Article Text

Download PDFPDF

Biology of the blood–nerve barrier and its alteration in immune mediated neuropathies
  1. Takashi Kanda
  1. Correspondence to Dr T Kanda, Department of Neurology and Clinical Neuroscience, Yamaguchi University Graduate School of Medicine, 1-1-1 Minamikogushi, Ube, Yamaguchi 7558505, Japan; tkanda{at}


The blood–nerve barrier (BNB) is a dynamic and competent interface between the endoneurial microenvironment and the surrounding extracellular space or blood. It is localised at the innermost layer of the multilayered ensheathing perineurium and endoneurial microvessels, and is the key structure that controls the internal milieu of the peripheral nerve parenchyma. Since the endoneurial BNB is the point of entry for pathogenic T cells and various soluble factors, including cytokines, chemokines and immunoglobulins, understanding this structure is important to prevent and treat human immune mediated neuropathies such as Guillain–Barré syndrome, chronic inflammatory demyelinating polyneuropathy, POEMS (polyneuropathy, organomegaly, endocrinopathy, monoclonal protein and skin changes) syndrome and a subset of diabetic neuropathy. However, compared with the blood–brain barrier, only limited knowledge has been accumulated regarding the function, cell biology and clinical significance of the BNB. This review describes the basic structure and functions of the endoneurial BNB, provides an update of the biology of the cells comprising the BNB, and highlights the pathology and pathomechanisms of BNB breakdown in immune mediated neuropathies. The human immortalised cell lines of BNB origin established in our laboratory will facilitate the future development of BNB research. Potential therapeutic strategies for immune mediated neuropathies manipulating the BNB are also discussed.

  • Neuropathy
  • Neurobiology
  • Guillain-Barre Syndrome
  • Neuroimmunology
  • Peripheral Neuropathology

Statistics from


The presence of the blood–nerve barrier (BNB) restricts the movement of soluble mediators and leucocytes from the blood contents to the peripheral nervous system (PNS) parenchyma. As the endoneurial homeostasis protected by the BNB is a prerequisite for the proper function of the PNS,1 pathological breakdown of the BNB may be a key event that induces various peripheral neuropathies. However, compared with its CNS counterpart, the blood–brain barrier (BBB), only limited knowledge has been accumulated with regard to the function, cell biology and clinical significance of the BNB. The aims of this article are to review the recent progress in research about the cell biology of BNB composing cells, to discuss the importance of BNB breakdown in immune mediated neuropathies to better understand their pathogenesis and to provide an overview of the development of novel therapies against these currently intractable disorders.

Structure and function of the BNB

The BNB is located at the innermost layer of the perineurium and at the endoneurial microvasculature in the PNS.2 Although the perineurial barrier is also an important structure, this article focuses on the BNB at the endoneurial microvessels because the infiltration of mononuclear cells and leakage of soluble factors across this microvascular barrier is a key step in the development of immune mediated neuropathies, including Guillain–Barré syndrome, chronic inflammatory demyelinating polyneuropathy (CIDP), multifocal motor neuropathy (MMN), etc.

Endothelial cells composing small endoneurial vessels are normally non-fenestrated and contain few pinocytotic vesicles, and the adjacent endothelial cells are connected by complex and continuous tight junctions. These endothelial features are shared by those of the BBB and are considered to constitute the anatomical basis of the BNB that isolates the endoneurium from the intravascular component, inhibiting non-specific transcellular passage and paracellular diffusion of hydrophilic molecules. In addition, the endothelial cells forming the BNB express various receptors and transporters which remove toxic metabolites to maintain PNS homeostasis, and help to incorporate necessary compounds into the PNS parenchyma.3 Thus the BNB is not just a ‘barrier’ or ‘wall’ but a competent interface which actively exchanges materials between the endoneurial microenvironment and the surrounding extracellular space or blood. The term ‘blood–nerve interface’ has also been proposed for this feature.1

The structure of the BNB is essentially different from that of the BBB (figure 1). The highly differentiated endothelial cells composing both the BNB and BBB are completely surrounded by a basement membrane with embedded pericytes. However, in the BBB, the entire abluminal aspect of this endothelial cell/pericyte/basement membrane complex is further ensheathed by a unique structure called the glia limitans perivascularis, consisting of a basement membrane composed of laminins, which is distinct from that of the endothelial basement membrane,4 and by an astrocytic endfoot layer. The latter two structures, glia limitans perivascularis and astrocytic endfoot, are not present in the BNB.

