 genes were examined by RNASeq. A differential expressed gene list was generated by limiting P.adj to <0.01, resulting in 609 genes in this category. Then 17 genes with no expression values detected in two or more samples were excluded, resulting in 592 differentially expressed genes. The genes were further sorted according to log2 fold changes from high to low, and separated into upregulated genes (injured/intact ratio>1; log2>0) and downregulated genes (injured/intact ratio<1; log2<0). For upregulated genes, we selected 15 genes based on their highest rankings. For downregulated genes, we chose the 6 genes with lowest fold changes. Additionally, we constructed a separate heatmap for the panel of genes used in the past to mark inflammatory responses (e.g., Tnfa, Adam8, Il6, Tnfaip2, Tnfrsf1a, 1b and 13b). After inputting the chosen upregulated genes and downregulated genes, a heatmap was generated using R package pheatmap,¹³ in which rows (represent gene expression) were scaled and hierarchically clustered, and a gap was inserted among columns to differentiate the intact control and injured groups.

Gene Ontology (GO)Analysis and Protein-Protein Interaction (PPI) Network Visualization.

Lists of upregulated and downregulated genes were analyzed separated for GO analysis was performed, and biological pathways were visualized with the R software and various R packages.¹³

Ranking PPI networks were downloaded from STRING database (https://string-db.org/, Version: 11.5) and imported into CytoScape software (Version: 3.9.1).¹⁴ The minimal interaction score was set at high confidence (0.70). Confidence scores are scaled between 0 and 1, with 1 corresponding to the estimated likelihood of a given association being true. Disconnected nodes in the network were not displayed. CytoHubba plugin was used to rank nodes with the MCC method.¹⁵ The parameters of nodes and edges were adjusted based on node rankings and the combined interaction scores predicted by STRING database, thus visualizing PPI networks in a more intuitive and clear way.¹⁴

RESULTS

Gene expression profiles of injured and intact IVDs were clearly different.

Gene expression profiles in the intact and injured IVD tissues 1-week post injury was compared. Among the 18,078 genes examined, 592 had adjusted p value (p.adj) <0.01. Among these, genes with highest injured/intact ratio and genes of interest were presented in the heatmap, to visualize the striking differences between the injured and intact IVDs. Novel genes upregulated in IVD tissues in response to injury were discovered: Chl1, Lum, Saa3, along with Cxcl1 included in the previously selected panel were among the upregulated genes with the highest injured/intact control ratio. None of the down-regulated gene depicted in the injured IVDs include 4833412C05Rik, Myoc, Serpina, Tcap: Myh, and Ckm have been previously described (Fig. 2). All genes used previously to mark injury and inflammation (Il6, Adam8, TNFa and Tnfaip2) were upregulated in response to disc injury (Fig. S1), although the magnitude of upregulation was not as high as the genes displayed in Fig. 2 [deposited at the GEO (Gene Expression Omnibus) database: http://www.ncbi.nlm.nih.gov/geo/; GSE#255262].

Figure 2. Differentially Expressed Genes Between Intact Control and Injured Intervertebral Discs (IVDs).

Each column represents data from one intact control (c) or injured (i) mouse mouse IVD. C1–3: intact controls; i1-4: injured mouse IVDs. Genes upregulated in injured IVDs: Chl1: Cell Adhesion Molecule L1 Like; Lum: Lumican; Saa: serum amyloid A; Cxcl1: chemokine (C-X-C motif) ligand 1; Il1rn: Interleukin 1 Receptor Antagonist; Clec4d: C-Type Lectin Domain Family 4 Member D; Mmp12: matrix metallopeptidase 12; Cd300if: CD300 Molecule Like Family Member F; Sirpb1c: Signal Regulatory Protein Beta 1; Tnc: Tenascin C; Lilra6: leukocyte immunoglobulin like receptor A6; Zc3h12d: Zinc Finger CCCH-Type Containing 12D. Genes downregulated in injured IVDs: 4833412C05Rik: RIKEN cDNA 4833412C05 gene; Myoc: myocilin; Serpina: serine peptidase inhibitor, clade A; Tcap: Telethonin; Myh: myosin heavy chain; Ckm: Creatine Kinase, M-Type.

Leukocyte migration and activation pathways were enriched among genes upregulated in injured IVDs.

Gene ontology (GO) for biological processes were analyzed among upregulated genes (injured/intact control ratio>1). Leukocyte migration and myeloid leukocyte activation are the top overrepresented biological processes. Not surprisingly, negative regulation of immune system processes also ranked very high, likely reflecting the effort to control inflammation during the repair process (Fig. 3B). Tnfa had the highest network connection among the upregulated genes (Fig. 3C).

Figure 3. Gene ontology and interaction networks for upregulated genes in injured intervertebral discs (IVDs).

