OUP user menu

Mediation of Protection and Recovery From Experimental Autoimmune Encephalomyelitis by Macrophages Expressing the Human Voltage-Gated Sodium Channel NaV1.5

Kusha Rahgozar BS, Erik Wright BS, Lisette M. Carrithers BA, Michael D. Carrithers MD, PhD
DOI: http://dx.doi.org/10.1097/NEN.0b013e318293eb08 489-504 First published online: 1 June 2013

Abstract

Multiple sclerosis (MS) is the most common nontraumatic cause of neurologic disability in young adults. Despite treatment, progressive tissue injury leads to accumulation of disability in many patients. Here, our goal was to developan immune-mediated strategy to promote tissue repair and clinical recovery in an MS animal model. We previously demonstrated that a variant of the voltage-gated sodium channel NaV1.5 is expressed intracellularly in human macrophages, and that it regulates cellular signaling. This channel is not expressed in mouse macrophages, which has limited the study of its functions. To overcome this obstacle, we developed a novel transgenic mouse model (C57BL6c-fms-hSCN5A), in which the human macrophage NaV1.5 splice variant is expressed in vivo in mouse macrophages. These mice were protected from experimental autoimmune encephalomyelitis, the mouse model of MS. During active inflammatory disease, NaV1.5-positive macrophages were found in spinal cord lesions where they formed phagocytic cell clusters; they expressed markers of alternative activation during recovery. NaV1.5-positive macrophages that were adoptively transferred into wild-type recipients with established experimental autoimmune encephalomyelitis homed to lesions and promoted recovery. These results suggest that NaV1.5-positive macrophages enhance recovery from CNS inflammatory disease and could potentially be developed as a cell-based therapy for the treatment of MS.

Key Words
  • Alternative activation
  • Experimental autoimmune encephalomyelitis
  • Macrophage
  • Multiple sclerosis
  • Sodium channel

Introduction

Multiple sclerosis (MS) is an inflammatory presumed autoimmune disease of the CNS and is a common cause of neurologic disability (1). Although several immune-modulating therapies are available to treat MS patients, there is very limited efficacy on measures of disease progression and accumulation of neurologic deficits (2). Some investigators have hypothesized that innate immune mechanisms mediate chronic tissue injury in MS lesions (3). This hypothesis is based in part on the pathology of lesions, which demonstrate the presence of macrophages and activated microglia (4). Although the lineage of microglia has been controversial, recent studies indicate that they are resident mononuclear phagocytic cells of the brain and are derived from primitive macrophages (5).

Mononuclear phagocytes are highly motile cells that mediate host defense and tissue homeostasis. Metchnikoff was the first to describe the crucial role of phagocytes in the clearance of infectious pathogens and injured cells (6). The biochemical mechanisms and physiologic relevance of macrophage phagocytosis of pathogens have been areas of intense study since then, but their role in what Metchnikoff termed “physiological inflammation” has until recently been largely ignored. Consistent with the emerging role of alternatively activated macrophages in tissue repair (7), he hypothesized that phagocytes not only mediated host defense but were also necessary for recovery from tissue injury.

We have focused on how novel sodium channel variants regulate human macrophage function (810). Intracellular expression and activity of NaV1.5 and NaV1.6 variants in macrophages regulate basic cellular functions that are necessary for optimal innate immune responses. The initial description of sodium channel expression in phagocytes occurred during studies of the neuroprotective effects of sodium channel blockers in the murine MS model experimental autoimmune encephalomyelitis (EAE). In this work, the expression of the voltage-gated sodium channel, NaV1.6, was demonstrated in microglia from brains of mice with EAE and humans with MS (11). We subsequently characterized NaV1.5 and NaV1.6 expression and function in macrophages (810). These studies demonstrated that NaV1.5 regulates phagocytosis in human macrophages, and that NaV1.6 regulates cellular movement through its association with the F-actin cytoskeleton and regulation of podosome formation. Black et al (12) demonstrated NaV1.5 expression in phagocytes within MS lesions, but this variant is not expressed in mouse macrophages. Unlike sodium channels in excitable tissues in which they are present on the plasma membrane, macrophages NaV1.5 and NaV1.6 are expressed on intracellular organelles where they regulate intracellular signaling in part through mitochondrial sodium-calcium exchange (8, 10).

Despite the in vitro findings, the relevance of macrophage sodium channels to immune-mediated disease and regulation of cellular phenotype remains unclear. This issue is particularly pertinent to the functional significance of human macrophage NaV1 .5 because it is not expressed in murine macrophages and could not be readily studied in mouse EAE models. For example, enhanced phagocytosis by macrophages might enhance recovery from tissue injury through clearance of extracellular debris and apoptotic cells (13). Alternatively, inflammatory macrophages could enhance tissue injury through local release of reactive oxygen and nitrogen species. In view of the in vivo complexity of macrophage phenotype and function, the phenotype of human NaV1.5-positive macrophages needs to be assessed in an in vivo model.

To examine in vivo function of macrophage NaV1.5, we developed a knock-in mouse in which the human macrophage NaV1.5 splice variant is expressed selectively in monocytemacrophage lineage cells (C57BL6c-fms-hSCN5A mice). These mice developed much less severe EAE despite the presence of CNS inflammatory infiltrates. The macrophages in these mice demonstrated an unexpected polarization to an arginase-positive phenotype during EAE; and, when adoptively transferred to mice with clinical disease, the cells entered the CNS and enhanced recovery.

Materials and Methods

Cells

Primary mouse bone marrow cells were obtained from femurs and/or tibia of transgenic and wild-type (WT) mice and were differentiated to bone marrow-derived macrophages (BMDMs) in RPMI media supplemented with 10% fetal bovine serum, sodium pyruvate, nonessential amino acids, and macrophage colony-stimulating factor (20 ng/mL) for 5 to 7 days. For adoptive transfer experiments, cells were dissociated with TrypLE (Invitrogen, Carlsbad, CA), washed, resuspended in PBS, and then injected intraperitoneally into recipient mice (5 × 105 cells per mouse).

