Inflammation in intervertebral disc degeneration and regeneration

Abstract

Intervertebral disc (IVD) degeneration is one of the major causes of low back pain, a problem with a heavy economic burden, which has been increasing in prevalence as populations age. Deeper knowledge of the complex spatial and temporal orchestration of cellular interactions and extracellular matrix remodelling is critical to improve current IVD therapies, which have so far proved unsatisfactory. Inflammation has been correlated with degenerative disc disease but its role in discogenic pain and hernia regression remains controversial. The inflammatory response may be involved in the onset of disease, but it is also crucial in maintaining tissue homeostasis. Furthermore, if properly balanced it may contribute to tissue repair/regeneration as has already been demonstrated in other tissues. In this review, we focus on how inflammation has been associated with IVD degeneration by describing observational and in vitro studies as well as in vivo animal models. Finally, we provide an overview of IVD regenerative therapies that target key inflammatory players.

1. Introduction

Between 70 and 85% of all people have low back pain (LBP) at some time in their life. LBP can limit the activity in people younger than 45, causing a tremendous socio-economic impact [1]. The aetiology of LBP is unclear but in 40% of cases it is related to intervertebral disc (IVD) degeneration [2]. In 90% of sciatica cases, it is also associated with a herniated IVD that compresses a nerve root causing pain [3]. Novel strategies such as gene therapy, growth factor injection, cell-based therapies and tissue engineering approaches are being developed towards impairing degeneration or promoting regeneration of the IVD [4]. However, to achieve full IVD regeneration it is also necessary to recover the biomechanical properties of a native IVD and restore the biological behaviour of resident cells, including production of healthy extracellular matrix (ECM), while ensuring reduction of IVD-associated pain.

Traditionally, inflammation has mostly been seen as detrimental and correlated with disease progression, but it remains unclear whether it is a cause or consequence of IVD degeneration and herniation. A balanced inflammatory response may be required for restoring IVD function as recently suggested for other tissues [5,6]. In this review, we will discuss the inflammatory reaction in IVD, both in homeostasis and IVD degeneration, and comprehensively cover the strategies applied to inflammatory cells and factors targeted towards IVD regeneration.

2. The origin of inflammation in intervertebral disc

Inflammation has mostly been regarded as a response to infection or tissue injury, but the scientific community has been increasingly researching its physiological role in maintaining tissue homeostasis [7]. In general, the mechanisms of inflammation are dependent on the inducing agent and context, with the inflammatory response in the context of infection having been investigated the most. In the infection instigated inflammatory response, plasma and leucocytes are recruited to the site of infection and soluble mediators that lead to recruitment and activation of other cell types are secreted. A complex cascade of events is triggered that eventually leads to clearing infection from the tissue and resolution of inflammation. In response to tissue injury, there also exists a vascular response and an orchestrated recruitment and activation of various cell types. However, the IVD is an avascular tissue and therefore it is unsurprising that the inflammatory response is different in this context.

The process of IVD degeneration implies a cascade of structurally disrupting events, normally starting with declining nutrition of cells within the central IVD, followed by accumulation of cell waste products and degraded matrix molecules. This creates an increasingly acidic environment which further compromises cell viability [8]. Various causes have been hypothesized to play a role in the pathogenesis of degenerative disc disease (DDD), such as endplate calcification, leading to an impairment of normal nutrition routes, excessive mechanical loading, genetic pre-disposition, unhealthy habits, ageing and spine infection [9–16]. Regardless of the cause, inflammation is an omnipresent player, and its association to LBP is clear [17]. Yet, it remains uncertain what may trigger the recruitment of immune cells to IVD and the associated inflammatory response (figure 1).

Figure 1.

Inflammation in the IVD. It is unclear whether inflammation is the cause or consequence of disc degeneration and herniation, and what may trigger activation and recruitment of different immune cells. The normal ageing process allied to some genetic pre-disposition causes the IVD to degenerate giving rise to profound changes in the ECM—loss of proteoglycan content, dehydration, malnutrition, decrease of native cell population, matrix breakdown and calcifications. In this scenario, the natural response to mechanical loading is compromised and the IVD becomes prone to microfissures and consequent ingrowth of blood and nerve vessels. (1) Disc herniation may also occur when the AF is no longer able to sustain the NP. ECM fragments and microcrystals may internally elicit an inflammatory response, stimulating endogenous IVD cells to produce pro-inflammatory mediators, that will further feed the cascade of tissue degeneration—IL-1β, IL-8, IL-6. (2) NP is recognized as non-self by the immune system. Hence, its exposure (both in microfissures and herniation) may propagate an immunologic response, with recruitment of macrophages, lymphocytes and other possible inflammatory cells, in order to eliminate the foreign body. Discogenic pain has been many times attributed to TNF-α, PGE2, NO and IFN-γ secretion by macrophages, concomitantly with NGF and substance P production, accompanying the processes of nerve ingrowth and angiogenesis inwards the degenerated IVD. Activated B and T lymphocytes are also recruited to the site, contributing to the positive pro-inflammatory feedback loop established. It is not well understood how endogenous IVD cells interact with exogenous inflammatory cells and whether they positively contribute to tissue resorption and regeneration or not. Spontaneous disc regression is currently believed to be a consequence of macrophage activity.

One hypothesis for the cause of the IVD inflammatory response relies on endogenous factors, such as crystals and ECM breakdown products, which could induce the inflammatory response [7]. Crystal deposits of calcium pyrophosphate dihydrate (CPPD), cuboid microstructures (characterized as magnesium whitlockite) and hydroxyapatite (HA) have been observed in degenerated IVD specimens [15,18–20]. In articular cartilage, regions with crystals showed altered amounts of collagen, calcium-binding proteins, decorin and large proteoglycan content, as well as abnormal pericellular matrix deposition [21,22]. Phagocytosis of crystals present in joints and pericellular tissues can trigger activation of the NOD-like receptor family pyrin domain containing 3 (NALP3) inflammasome. This cytoplasmic multimolecular protein complex regulates activity of caspase-1 and maturation and release of interleukin (IL)-1β [7,23], the latter being commonly found in degenerated IVD [24].