Figure 1

Basic features of the blood–nerve barrier (BNB) and blood–brain barrier (BBB). Endothelial cells (yellow) of the BNB (left) and BBB (right) are connected by tight junctions (red rectangles) and embedded in a single basement membrane (pink) with surrounding pericytes (blue). In the BBB, the second basement membrane (light green; glia limitans perivascularis) wraps the whole endothelial cell/first basement membrane/pericyte complex, and the astrocytic endfoot layer (dark green) surrounds the outer surface. The empty space between the first and second basement membranes is called the perivascular space, which is observed in postcapillary venules and drains into the CSF, and there are occasional antigen presenting cells contained in this space.

As it was believed that astrocytes play a major role in BBB maintenance by regulating local water transport and by producing various growth factors and cytokines relevant for barrier maturation and maintenance, the BNB, lacking in astrocytes, has long been thought to be ‘leakier’ and ‘weaker’ than the BBB.5 However, Poduslo et al6 demonstrated that the BNB and BBB may be equally effective in excluding small and large molecules from the neural parenchyma in vivo. This means that the healthy PNS parenchyma is tightly protected by the BNB. Hence, similar to the CNS malfunction that arises due to BBB impairment, destruction of the BNB may impair endoneurial homeostasis and allow toxic substances and immunoglobulins to enter the endoneurium, leading to the onset and further worsening of immune mediated peripheral neuropathy.

Cellular biology of the BNB

To understand the biological basis of the BNB, immortalised BNB cell lines that possess in vivo characteristics are the most potent tool. However, compared with the recent developments in BBB research based on primary cultured and immortalised endothelial cells of BBB origin,7 ,8 in vitro studies of the BNB have fallen far behind. In the 1990s, two reports concerning the primary culture of rat9 and bovine10 endoneurial endothelial cells were published, and we conducted in vitro studies using bovine primary culture cells.11 ,12 However, bovine primary culture cells apparently have some limitations with regard to evaluating the molecular evolution of BNB composing cells, and we have moved forward to using cells from rodents and humans. Using rat or human sciatic nerve tissue as a starting material, we sequentially established an immortalised rat endothelial cell line (TR-BNBs3), followed by a rat pericyte cell line,13 a human endothelial cell line14 ,15 (figure 2) and a human pericyte cell line.16

Figure 2

A human immortalised endothelial cell line of endoneurial origin. A fluorescent micrograph of the DiI-Ac-LDL incorporating cells. Bar=100 μm.

The endothelial cells forming the BBB express many tight junction associated molecules, including occludin,17 claudin-5,18 claudin-12,18 ZO-1, ZO-2 and JAM-A.19 These molecules compose tight junctions and limit the paracellular permeability in order to maintain the brain microenvironment, thus leading to barrier properties. Our human and rat endothelial cell lines of BNB origin share these tight junction molecules. Compared with the rat and human immortalised BBB cell lines also established in our laboratory,8 ,20 most of the molecular (tight junction proteins, transporters, etc) and physiological (permeability measured by 14C-inulin and transendothelial electrical resistance) characteristics were similar, except that the BNB endothelial cells did not express efflux transporters OAT3 and Oatp2, which, respectively, carry homovanillic acid and dehydroepiandrosterone from the nervous system to the blood. The absence of OAT3 and Oatp2 in endothelial cells of BNB origin seems reasonable, and may reflect an important difference between the microenvironments of the CNS and PNS—namely, that the former has synapses while the latter does not. This finding also supports the concept that the BNB is a distinct structure that differs from the BBB at the cellular and molecular levels.