THP-1 cells, a human monocytic cell line, were maintained in RPMI-1640 media supplemented with 10% fetal bovine serum, sodium pyruvate, and nonessential amino acids. Differentiation to a macrophage phenotype was induced by treatment with 12-O-tetradecanoylphorbol-13-acetate (10 ng/mL) for 72 hours.

cDNA Cloning

SCN5A is the gene that encodes NaV1.5. A polymerase chain reaction (PCR)-based approach was used to sequence SCN5A cDNA generated from differentiated THP-1 cells (9). Messenger RNA was isolated and reverse transcribed. To obtain the full-length coding region of SCN5A, 2 separate PCR products were generated and then stitched together. The first PCR product was generated using exon 3 forward and exon 27 reverse primers and the second one using exon 2 forward and exon 3 reverse primers (Figure, Supplemental Digital Content 1, http://www.jnen.oxfordjournals.org/lookup/suppl/doi:10.1097/NEN.0b013e318293eb08/-/DC1). These 2 PCR products were gel purified and used as a template for another PCR reaction using exon 2 forward and exon 27 reverse primers to obtain the full-length cDNA (6 kb). Polymerase chain reaction was performed using platinum Taq DNA polymerase high fidelity (Invitrogen) according to the recommendations of the supplier. Cycle conditions were as follows: denaturation at 94°C for 30 seconds, annealing at 55°C for 30 seconds, and extension at 68°C for 7 minutes for the first PCR product and 1 minute for the second one (40 cycles). The full-length PCR product was sequenced in the forward and reverse directions at the Keck Biotechnology core facility at Yale University School of Medicine using the listed primers. DNASTAR (Lasergene) was used for the analysis of sequence data.

Human Macrophage SCN5a Transgene Construct

The cDNA for the full-length coding region of the SCN5A splice variant was spliced downstream of a mouse c-fms promoter. This promoter contains a 3.5-kb 5′ flanking sequence of the c-fms gene and the downstream intron 2 (14). Pronuclear injection, generation of founder mice, and Southern blot analysis were performed at inGenious Laboratories, Stony Brook, NY. This construct specifically targets expression of the human macrophage SCN5A splice variant to the monocyte-macrophage lineage.

Quantitative PCR

RNA purification, reverse transcription, quantitative PCR, and data analysis were performed as described previously (10). Data were acquired on a Cepheid SmartCycler and analyzed by the ΔΔCt method. The following TaqMan primers were obtained from Applied Biosystems: dHs00165693_m1 (human SCN5A), Mm99999915_g1 (mouse gapdh), Mm99999915_g1 (mouse cdh1), and Mm00475988_m1 (mouse arg1).

Mice

Experimental mouse strains were bred at our on-campus breeding facility (Biotron facility), where our transgenic colony is maintained. Experimental mice were transferred to an approved University of Wisconsin animal facility for the performance and monitoring of all in vivo experiments. Age (12–16 weeks)- and sex-matched transgene-positive and negative (WT) littermates were used for EAE experiments, and transgene-positive mice were used as bone marrow donors. Seven to 13 mice were used for each EAE condition, and 3 to 4 mice for each histologic and flow cytometry condition (same age and sex ratios as for EAE experiments).

EAE Induction

A commercially available kit (Hooke Laboratories, Lawrence, MA) was used to induce EAE. Mice were immunized with myelin oligodendrocyte glycoprotein peptide 35–55 (MOG35–55) emulsified in complete Freund adjuvant by injecting them subcutaneously at 2 sites on the back (0.1 mL of emulsion per site; 1 mg/mL MOG35–55, and 2.5 mg/mL killed H37Ra Mycobacterium tuberculosis). On the same and subsequent days, an intraperitoneal injection of pertussis toxin in PBS, at 100 ng per mouse per dose (0.1 mL), was performed. Mice were examined daily and were given additional soft food (Dough Diet) and a hydration source (Napa Nectar) in the bottom of their cage if needed. Mice were followed for up to 40 days. Animals were observed on a daily basis for signs of clinical EAE. The animals were graded by a blinded examiner as follows: 1, limp tail; 2, partial hind limb paralysis; 3, complete hind limb paralysis; 4, hind and front limb paralysis; and 5, moribund. Mice were killed if they developed a score of 4. Animals were cared for in accordance with the University of Wisconsin-Madison IACUC guidelines and the National Research Council's Guide.

Using this protocol in our colony, onset of disease usually occurs at approximately Day 14; onset of peak disease occurs 5 days after disease onset; peak disease onset is on average Day 19; disease stabilization occurs 10 days after disease onset (average of Day 24). Late EAE in adoptive transfer experiments was defined as 14 days after disease onset (Day 28). For adoptive transfer experiments, mice were randomized on a rolling basis as they developed EAE. Bone marrow-derived macrophages were grown, prepared, and administered as described above on the first day of clinical signs.

Histology

Perfusions, tissue preparation for frozen sections, sectioning, and staining for immunofluorescence were performed as described (8, 15). The following antibodies were used for immunohistochemistry: rabbit anti-human NaV1.5 (Alomone Labs, Jerusalem, Israel); anti-arginase 1 (clone19/Arginase I; BD Biosciences, San Jose, CA); anti-F4/80 (CIA3-1), rabbit anti-E cadherin, mouse (MBP101) and rabbit anti-myelin basic protein, and rabbit anti-200-kd neurofilament were from Abcam (Cambridge, MA); anti-phospho-PHF-tau (paired helical filaments tau, pSer202/Thr205, clone AT8) was from Thermo Fisher; and anti-CD11b (M1/70) and anti-CD4 (GK1.5) were from eBioscience (San Diego, CA). Alexa dye (488 or 555) labeled secondary antibodies (donkey anti-rabbit, rat, or mouse) were from Invitrogen. Isotype controls were obtained from eBioscience. Processing and staining for hematoxylin and eosin staining were performed at the University of Wisconsin Department of Pathology core facility.

Fluorescence Microscopy

Fluorescence images were acquired and analyzed using a Zeiss Axiovert 200 fluorescent microscope equipped with Axiovision version 4.8 software. Quantitative analysis of hyperphosphorylated tau staining was performed as a modification of our previous approach (16, 17). Frozen longitudinal sections (8 μm) of mouse spinal cord samples were separated by more than 20 μm to avoid staining the same cell twice. Sections were stained for immunofluorescence with anti-neurofilament and anti-phospho tau and subsequently with the nuclear stain 4′, 6-diamidino-2-phenylindole (DAPI). Whole spinal cords from mice immunized with MOG 35–55 were analyzed (onset of peak disease, Day 19), and regions of inflammatory infiltrates, as defined by DAPI staining and analysis of serial sections stained for immune cell markers, were imaged using a 20 × objective (Zeiss) within a 400 × 400 μm2 microscopic field. Three complete sections from 3 separate spinal cords were analyzed for each condition. The approximate area of infiltrate per longitudinal cross section in the WT spinal cord was 31.2 ± 3.7 mm2 versus 8.2 ± 1.3 mm2 in the transgenics. Images were analyzed using the AutoMeasurement component of Axiovision software; areas of phospho tau staining were expressed as square micrometers per square millimeter tissue.