ECM breakdown products generated during tissue dysfunction or damage may also promote an inflammatory response as has been shown in various models [25,26]. The IVD is mostly composed of ECM molecules, including collagens, proteoglycans and other matrix proteins (see table 1 for more information), which are continuously synthesized and degraded by local existing proteases to maintain homeostasis (table 2) [66,74–76]. However, when an imbalance occurs, degradation products might trigger inflammation. For instance, fibronectin fragments alleviate metalloproteinase (MMP) inhibition and so promote monocyte migration in vitro [77]. This certainly facilitates their recruitment into the inflamed region. Fragments of laminin, collagen type XIV and fibrin can also modulate inflammatory cell infiltration and proliferation in other systems [26]. In cartilage explant cultures, fibromodulin fragments are produced following IL-1 stimulation [78]. In different in vitro settings, fragments originated from elastin, laminins, collagen (type I and IV), fibronectin, ectactin/nidogen, thrombospondin and hyaluronan also induce protease and cytokine production, independent of their chemotactic activity [79]. Some of the aforementioned studies were performed in highly vascularized model systems (i.e. cardiovascular, lung or renal tissues). While these systems are very different from healthy adult IVDs, which are largely avascular, their findings might still be of relevance when studying phenomena associated with disc herniation or sequestration, in which blood vessels are much more abundant [80].

Table 1.

Main extracellular matrix components of a young and healthy IVD.

name	distribution/localization	putative/possible function
collagens
fibril-forming collagens
type I	AF and NP	confers tensile stiffness allowing torsion and flexion [27–30]
type II	AF and NP	confines PG within the matrix to retain more water to allow larger deformations and withstand greater compressive loads [31,32]
type III	NP and outer AF	organizes pericellular environment; allows extensibility of tissue [33–35]
type V	AF and NP (increased in AF cells when compared to NP cells)	regulates fibril diameter (smaller if this collagen is more abundant) influencing mechanical properties [35,36]
type XI [37]	all over, mostly NP	regulates fibril diameter (smaller if this collagen is more abundant) influencing mechanical properties [35,38]
beaded-filament forming collagens
type VI	all over, mostly NP	helps cell fixation to the matrix and facilitates collagen bundles' sliding and lubrication [39,40]
FACIT collagens
type IX	NP	maintains matrix integrity [41,42]
type XII	AF	might regulate fibrillogenesis [29,43]
type XIV	AF	might regulate fibrillogenesis [29,43]
proteoglycans
aggregating PGs
aggrecan	AF and NP	maintains IVD's osmotic pressure; may act as an anti-angiogenic factor due to its inhibition of endothelial cell migration [40,44,45]
versican	all over, mostly AF	favours the attachment of adjacent lamellae, contributes to resistance to compressive forces and facilitates cell migration, since it is an anti-adhesive molecule [35,40]
non-aggregating PGs
small leucine-rich proteoglycans (SLRPs)
decorin	outer AF and fibrillar NP	regulates collagen fibril diameter and spacing, maintaining uniform patterning; GFs' reservoir (TGF-β), modulating ECM metabolism [46–48]
biglycan	outer AF and NP (fibrillar and pericellular region)	GFs' reservoir (TGF-β), modulating ECM metabolism [46,47]
asporin	outer and inner AF, rarely NP	GFs' reservoir (TGF-β), modulating ECM metabolism; may play a major role in modulating chondrocyte matrix homeostasis [49,50]
fibromodulin	AF and NP	regulates collagen fibril diameter and spacing, maintaining uniform patterning; GFs' reservoir (TGF-β), modulating ECM metabolism [44,47]
lumican	AF and NP	regulates collagen fibril diameter and spacing, maintaining uniform patterning [44,51]
prolargin (encoded by PRELP*)	all over, mostly AF	anchors basement membranes to the underlying connective tissue [44,52]
chondroadherin	AF and NP	binds integrin ad collagen; regulates cell metabolism and ECM structure, promoting matrix homeostasis [44,53,54]
osteoglycin/mimecan	AF and NP	unknown [44]
other matrix proteins
other PGs
perlecan	AF and NP	has a role in cell proliferation and differentiation by acting as co-receptor for FGFs; matrix organization and stabilization; role in FGF signalling [40,55]
fibronectin	all over the disc	preserves structural integrity of the ECM; involved in cell adhesion through interaction with integrins [56–58]
elastin	all over the disc	preserves structural integrity of the ECM; helps to regain disc height and shape after deformation [59–61]
COMP	all over, mostly AF	preserves structural integrity of the ECM; binds other matrix proteins and catalyses polymerization of type II collagen fibrils; prevents vascularization of cartilage [62,63]
thrombospondin	AF	preserves structural integrity of the ECM; mediates cell adhesion, matrix–matrix interactions, cell migration and proliferation in other tissues; might prevent vascularization of the tissue; activates TGF-β complex [64,65]

Table 2.

Main IVD proteinases.

name	distribution/localization	putative/possible function
aggrecanases
ADAMTS1, 4, 5, 9 and 15	ADAMTS1: NP and AF ADAMTS4: low levels NP and AF ADAMTS5: low levels NP and AF ADAMTS9: NP and AF ADAMTS15: low levels NP and AF	degrades aggrecan [66–69], as well as versican, biglycan, fibromodulin, COMP, TSP1, TSP2, nidogen, among other substrates [70]
collagenases
MMP1, 8 and 13	MMP1: low levels, mostly inner AF and NP MMP8: low levels MMP13: low levels, mostly NP	cleaves fibrillar collagen [66–68]
gelatinases
MMP2 and 9	MMP2: low levels, mostly inner AF and NP MMP9: low levels AF and NP	degrades denatured collagen and basement membrane collagen [68]
stromelysin
MMP3 and 10	MMP3: low levels, mostly in the adult NP MMP10: only checked in the NP	digests non-collagenous matrix proteins and denatured collagen [66,68,71]
matrilysin
MMP7	NP and inner AF	degrades aggrecan and collagen type II [72]
other MMPs
MMP19	AF and NP	cleaves aggrecan, COMP, types I and IV collagen, and fibronectin and acts on tenascin; can interfere with stabilization of capillary-like structures, possibly playing a role in the avascular status of the disc; regulates IGF-mediated proliferation in other tissues by proteolysis of IGFBP3 [73]

