A new animal model of spontaneous autoimmune peripheral polyneuropathy: implications for Guillain-Barré syndrome

L31 mice with constitutive expression of co-stimulatory B7.2/CD86 were generated with MHC I promoter and Igμ enhancer [14]. L31/CD4^-/- mice have been described previously [13] and were maintained by successive backcross to C57BL/6 mice for more than 7 generations. C57BL/6 mice bred and housed in the same facility were used as controls. All procedures were in accordance with the guidelines of the Canadian Council on Animal Care, and approved by the animal care committee of McGill University. In total, 49 L31/CD4^-/- and 18 C57BL/6 mice (controls), aged at 2-6 months, were included in the study. 29 L31/CD4^-/- symptomatic mice and 6 C57BL/6 mice were monitored for their clinical scores and behavioral responses. 20 L31/CD4^-/- mice were euthanized before disease onset, as pre-symptomatic mice for tissue collection.

All instruments and reagents have been used in the study were listed in Additional file 1: Table S1.

All mice included in the current study had their disease onset between 2-4 months after birth. Animals whose clinical score ≥4 were excluded from sensory tests.

Clinical scores for motor deficits: 0 = normal; 1 = reduced tonus of tail and/or limp tail; 2 = weakness of one hind limb, staying in clasping or outstretching position when lifted by the tail; 3 = weakness of two hind limbs, both staying in clasping or outstretching position when lifted by the tail; 4 = paresis or splaying of one hind limb; 5 = paresis or splaying of two hind limbs; 6 = moribund or death. L31/CD4^-/- transgenic mice with clinical score at 0 were defined as pre-symptomatic mice.

von Frey Test was performed to test paw sensitivity to mechanical stimuli. Calibrated monofilaments were applied to the plantar surface of the hindpaw and the 50% threshold to withdraw was calculated as previously described [15]. A decrease in threshold suggests the development of mechanical allodynia.

Regular immunofluorescent staining was performed as previously described [16]. Tissues were incubated with primary antibodies overnight at 4°C, followed by fluorochrome-conjugated secondary antibodies for 60 min at room temperature, and then counterstained with marker for nucleus 4′, 6-Diamidino-2-phenylindole dihydrochloride (DAPI) and/or FluoroMyelin. Details of all antibodies used in the study were listed in Table 1. Images were acquired using an Olympus BX51 microscope equipped with a color digital camera or Olympus confocal laser-scanning microscope.

Single-cell suspensions and blocking were prepared as described previously [16]. Samples were then stained with specific fluorochrome-conjugated antibodies for 25 min at 4°C. Staining specificity was identified by omitting antibodies, and correlation of spectral overlap was done by using negative and positive compensation beads. Cellular events were acquired using a LSR Fortessa flow cytometer and data was analyzed using Flow Jo software. Detail information of antibodies used here was listed in Table 1.

Blood nerve barrier (BNB), blood spinal cord barrier (BSCB) and blood brain barrier (BBB) permeability was assessed using a micromolecular tracer sodium fluorescein (NaFlu). The protocol was adopted and modified from previous studies [17]. Briefly, NaFlu was administrated intravenously (10%; 2 ml/kg) and allowed to circulate for 30 min. Then, mice were transcardially perfused with cold saline for 10 min to remove intravascular NaFlu. The choroid plexuse, meninge and epineurium were removed before nervous tissues were dissected into different regions: brain, cervical and lumbar spinal cords, and sciatic nerves. Following tissue homogenization, the concentration of NaFlu in supernatant was measured with spectrophotofluorometer. NaFlu was expressed as g/g of tissue.

B7.2 expression was verified by flow cytometry on macrophages and microglia, isolated from sciatic nerves and lumbar spinal cords (Additional file 2: Figure S1). Increased expression of B7.2 was detected in macrophages of L31/CD4^-/- mouse sciatic nerves (Figure 1A-B), which was derived most likely from B7.2 transgene, as seen in spinal microglia, resident macrophages in the CNS [13]. It appears that B7.2 transgene primed macrophages by increasing macrophage cell density (Figure 1A). Immunohistochemistry analysis confirmed that L31/CD4^-/- mice had more Iba-1⁺ macrophages in sciatic nerves, and these were elongated resident macrophages [18]. Many of them were double-labelled by Iba-1 and B7.2 (Figure 1B). Similar changes were detected in the spinal cords (Figure 1C-D). Pre-symptomatic L31/CD4^-/- mice had more CD11b⁺ CD45⁺ microglia than wild type mice (Figure 1C). Many of them expressed B7.2, while the expression of B7.2 was almost undetectable in wild type mice (Figure 1C-D).