Flow Cytometry

Cell preparation, staining, and analysis for flow cytometry were performed as described previously (15, 18). For analysis of CNS mononuclear cells, whole spinal cord was used (19). Data acquisition was performed on an LSRII flow cytometer (Becton-Dickinson) at the University of Wisconsin Carbone Cancer Center core facility at the Wisconsin Institute for Medical Research. The following antibodies were used: rabbit anti-NaV1.5 (Alomone); sheep anti-Arginase 1 (fluorescein; R&D Systems, Minneapolis, MN); APC-labeled anti-F4/80 (clone BM8), APC-eFluor 780 CD4 (GK1.5), PE-CD8, eFluor 450 CD11b (M1/70), PerCP-efluor 710 CD324 (DECMA-1), and fluorescein isothiocyanate-interleukin 10 (IL10) were from eBioscience. Secondary antibodies for anti-NaV1.5 were either Alexa 488 goat-anti-rabbit (Invitrogen) or PE-labeled goat anti-rabbit Fab fragments (eBioscience). Similar results were obtained for both approaches. Isotype controls were obtained from eBioscience.

Data Analysis

Data were analyzed using Axiovision 4.8 (Zeiss) and FlowJo (Treestar, Ashland, OR) software. Statistical analysis (SEM calculation, analysis of variance, and t test) was performed using Kaleidagraph 4.1 (Synergy).

Results

cDNA Cloning of Macrophage SCN5A Reveals That It Is a Novel Splice Variant

Complementary DNA cloning of human macrophage SCN5A (hSCN5A) demonstrated that it was a novel splice variant that lacked exon 25, contained exon 7, and was most homologous to a predicted splice variant identified as SCN5A transcript variant 4 (Table 1; Figure, Supplemental Digital Content 1, http://www.jnen.oxfordjournals.org/lookup/suppl/doi:10.1097/NEN.0b013e318293eb08/-/DC1). Exons 6 and 7 are alternate, duplicate coding exons; however, exon 25 is an in-frame coding region that encodes an extracellular portion of the channel. Deletion of exon 25 results in an 18-amino acid deletion in NaV1.5 (Figure, Supplemental Digital Content 2, http://www.jnen.oxfordjournals.org/lookup/suppl/doi:10.1097/NEN.0b013e318293eb08/-/DC1).

View this table:
TABLE 1

Macrophage SCN5A Is a Splice Variant Homologous to Transcript Variant 4

Accession No.IsoformTranscript VariantDeleted Exons% Identity With New Sequence
NM 198056a1699.3
NM 000335b (hH1c)2699.3
NM 001099404c3799.4
NM 001099405d47 and 2599.5
*KC858891 (Macrophage variant)7 and 25
  • Total RNA was isolated from differentiated and primed THP-1 cells and reverse-transcribed to cDNA. PCR amplification of SCN5A was performed using gene specific primers to generate a 6-kb fragment that was purified and cloned into a TA cloning vector. Nucleotide sequencing was performed at the Keck facility at the Yale University School of Medicine. Sequencing was performed from both the forward and reverse ends to provide sequence confirmation. Sequence comparisons were performed using DNASTAR software (Lasergene).

  • * This nucleotide sequence has been submitted to the Gen-BankTM/EBI Data Bank.

Generation of C57BL6c-fms-hSCN5A Mice

Because NaV1.5 is not expressed in murine macrophages, it was necessary to develop a transgenic model to study its in vivo function (Fig. 1). We selected the c-fms promoter, which encodes the macrophage colony-stimulating factor receptor, to direct expression of the human SCN5A channel variant in all macrophage subsets, including monocytes, macrophages, and phagocytic bone marrow-derived microglia (14). As indicated in a recent review, it is not intrinsically possible to limit transgene expression solely to classic macrophages with any of the macrophage-specific promoters that have been used, particularly for cell ablation studies (20). The advantage of the c-fms promoter construct for gene knock-in studies is that it results in reliable expression of genes in F4/80-positive macrophages with a lower level of expression in other myeloid cells. Mouse founder lines were screened by Southern blotting (inGenious Laboratories). All founder lines that screened positive by Southern blotting were healthy and viable; the line that best demonstrated human SCN5A expression in BMDMs was used for subsequent experiments.

FIGURE 1

Human macrophage SCN5a transgene construct. The gene SCN5A encodes the sodium channel NaV1.5. The human macrophage splice variant contains a deletion of exon 25. The cDNA for the full-length coding region of this splice variant was spliced downstream of a mouse c-fms promoter. This promoter contains a 3.5-kb 5′ flanking sequence of the c-fms gene and the downstream intron 2 (14). This construct specifically targets expression of the human macrophage SCN5A splice variant to the monocyte-macrophage lineage.

Human Macrophage NaV1.5 Is Expressed in Monocyte-Macrophages From C57BL6c-fms-hsCN5A Mice

Heterozygous transgenic mice had a normal appearance, life span, thymic and splenic development, and cell number (data not shown). Analysis of peripheral blood C1 1b-positive, F4/80-positive cells and BMDMs by flow cytometry and immunohistochemistry demonstrated expression of human macrophage NaV1.5 (hNaV1.5), the protein encoded by hSCN5A (Fig. 2A–C).

FIGURE 2

The human macrophage NaV1.5 variant (hNaV1.5) is expressed in mouse peripheral blood mononuclear cells (PBMCs) and bone marrow-derived macrophages (BMDMs) from C57BL6c-fms-hSCN5A mice. (A) PBMCs were isolated and stained for surface markers and intracellular hNaV1.5 expression. Rabbit isotype control staining for NaV1.5 is also shown in the lower tracing. (B) BMDMs from mature mice (hSCN5A) were differentiated for 10 days in the presence of macrophage colony-stimulating factor and then analyzed by immunohistochemistry. Low-power views show coexpression of hNaV1.5 and the mouse macrophage marker F4/80 in BMDMs. The higher power view (bottom right) demonstrates a vesicular-type staining pattern for hNaV1.5, similar to that observed in primary human macrophages and cell lines. Scale bars = 50 μm (low power) and 10 μm (high power). (C) Flow cytometry analysis confirmed hNaV1.5 expression in most F4/80-positive BMDMs. Staining for hNaV1.5 was assessed using a rabbit anti-human NaV1.5; rabbit isotype control staining is shown in the right tracing.