It appears that fragment release initiates and propagates the inflammatory response locally. Many of these fragments (e.g. originated from biglycan, fibronectin, hyaluronan) signal through toll-like receptor-2 (TLR2) and/or TLR4 in other model systems [81,82]. TLR4, in particular, is a well-known pattern recognition receptor involved in innate immune responses that has been implicated in inflammatory degeneration [83]. In human IVD cells, hyaluronic acid fragments (fHA) lead to increased mRNA expression levels of inflammatory and catabolic genes IL-1β, IL-6, IL-8, cyclooxygenase (COX)-2, metaloprotease-1 and -13, and IL-6 [84]. However, while IL-6 production is dependent on TLR2 it is independent of TLR4. It should be noted that low and high molecular weight molecules can have different effects, even through the same pathways. For example, high molecular weight hyaluronan protects epithelial cells against pro-apoptotic stimuli through NF-κB activation, in a TLR dependent way. Low molecular weight degradation products can induce inflammation, promoting macrophage mediated production of IL-1β and tumour necrosis factor alpha (TNF-α), through activation of the NF-κB/IκBα complex [82]. It is difficult to assess the overall role of ECM proteins within an immune setting because of their dual roles and because many proteases and a variety of fragments are released simultaneously. This difficulty is exacerbated by the scarcity of in vivo data, owing to limitations in the techniques used to detect fragments and immune cells, which are present at low concentrations and are short lived [85].

Numerous studies suggest that the IVD might endogenously include inflammatory-like cells [86,87]. In particular, it has been shown in vitro that a population of IVD cells can phagocytize beads and apoptotic bodies [86]. In turn, human surgical non-herniated nucleus pulposus (NP) samples presented a high number of resident CD68+ cells [87]. Furthermore, a recent robust analysis of cytokine/chemokine expression profile of human NP cells has presented clear evidence that NP cells, or at least some of them, are producers of specific inflammation-associated molecules, even in basal conditions (non-degenerated NP) [88]. In addition, infiltrated leukocytes (CD11b-positive cells) were found even in prolapsed IVDs, where NP is supposedly intact and isolated from any vascular source of immune cells [88]. The question of whether these cells could be resident macrophages or macrophage-like cells remains. Although pleiotropic, cytokines and chemokines have three modes of action: (i) stimulating the production of other inflammatory mediators and MMPs, (ii) enhancing matrix degradation, and (iii) recruiting inflammatory cells and activating phagocytosis [89–94]. Together, these effects can contribute to disease progression in the IVD.

It should be stressed that extreme mechanical loading has also been shown to alter ECM properties (through proteinase activation) and promote inflammation, contributing to IVD degeneration [95]. Apart from in vitro studies, organ cultures of bovine caudal IVDs have shown that compression induces apoptosis, produces inflammatory mediators and alters matrix integrity, leading to development of the disease [96].

In a more advanced degenerative stage, the well contained and apparently ‘sealed’ NP (immune privileged) becomes exposed to immune cells, which, responding to an inflammatory stimulus, may arise from newly formed blood vessels that invade pathological clefts and tears found in the annulus fibrosus (AF). Nociceptive nerve fibre ingrowth also accompanies angiogenesis and is believed to be the origin of discogenic pain that actively contributes to LBP [97–100]. Indeed, while the probable sites for focal damage and inflammation are vertebral endplates and AF (the only sites where the IVD is vascularized) [101], the NP is capable of attracting leukocytes and increasing vascular permeability when implanted subcutaneously [100]. An increase in expression of some cytokines and MMPs in herniated IVD may occur when molecules seen as ‘non-self’ by immune cells become exposed. This may also be linked to the phenomenon of spontaneous regression or disappearance of extruded IVD fragments, which has been attributed to matrix degradation and phagocytosis by recruited/infiltrated macrophages [102–106]. In cases with transligamentous extrusion, which can occur when the NP is potentially more exposed to immune cells, regression occurs [107]. Furthermore, the survival rate of subcutaneously transplanted rat NP cells is higher in immunocompromized NOD mice, and both rat macrophages and NK cells lyse autologous NP cells in vitro, indicating that immune cell populations respond to NP tissues [108].

In the next sections, we will review observational, in vitro and in vivo studies of the inflammatory milieu in IVD.

3. Inflammatory key players in intervertebral disc

3.1. Observational studies

A range of cytokines have been found in human IVDs in varying amounts, depending on whether the IVD is healthy, degenerated or herniated. Table 3 groups by methodology (observational, in vitro or pathway analysis) some of the most important studies that have clarified which inflammatory factors are expressed during homeostasis and with degeneration. Importantly, the identity of the cells producing these mediators (i.e. NP cells, AF cells, native IVD cells only, native cells plus infiltrating inflammatory cells) is highlighted.

Table 3.

Inflammatory mediators found in the human IVD.

