All studies were performed according to institutional guidelines and animals were handled according to protocols that were approved by the Animal Welfare Committee at the University of Texas Health Science Center at Houston (Protocol number: HSC-AWC-11-046) and that are in accordance with Federal guidelines. The studies using human autopsied nerve samples were approved by the Committee for the Protection of Human Subjects at the University of Texas Health Science Center at Houston (Approval number: HSC-GEN-08-0233) and it qualifies for exempt status (category#4) according to 45 CFR 46.101(b). {CATEGORY #4 : Research, involving the collection or study of existing data, documents, records, pathological specimens, or diagnostic specimens, if these sources are publicly available or if the information is recorded by the investigator in such a manner that subjects cannot be identified directly or through identifiers linked to the subjects.}

Mutant and transgenic mice used in nerve injury models (described below), are listed in Table 1. Osteopetrotic (op/op), C3-null, Fcer1g-null, Fcer1α-null, Fcgr2b-null, Fcgr3-null were from Jackson Laboratories. Two knockout mice lines, B4galnt1-null and St8sia1-null, with altered ganglioside expression in the nervous system were bred in house. Wild type control mice (WT mice) used in the studies were littermates (such as the controls for Fcgr4-null, op/op mice, and B4galnt1-null, etc.) or the corresponding background matched control animals suggested and provided by the vendor (such as the controls for Fcer1g-null, Fcer1α-null, Fcgr2b-null, and Fcgr3-null, etc.).

Fcgr4-null mice lack activating FcγRIV only but express all other FcγRs. Heterozygous Fcgr4-null mice were received from the European Conditional Mouse Mutagenesis Program consortia distributed by the Helmholtz Zentrum München, Munich, Germany. The original ES cell containing the targeting construct was developed by the international mouse knockout consortium resulting in a non-conditional deletion of the Fcgr4 gene and were initially supplied on the mixed C57BL/6;129S5/SvEvBrd background [19]. Deletion was confirmed by genomic PCR before crossing back onto the C57BL/6 background for at least 10 generations and inter-crossing to generate homozygous knock-out mice. Deletion of FcγRIV at the protein level was confirmed in the blood by flow cytometry (assessing CD11b positive monocytes) and in the spleen, liver and lymph node by immunohistochemistry (Glennie et al. unpublished observations).

Three previously described IgG monoclonal antibodies (mAbs) against gangliosides GD1a and/or GT1b {GD1a/GT1b-2b, GD1a-1 (E6 clone), and GD1a-2b} were used in our study [20], [21].These mAbs are designated by their ganglioside specificity and IgG isotype; e.g., GD1a/GT1b-2b refers to a mAb with GD1a and GT1b specificity and IgG2b isotype. Control mouse IgGs were used as sham Abs.

Nerve crush provides a convenient and well characterized system to study the regeneration of injured axons mimicking the regenerative response of transected/degenerating axons in GBS. Briefly, in 8 to 12-week-old age male animals sciatic nerve was crushed at mid-thigh level on one side, as described [14]. Animals were administrated various doses of specific anti-glycan mAbs or control Abs by intra-peritoneal (i.p.) route (1 mg on days 0, 3, 7, 10, and 14 after crush). Behavior test (pinprick test; twice per week) and electrophysiology studies (on day 16 post surgery) were conducted on all animals. Studies were terminated on day 17 after the crush. All animals were perfused, sciatic and tibial nerves were harvested and post-fixed in a mixture of 3% glutaraldehyde and 4% paraformaldehyde.

To evaluate sensory functional recovery, pinprick tests were conducted on animals with sciatic nerve injury. Pin prick tests were performed 1 day prior to the surgery and on indicated days post nerve crush, as described [22]. Briefly, the lateral part of the plantar surface of the hind paw, where sciatic nerve branches sense, was divided into 5 areas. A needle was gently applied to each of those areas. It was scored 1 for this area if the animal quickly lifts, licks, or shakes its paw after the needle prick. No response was considered as 0. All testing was performed blindly.

Sciatic nerve conduction studies were performed on all animals, as described [14]. Briefly, sciatic nerves were stimulated at sciatic notch and compound muscle action potential amplitudes were recorded in the hind-paw on day 16 post nerve crush.