Bone marrow-derived macrophages from C57BL6c-fms-hSCN5A mice also demonstrated expression of hSCN5A mRNA. Human SCN5A mRNA expression in transgenic BMDM was similar to that previously observed in primary monocyte -derived human macrophages (8). The expression of hSCN5A in transgenic cells was 10.1 ± 0.3 copies/GAPDH ×10−5; hSCN5A expression was not detected threshold cycle >50) in WT cells. In addition, one unexpected result was that, in the absence of any polarizing cytokines, quantitative PCR analysis demonstrated an approximately 3-fold increase in arginase (arg1) and E-cadherin (cdh1) mRNA expression (Fig. 3A). Increased expression of these molecules serves as a marker for alternative macrophage activation (7, 21). In addition, E-cadherin is expressed within phagocytic nodules and early granulomas to facilitate cell-cell adhesion (22). To assess markers of alternative activation further, we also performed intracellular staining for arg1 and IL10 (Fig. 3B). As expected, the percent of arg1-positive BMDMs was increased in the transgenic condition as compared with that in WT controls. In addition, hNaV1.5-positive BMDMs expressed intermediate and high levels of IL10, whereas control cells demonstrated a low expression.

FIGURE 3

Quantitative PCR (qPCR) and fluorescence-activated cell sorting analysis of macrophage markers. (A) The hSCN5A encodes the human NaV1.5 protein. Bone marrow-derived macrophages (BMDMs) were differentiated in macrophage colony-stimulating factor for 5 to 7 days and then analyzed for mRNA expression by qPCR. Threshold cycle (Ct) values were normalized using glyceraldehydes-3-phosphate dehydrogenase (GAPDH) as a control for each experiment; the ΔΔCt method was used to calculate copy number relative to GAPDH expression. Arginase (arg1) and E-cadherin (cdh1) expression in macrophages derived from C57BL6c-fms-hSCN5A mice were increased approximately 3-fold versus wild-type mice. The measured mRNA levels were arg1, 12.03 ± 0.69 copies/GAPDH ×10−5 in the transgenic group and 4.5 ± 0.17 copies/GAPDH ×10−5 (n = 4, p < 0.001); cdh1, 1.27 ± 0.13 copies/GAPDH ×10−5 in the transgenic group and 0.45 ±0.03 copies/GAPDH ×10−5 in the wild-type group (n = 4, p<0.05). (B) The percent of arginase protein-positive cells was increased in hSCN5A-positive mouse BMDMs as determined by intracellular staining and flow cytometry (left tracing). In addition, the percent of high and intermediate interleukin 10 (IL10)-expressing cells was greater in the transgenic cells as compared with that in littermate controls (right tracing).

Decreased EAE Severity in C57BL6c-fms-hSCN5A Mice

We used the MOG35–55 immunization model to assess disease course in age- and sex-matched transgenic mice and WT littermates. C57BL6c-fms-hSCN5A mice demonstrated reduced peak disease severity, cumulative disease score, and disease incidence (Fig. 4; Table 2). These results suggested that the transgenic mice have either an impaired immune response or enhanced anti-inflammatory regulatory mechanisms.

View this table:
TABLE 2

Experimental Autoimmune Encephalomyelitis Clinical Data

StrainNo. Mice (male/female)Disease IncidencePeak Disease ScoreCumulative Disease Score Through 30 Days
C57BL68/513/132.8 ± 0.439.1 ± 7.8
C57BL6c-fms-hSCN5A7/43/110.9 ± 0.6*10.9 ± 7.1*
  • All recipient and donor mice were 12 to 16 weeks of age and were age- and sex-matched littermates.

  • * p ≤ 0.05

FIGURE 4

C57BL6c-fms-hSCN5A (hSCN5A) mice demonstrate decreased severity of experimental autoimmune encephalomyelitis (EAE) as compared with wild-type mice. EAE was induced, and the mice were evaluated clinically as described in the Materials and Methods section. Clinical data are shown Table 2.

Reduced CD4 T-Lymphocyte Infiltration and Presence of Mononuclear Phagocytic Cell Clusters During EAE in C57BL6c-fms-hSCN5A Mice

To assess the possible mechanisms of disease protection in C57BL6c-fms-hSCN5A mice, we analyzed inflammatory cell infiltrates from transgenic mice and sex-matched WT littermates by flow cytometry and histologic analysis. At the approximate onset of peak disease (∼Day 19), transgenic and WT mice demonstrated similar numbers of total mononuclear cells but an altered ratio of CD4/CD11b cells (Fig. 5). Spinal cord infiltrates from transgenic mice demonstrated a reduced but detectable frequency of CD4-positive T lymphocytes and an increased frequency of CD11b-positive immune cells. Total mononuclear cell counts per spinal cord were 65,338 ± 4,686 cells in the transgenic and 55,778 ± 5,271 cells in the WT mice (n = 3, not significant). The cell frequencies were CD11b, 79.3% ± 1.2% in the transgenic versus 52.5% ± 5.4% in the WT (n = 3, p < 0.01); CD4, 5.6% ± 0.51% in the transgenic versus 18.7% ± 1.2% in the WT (n = 3, p < 0.01); there were no significant differences in CD8 frequencies. These results suggested that immune cells from C57BL6c-fms-hSCN5A mice can respond to immunization and home to inflamed spinal cord, but that there is a reduced CD4 T cell-mediated injury response in the target organ.