mediator	condition	producing cells	references
inflammatory factors that are expressed during homeostasis
observational studies
TNF-α	post-mortem and non-degenerate samples	IVD cells	[24,71,109]
IL-1β		IVD cells	[71,109–112]
IL-1α, IL-1Ra, IL-1RI, and ICE		NP and AF cells	[112]
IL-6, IL-8, RANTES		AF and NP cells	[110]
NGF		NP and AF cells in monolayer and alginate bead culture	[71,113]
NGF receptor (trkA)		NP and AF cells in monolayer and alginate bead culture	[113]
substance P			[71]
PLA2		NP and AF cells	[114]
CCL3 and CCL4		IVD cells	[115]
NOTCH		IVD cells	[109]
MMPs		IVD cells	[71,92]
inflammatory factors that are expressed with degeneration
observational studies
TNF-α	herniations (including subligamentous and transligamentous), protrusion, extrusion, sequestration, spondylosis, scoliosis, degenerated or discogenic pain	IVD cells and infiltrating cells	[24,71,116–121]
IL-1β			[71,111,117,118,120,121]
IL-1α			[117–119]
IL1-Ra, NO			[118]
IL-6			[17,117,118,122]
IL-8			[17,119]
IL-12, IL-17, IFN-γ			[122]
IL-20 (and its receptors)			[90]
IL-10, TGF-β, RANTES			[119]
IL-16, CCL2, CCL7, CXCL8			[88]
substance P			[71,118]
PGE2			[17,118]
COX-2			[121]
PLA2			[114,121]
NGF			[71]
VEGF, bFGF			[116]
GM-CSF			[117]
CD20, CD45RO, CD68			[107]
MMPs			[71,107,116,118]
in vitro studies
TNF-α	degenerate, sciatica, discogenic pain, extrusion, sequestration	IVD cells and infiltrating cells treated with different inflammatory stimulus (including IL-1β, TNF-α, substance P, IL-17, IL-20, IFN-γ, LPS) in monolayer or three-dimensional cell culture, or co-cultured with macrophage-like cells, or exposed to high mechanical strain	[95,121,123–125]
IL-1β	discogenic pain and post-mortem		[110,112,124–126]
IL-1α, IL-1Ra, IL-1RI, and ICE	degenerate and post-mortem		[112]
IL-6	discogenic pain, scoliosis, sciatica, extrusion, sequestration, degenerate, myelopathy or radiculopathy and post-mortem		[89,90,95,110,124,126–129]
IL-8	discogenic pain, scoliosis, sciatica, extrusion, sequestration, degenerate, myelopathy or radiculopathy and post-mortem		[90,95,110,124,126–128]
IL-17A	protrusion, extrusion and scoliosis		[123]
IL-15, IFN-γ, CXCL9, TLR-2, TLR-4, MCP-3	post-mortem		[95]
RANTES	discogenic pain and post-mortem		[110]
MCP-1	discogenic pain, scoliosis, sciatica, extrusion, sequestration and post-mortem		[90,95,127]
TFGF-β1	discogenic pain, scoliosis, sciatica and post-mortem		[95,127]
substance P	myelopathy or radiculopathy		[126]
bFGF	scoliosis, sciatica and discogenic pain		[127]
PGE2	degenerate, scoliosis, sciatica and discogenic pain		[89,121,128–130]
COX-2	extruded and sequestrated IVD tissue		[121]
NGF	post-mortem		[95,113]
NGF receptor (trkA)	post-mortem		[113]
PGF2α	degenerate		[128]
NO	degenerate, scoliosis		[89,124,129,130]
ICAM-1 (CD54)	degenerate and scoliosis		[89]
MMPs	degenerate, non-degenerate, extrusion and sequestration		[90,92,112,130]
NOTCH	protrusion		[109]
pathway analysis
NF-κB, MAPK and C/EBPβ→ CCL3	discectomy	IVD cells and infiltrating cells	[115]
NF-κB→ADAMTS-4 and -5	DDD and myelopathy	NP and infiltrating cells treated with IL-1β and TNF-α	[131]
NF-κB→Sox9 and collagen type II	spine trauma	IVD cells treated with IL-1	[132]
NF-κB and MAPK→NOTCH pathway	protrusion	NP cells	[109]

3.1.1. Post-mortem samples

Separating NP from AF tissue upon discectomy is a very challenging task, particularly when dealing with degenerated human IVD tissue. In cases of disc herniation (e.g. extruded, sequestered), the IVD is invaded by other cell types, confounding analysis of molecules released by regions of the IVD. Samples taken post-mortem, which are not contaminated by infiltrating inflammatory cells (or at least not to the same extent as herniated discs), are therefore superior when investigating IVD homeostasis. For instance, TNF-α was substantially expressed in autopsy material in fetal/infantile and older adult NP, whereas it was sparsely expressed in adolescent and young adult NP. It was not found in the AF of young adults (below 25 years), but significantly increased in older individuals [24]. Also, calcium-dependent phospholipase A2 (PLA2), a regulator of prostaglandin E2 (PGE₂) production, has been found in both cadaveric and surgical samples, and IVDs of middle-aged cases had higher PLA2 activity than those of younger and older subjects, indicating an important physiological role in maintaining homeostasis [114]. Like TNF-α or PLA2, many other inflammatory key players have been localized in non-degenerated human IVD tissue (table 3). Importantly, IVD native enzyme activity has also been studied. It was shown in the intact IVD that IL-1 is a key cytokine mediating IVD matrix degradation, by measuring enzyme activity (in situ zymography (ISZ)) against gelatin, collagen II and casein matrices [92]. Also, MMP-10 expression (at mRNA and protein levels) was increased in the symptomatic degenerate IVD, when compared to non-symptomatic one—possibly contributing to matrix degradation and initiation of nociception [71]. An additional perspective is given by studies that identified factors not naturally produced by native IVD cells: immunoreactivity for IL-4, IL-6, IL-12 and interferon (IFN)-γ was modest in surgical IVD tissue, being higher in herniated IVD samples and virtually non-existent in the control samples taken from post-mortem non-degenerated IVDs [122]. The majority of these reports used post-mortem IVD samples as healthy controls, exposing the role of IVD native cells in IVD homeostasis. This knowledge is of potential interest for the development of endogenous therapeutic routes to restore homeostasis in DDD.

3.1.2. Degenerated samples

Regarding human degenerated IVD samples, early studies detected the presence of IL-1, intracellular adhesion molecule-1 (ICAM-1), lymphocyte function-associated antigen (LFA) and fibroblast growth factor (FGF) [133]. Immunoreactivity for some cytokines (IL-4, IL-6, IL-12, IFN-γ, TNF-α) seemed modest but evident in herniated and degenerated discs, with substantial macrophage infiltration [116,122]. Also, pathologic discs highly expressed IL-17, suggesting the involvement of Th17 lymphocytes in disc herniation [123]. Others have shown a higher expression of TNF-α, IL-1β, IL-6, IL-8, IL-20, PGE₂ and nitric oxide (NO) in herniated discs [17,90,117]. Some inflammatory key players have also been associated with pain in human IVD: RANTES and IL-1β expression was significantly higher in painful versus painless discs, contrarily to IL-6 and IL-8 [110]. A strong difference was observed in the levels of nerve growth factor (NGF), neurofilament-68, growth-associated protein (GAP)-43, and substance P in invading nerve fibers, in and around the outer layer of uncontained herniated versus spondylotic IVDs [116]. Another study that evaluated 91 cytokine- and chemokine-associated genes in human NP cells showed that NP cells are a source of IL-16, CCL2, CCL7 and CXCL8 [88]. Some of the pro-inflammatory cytokines usually present at increased levels in human degenerated discs, such as IL-1β and TNF-α, may mediate catabolic effects, decreasing proteoglycan production and enhancing MMP expression [71,111,134,135].