Pericytes are polygonal cells located at the periphery of the microvessel wall and wrap it with their processes (figure 1). Although having largely been ignored in the clinical literature for a long time, brain pericytes are now recognised as important cellular constituents of the BBB and actively communicate with other cells of the neurovascular unit, such as endothelial cells, astrocytes and neurones.21 Pericytes also exhibit phagocytic activity and may function as pluripotent stem cells, capable of forming neurones and glia.22 Hence peripheral nerve pericytes may take part in BNB maintenance, tissue repair and revascularisation after PNS injury. Another important piece of information confirmed by these immortalised BNB cell lines is that peripheral nerve pericytes maintain the physiological BNB function and enhance the expression of endothelial tight junction molecules through their secretion of various soluble factors, especially basic fibroblast growth factor.16 It can thus be concluded that in the BNB, peripheral nerve pericytes play the same role as astrocytes in the BBB. Peripheral nerve pericytes also produce several neurotrophic factors, such as nerve growth factor, brain derived neurotrophic factor and glial cell derived neurotrophic factor,16 which may facilitate axonal regeneration in peripheral neuropathy. These neurotrophic factors easily gain access to axons and Schwann cells because pericytes are located just behind the continuous endothelial cells and nothing interferes with the diffusion of neurotrophic factors released by pericytes. This means that early and complete recovery of BNB function in immune mediated neuropathy by immunomodulatory therapy could be beneficial in two aspects: in stopping the inflammatory cell/humoral factor intrusion into the PNS parenchyma and in facilitating axonal regeneration after the recovery in the milleu intérieur of the PNS, with the aid of neurotrophic factors secreted from healthy pericytes.

BNB alteration in immune mediated neuropathies

Because the healthy PNS is tightly sealed by the BNB, intrusion of pathogenic T cells, as well as humoral factors (including immunoglobulin), into the PNS parenchyma follows BNB impairment. In particular, microvessels in several important structures in the PNS, including the dorsal root ganglia and nearby spinal roots1 ,23 as well as neuromuscular junctions (NMJ),24 are devoid of barrier properties; hence these sites are believed to be especially vulnerable in inflammatory neuropathies25 ,26 because macromolecules such as immunoglobulins can easily penetrate into the PNS parenchyma through their leaky microvasculatures. However, in NMJ, antibodies against presynaptic axons were promptly incorporated into axons and underwent retrograde transport.27 This rapid clearance of autoantibodies, which has rarely occurred in nodes of Ranvier, may result in relative sparing of NMJ as the site of complement mediated injury in immune mediated neuropathies, including Guillain–Barré syndrome,28 despite the absence of BNB at NMJ.

Biomarkers that enable detection of BNB breakdown are not well established. Although elevated levels of soluble adhesion molecules,29 chemokines30 and matrix metalloproteinases31 in serum and CSF observed in patients with neuropathy may be indicative of T cell migration across the BNB, only pathological examination and MRI are reliable methods that can presently be used to assess BNB derangements in human immune mediated neuropathies. Enhancement of the spinal root, cauda equina and peripheral nerve trunk in T1 weighted MRI with gadolinium enhancement32 is occasionally observed in immune mediated neuropathies, and is interpreted as a hallmark of BNB breakdown. This finding should be carefully interpreted in the spinal root because the BNB is physiologically lacking in the dorsal root ganglia and nearby spinal roots, as mentioned above.

Pathology of BNB derangements

Morphological abnormalities of endothelial cells constituting the BNB have occasionally been reported in various neuropathies. As autopsy specimens are not adequate to evaluate the fine ultrastructural changes of the BNB, including endothelial fenestration and tight junction impairment, most of the reliable documentation has been based on nerve biopsy specimens. In immune mediated neuropathies, fenestration of endoneurial microvessels,33 ,34 gaps between adjacent endothelial cells35 and the disappearance of tight junctions have been reported in CIDP34 and macroglobulinaemic neuropathy.33 ,35 ,36 Monoclonal IgM in two35 ,36 out of these three cases showed antimyelin associated glycoprotein (MAG)/sulfoglucuronosyl paragloboside (SGPG) activity. As we previously reported that anti-MAG/SGPG IgM was toxic and induced permeability change in primary cultured bovine endothelial cells of BBB origin,37 the detrimental effect of anti-MAG/SGPG antibody against BNB is highly plausible. We also found that autoimmune demyelinating neuropathy patients with anti-glycosphingolipid (GSL) antibodies showed more severe BNB disruption than those without anti-GSL antibodies or non-autoimmune neuropathy patients.34 The most commonly observed change in endoneurial microvessels in anti-GSL antibody positive autoimmune demyelinative neuropathy patients was the finding of continuous spaces lacking tight junctions between endothelial cells. These morphological abnormalities may be the result of direct attacks by anti-GSL antibodies against GSL epitopes on the luminal surface of the endothelial cells forming the BNB, because primary cultured bovine endothelial cells of BNB origin are known to contain various GSLs, including GM1, GD1a and GD1b.10 The effect of anti-GSL antibodies11 and the sera from Guillain–Barré syndrome patients12 on opening the BNB was also confirmed by in vitro studies using bovine BNB derived endothelial cells.