A: Gene ontology analysis; P.adj: adjusted p-value. Q value: estimated false discovery rate. B: interaction network; Rank: upregulated genes in injured IVDs with descending number of network connections. Tnfa: tumor necrosis factor-alpha; Il6: interleukin 6; Cxcl: C-X-C motif chemokine ligand; Mmp: matrix metallopeptidase; Vcam: vascular cell adhesion molecule; Ccr2: C-C motif chemokine receptor 2; Tnfrsf: tumor necrosis factor receptor superfamily.

Cytoskeleton organization pathways were enriched among downregulated genes in injured IVDs.

Pathways related to cytoskeletal organization were enriched among downregulated genes in the injured discs. These processes are likely related to the repair process, replacing IVD tissues with fibrocartilage tissues (Fig. 4A). Acta1 (actin alpha 1), a gene for cytoskeletal structural component, has the highest number of network connections (Fig. 4B). When all different genes were analyzed together, muscle system process was the most enriched pathway and Tnfa was the most connected gene (Fig.S2).

Figure 4. Downregulated genes in injured compared with intact control intervertebral discs (IVDs).

A: gene ontology analysis; P.adj: adjusted p-value; Q value: estimated false discovery rate. B: interaction network analysis; Rank: genes downregulated in injured discs with descending number of network connections; Acta1: actin alpha 1; Myh: myosin heavy chain; Myl: myosin light chain; Actn: actinin; Atp2a1: SERCA Ca(2+)-ATPases; Tcap: Telethonin; Ckm: creatine kinase, muscle.

DISCUSSION

RNA sequencing of 18,078 genes in a mouse tail disc injury model compared with non-injured controls is presented here. Of those genes, 592 were differentially expressed in the injured model. Interestingly, 14 of the 15 most up-regulated genes have not been described previously in similar animal disc injury models. Gene ontology analysis of upregulated genes revealed that leukocyte migration was the most enriched biological process, and the most enriched downregulated pathways were related to cytoskeletal organization. Targeting these pathways may mitigate the degenerative cascade following disc injury.

The IVD degenerates with injury and aging. It is rarely feasible to examine human IVD tissues immediately post injury, but animal models provide an opportunity to examine the cellular molecular events in the acute phase. Here, we examined the transcriptome of mouse IVD tissue 1-week post injury. Consistent with previous findings, Il6, Adam8, Cxcl1, and Tnfaip2 gene expression were elevated post injury.¹⁶ Tumor necrosis factor alpha (TNFa) and TNF receptor superfamily (Tnfrsf1a, 1b and 13b) were among the most upregulated genes (Fig S1). This is not surprising, since TNFa is known to play a role in IVD degeneration and back pain.^{17, 18} Both Tnfrsf1a and 1b were also upregulated post injury. These two receptors have opposing effects, with TNFRSF1a (also known as TNFR1) proinflammatory and TNFRSF1b (aka TNFR2) pro-regenerative.¹⁹ The net effects of blocking TNFa may depend on the relative number of TNFRSF, and their affinity to the ligand.

Fourteen of the 15 genes most upregulated in IVD tissues in response to injury have not been described previously in the IVDs (include Lilra6, Chl1, Lum, Postn, Saa3, Mmp12, etc.). Some of these novel genes may be better markers for disc injury than we used before because of low individual variabilities. For example, Chl1 and Lum were elevated in 3 of 4 injured mouse discs, while Il6 and Cxcl1 were markedly elevated in 2 of 4 mice. Other novel genes found in this study may reveal additional mechanisms of disc degeneration following injury. For example, MMP12 may contribute to angiogenesis in the degenerative IVD because it activates endoglin,²⁰ which plays a crucial role in angiogenesis. Leukocyte migration and myeloid leukocyte activation are the most enriched biological processes among the upregulated genes. This is consistent with macrophage in the injured IVDs,¹⁶ despite the dense annulus fibrosus tissues rich in proteoglycans known to deter leukocyte infiltration. Tnfa and Il6 have the highest number of network connections. These pathways and molecules may serve as therapeutic targets since they may have the highest biological impacts.

Novel genes downregulated in the injured IVDs include 4833412C05Rik, Myoc, Serpina, Tcap: Myh, and Ckm. These genes have not been used as markers for disc injury/degeneration previously. The most enriched downregulated pathway was related to cytoskeletal organization. It will be a worthwhile future direction to examine their biological significance. The downregulation of genes linked to cytoskeletal organization potentially impacts cell structure and function in the IVDs, and might explain why the tissue undergo irreversible degeneration.