3.2. In vitro studies

Different in vitro studies have focused on studying the sources and role of some inflammatory mediators associated with herniated and degenerated IVD. An increase in IL-6, IL-8 and PGE₂ was observed in control and degenerated human IVD tissues upon lipopolysaccharide (LPS) stimulation [127]. Furthermore, it was shown that substance P, expressed by IVD cells, upregulates IL-1β, IL-6 and IL-8 in both NP and AF, and RANTES and TNF-α in AF only [126]. Also, NP cells were shown to express the CCL3 ligand (also known as macrophage inflammatory protein (MIP)-1α), which is well known for its chemotactic and pro-inflammatory effects, through activation of the MAPK, NF-κB and C/EBP signalling pathways after treatment with IL-1β or TNF-α [115]. These studies suggest a contribution of native IVD cells to the inflammatory milieu. However, others have defended the hypothesis that immune cells, such as macrophages, neutrophils and T cells, can be recruited to degenerated IVD [135]. This hypothesis is supported by evidence that Th17 cells expressing CCR6 are recruited to degenerated IVD by CCL20 secretion from NP [123], and that macrophages can migrate after stimulation with conditioned medium from rat NP cells treated with IL-1β or TNF-α [115].

TNF-α, which is one of the most studied pro-inflammatory molecules, is known to promote aggrecan degradation, disc catabolism and expression of pro-inflammatory cytokines and NGF, without any recovery [136]. TNF-α is an adipokine that has been associated with higher numbers of bovine IVD senescent cells and is therefore implicated with the inability of degenerated IVD to repopulate by itself [137]. Curiously, although many studies have focused on the role of TNF-α in IVD degeneration [136], Hoyland and co-authors suggest instead that IL-1β is the key regulator of matrix degradation in degenerated IVD: IL-1 has greater expression in the IVDs clinically associated with chronic LPB and treatments against IL-1β were shown to inhibit matrix degradation [92]. IL-1 is upregulated in degenerated human discs, inducing MMP7, MMP13 and ADAMTS (a disintegrin and metalloproteinase with thrombospondin motifs)—suggesting a deregulation of the normal IVD homoeostasis [138]. TNF-α blockers had no effect on matrix-degrading activity, suggesting that its upregulation in DDD is not associated with matrix degradation [92] but instead with neighbouring nerve root irritation, which is confirmed by other studies [134,139]. Hence, TNF-α might be contributing to discogenic pain in cases where nerve ingrowth into IVD degenerative fissures occurs [140,141].

3.3. In vivo studies

Although some IVD degeneration related inflammatory mediators identified in vitro have not yet been studied in vivo, recent evidence has shed light on the role of several molecules. The majority of studies have been conducted in rat, rabbit and porcine animal models.

In a rat animal model of IVD herniation, NP exposure led to increased IL-6, TNF-α and IFN-γ levels. Other cytokines (IL-1β, IL-10, IL-1α and IL-2), already increased by the surgical procedure, were not altered [142]. In another study, TNF-α was identified in rat herniated IVDs and associated with radicular pain [139]. Also, a rat model of caudal annular incision demonstrated a transient peak in IL-1β 4 days following injury. This model was characterized by NP size decrease, annular collagen layer disorganization, and cellular metaplasia of annular fibroblasts to chondrocyte-like cells. However, no significant changes in TNF-α or IL-6 were seen [143]. In a lumbar rabbit annular incision model, no alteration in IL-1α or TNF genes was observed in whole IVDs at either one or three weeks after injury [144]. In the same model, IL-1β, transforming growth factor (TGF)-β1 and iNOS (inducible nitric oxide synthase) gene expression increased after three weeks, but decreased between weeks 6 and 12, having a second peak at 24 weeks—possibly showing a long-term pro-inflammatory action [145]. In a rabbit model of IVD herniation, the presence of TNF-α, IL-1β and MCP-1 (which has been demonstrated to be a potent macrophage chemoattractant [146]) was also analysed: TNF-α and IL-1β were detected after day 1 (via immunohistochemistry) followed by MCP-1 3 days post-injury. Infiltrating cells, mainly macrophages, were also observed after day 3 [147]. Interestingly, in a lumbar porcine model of annular incision, a significant increase in IL-8 accompanied by a decrease in IL-1 was observed in IVDs subjected to discectomy at 12 weeks post-injury, while no difference was observed in disc morphology, proteoglycan content, or in levels of IL-6 and TNF-α expression between untreated and injured IVDs. Whereas both IL-1 and IL-8 have pro-inflammatory properties, the authors propose that such discectomy procedure may be capable of initiating a repair response in the IVD, given that expression of IL-8 (an anabolic agent) is increased when the catabolic IL-1 decreases [148].

The TLR4-ligand LPS triggered inflammation when injected in a rat caudal IVD [149]. LPS injection in a rat IVD led to an increase in the levels of IL-1β, TNF-α, HMGB1 (high-mobility group box 1) and MIF (macrophage migration inhibitory factor), which correlated with morphological changes in tissue organization, namely interruption of NP/AF border, contraction of NP shape and decrease of IVD height [149]. Recently, different components of the tissue renin-angiotensin system (tRAS) (angiotensin converting enzyme, Ang II, Ang II receptor type 1, Ang II receptor type 2 and cathepsin D), that contribute to inflammation in many organs, have been found for the first time in the normal rat IVD, at both mRNA and protein levels [150]. However, the association between tRAS and IVD degeneration and its relationship to IVD inflammation has yet to be elucidated.