FIGURE 5

Flow cytometry analysis of CNS mononuclear cells during experimental autoimmune encephalomyelitis (EAE) in C57BL6c-fms-hSCN5A (hSCN5A) and wild-type (WT) mice. Single-cell preparations from mouse spinal cord were performed at onset of peak disease (5 days after disease onset, approximately Day 19), stained for cell markers, and analyzed. (A) A representative plot is shown that demonstrates an increased ratio of CD11b-positive/CD4-positive cells in hSCN5A mice versus WT littermate controls. (B) Quantitative analysis of cell counts and frequencies revealed that there were similar numbers of cells within the mononuclear immune cell gate; there were altered cell frequencies of CD11b-positive and CD4-positive but not CD8-positive populations. The cell counts for total mononuclear cells per spinal cord were 65,538 ± 4,686 for the hSCN5A transgenic and 55,778 ± 5,271 for the WT (n = 3, not significant). The percent CD11b positive was 79.3 ± 1.2 for the transgenic and 52.5 ± 5.4 for the WT (p < 0.01). The percent CD4 positive was 5.6 ± 0.5 for the transgenic and 18.7 ± 1.2 for the WT (p < 0.01).

Histologic analysis of spinal cord infiltrates at the same time point demonstrated extensive inflammation and tissue injury in WT mice but more limited, and clearly demarcated, demyelinating lesions in C57BL6c-fms-hSCN5A mice (Fig. 6). In the transgenic mice, the lesions had a more “punched-out” appearance and consisted predominantly of F4/80-positive phagocytic cells (Fig. 6C), and many of these were immunopositive for hNaV1.5 (Fig. 7). Figures 6C and 7 show higher power views of the lesions shown in Figure 6B. Cells within these clusters also expressed CD11b.

FIGURE 6

Histologic analysis of spinal cord inflammatory infiltrates showed a reduction in lesional CD4-positive T lymphocytes in C57BL6c-fms-hSCN5A (hSCN5A) mice and reveals the presence of phagocytic cell clusters. Mice were sacrificed for tissue analysis at onset of peak disease (5 days after disease onset). (A) Spinal cord white matter lesions were analyzed in longitudinal sections to maximize the area analyzed; the lesions often had an elongated oval shape that extended over multiple segments. Representative lesions from a WT mouse are shown at low power. The hematoxylin and eosin stain (left) demonstrates a moderately sized ovoid lesion containing mononuclear cells; the lesion is outlined in black. The immunofluorescent stains for myelin basic protein ([MBP] green) and the nuclear DAPI stain (blue) in frozen sections (right) show more extensive inflammation. (B, C) Higher power images of immunofluorescent staining of frozen sections for MBP (green in [B] and [C]), CD4 (red in [B]), and F4/80 (red in [C]). There are discrete well-defined regions of demyelination in hSCN5A mice that are associated with dense clusters of F4/80-positive phagocytic cells ([C] lower micrographs) and very few CD4-positive cells ([B] lower micrographs, white arrow). In contrast, wild-type mice have more diffuse, linear lesions associated with more numerous CD4-positive cells ([B] upper micrographs) and fewer F4/80-positive cells ([C] upper micrographs). Areas of detail in (B) are designated by the white outline in the adjacent micrograph. Scale bar = (A) 100 μm; (B) 50 μm; (C) 20 μm.

FIGURE 7

Human NaV1.5 expression in spinal cord demyelinating lesions in C57BL6c-fms-hSCN5A (hSCN5A) mice with experimental autoimmune encephalomyelitis. Many of the cells within the phagocytic nodular lesions in transgenic mice are NaV1.5 positive (red, middle panel). Myelin basic protein (MBP) immunostaining is shown in green. Scale bar = 20 μm.

Wild-type lesions demonstrated large linear regions of injury and showed a greater number of CD4-positive cells, with fewer F4/80-positive cells (Fig. 6B, C). These observations not only confirmed the results from flow cytometry but suggest that hNaV1.5-positive phagocytes have enhanced capacity to form discrete clusters within lesions. In a general sense, phagocytic clusters might be a precursor to granuloma formation and contain and limit local inflammatory responses (23). Such clusters within the CNS might represent a restricted host response to injury that prevents disseminated inflammation and resolves without residual injury (24, 25).

Decreased Axonal Injury in C57BL6c-fms-hSCN5A Mice

We further examined lesional tissue injury by staining for phospho-PHF-tau and neurofilament to assess axonal injury at onset of peak disease (Day 19) (16).

In spinal cord lesions from WT littermate controls, there was diffuse axonal injury, as determined by axonal swelling and the presence of phospho-PHF tau that colocalized with neurofilament staining; whereas in transgenic mice, only a minimal degree of phospho-PHF tau staining was detected and was found primarily in phagocytic cells within clusters (Fig. 8). Within regions that contained inflammatory infiltrates, the area of phospho-PHF tau staining was 17,484 ± 2,868 μm2/mm2 tissue in the WT and 223 ± 75 μm2/mm2 tissue in the transgenics (p < 0.01). These results suggested that there was decreased axonal injury in the C57BL6c-fms-hSCN5A mice.

FIGURE 8

Analysis of spinal cord lesions during experimental autoimmune encephalomyelitis reveals decreased axonal injury in C57BL6c-fms-hSCN5A (hSCN5A) mice. Inflammatory spinal cord lesions were analyzed for evidence of axonal injury using phospho-PHF tau (AT8; green) and neurofilament (NF; red) immunohistochemistry. Wild-type (WT) mice (upper micrographs) demonstrated more severe axonal injury as compared with C57BL6c-fms-hSCN5A mice (lower micrographs). Lesions in WT mice show increased numbers of swollen axons and hyperphosphorylated NF-positive axons. Quantitative analysis (bar graph, right) shows that the area of phospho-PHF tau immunoreactivity was 17,484 ± 2,868 μm2/mm2 tissue in the WT and 223 ± 75 μm2/mm2 tissue in the transgenics (p < 0.01). Scale bar = 50 μm.

Arginase-1 and E-Cadherin Expression in Phagocytic Cell Clusters

Increased expressions of arginase and E-cadherin serve as markers for alternative macrophage activation (7, 21), and E-cadherin (cdh1) is also expressed within phagocytic clusters and early granulomas to facilitate cell-cell adhesion (22). This analysis was performed on sections from mice from a slightly later period (recovery or stabilization, 10 days from disease onset, approximately Day 24), so that sufficient numbers of myeloid cells would be present within WT lesions.

Consistent with our hypothesis that macrophage NaV1.5 expression facilitates alternative activation, phagocytic cell clusters within lesions from C57BL6c-fms-hSCN5A mice demonstrated high levels of expression of cdh1 and arginase (arg1) in CD11b-positive cells (Fig. 9). In contrast, in WT littermate controls, arg1 expression was only detected in nerve fibers, and cdh1 was expressed at low levels in only a few CD11b-positive cells. These results suggested that lesional phagocytes have an altered phenotype in C57BL6c-fms-hSCN5A mice.