Sciatic nerve (∼10 mm distal to the crush site) and tibial nerve segments (∼20 mm distal to the crush site) or grafted nerve segment were embedded in Epon, as described [23]. 1µm- toluidine blue stained sections were obtained and one entire section of the whole nerve segment/animal was used for quantification, as described [15], [24]. All myelinated axons in a single whole cross section of the nerve were counted at light level (40X) by using a motorized stage and stereotactic imaging software (Axiovision; Zeiss), as described [14], [24]. This morphometric avoids random sampling and includes all regenerating nerve fibers. 12 nerves/group were used for each individual nerve crush experiment and 6 nerves/group for each nerve grafting experiment (also notated in the respective figure legends).

The nerve graft studies were performed on age and gender matched WT control and Fcer1g-null mice. The sciatic nerve grafts (∼2 cm long) collected from WT or Fcer1g-null mice were implanted to the proximal stumps of transected sciatic nerves of host animals (WT or Fcer1g-null mice) with 10-0 prolene sutures. Different groups include: (1) WT nerve grafts in WT hosts; (2) Fcer1g-null nerve grafts in WT hosts; (3) WT nerve grafts in Fcer1g-null hosts; (4) Fcer1g-null nerve grafts in Fcer1g-null hosts. After the surgery, all animals were administered 5 doses of GD1a/GT1b-2b mAb (1mg, i.p.) The grafts were harvested 3 weeks after the surgery and middle segment of each graft was processed and embedded in Epon and morphometry was performed on 1-µm cross sections.

The tissue lysates were extracted from intact and injured mouse nerves at different time points after nerve crush, or human nerves from controls and GBS patients. Protein concentration was determined and 2 µg of total protein/lane were electrophoresed, transferred to PVDF membranes, and probed with anti-Fcγ common chain [25] (US Biological) and anti-α-Tubulin or anti-β-actin (loading/internal control) (Cell Signaling) Abs.

Injured or intact mouse sciatic nerves, or human nerves from controls and GBS patients were fixed and cryoprotected. Cryosections were double labeled for Fcγ common chain (shared by all activating FcγRs) and CD68 (Macrophage marker; AbD serotech), or glial fibrillary acidic protein (GFAP) (Schwann cell marker; Santa Cruz Biotechnology), or Iba1 (microglial cell marker; Abcam) with specific Abs. The sections were then developed with specific fluorescently conjugated secondary Abs and examined by fluorescent microscopy (Zeiss).

Human cauda equina, were obtained from autopsies of GBS patients (various intervals after onset of GBS, i.e., acute and post-acute phases) or patients without neuropathic disorders (controls). The human tissues were then post-fixed in 4% paraformaldehyde for 24–48 hours. Fixed tissues were cryoprotected and cryosectioned, 10–15 µm cross sections were air-dried on glass slides for immunocytochemistry studies.

All primer sets were from Life Technologies: 18s, Hs99999901_s1; FcγRI, Mm00438874_m1; FcγRII, Mm00438880_g1; FcγRIII, Mm00438883_m1; FcγRIV, Mm00519988_m1. Total RNA was extracted from injured or intact nerve tissues, according to manufacturer’s instruction (Invitrogen). cDNA was generated through reverse transcription from 0.2 µg of each RNA sample by using High Capacity RNA-to-cDNA master mix (Life Technologies). Real-time PCR was performed on ABI Step-One Plus (Applied Biosystems) using TaqMan Fast Advanced Master Mix (Life Technologies). 18s was used as normalization control. All PCR were conducted in triplicate and repeated at least three times.

Serum samples from WT and Mutant op/op mice collected at various intervals after administration of anti-glycan mAbs were used for ELISA, as described [20].

All numerical data are presented as mean ± s.e.m. Differences between groups were determined using Student’s t test or ANOVA with corrections for multiple comparisons, p values < 0.05 were considered statistically significant.

We examined the inhibitory effects of three GD1a-reactive mAbs in the nerve crush model because some patients with GBS and anti-GD1a Abs have poor recovery [4], [6], [9]. These mAbs were tested in WT, B4galnt1-null (lack all complex gangliosides), and St8sia1-null (lack b-series but overexpress a-series gangliosides including GD1a) (Figure 1). Our results showed that GD1a/GT1b-2b mAb induce significant inhibition of axon regeneration in WT and St8sia1-null animals but not in B4galnt1-null mice that did not express corresponding glycan antigens (Table 2), as reported previously [14]. Two anti-GD1a mAbs did not inhibit axon regeneration in WT or B4galnt1-null, whereas these mAbs induced severe inhibition in St8sia1-null mice (Table 2). These findings were consistent with our previous studies showing that higher GD1a density present in St8sia1-null animals {building up behind the biosynthetic block (Figure 1)} was necessary to induce anti-GD1a Ab-mediated axonal injury in mice [26]. These results emphasize the importance of immune complex formation in Ab-mediated inhibition of axon regeneration.