Most of these in vivo studies mainly identify and quantify inflammatory mediators in IVD, but fail to produce a mechanistic explanation of their role in either IVD degeneration or regeneration. One of the reasons for this failure may be the difficulty of unravelling the complex inflammatory mechanisms in injury models of IVD pathology. In this context, animal models of spontaneous IVD degeneration such as the sand rat [151] and both younger chondrodystrophic (with cervical or thoracolumbar IVD disease) and older non-chondrodystrophic (caudal cervical or lumbosacral IVD disease) dogs [152,153] could bring new insights to the clinic. However, as far as we are aware, the inflammatory response has not been addressed in these models. Importantly, differences between species could also bring some intricacy to this issue. For instance, notochordal cells (NC) seem to disappear in the adult human IVD, while in many other species they are retained throughout adulthood [154]. As more models become available, it is central to translate information between species and to interpret the models appropriately to understand in greater depth the process of inflammation.

4. Strategies to target and modulate inflammation towards intervertebral disc regeneration

Promoting IVD regeneration relies on restoring naive IVD properties by: (i) recovering IVD biomechanics, (ii) re-establishing cell biological activity, including production of healthy ECM, and (iii) reducing IVD-associated pain. Biological approaches focusing on IVD regeneration or IVD-associated pain relief begun in the early 1990s, and have since increased in number and diversity as reviewed elsewhere [4].

A well-balanced approach supporting tissue regeneration and control of inflammatory response could be successful in reducing IVD-associated pain. Although the molecular mechanisms behind IVD pathology and inflammatory response remain to be elucidated in detail, some inflammatory-related molecules are key targets of novel therapies in DDD [155,156]. In this section, an overview of the strategies targeting inflammatory mediators towards IVD regeneration will be given.

4.1. Injection of molecules

The more direct approaches to control inflammation in IVD are to inject regulating agents close to the IVD. Example agents are TNF-α blockers, such as infliximab, adalimumab, etanercept [136,157–160], or IL-1 inhibitors, such as IL-1 receptor antagonist (IL-1Ra) [161]. Other TNF-α blockers include a monoclonal antibody tested in herniated IVD patients, who showed less leg and back pain after antibody administration [162], and a p38-TNF-α inhibitor, which was tested in the spine to address neuroinflammation but not specifically for IVD [163]. Other strategies involve injecting corticosteroids into the IVD [164], or the anti-cholesterolemic drug simvastatin [165], which appeared to retard IVD degeneration in animal models.

The therapeutic potential of IL-1Ra for sustained attenuation of IL-1β has also been explored using poly(lactic-co-glycolic acid) (PLGA) microspheres as a delivery system [166]. IL-1Ra-PLGA microspheres inhibited NO production in NP cell cultures and partially restored the levels of iNOS, ADAMTS-4, MMP-13, IL-1β, IL-6 and TLR-4, which were increased in the presence in IL-1β [166].

Another candidate to control inflammation in IVDs is COX-2, which regulates PGE₂ synthesis in inflammatory conditions. Epidural injection of COX-2 inhibitors was shown to reduce pain in a rat model of IVD herniation [167]. Another approach uses platelet-rich plasma (PRP) as a therapy for degenerated IVD [168]—PRP was able to rescue chondrocyte degeneration induced by IL-1β and TNF-α [169].

Other approaches to reduce IVD-associated pain have been suggested. Resveratrol, a naturally occurring polyphenol present in red wine, was able to reduce IL-6, IL-8, MMP1, MMP3 and MMP13 expression when injected into the IVD [170]. Rhein (4,5-dihydroxyanthraquinone-2-carboxylic acid), an anthraquinone molecule derived from the rhizome of Rheum palmatum that exhibits anti-inflammatory activity and is used in the treatment of osteoarthritis and pain relief, was hypothesized to be a therapeutic agent for IVD through the regulation of IL-1 activity [171]. Fullerol, a derivative from fullerene and known anti-oxidant, retards cellular apoptosis and suppresses dorsal root ganglion (DRG) and neuron TNF-α-induced inflammatory responses, which when incorporated into nanoparticles is relevant for LBP treatment [172].

A very recent and promising approach to target inflammation in IVD is the inhibitor of IκB kinase-β (IKKβ), involved in NF-κB activation. The intradiscal injection of IKKβ reduced the levels of TNF-α, IL-1β and IL-6 of an injured IVD while suppressing high levels of neuropeptides within DRG neurons [173].

Despite good results in other therapeutic areas, injected molecules might be inefficient in DDD due to the short half-life of proteins in solution and the limited effect of a single protein in a complex process such as IVD degeneration [174]. Also, the risk of inducing IVD degeneration by puncturing of the IVD should be considered [175], although more recent studies describe alternative routes for molecule delivery [176]. Furthermore, given the predominantly avascular nature of IVD, systemic delivery of soluble molecules is unlikely to be effective in this situation. This view is supported by the report that the concentration of antibiotics in IVDs was undetectable in the NP of patients with IVD infection under systemic administration of antibiotics [177]. In addition, it has been shown that the rate of diffusion of antibiotics into the IVD is reduced by endplate calcification, increase in IVD size and solute molecular weight [178,179]. Moreover, although some nutrients' diffusion to the NP may occur via the endplates, the short half-life of pharmaceutical drugs or proteins can result in limited therapeutic doses that reach the NP [166].

4.2. Gene therapy

Gene therapy promises more prolonged effects in DDD, by introducing the possibility of locally modulating the expression of a specific gene and the consequent production of its protein [174]. As early as 1997, a study suggested genetic modifications as a therapy for DDD [180]. At that point, a retrovirus vector was proposed to transduce bovine chondrocytic endplate cells with IL-1Ra [180]. Cell transfection resulted in IL-1Ra production in 48 h, and injection of transfected cells in degenerated NP explants considerably reduced expression of several enzymes (such as MMP3) for two weeks after injection. This strategy aims at decreasing IL-1 mediated matrix degradation and stopping DDD progression [181]. In vivo, TGF-β1 transfection of rabbit IVD cells also enhanced proteoglycan synthesis for six weeks [182]. In agreement with this result, cells from human degenerated IVD transfected with TGF-β1 increased both proteoglycan and collagen production [183,184].