Genes with no expression values detected in two or more samples (a total of 17, including Cxcl5 and Cd300c2) were not included the heatmap and network analysis. These genes were excluded because their expression values at baseline were very low, so the fold changes are unreliable. However, these may be genes only expressed following injury and worth further studies. Another limitation of the current study is that only young adult male mice on the DBA background were used. Since genetic background, sex,²¹ and age are known to affect gene expression levels, these factors should be considered when data is used for future work. One limitation is that the mouse tail IVD injury only reflect some aspects of the human annular tear/disc herniation. The tail IVD is substantially different in load-bearing from that of human lumbar spine. In addition, the human disc injury often results from a compression or rotation type of mechanism. Caution must be applied in seeking to generalize the findings to other species or different types of IVD injury.

Another limitation is that the mice were treated with buprenorphine for pain related to the procedure. Opioid analgesics may affect the biomarkers used since opioids are known to be immune modulators. For example, fentanyl, methadone, loperamide and beta-endorphin induced a remarkable production of IL-4 by human T lymphocytes.²² Future work aimed at confirming the novel gene expression and protein distribution will include sham-injured mice exposed to the same medications. Finally, the genes associated with leukocyte migration may be expressed by the IVD cells, or by infiltrating leukocytes. In the future, Real-time PCR will validate the specific gene expression changes in response to disc injury. RNAScope (RNA in situ hybridization) and immunostaining will enable the changes in gene expression and protein production to be placed in a morphological context.

CONCLUSIONS

With a nonbiased approach, novel genes were found which could serve as markers for future studies. Genes used previously as markers for disc injury and inflammation have been confirmed valid. Leukocyte migration pathway was the most enriched biological process, and Tnfa had the most network connections among upregulated genes. Treatments targeting key pathways/molecules may be effective post injury, since they may have the highest impact on disc inflammation and repair.

Supplementary Material

Figure S1

Previously Described Genes Differentially Expressed Between Intact Control and Injured Intervertebral Discs (IVDs). Each column represents data from one intact control (c) or injured (i) mouse mouse IVD. C1–3: intact controls; i1-4: injured mouse IVDs. Tnfrsf: tumor necrosis factor (TNF) receptor superfamily; Il6: interleukin 6; TNFa: Tumor Necrosis Factor alpha, also known as Tnfsf1a; Adam8: ADAM metallopeptidase domain 8; Tnfaip2: TNF alpha induced protein 2.

Figure S2

Gene ontology and network analysis for differentially expressed genes in injured compared with intact intervertebral discs (IVDs). A: pathway analysis; P.adj: adjusted p-value. Q value: estimated false discovery rate. B: interaction network for all differentially expressed genes; Rank: upregulated genes in injured IVDs with largest number of network connections. Tnfa: tumor necrosis factor-alpha; Il: interleukin; Acta: actin alpha; Cxcl: C-X-C motif chemokine ligand; Myh: myosin heavy chain; Myl: myosin light chain; Actn: actinin; Mmp: matrix metallopeptidase; Vcam: vascular cell adhesion molecule.

Summary.

What is Known.

Inflammatory marker gene expression changes post disc injury.

What is New.

Whole transcriptome study following mouse tail IVD injury has been described here. Novel molecular markers, enriched biological pathways and highly connected molecules have been identified, which may help design treatments for discogenic pain post injury.

Abbreviations:

Tnfrsf

tumor necrosis factor (TNF) receptor superfamily

Il6

interleukin 6

TNFa

Tumor Necrosis Factor alpha, also known as Tnfsf1a

Adam8

ADAM metallopeptidase domain 8

Tnfaip2

TNF alpha induced protein 2

Chl1

Cell Adhesion Molecule L1 Like

Lum

Lumican

Saa

serum amyloid A

Cxcl1

chemokine (C-X-C motif) ligand 1

Il1rn

Interleukin 1 Receptor Antagonist

Clec4d

C-Type Lectin Domain Family 4 Member D

Mmp

matrix metallopeptidase

Cd300if

CD300 Molecule Like Family Member F

Sirpb1c

Signal Regulatory Protein Beta 1

Tnc

Tenascin C

Lilra6

leukocyte immunoglobulin like receptor A6

Zc3h12d

Zinc Finger CCCH-Type Containing 12D

4833412C05Rik

RIKEN cDNA 4833412C05 gene

Myoc

myocilin

Serpina

serine peptidase inhibitor, clade A

Tcap

Telethonin

Myh

myosin heavy chain

Ckm

Creatine Kinase, M-Type

Vcam1

vascular cell adhesion molecule 1

Ccr

C-C motif chemokine receptor

Acta1

actin alpha 1

Myh

myosin heavy chain

Myl

myosin light chain

Actn

actinin alpha

Atp2a1

SERCA Ca(2+)-ATPases

Tcap

Telethonin

Ckm

creatine kinase, muscle

Data availability statement:

all data are available upon request to corresponding author (Zhang). RNASeq dataset has been deposited to the Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/geo/; GSE#255262).