First, we examined the expression of FcγRs in mouse nerve crush model and compared it to GBS tissues. Quantitative PCR studies showed highly significant upregulation (range 3–20 fold) of different FcγRs (activating and inhibitory) at mRNA level in injured nerves compared to uninjured control nerves. mRNA of FcγRII and FcγRIII had comparatively higher expression levels throughout the study period compared to other FcγRs (Figure 2A). Upregulation of activating FcγRs was confirmed at protein level by ICC and immunoblotting on uninjured and injured nerve segments. FcγRs were undetectable in uninjured nerves and their expression was significantly upregulated in injured nerves (Figures 2B and 2C). There are two macrophage populations, namely resident and hematogenous monocyte-derived macrophages, which respond and participate in degenerative and regenerative responses after mammalian neural injury [27]. Reliable immunologic and phenotypic markers that can differentiate between the two macrophage populations are not readily available. However, resident macrophages such as microglia usually act as an early responder to the nerve injury, prior to the massive influx of hematogenous macrophages [28] and temporal profiling allows distinguishing these populations. The ICC studies showed that activating FcγRs were expressed by resident (Schwann and microglial cells) and recruited (macrophages) glia in injured nerves (Figure 2D). The upregulation of activating FcγRs on resident glia including microglia and Schwann cells was confirmed at time points prior to the significant recruitment of circulating macrophages in injured nerves.

Examination of GBS nerves, obtained at different time points after onset, showed that there was significant upregulation of Fcγ common chain in acute (data not shown) and post-acute phase of GBS nerves, whereas control/uninjured human nerves did not show staining for Fcγ common chain (Figure 3A). Western blotting studies confirmed ICC findings (Figure 3B). These results provide evidence that activating FcγRs are upregulated in GBS nerves and injured nerves of experimental animals and are available to participate in Ab-mediated inflammation.

We tested the inhibitory effects of GD1a/GT1b-2b in nerve crush model on axon regeneration in Fcer1g-null mice, which lack all activating FcγRs and only express inhibitory FcγRIIB [29]. Our results showed that Fcer1g-null mice were resistant to the severe inhibitory effects seen in background-matched WT mice, as assessed by behavioral, electrophysiological, and morphometric measures. Behavioral studies (Pinprick test) showed that GD1a/GT1b-2b mAb significantly reduced sensory functional recovery after nerve injury in WT mice compared to WT animals treated with sham Abs (Figure 4A), whereas the sensory functional recovery was similar in Fcer1g-null mice treated with GD1a/GT1b-2b and sham Abs (Figure 4B). Quantitative electrophysiology (sciatic nerve conductions) indicated that GD1a/GT1b-2b mAb adversely affected the motor nerve regeneration and target (muscle) reinnervation in WT animals but not in Fcer1g-null mice (Figure 4C). Morphological studies also demonstrated that the GD1a/GT1b-2b mediated inhibition of axon regeneration found in WT mice was reversed in Fcer1g-null mice (Figures 4D-4G). There was significant decrease in regenerating myelinated nerve fibers (MF) in GD1a/GT1b-2b-treated WT animals at sciatic (SN) (336±38) and tibial (TN) (36±5) nerves compared with sham Ab-treated WT mice in sciatic (2253±152) and tibial (596±69) nerves (Figures 4D, 4F, and 4G). In contrast, no significant difference in MF regeneration was found in Fcer1g-null mice treated with GD1a/GT1b-2b (SN, 1995±187 and TN, 577±55) or sham Ab (SN, 2032±156 and TN, 642±60) (Figures 4E, 4F, and 4G).