Gene therapy in a clinical setting may be limited by the safety of the gene transfer vector. Aspects such as exposure to high doses, long-term use, misplaced injections and the possibility of oncogenesis are key concerns when treating a chronic disease like DDD [185]. Progress in the development of more reliable viral vector constructs and in a better control of transgene expression would improve the safety of these therapies. Also, elucidation of molecular mechanisms behind the degenerative process and characterization of cell populations in IVD, as well as their role in ECM production, could bring new advancements to this field [174].

4.3. Cell-based therapies

Several cell-based therapies to stimulate IVD regeneration have been proposed in recent years: haematopoietic stem cells (HSC) [186], fetal spine cells [187], immortalized NP-cell lines [188], autologous IVD chondrocytes [189], embryonic stem cells (SC) [190], induced pluripotent SC [191], olfactory SC [192] and MSCs (derived either from bone marrow [193] or from umbilical cord blood [194]) have all been suggested to have potential for IVD repair/regeneration. NP progenitor cells were isolated from the NP (with approximately 1% frequency) and differentiated into chondrogenic and neurogenic lineages, suggesting potential for IVD regeneration [195]. Besides IVD regeneration, progenitor cells might play a protective role in regulating inflammation in IVD: rabbit NC were shown to reduce the levels of pro-inflammatory cytokines, IL-6 and IL-8, as well as iNOS, in in vitro co-cultures of AF cells with macrophages [196].

MSCs are one of the most attractive candidate cell types for IVD regeneration, partly because they could be autologous transplants. In a canine model, MSCs were able to increase collagen type II expression while decreasing cell apoptosis in IVD [197]. In rabbits, MSCs were able to remain in the IVD up to 24 weeks [198]. However, the number of transplanted MSCs is crucial; in the canine model 10⁶ MSCs per IVD was ideal, since 10⁵ MSCs resulted in decreased cell viability while 10⁷ MSCs induced cell apoptosis [197]. Besides MSC multi-differentiation capacity, an associated immuno-modulatory effect has been suggested [199]. MSC role in inflammation is based on their active role as cytokine-release factories that interact directly with injured cells [200]. In this novel scenario, MSCs were shown to secrete IL-1Ra in a mouse model of lung injury [201] or produce a potent anti-inflammatory protein (TNF-α stimulated gene/protein 6, TSG-6) in a mouse infarct model [202]. Interestingly, TSG-6 was also identified as a key player in a rat model of corneal injury after MSC systemic administration [203]. After implantation of MSCs into beagle nucleotomized IVDs, the expression of Fas ligand (FasL) (a protein found in other immune privileged sites) was restored. It was suggested that MSCs either differentiated into cells expressing FasL, or stimulated the few remaining NP cells to produce this molecule—thus contributing to the recovery of immune privilege in degenerated IVDs [204]. Although the beneficial effects of these cells have been demonstrated in several models, the mechanisms behind MSC-based therapies are not clear.

In vitro studies showed that MSCs repress IgG production of peripheral blood lymphocytes co-cultured with IVD fragments from the same donors [205]. MSC influence in IVD inflammatory response has not been fully dissected until now due to its multi-factorial complexity and time dependence [205]. Human MSCs were able to downregulate gene expression of pro-inflammatory cytokines (IL-3, IL-6, IL-11, IL-15, TNF-α) and MMPs when in co-culture with rat NP cells [198].

In humans, two clinical trials took advantage of autologous MSCs, albeit with controversial results. MSCs were either directly injected in NP [206] or implanted in the IVD after seeding in collagen sponges [207]. In the first case, when MSCs were directly injected in patients diagnosed with DDD, but with preserved external AF and persistent LBP, the lumbar pain was strongly reduced after three months. However, no improvement on IVD height was detected by imaging [206]. Injection of MSCs in degenerated IVDs seems to promote an analgesic effect, due to trophic effects, which can occur quicker than detecting possible regenerative effects [206]. Given this, the authors suggest that the MSCs exhibited immuno-modulatory properties. The second case reports the implantation of MSCs after seeding in collagen sponges [207]. Two years post-surgery, the published results reported relief or disappearance of LBP and improvement of the vacuum phenomenon (gas in the intervertebral space, associated with intervertebral regressive degeneration). However, besides the small number of patients used (two), this study also lacks experimental details such as the controls used and effective number of cells transplanted [207]. More recent studies propose IVD injection of umbilical cord-derived MSCs as a promising therapy to overcome chronic discogenic LBP [208]. In this study, pain and lumbar function were recovered after cell transplantation and preserved over a 2-year follow-up period; however, only two patients were studied. Another recent study injected bone marrow concentrate cells into 26 patients [209]. In the 1-year follow-up study, the majority of the patients showed improvement of pain score and reduced impairment, with only some of them presenting IVD rehydration. The authors emphasize the use of critical unmanipulated cell doses. Usually, MSC-based therapies involve cell expansion in vitro to obtain sufficient cell numbers, but this in vitro manipulation risks modifying their receptor expression and can introduce contaminants.

One interesting feature of MSCs is their capacity to migrate into injured tissues and participate in the regenerative process, interacting with the surrounding environment through secretion of numerous molecules such as growth factors, cytokines and chemokines [210]. Nevertheless, contrary to leukocyte migration and haematopoietic SC homing, the mechanisms that regulate MSC migration to injured sites are not well characterized [211]. In vitro, in a pro-inflammatory environment stimulated by TNF-α, MSCs migrate towards SDF-1, RANTES and MDC gradients, amongst others [212]. Furthermore, MMPs and their inhibitors have also been shown to enhance MSC migratory capacity [213]. In a recent study, MSCs were recruited in vitro by conditioned medium from IVDs cultured under degenerative-simulated conditions [214]. CCL5/RANTES has been identified as a key chemoattractant released by degenerative IVD in organ culture [215]. Moreover, CXCL12/SDF-1 delivery in IVD organ cultures promotes MSC recruitment towards NP, especially if MSCs were harvested from young donors [216]. This does not exclude the possibility that other cytokines involved in IVD degeneration pathogenesis, namely TNF-α and IL-1β, might play a role in regulation of MSC recruitment to the IVD [181]. A hypothetical migration route of high-proliferative cells lateral to the epiphyseal plate and the outer border region of the IVD was recently described [217]—if this route is confirmed, new strategies envisaging IVD regeneration may be attempted.