FIGURE 9

E-cadherin and arginase-1 expression in phagocytic clusters of C57BL6c-fms-hSCN5A (hSCN5A) mice. (A, B) Mononuclear cells within phagocytic clusters of C57BL6c-fms-hSCN5A mice (lower micrographs) demonstrate immunoreactivity for E-cadherin (Cdh1) ([A] red) and arginase-1 (Arg1) ([B] red), 2 markers of alternatively activated macrophages. Inflammatory infiltrates in wild-type (WT) mice (upper micrographs), as defined by CD11b staining (green), demonstrated substantially less Cdh1 staining within CD11b-positive cells as compared with transgenic cells. Arg1 expression was not detected in WT CD11b-positive cells but was present in neural cellular processes (B) in all conditions. There was enhanced expression of Arg-1 in resident cells adjacent to some phagocytic clusters in transgenic mice (lower micrographs). Scale bars = (A) 50 μm; (B) 20 μm.

We also observed increased expression of arg1 in adjacent resident cells in some lesions (Fig. 9, lower micrographs). Arginase 1 has been shown previously to be expressed in neurons and can enhance axonal regeneration after injury through increased polyamine synthesis (26, 27). Arginase 1 also inhibits the synthesis of nitric oxide by the inducible nitric oxide synthase pathway and, through this mechanism, can decrease local tissue injury in EAE and other disease models (28).

Recovery From EAE After Adoptive Transfer of hNaV1.5-Positive BMDMs From C57BL6c-fms-hSCN5A Mice

Based on analysis of the disease course and phenotype of inflammatory cell infiltrates, we hypothesized that hNaV1.5-positive macrophages from our transgenic mice mediated protection from EAE. Alternative hypotheses included transgene-mediated alterations in peripheral antigen presentation and cell movement. However, in our experiments, C57BL6c-fms-hSCN5A mice still developed inflammatory lesions associated with milder or absent clinical disease. These results suggested that the presence of the transgene did not prevent the development of a peripheral immune response or inhibit trafficking of immune cells to the CNS. Second, based on our in vitro analysis of hNaV1.5-positive BMDMs, we hypothesized that peripherally derived alternatively activated cells would be sufficient to mediate disease protection, and that resident tissue NaV1.5-positive macrophages, activated microglia, would not be necessary to initiate these protective mechanisms.

To test our hypotheses, we adoptively transferred hNaV1.5-positive BMDM (C57BL6c-fms-hSCN5A age- and sex-matched donors) into WT littermates with early signs of EAE (EAE score 1–2 at disease onset, Days 14–17). Disease onset was selected as the time for adoptive transfer to demonstrate the potential therapeutic efficacy of these cells during a treatment paradigm and to avoid any potential effects on peripheral immune responses after immunization.

Mice that received hNaV1.5-positive BMDMs demonstrated improved recovery from EAE as compared with the control group. Significant differences between the hNaV1.5-positive BMDMs and the vehicle group were observed during the recovery period, starting at Day 26, and continued through the end of the observation period (Fig. 10A; Table 3). The differences between groups at peak disease onset (5 days after disease onset) were not statistically significant (3.7 ± 0.5 in the vehicle group and 2.6 ± 0.4 in the adoptive transfer group). At this peak severity level of disease in MOG-induced EAE, few animals show substantial recovery and usually develop a chronic stable disease course (Fig. 4; Table 2). It is noteworthy, therefore, that a relatively small number of adoptively transferred cells could mediate clinical recovery during experimental conditions where the mice develop relatively severe disease (12 of 15 mice, 6 per group, reached a score of 3).

View this table:
TABLE 3

Adoptive Transfer Treatment Paradigm

Strain/TreatmentNo. Mice (male/female)Incidence of Peak Clinical Score ≥3.0Incidence of Recovery ≥1.0 in Clinical ScoreCumulative Disease Score Through 40 Days
C57BL6/Vehicle4/36/71/786.1 ± 11.9*
C57BL6/hSCN5A AT5/36/87/852.9 ± 9.6
C57BL6/WT AT4/37/70/7122 ± 12.3*
  • Bone marrow-derived macrophages (hSCN5A transgenic [hSCN5A] or wild-type [WT] cells) or vehicle (PBS) was transferred at disease onset to WT recipients. All recipient and donor mice were 12 to 16 weeks of age and were age- and sex-matched littermates.

  • * p ≤ 0.05 vs hSCN5A adoptive transfer group.

FIGURE 10

Enhanced clinical recovery from experimental autoimmune encephalomyelitis (EAE) after adoptive transfer of hNaV1.5-positive bone marrow-derived macrophages (BMDMs) at disease onset. (A) EAE was induced in C57BL6 wild-type (WT) littermatesof transgene-positive donors that were randomized to receive a single dose of vehicle or hNaV1.5-positive BMDM (5 × 105 cells) at disease onset (Days 13–17). Significant differences between the groups were observed during the recovery period, starting at Day 26 and continuing through the end of the observation period. The differences between groups at peak disease onset (Day 22) were not significant (3.7 ± 0.5 in the vehicle group and 2.6 ± 0.4 in the adoptive transfer group). (B) The use of WT BMDMs (5×105) as treatment at disease onset resulted in disease exacerbation, with a peak disease score of 5 in all mice.

By definition, in these experiments, there was 100% disease incidence. Male mice were included in these experiments and those shown in Figure 4 because they tend to develop more severe disease in this EAE model. Twelve of 15 total mice (6 in each group) and 9 of 9 males reached a score of 3. There were 4 males and 3 females in the vehicle group and 5 males and 3 females in the treatment group. Improvement of at least 1.0 in disease score occurred in 1 of 7 of control mice and 7 of 8 of treated animals. Cumulative disease score through 40 days was 86.1 ± 11.9 in the vehicle treated group and 52.9 ± 9.6 in the adoptive transfer group (p < 0.05). These findings suggested that transferred hNaV1.5-positive BMDMs could home to affected areas of the CNS in WT mice, maintain an arginase-positive phenotype, as demonstrated in C57BL6c-fms-hSCN5A mice with EAE, and mediate disease recovery.

In contrast, transfer of WT BMDMs from littermate controls resulted in disease exacerbation (Fig. 10B). All mice (4 males and 3 females) developed a peak score of 5; the cumulative disease score was 122 ± 12.3 (p < 0.05 as compared with transgenic adoptive transfer).