Almost all L31/CD4^-/- mice developed motor deficits (Additional file 3: Video 1 and 2, Figure 2A) within 4 months after birth. Majority (55%) demonstrated hind limb weakness (score 2-3), while walking was not altered significantly. However, 14% mice had difficulties walking properly. One or two hind limbs were found with paresis, in a splaying position (score 4-5). Whatever the severity of the illness, the symptoms appeared suddenly, reaching a maximum within 3-7 days. Most animals, after the appearance of neurological signs, remained stable. Among 29 L31/CD4^-/- symptomatic mice monitored for behavior, 4 died of illness (14%). No apparent recovery was found within the observation period (maximum 2 months following disease onset).

Poor coordination was observed in both pre-symptomatic (clinical score = 0, ≥ 2 months old) and symptomatic L31/CD4^-/- mice (clinical score ≥1). The latency of fall in Rotarod test dropped from 98.60 ± 12.01 sec in wild type mice to 45.50 ± 7.27 sec in pre-symptomatic mice. Symptomatic mice could only stand on the running rod for 12.5 ± 2.72 sec (Figure 2B).

L31/CD4^-/- mice also developed sensory disorders. Symptomatic animals exhibited numbness to von Frey hair as paw withdrawal thresholds reached at 3.48 ± 0.15 g, while wild type and pre-symptomatic mice responded at 0.78 ± 0.07 g and 0.83 ± 0.21 g, respectively (Figure 2C). However, they were hypersensitive to cold and heat stimulation. Duration of withdrawal in acetone test was increased (Figure 2D) and the hot plate latency was decreased (Figure 2E) in symptomatic mice. Interestingly, thermal hypersensitivity appeared prior to motor dysfunctions, since alteration in response to cold and heat stimulation was observed in pre-symptomatic mice (Figure 2D-E).

Immunohistochemistry studies demonstrated that myelin (Figure 3B, F, K) and axonal structures (Figure 3F, K) in pre-symptomatic mice remained in healthy condition. Same as in wild type mice (Figure 3A, E, J), myelin sheaths in pre-symptomatic mice appeared as wide straight lines in longitudinal sections (Figure 3B) and wrapped around PGP9.5⁺ axons in longitudinal (Figure 3F) and cross sections (Figure 3K). However, focal (Figure 3C) or severe, diffused (Figure 3D) myelin loss was detected in symptomatic L31 CD4^-/- mice. Destruction of the myelin in symptomatic mice was most frequently accompanied by axonal damage where the number of PGP9.5⁺ axonal fibers was significantly reduced (Figure 3G, H, L). Nevertheless, some naked PGP9.5⁺ axonal fibers were found in the area where myelin disappeared or fragmented (Figure 3G, H, M, N), indicating demyelination might occur, at least in some area, prior to the axonal damage. Image with high magnification in Figure 3I revealed a cell-shaped myelin labeling with a DAPI⁺ nucleus inside, suggesting presumably it could be a macrophage engulfing myelin debris, while two nude axons were apposed closely. The detection of ATF3, a marker for neuronal degeneration, in DRG sensory neurons (Figure 3O), as well as in spinal cord motor neurons (Figure 3P) suggests that both sensory and motor neurons were affected.

In parallel to the demyelination and axonal damage, there was a massive infiltration of immune cells into the nerves (Figure 4). They consisted of mainly round-shaped Iba-1⁺ macrophages (Figure 4A) and CD8⁺ T cells (Figure 4B). Apart from a slight increase of elongated resident macrophages (Figure 4A, insert in the middle panel), no round-shaped macrophages or CD8⁺ T cells were found in the sciatic nerves of pre-symptomatic mice (Figure 4A-B). Immune cell infiltration was confirmed by quantitative flow cytometry analysis (Figure 4C, Additional file 4: Figure S2). In symptomatic mice (Figure 4D), infiltrates were either grouped in small foci or distributed diffusely. In the vicinity of cellular clusters, myelin sheaths were disintegrated or replaced by oval-shaped vacuoles. Fragmented myelin was found within Iba-1⁺macrophages (yellow signals within Iba-1⁺ cells). There were less infiltrates in areas where myelin remained in healthy-shape (Figure 4D-left). Majority of infiltrated Iba-1⁺ macrophages had high levels of B7.2 expression (Figure 4D-middle) and CD8⁺ T cells were in close apposition with Iba-1⁺ macrophages (Figure 4D-right).