5. Future perspectives

Recent findings from in vitro studies, animal models and clinical trials have started to unveil the role of inflammation in IVD degeneration. However, no evidence for a beneficial role of inflammation in maintaining homeostasis has been presented, owing to the difficulty in studying IVD tissue homeostasis. In other tissues, such as bone [218,219] or cardiovascular tissue [220], the control of inflammation has already proven to be critical in shifting the degeneration/regeneration balance towards regeneration. In particular, our group has focused on modulating inflammation in bone [5,221–224]. Hence, we believe that novel therapies for DDD should aim at restoring the homeostatic inflammatory conditions in the disc, rather than totally inhibiting inflammation, thus enabling endogenous repair mechanisms to operate.

Our group has recently shown that incorporating fibrinogen, a well-known inflammatory protein, into a biomaterial leads to increased bone formation [5]. In vitro studies have shown that fibrinogen-modified biomaterial stimulates NK cell-mediated MSC recruitment without affecting the MSC differentiation marker alkaline phosphatase [222]. Moreover, a broad analysis of macrophage-secreted factors showed that fibrinogen modified macrophage response, leading to a downregulation of the expression of inflammatory cytokines and a stimulation in the production of growth factor [224]. Factors such as MIP-1δ, platelet-derived growth factor-BB, bone morphogenetic protein (BMP)-5 and BMP-7 were significantly promoted by fibrinogen [224], which may impact tissue regeneration.

Recent advances in development biology also highlight the crucial role of immune cells. An efficient nuclear reprogramming to obtain induced pluripotent stem (iPS) cells was shown to require activation of an innate response [225] and was achieved via activation of TLR3 in the work of Yamanaka and colleagues [226]. The importance of this finding in physiological situations remains unclear, but it is becoming increasingly evident that activation of the immune response, particularly the innate response, may contribute to regulation of stem cell behaviour [225,227]. Moreover, it has been known for a long time that the post-inflammatory wound repair process recapitulates basic phenomena that occur during embryogenesis [228]. Therefore, future studies should focus on trying to understand what happens early in development, to discover more cues on how to modulate inflammation in a disease situation.

Although diverse studies have presented data on inflammatory key players in IVD, the inherent variability and contradictions arising from the different in vitro culture conditions and animal models used in these studies may be hampering translation of the research to a clinical setting. The standardization of methods, the correlation of results with different IVD clinical problems, the use of alternative ex vivo models (based on organotypic cultures or bioreactors) and the use of more physiologically accurate in vivo models of IVD degeneration could bring further advances to the IVD research field.

For IVD regeneration therapies to succeed, it will be important to address IVD degeneration together with inflammation. Until now, most studies have focused on only one of these two aspects. An integrated strategy, which addresses both the synergistic interplay that exists between the multiple factors associated with IVD degeneration and balances the inflammatory response, could be a step closer to the success of IVD regenerative strategies and bring relief for those suffering from LBP.

Glossary

ADAMTS

A disintegrin and metalloproteinase with thrombospondin motifs

AF

Annulus fibrosus

BMP

Bone morphogenic protein

C/EBPβ

CCAAT/enhancer-binding protein beta

CCL

Chemokine ligand

CEP

Cartilaginous endplates

COX

Cyclooxygenase

CPPD

Calcium pyrophosphate dihydrate

CXCL9

Monokine induced by gamma interferon

DDD

Degenerative disc disease

DRG

Dorsal root ganglion

ECM

Extracellular matrix

FasL

Fas ligand

FGF

Fibroblast growth factor

fHA

Hyaluronic acid fragments

GAG

Glycosaminoglycan

GAP

Growth-associated protein

GDF-1

Growth and differentiation factor-1

GM-CSF

Granulocyte macrophage colony-stimulating factor

HA

Hyaluronic acid

HMGB1

High-mobility group protein box 1

HSC

Haematopoietic stem cells

ICAM-1

Intracellular adhesion molecule-1

ICE

IL-1β-converting enzyme

IFN-γ

Interferon-γ

IKKβ

3-Phosphoinositide-dependent protein kinase-1-mediated IκB kinase β

IL

Interleukin

IL-1Ra

Interleukin-1 receptor antagonist

iNOS

Inducible nitric oxide synthase

iPS

Induced pluripotent stem cells

IVD

Intervertebral disc

LBP

Low back pain

LFA

Lymphocyte function-associated antigen

LPS

Lipopolysaccharide

MAPK

Mitogen-activated protein kinases

MCP

Monocyte chemoattractant protein

MDC

Macrophage-derived chemokine

MIF

Macrophage migration inhibitory factor

MIP

Macrophage inflammatory protein

MMP

Metalloproteinase

mRNA

Messenger ribonucleic acid

MSC

Mesenchymal stromal cells

NALP3

NACHT, LRR and PYD domains-containing protein 3

NC

Notochordal cells

NF-κB

Nuclear factor kappa-light-chain-enhancer of activated B cells

NGF

Nerve growth factor

NLRP3

NOD-like receptor family, pyrin domain containing 3

NO

Nitric oxide

NP

Nucleus pulposus

PGE₂

Prostaglandin E2

PGF2α

Prostaglandin F2 alpha

PLA2

Calcium-dependent phospholipase A2

PLGA

Poly(lactic-co-glycolic acid)

PRP

Platelet-rich plasma

RANTES

Regulated on activation, normal T-cell expressed and secreted

Rhein

4,5-Dihydroxyanthraquinone-2-carboxylic acid

SC

Stem cells

Sox-9

Transcription factor Sox-9

TGF

Transforming growth factor

TGS-6

TNF-α stimulated gene/protein 6

TIMP

Tissue inhibitor of metalloproteinase

TLR

Toll-like receptor

TNF-α

Tumour necrosis factor alpha

TrkA

High affinity NGF receptor

tRAS

Tissue renin-angiotensin system