Adoptively Transferred hNaV1.5-positive BMDMs Migrate Into the Spinal Cord During EAE

After treatment with adoptively transferred cells, flow cytometry analysis and histologic analysis of spinal cord tissue during later EAE (14 days after disease onset) demonstrated the presence of a small number of hNaV1.5-positive macrophages in isolated mononuclear cells and at sites of inflammatory infiltrates (Figs. 11, 12). Within a myeloid cell-predominant gate of analyzed mononuclear cells, approximately 5% to 10% of CD11b-positive cells expressed high levels of hNaV1.5 in mice treated by adoptive transfer; these high-expressing cells were not observed in vehicle-treated WT mice (Fig. 11). The average count of CD11b-positive, hNaV1.5-positive was 2,650 ± 530 cells per spinal cord (n = 4), which represented approximately 0.5% of adoptively transferred cells. There were no statistically significant differences between the treated and untreated conditions in the CD4-positive, CD11b-positive or total mononuclear cell number. The total mononuclear cell counts at this time point were 57,836 ± 11,258 cells per spinal cord in the adoptive transfer condition and 51,358 ± 9,176 cells per spinal cord in the vehicle-treated condition (n = 4; p = not significant).

FIGURE 11

Detection of hNaV1.5-positive macrophages in the spinal cord after adoptive transfer. Approximately 2 weeks after adoptive transfer at onset of experimental autoimmune encephalomyelitis, spinal cord mononuclear cells were analyzed by flow cytometry. Within the mononuclear cell gate, myeloid (R1) and lymphocyte (R2) populations could be identified based on cell size (left). Within the myeloid gate, approximately 5% to 10% of cells demonstrated a high level of hNaV1.5 expression in the adoptive transfer group (middle). These high-expressing cells also expressed higher levels of E-cadherin (cdh1, CD324) as compared with the total CD11b-positive population, consistent with an alternatively activated phenotype (right). hNaV1.5-positive, CD11b-positive cells were not observed in untreated controls (middle). Similar numbers of CD4 T lymphocytes were observed in the treated and untreated groups (lower middle). FSC, forward scatter; SSC, side scatter.

FIGURE 12

hNaV1.5-positive bone marrow-derived macrophages can be detected within spinal cord phagocytic clusters after adoptive transfer. Histologic analysis of spinal cord was performed approximately 2 weeks after adoptive transfer (killed on Days 28–30). Longitudinal sections were stained for expression of CD11b and hNaV1.5. Small CD11b-positive (green) phagocytic cell clusters were observed in the white matter (the white line demarcates the meningeal border). A single high-expressing hNaV1.5-positive, CD11b-positive cell was observed within 2 different representative cell clusters (upper and lower panels). Scale bar = 20 μm.

Immunofluorescent staining for hNaV1.5 also demonstrated the presence of high-expressing cells within spinal cord phagocytic clusters in mice that received adoptively transferred cells (Fig. 12). The clusters that contained hNaV1.5-positive cells appeared to have a single positive cell associated with other CD11b-positive cells and with extensive cellular processes. These results confirmed the results from flow cytometry, demonstrated that phagocytic clusters form after adoptive transfer, and revealed that hNaV1.5-positive, CD11b-positive cells are present within the clusters. The combined flow cytometry and histologic data demonstrated that adoptively transferred hNaV1.5-positive BMDMs can home to the inflamed CNS and reside within phagocytic clusters that are similar in appearance to those observed in C57BL6c-fms-hSCN5A mice (Figs. 6, 7, 9). Given the high cadherin expression (Fig. 11) and extensive cellular processes within these cell clusters (Fig. 12), the results also suggested that hNaV1.5-positive BMDMs may enhance the formation of phagocytic clusters.

Increased Arginase-1 Expression in CNS and Peripheral Myeloid Cells After Adoptive Transfer of hNaV1.5-Positive BMDMs

We also examined the expression of Arg1 in isolated CNS mononuclear cells (EAE Days 28–30) after adoptive transfer of hNaV1.5-positive BMDMs from C57BL6c-fms-hSCN5A mice at disease onset. There was an approximately 2-fold increase in percent CD11b-positive, Arg1-positive cells in CNS mononuclear spinal cord cells (Fig. 13). The percent of CD11b-positive, Arg1-positive cells in the adoptive transfer group was 23.2% ± 1.1% and 12.2% ± 1.2% in the vehicle-treated control group (n = 4; p < 0.01). This increase in CD11b-positive, Arg1-positive cells cannot be explained entirely by the presence of adoptively transferred cells because of the relatively small number of CD11b-positive, hNaV1.5-positive cells within the CNS after adoptive transfer. Although the mechanism is unclear, these results suggest that hNaV1.5- positive macrophages may not only directly regulate recovery from tissue injury but also might mediate a bystander effect and induce an arginase-positive phenotype in host cells.

FIGURE 13

Increased arginase-1 expression in CD11b-positive cells after adoptive transfer of NaV1.5-positive bone marrow-derived macrophages. (A) Mononuclear cell populations were analyzed by flow cytometry approximately 2 weeks after adoptive transfer. (B) There was an increase in the percent of CD11b-positive, Arg1-positive cells in spinal cords in the adoptive transfer group as compared with untreated controls (left). The total mononuclear cells counts at this time point were not significantly different (right).

Discussion

We here demonstrate that the human NaV1.5 macrophage variant is a novel splice variant that regulates cellular phenotype and disease pathogenesis in vivo. This channel variant is not expressed in mouse macrophages. To study its in vivo function, we cloned the human macrophage variant of SCN5A, the gene that encodes the NaV1.5 channel, and developed a novel transgenic mouse (C57BL6c-fms-hSCN5A) that selectively expresses the channel variant in phagocytic cells, particularly CD11b-positive, F4/80-positive macrophages. These transgenic mice developed much less severe EAE and had smaller inflammatory CNS lesions associated with a reduction in CD4 T-lymphocyte infiltration and tissue injury as compared with littermate WT controls. The lesions in transgene-positive mice were characterized by small phagocytic cell clusters and expression of arginase, a marker of alternatively activated macrophages. Based on those results, we hypothesized that human macrophage NaV1.5 expressed in peripheral and central mononuclear phagocytes mediates anti-inflammatory mechanisms that ameliorate CNS inflammation and promote tissue repair. However, we could not rule out an effect of the gene knock-in on peripheral and central antigen presentation. To address that issue in subsequent EAE experiments, we adoptively transferred BMDMs from C57BL6c-fms-hSCN5A mice into WT recipients at the time of disease onset. Analysis of these mice demonstrated enhanced clinical recovery, the presence of transplanted cells in recipient spinal cord, and increased numbers of arginase-positive cells in mononuclear CNS infiltrates. Based on these results, we hypothesize that hNaV1.5-positive macrophage cell therapy might be an effective treatment for patients with MS.