The inhibition of axon regeneration induced by GD1a/GT1b-2b was abrogated in Fcer1g-null mice, which exclusively express inhibitory FcγRIIB. This led to the hypothesis that the inhibitory FcγRs are not involved in the anti-glycan Ab-mediated inhibition. Availability of Fcgr2b -null mice (which lack inhibitory FcγRIIB but express all activating FcγRs) allowed reconstitution studies. We found that axon regeneration in Fcgr2b -null mice was severely inhibited by GD1a/GT1b-2b mAb. The numbers of regenerating MF in sciatic (57±19) and tibial (9±8) nerves in GD1a/GT1b-2b-treated Fcgr2b -null mice were significantly decreased compared with those in sham Ab-treated sciatic (2177±225) and tibial (678±97) nerves (Figures 5A-5C). GD1a/GT1b-2b induced even more severe inhibition of axonal regeneration in Fcgr2b -null mice (SN, 57±19 and TN, 9±8) than what it did in the control animals (SN, 299±22 and TN, 36±6). Electrophysiological studies were consistent with morphological findings showing that the treatment with GD1a/GT1b-2b mAb abolished evoked CMAP responses in Fcgr2b -null mutants (Figure 5D). These studies in Fcer1g-null and Fcgr2b -null mice indicated that anti-glycan Ab-mediated inhibition of axon regeneration in this model was completely dependent on the expression of activating FcγRs.

Activating FcγRs include FcγRI, FcγRIII, and FcγRIV, which differ in their Ab binding affinities and isotype specificity due to their different molecular structures [30]. In order to evaluate the role of different individual activating FcγRs in the Ab-mediated axonal inhibition, the inhibitory effect of GD1a/GT1b-2b was examined in three different transgenic lines Fcer1α-null, Fcgr3-null, and Fcgr4-null, which lack activating FcγRI, FcγRIII, or FcγRIV, respectively.

Our results showed that Fcer1α-null and Fcgr4-null mice were susceptible to anti-glycan Ab-mediated inhibition of axon regeneration (Figures. 6 and 7), whereas this inhibitory effect was almost completely reversed in Fcgr3-null mice (Figure 8). We found that anti-glycan Ab-mediated inhibition of axon regeneration was independent of activating FcγRI. The morphological data showed that the treatment with GD1a/GT1b-2b significantly reduced the numbers of regenerating MF in sciatic (298±35) and tibial (34±14) nerves in Fcer1α-null mice compared with sham Ab treatment (2197±93 at sciatic and 542±80 at tibial nerve levels) (Figures 6A-6C). There was no significant difference in anti-glycan Ab induced inhibition of axon regeneration between Fcer1α-null mice (SN, 298±35 and TN, 34±14) and WT control animals (SN, 286±39 and TN, 25±3). Electrophysiological studies were consistent with morphological findings showing that the treatment with GD1a/GT1b-2b mAb abolished evoked CMAP responses in Fcer1α-null mutants (Figure 6D). Next, our studies indicated that anti-glycan Ab-mediated inhibition of axon regeneration was also independent of activating FcγRIV. We found that there was a significant decrease in numbers of regenerating MF in GD1a/GT1b-2b-treated sciatic (282±82) and tibial (15±3) nerves in Fcgr4-null mice compared with sham Ab treated sciatic (2952±64) and tibial (795±68) nerves (Figures 7A-7C). The anti-glycan Ab mediated inhibition of axon regeneration was similar between Fcgr4-null mice (SN, 282±82 and TN, 15±3) and their littermate control animals (SN, 320±50 and TN, 21±2). Electrophysiological studies were consistent with morphometry showing that the treatment with GD1a/GT1b-2b mAb abolished evoked CMAP responses in Fcgr4-null mutants (Figure 7D).

In contrast to Fcer1α- and Fcgr4-null mice, we found that FcγRIII-null mice were resistant to anti-glycan Ab-mediated inhibition of axon regeneration. Morphometry showed that GD1a/GT1b-2b-induced inhibition of axon regeneration observed in WT was largely reversed in Fcgr3-null mice at sciatic (Figures 8A and 8B) and tibial (Figure 8C) nerve levels. In this set of studies, number of regenerating MF in WT mice treated with GD1a/GT1b-2b were 286±39 at sciatic and 25±3 at tibial nerve levels compared to sham Ab-treated group (SN, 2128±223; TN 627±89). In comparison, number of regenerating MF in GD1a/GT1b-2b-treated Fcgr3 -null mice were1125±162 at sciatic and 371±23 at tibial nerve levels, whereas number of regenerating MF in sham Ab-treated Fcgr3 -null mice were 1807±178 at sciatic and 519±39 at tibial nerve levels. Electrophysiology showed that Fcgr3 -null mutants treated with GD1a/GT1b-2b mAb had significant recovery of evoked CMAP amplitudes compared to WT mice treated with GD1a/GT1b-2b mAb, thus, confirming the functional recovery of regenerating fibers (Figure 8D). Overall, these studies indicate that activating FcγRIII is the dominant FcγR involved in GD1a/GT1b-2b mAb-mediated inhibition.