Myelin structures in the spinal cords of pre-symptomatic mice remained intact. Very little spotted demyelinating lesions could be found only in severely diseased animals. Even then, they were restricted to the dorsal columns and the areas surrounding ascending spinothalamic tracts (Figure 5A, D and E - pink zone). Clusters of infiltrated CD8⁺ T cells were seen within demyelinated spots or damaged motor neurons (Figure 5A). A few isolated CD8⁺ T cells were also detected in pre-symptomatic mice where myelin architecture was not compromised (Figure 5A). Robust spinal microglia activation was detected in symptomatic mice, predominantly in the grey matter (Figure 5B and 5E-green zone). Iba-1⁺ microglia were clustered at the dorsal horns where central afferents of sensory neurons terminate and in the ventral horns surrounding motor neurons (Figure 5B). The topography and morphological changes are exactly the same as the spinal microglial activation triggered by a physical injury to the sciatic nerve [19], which suggests that microglial activation in the grey matter of symptomatic mice could be a response secondary to the PNS damage. However, in the white matter, activated microglia were found restricted to the zone where myelin was destroyed (Figure 5C, arrows). The increase of CD11b⁺CD45⁺ microglia and CD8⁺ T cells were quantified by flow cytometry (Figure 5D, Additional file 5: Figure S3). Either in grey matter or white matter, activated microglia had high levels of B7.2 expression as many B7.2⁺ cells were colocalized with ramified Iba-1⁺ cells. Due to the incompatibility of the antibodies, CD8 and B7.2 antibodies could not be used on the same tissue section, however presumably round shaped B7.2 single labelled cells were infiltrated CD8⁺ T cells (Figure 5E).

The homeostasis of the nervous tissue is protected by neural barriers, from periphery to central: BBB, BSCB and BNB. As the integrity of the barriers is crucial for immune cell recruitment, we examined the functional status of the barriers with a fluorescent tracer, NaFlu. Compared with intact sciatic nerves from wild type mice, the content of NaFlu in diseased sciatic nerves was dramatically increased (Figure 6A-B), suggesting a disruption of the BNB. Interestingly, the levels of NaFlu in the brain and the spinal cord of symptomatic mice remained similar to those of wild type mice (Figure 6A), further confirming the pathology was primarily in the PNS.

If the BNB restricts the entrance of circulating immune cells into the nerves, then vessels at 3 places of the PNS, including spinal roots, DRG and nerve terminals, are physiologically permeable to various components in the circulation, where most immune cells could definitely have direct access to the endoneurium. While there were some resident Iba-1⁺ macrophages and no CD8⁺ T cells in the DRG and the roots of wild type mice (Figure 7A), immune infiltrates were apparent in these neural structures of pre-symptomatic animals (Figure 7B), which was not the case in sciatic nerves, nor in spinal cords. It was possible to detect the slightest degree of infiltration, presumably the earliest pathologic changes (Figure 7B, upper panels) and also significant amount of both macrophages and CD8⁺ T cells with high levels of B7.2 expression (Figure 7B, bottom panels) suggesting a status close to disease onset. Not surprisingly, the DRG and roots of symptomatic mice were virtually flooded with both macrophages and CD8⁺ T cells, where DRG sensory neurons were barely observable (Figure 7C).

The most common initial symptoms of GBS are numbness, paresthesia, weakness and pain in the limbs or some combination of these symptoms. The main feature is progressive bilateral and relatively symmetric weakness of the limbs, and the weakness progresses over a period of 12 hours to 28 days before a plateau is reached [20]. Our results demonstrated that L31/CD4^-/- mice replicated many clinical aspects of this complex human autoimmune disease, including both sensory and motor deficits, and a temporal course of an acute onset and a stable progression [21‐23]. L31/CD4^-/- mice also imitated pathological changes in GBS, with characteristic lymphocytes and macrophages-associated demyelination and axonal damage at the peripheral nerves, suggesting these mice might represent more closely axonal variants of GBS. Although it remains to be determined what are the specific antigens (myelin or axon or both) for autoimmune response, demyelination in peripheral nerves of L31/CD4^-/- mice is still distinguishable with myelin loss secondary to Wallerian degeneration by the massive infiltration of T cells which was initiated prior to the damage of the nervous tissues. Comparable to what we observed restricted damage in spinal cords, loss of myelinated fibers in posterior tracts or inflammatory infiltrates/spinal microglia activation in the grey matter have been reported in human GBS [24, 25]. Similar neurological deficits and pathological changes in peripheral nerves were also detected in L31 mice (data not shown). The L31/CD4^-/- mouse model had many similarities with the EAN [3, 26], one of the essential tools in the research of autoimmune demyelinating diseases of the PNS. However the use of EAN was restricted due to the administration of powerful immune adjuvants to break-down self-tolerance. The field lacks of spontaneous and translatable models for mechanistic studies. The impressive resemblance between the L31/CD4^-/- mouse model and human GBS implies it as an unequivocal experimental setting for spontaneous autoimmune peripheral polyneuropathy, especially as it allows taking advantage of the wealth of genetic resources on the C57BL/6 background.