Our hypothesis contrasts with many earlier models of inflammatory pathogenesis in MS. For example, a common model of disease pathogenesis in MS posits that lymphocyte-initiated inflammation leads to recruitment of monocytes, differentiation into macrophages within the CNS microenvironment, local activation and recruitment of microglia, and subsequent phagocyte-mediated demyelination and axonal injury (1). This theory is based, in part, on the observation that macrophages and activated microglia are abundant within MS lesions (4). However, the presence of mononuclear phagocytes within brain lesions does not necessarily demonstrate that they have only a pathogenic role. In addition, other immune cell subtypes can directly mediate tissue injury during autoimmunity. These subtypes include cytotoxic CD8 T lymphocytes and B lymphocyte-dependent autoantibody production coupled with complement fixation. An unsupported component of this model was that it postulated that macrophages were not only scavengers of debris and apoptotic cells, but that they somehow directly engulfed myelin from healthy axons. A more likely pathogenic mechanism in MS would be acute or chronic activation of macrophage pattern-recognition receptors that can lead to initiation of inflammatory T-cell responses and persistent localized release of inflammatory cytokines (29).

Consistent with earlier hypotheses of their cellular function (6), recent studies demonstrate that macrophages, including peripherally recruited cells and tissue resident cells such as microglia (5), mediate recovery from injury. In the CNS, depletion or inhibition of mononuclear phagocytes leads to increased disease severity in EAE (30), neurodegenerative (31), and stroke (32) disease models. A recent genetic study in families with Alzheimer disease also suggests that defective phagocytic mechanisms mediated by a variant of the TREM2 (triggering receptor expressed on myeloid cells 2) gene increase disease risk (33). Although it might be considered counterintuitive that the knock-in of a human gene that enhances macrophage phagocytic function could result in reduced disease severity in EAE or other CNS inflammatory disease model (12), our results in macrophage hNaV1.5-positive transgenic mice are consistent with these related studies of phagocytic cell function in chronic CNS diseases.

In addition, a large study of the sodium channel blocker, lamotrigine, for treatment of secondary progressive MS did not meet its primary outcome measure, that is, reduced clinical progression (34). Many of the patients stopped the medication on their own because of increased weakness, and there was increased brain volume loss in the treatment group that was only partially reversible after cessation of treatment. The rationale for that trial is that sodium channel blockers may have a neuroprotective effect on demyelinated axons. However, any potential benefit could have been negated by the inhibition of protective macrophage-microglia responses.

Macrophages can mediate recovery from tissue injury by at least 2 mechanisms: phagocytosis of debris and apoptotic cells within lesions (13) and differentiation to an alternatively activated phenotype (35). In a simplistic sense, macrophages have been classified as M1 inflammatory or M2 alternatively activated cells. Interferon-γ mediates polarization to an M1 phenotype and significantly enhances phagocytic activity (36), whereas interleukin 4 and interleukin 13 polarize cells to alternatively activated macrophages. Alternatively activated macrophages demonstrate anti-inflammatory and regenerative properties in mouse EAE models and express high levels of arginase (30, 37). Alternatively activated macrophages within lesions mediate these effects through local release of cytokines and regulation of tissue repair (38). Although this classification scheme has been useful in determining the role of cell subsets in cellular function, macrophage phenotype in vivo is more complex and a range of intermediate phenotypes exist that are clinically relevant (39).

It is unclear how NaV1.5 regulates macrophage phenotype at a cellular and molecular level. Persistent expression of NaV1.5 and alterations in intracellular calcium signaling (10) may enhance polarization to an arginase-positive phenotype. In addition, modeling of mouse macrophage differentiation suggests that conversion of arginase-positive cells to an M1 phenotype readily occurs in inflammatory condition, but M1 polarization appears to be a more stable phenotype (40). Macrophage NaV1.5 may enhance the maintenance of the arginase-positive phenotype within inflammatory lesions and overcome some of the limitations of immune regulatory cell plasticity (41). Although additional characterization of these molecular mechanisms is required before clinical translation, therapeutic use of macrophages requires the identification of specific cell subtypes that are therapeutically relevant. Use of nonselected macrophage populations could potentially lead to macrophage activation syndrome (42) or a sepsis-like process; the development of severe EAE in mice treated with unselected WT BMDMs in our experiments demonstrates some of the challenges of cell therapy (Fig. 10B). Stability of phenotype in vivo will be a critical criterion in the development of this therapeutic approach.

Despite protective endogenous mechanisms, chronic progressive injury continues in many MS patients and pathologic inflammation persists within existing lesions. From a clinical perspective, 2 unanswered questions are “Why are these repair mechanisms defective in many patients with neurologic disease?” and “How can these repair mechanisms be enhanced?” Novel strategies are required to deliver sufficient numbers of therapeutically effective cells to disseminated lesion sites, maintain their pro-repair and anti-inflammatory phenotype, minimize adverse events, and avoid the use of long-term immune suppressive treatments. Our long-term goal is to develop human NaV1.5-positive macrophages as a cell-based therapy that meets these criteria.

Acknowledgement

We thank Gouri Chatterjee for assistance with cDNA cloning and Toshi Kinoshita for help with hematoxylin and eosin staining. We acknowledge the assistance of Khen Macilvay and the University of Wisconsin Carbone Cancer Center Flow Cytometry Laboratory with acquisition of flow cytometry data.

Footnotes

  • The contents of this article do not represent the views of the Department of Veterans Affairs or the US Government.

  • This work was supported by a VA Merit Award from the BLR&D service to Michael Carrithers and the University of Wisconsin.

  • The nucleotide sequence reported in this paper has been submitted to the Gen-BankTM/EBI Data Bank with accession number(s) KC858891.

  • Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal's Web site (www.jneuropath.com).

References

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
View Abstract