Op/op mice with macrophage and microglia deficiency were examined to determine whether anti-glycan Abs mediated inhibition of axon regeneration depends on FcγRs-bearing macrophage/microglial population. Inhibitory effects of GD1a/GT1b-2b mAb on axon regeneration were compared in op/op mice and WT littermates in nerve crush model.

Our results showed that op/op mice had significantly more regenerated axons compared to WT animals indicating the involvement of macrophage/microglial cells in the anti-glycan Ab-mediated inhibition (Figures 9A-9C). Morphometry showed that number of regenerating MF in op/op mice treated with GD1a/GT1b-2b was significantly higher than WT mice treated with the same mAb at sciatic (Figure 9B) and tibial (Figure 9C) nerve levels. Quantification of MF fiber showed that WT mice treated with sham Abs had 2151±181 at sciatic and 601±72 at tibial nerve levels, whereas WT mice treated with GD1a/GT1b-2b had 299±37 at sciatic and 27±5 at tibial nerve levels; op/op mice treated with sham Abs had 2203±252 at sciatic and 651±87 at tibial nerve levels. In comparison, op/op mice treated with GD1a/GT1b-2b had 1572±171 at sciatic and 439±59 at tibial nerve levels. Serological studies showed that the levels of circulating GD1a/GT1b-2b mAb in op/op mice were comparable to their WT counterparts (Figure 9D), demonstrating that IgG homeostasis was not altered in these mice.

Our results showed that both endoneurial glia (Schwann and microglial cells) and recruited macrophages express activating FcγRs (Figure 2D), therefore, we asked which FcγRs-expressing cells participate in producing an inflammatory inhibitory milieu in the injured nerve.

A nerve grafting paradigm, in which nerve segments of donor mice (WT or Fcer1g-null/ activating FcγRs-null) were transplanted into host animals (WT mice or Fcer1g-null), that allows determination of the contribution of circulating macrophages (recruited in injured nerves from hosts) and resident endoneurial glial cells (in grafted nerve segments from donors) in Ab-mediated inhibition of axon regeneration was used to address this question. Previous studies had established that grafted nerve segments retain donor glia [31], [32]. These chimeric animals were administered GD1a/GT1b-2b Ab and axon regeneration was assessed in the grafted nerve segments. We found that Fcer1g-null hosts implanted with nerve grafts from Fcer1g-null donors were not susceptible to Ab-mediated inhibition, whereas WT hosts implanted with WT nerve grafts showed severe inhibition (Figures 9E and 9F). Notably, the axon regeneration in WT or Fcer1g-null grafts implanted in Fcer1g-null hosts was more pronounced compared to WT or Fcer1g-null grafts implanted in WT hosts. Additionally, WT grafts implanted in Fcer1g-null hosts had modest but significant reduction in number of regenerating axons compared to Fcer1g-null grafts implanted in Fcer1g-null hosts (Figures 9E and 9F). Morphometry showed the numbers of MF in different chimeras is as follows: 1) Fcer1g-null grafts in mutant Fcer1g-null hosts had 1609±206 MF; 2) WT grafts in Fcer1g-null hosts had 996±162 MF; 3) WT grafts in WT hosts had 4±1 MF; and 4) Fcer1g-null grafts in WT hosts had 6±4 MF.

Overall, these studies showed that the macrophages recruited from the circulation of the host animals were the dominant cell type and endogenous nerve glia had minor contribution to the activating FcγRs-induced inflammation and Ab-mediated inhibition of axon regeneration.

Complement-fixing pathogenic Abs can cause nerve injury through the activation of classical complement pathway. We previously demonstrated that membrane attack complex (C5b-9) is not involved in anti-glycan Ab-induced inhibition of axon regeneration [14]. In this study we examined the role of C3 complement component, a critical component of classical complement pathway, in Ab-mediated inhibition. The effect of anti-glycan mAb, GD1a/GT1b-2b, was examined in C3-null mice. We found that GD1a/GT1b-2b mAb produced severe inhibition of axon regeneration in C3-null mice (Figure 10). The numbers of regenerating MF in GD1a/GT1b-2b-treated sciatic (354±90) and tibial (42±9) nerves were significantly reduced compared with that in sham Ab treated sciatic (1501±84) and tibial (409±25) nerves. These data provide further evidence that complement is not involved in this Ab-mediated inhibition of axon regeneration in our models.