Co-stimulatory signals play a key role in regulating T cell activation and are believed to have a decisive influence in the induction and penetration of cellular effecter mechanisms in spontaneous autoimmune peripheral polyneuropathy. Sensitization of APCs by increasing the expression of B7.2 through genetic approaches simulates certain common pre-dispositive conditions in GBS, e.g., infection, which is usually the most potent stimulus for B7 expression [27, 28]. T cells are required in the pathogenic sequence of immune-mediated nerve damage [29], but the relative contributions of CD4⁺ and CD8⁺ T cells are controversial. Although most inflammatory infiltrates in the PNS of rats with EAN are CD4⁺ cells [30], further studies indicated that CD8⁺ cells may also be involved in the pathogenesis [31]. The necessity of both CD4⁺ and CD8⁺ cells in the genesis of EAN was confirmed by using CD4 and CD8 knock-out mice, where the severity of clinical scores and histopathological manisfestations of P0 peptide induced EAN was significantly lower than those in their wild type counterparts [32]. In human GBS, very limited post mortem and biopsy data is available. However, one autopsy study provided clear evidence for a potential involvement of cytotoxic T cells in demyelination. Wanschitz et al [33] reported that in all 11 examined subjects, among infiltrated endoneurial T cells, the ratio of CD8⁺ to CD3⁺ was around 0.7 to 1.4. The density of CD8/CD3 double positive T cells increased significantly in cases with more than 4 weeks duration of illness. Many of these CD8⁺ cells were found granzyme B positive; Schwann cells and myelin sheaths were detected with upregulation of MHC class I molecules [33]. In coincidence with the observation in GBS patients, our previous and current studies with L31 and L31/CD4^-/- mice demonstrated that 1) altering co-stimulation signals primed APCs in the nervous tissue mice and drove a predominately CD8⁺ T cell mediated autoimmune reaction [12], 2) CD8⁺ cells are able to lead to demyelination in the peripheral nerves if they can find co-stimulation in the PNS; 3) The demyelinating disease can be initiated in the absence of CD4⁺ T cells, even displaying accelerated disease development and increased expressivity [13], which indicates an immunoregulatory role for CD4⁺ T cells in disease pathogenesis. Hence, L31 and L31/CD4^-/- mice are unique to understand molecular pathways by which CD8⁺ T cells are involved in demyelination and axonal damage. Albeit, a characterization of functional phenotypes of CD8⁺ cells in the demyelinating nerves at different stages of the disease is essential, which will provide insights to understand the roles of CD8⁺ cells in the pathogenesis. Furthermore, crucial questions with regard to the decisive contribution of macrophages, as well as interaction between macrophages and lymphocytes in the context of demyelination and axonal loss could be explored in L31 and L31/CD4^-/- mice. However, whether and to what extent antibody mediated humoral immunity is involved in the disease genesis of L31 and L31/CD4^-/- mice need to be determined in future experiments.

Most patients with GBS recover spontaneously, 62% had made a complete or almost complete recovery at one year [34]. However, neurological disorders in either L31 [12] or L31/CD4^-/- mice, with a sudden onset, seem to persist and remain stable. Whether severe myelin/axonal damage and absence of overt recovery in L31/CD4^-/- and L31 mice could be attributed to the lack of CD4⁺ T cell population in the symptomatic nerves is an intriguing topic for further investigation. Or it might be due to the fact that this model reflects more closely to axonal variants of GBS, since most AMAN patients have more delayed recovery than AIDP [35] and ASMAN has been associated with a more severe course and poorer prognosis [36].

It remains presently elusive how the cascade of autoimmune responses targeting PNS structures is ignited. The fact that infiltration of immune cells was initiated before disease onset, in the roots and the DRG where the BNB is virtually absent [37] led us to propose the following cascade: 1) Transgene or infection derived- over-expression of co-stimulator B7.2 sensitizes APCs in the nervous tissue; 2) T cells first access the nervous tissue during immune surveillance through the DRGs and the roots where the endoneurial vasculature is basically permeable to various components in the blood stream; 3) T cells, especially CD8⁺ T cells that have specificities for PNS self-antigens presented by primed macrophages are engaged in productive antigen recognition and are reactivated due to the high levels of B7.2; 4) In aid with massive infiltration of macrophages, CD8⁺ T cells exert their cytotoxic effector functions, leading to destructive autoimmunity; 5) CD4⁺ T cells, could in fact, exert immunoregulatory functions during the disease course.