These pathological changes of endoneurial microvessels are likely the result of alterations of the tight junction proteins in endothelial cells. We found downregulation of claudin-5 and altered localisation of ZO-1 in biopsied sural nerves obtained from CIDP patients.38 So far, that is the only report that has confirmed changes in tight junction proteins in immune mediated neuropathies using human materials. The changes may have been due to upregulation of proinflammatory cytokines and vascular endothelial growth factor (VEGF) in the sera of CIDP patients39 because the endothelial cells forming the BNB are the only cells in the PNS directly exposed to the serum contents. We recently confirmed downregulation of claudin-5 in a human BNB derived endothelial cell line by VEGF and proinflammatory cytokines,14 ,16 and demonstrated its upregulation after corticosteroid treatment.15

Multifocal motor neuropathy

T1 weighted MRI with gadolinium enhancement is the most reliable method to evaluate BNB breakdown in the peripheral nerve trunk where the tightness of the BNB is normally guaranteed. Kaji et al32 described the enhancement of the enlarged median nerve at the site of conduction block in one MMN patient and showed clustering of denuded or thinly myelinated axons around the endoneurial microvessel in another MMN patient. Although the ultrastructural alterations of endoneurial microvessels have not been described in detail, it may reflect focal BNB derangements in MMN. The presence of anti-GM1 autoantibodies in the sera of MMN patients may be related because GM1 is present in endothelial cells forming the BNB,10 and anti-GM1 monoclonal antibodies can open the BNB without the need for complement components.11 One of the most intriguing findings in this disorder is the presence of multifocal persistent conduction block which is not related to the common compression sites; however, no clear explanation of this multifocality has been provided so far. We speculate that putative humoral factor(s) present in patients’ sera firstly reacts with the endothelial cell composing BNB, followed by a vicious cycle of local cytokine release, upregulation of adhesion molecules and T cell recruitment, which may result in persistent conduction block at the site of broken BNB. Screening of MMN sera for compounds that lead to BNB breakdown is now underway in our laboratory.

POEMS (Crow–Fukase) syndrome

In the context of microvascular changes in immune mediated neuropathies, POEMS (polyneuropathy, organomegaly, endocrinopathy, monoclonal protein and skin changes) syndrome cannot be ignored. In POEMS syndrome, hyperplasia of endoneurial microvessels40 and endoneurial oedema41 have frequently been reported. These pathological alterations may be closely related to the high concentration of serum VEGF, a potent multifunctional cytokine that induces angiogenesis and microvascular hyperpermeability.11 Second, the microangiopathy caused by the above VEGF effect reduces the oxygen supply, which leads to the robust expression of HIF-1α by all of the constituents of the nerve, with a secondary increase in local VEGF expression causing a self-perpetuating VEGF mediated toxic gain of function.42 Serum VEGF induced BNB dysfunction may occur throughout the whole PNS, not necessarily confined to the sites of weak BNB: this may explain the neurophysiological findings that the nerve trunk is predominantly affected in POEMS syndrome.43 ,44 Scarlato et al also showed a ‘gap’ between adjacent endoneurial endothelial cells, suggesting BNB impairment in one case, but this change is different from the change that we previously reported in inflammatory neuropathies,34 and pathological alterations directly indicating BNB destruction (loss of tight junctions, gaps between adjacent endothelial cells, etc) are relatively rare in biopsied sural nerve specimens obtained from patients with POEMS syndrome (unpublished observations).

Diabetic neuropathy

While diabetic neuropathy is not usually categorised as an immune mediated neuropathy, microvascular changes are one of the pathological hallmarks of this disorder, and autoimmune abnormalities are evident in at least some patients with diabetic neuropathy. Microvascular abnormalities in diabetic neuropathy patients have long been reported, including loss of tight junctions,45 hypertrophy of the microvascular basement membrane46 and loss of microvascular pericytes.46 Among these changes, basement membrane hyperplasia around endoneurial microvessels possessing BNB function is the most conspicuous pathological abnormality in diabetic patients. This may cause BNB derangement as well as hypoxia in the endoneurial space, which result in further worsening of peripheral neuropathy; however, the pathogenesis remains unclear. Using our human in vitro BNB model, we recently found that advanced glycation end products induce basement membrane hypertrophy and BNB disruption by increasing the autocrine secretion of VEGF, as well as by augmenting transforming growth factor β signalling from microvascular pericytes under diabetic conditions.14

Future strategies to manipulate the BNB

The basic structure of the BNB is less complicated than that of the BBB because it lacks the glia limitans perivascularis and astrocytic endfoot. This has two major implications from a therapeutic point of view: pathogenic T cells can enter the PNS parenchyma more easily than the CNS because there is no acellular barrier (glia limitans) present to prevent non-activated T cell intrusion. As the glia limitans contains laminin isoforms which are different from those of the endothelial basement membrane,4 ,47 immature encephalitogenic T cells cannot bind to the glia limitans and thus cannot enter the CNS parenchyma. To breach this acellular barrier provided by the glia limitans, T cells need to re-encounter a cognate antigen in the context of major histocompatibility complex class II bearing antigen presenting cells in the perivascular space (the space between the endothelial basement membrane and the glia limitans) and acquire matrix metalloproteinase -2 and -9 activity.48 To enter the PNS parenchyma, T cells do not need to undergo this complex process, so blockade of T cell invasion just at the endothelial monolayer may be more important in the treatment of immune mediated neuropathies such as CIDP.

Recruitment of circulating T cells into the tissue parenchyma has been shown to begin by the sequential interaction of different adhesion and/or signalling molecules on T cells and endothelial cells in a multistep cascade.49 These sequential steps include rolling, adhesion, crawling and diapedesis of lymphocytes, and each process has its own molecular pathway. The same cascades may be relevant in the T cell/endothelial cell interaction in the BNB although there is no direct evidence that these processes are identical in the BNB and BBB or other organs.

Natalizumab, a humanised monoclonal antibody against the α4 integrin which strongly interferes with the ‘adhesion’ process, is now widely used for the treatment of relapsing forms of multiple sclerosis. This drug is theoretically also effective against immune mediated neuropathies, including CIDP,50 but a therapeutic trial in 61-year-old intractable CIDP failed.51 This result does not necessarily indicate that natalizumab will not be effective against immune mediated neuropathies, and further controlled studies are necessary. We recently found that corticosteroids, the firstline treatment for CIDP, enhance BNB function via upregulation of claudin-5.15 Further screening of candidate drugs that can upregulate claudin-5 and augment BNB integrity using our in vitro human BNB model are now underway.

The second implication of this simpler structure is that only pericytes should be considered as the influential cell population affecting endothelial function. Endothelial cells and pericytes are embedded together in one basement membrane (endothelial basement membrane) and thus these two cells are in close proximity so the paracrine secretion of growth factors and cytokines from nearby biologically active pericytes may strongly influence endothelial function. These factors may be key to maintaining BNB function.13 ,16 The paracrine secretion of transforming growth factor β from endoneurial pericytes disrupts the BNB and causes basement membrane hypertrophy around the endoneurial microvessels in diabetic neuropathy.14 A possible method to manipulate the BNB for therapeutic purposes is to modify endothelial function using oligonucleotides, siRNAs and virus vectors. Because endothelial cells forming the BNB are the only cells that come into direct contact with the blood constituents in the PNS, endothelial cells can be easily manipulated via the system circulation. Another possible method is to modify BNB pericytes: small hydrophobic substances that can reach the pericyte membrane through the endothelial monolayer and strengthen pericytic activity, including release of various cytokines/chemokines that influence endothelial function, may also be useful as drug candidates to control BNB function. The latter strategy may be more promising because indirect manipulation of endothelial cells, the principal regulators of BNB integrity, via humoral factors from pericytes would be more physiologically relevant than their direct modification.



  • Funding None.

  • Competing interests None.

  • Ethics approval The study was approved by the ethics committee of Yamaguchi University Graduate School of Medicine.

  • Provenance and peer review Commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.