Neutralization Of Class 3 Semaphorins

Published Date: 02 Nov 2017

Student: Renée R.C.E. Schreurs

Supervisor: Vasil Mecollari

ABSTRACT Spinal cord injury is an expensive and debilitating condition that has already affected millions of sufferers worldwide. Injury to the central nervous system (CNS) causes the formation of regeneration barriers such as the glial scar that hamper regrowth of axons through the scar tissue via the expression of axon growth-inhibiting molecules. Class 3 secreted semaphorins (Sema3s) are an example of such molecules and the focus of this study. The interaction between Sema3s and one of its receptors, neuropilins (NRPs), at the site of CNS injury has been implicated in the inability of axon to regenerate. This study aimed to investigate whether the repulsive action of Sem3s in spinal cord regeneration can be reduced by utilizing a strategy that neutralizes the different Semas with NRP1 and NRP2 receptor bodies. Female rats received a dorsal partial hemisection lesion at C4, resulting in a left-forelimb deficiency, and simultaneous injections at the injury site with one of six NRP receptor body treatments. The recovery of their locomotion was then tracked for twelve weeks using the BBB, FLAS, and novel CatWalk method. Subsequently the animals’ red nucleus was radioactively traced in order to track regenerating axons. Furthermore, the rat’s spinal cords were subjected to histology. The results indicate that treatment of the pinal cord with soluble NRP receptor bodies leads to improved behavioral outcome, but no clear selectivity among the different isoforms is evident. Furthermore, the Sema-NRP interaction effect seems not to be exclusive to axon regeneration, but seems to also interact with other cell types that leads to a difference in the extent of tissue sparing, may play a part in the modulation of the immune response, and changes the molecular composition of the scar environment.

1. Introduction

Traumatic injury to the spinal cord is a debilitating condition with limited chances of recovery that has already affected the lives of millions of people. Most commonly sustained through falls, sporting activities, violent attacks, and vehicle accidents, sufferers are often young and in need of medical care for the remainder of their lives1. Although injured peripheral nerves have the ability to regenerate, injury to the central nervous system (CNS) induces tissue damage, which creates regeneration barriers2,3. One of the main barriers is the glial scar4,5.

The glial scar consists of two main zones, the lesion core and the surrounding area. After injury to the CNS, an inflammatory reaction is triggered. Consequently, the surrounding neural tissue secretes various cytokines and chemokines, activating astrocytes, microglia, and oligodendrocyte precursor cells, all of which constitute the glial scar around the lesion site3-6. Over time, fibroblasts will begin to intrude, proliferate, and secrete extracellular matrix molecules to form the fibrotic center of the scar7. Although scar formation plays an important role in protecting the damaged tissue and limiting further cellular degeneration, the cells forming these scars secrete axonal growth-inhibiting molecules, which are believed to prevent axonal regeneration and functional recovery in the injured CNS3,7,8.

Intrinsically, CNS neurons have the ability to regenerate injured axons when exposed to the appropriate extracellular signals9. During development, axons are guided to their appropriate targets by both attractive and repulsive mechanisms. Interestingly, the axonal growth-inhibiting molecules believed to prevent regeneration after CNS injury are in many instances the same as the repulsive axon guidance cues expressed during development10. For instance, the fibrotic scar expresses various potent axonal growth-inhibitory molecules that act as chemical barriers for axon regeneration11. These include chondroitin sulfate proteoglycans, tenascin, and semaphorins7,10-12, the latter of which are the main focus of this study.

The semaphorin family of axon guidance molecules consists of at least twenty members, divided over eight classes13. Classes 1 and 2 represent invertebrate semaphorins. Classes 3 to 7 represent vertebrate semaphorins. Class 8 represents viral semaphorins (fig. 1). Of the vertebrate classes, 4 to 7 contain membrane-anchored semaphorins, and class 3 contains secreted semaphorins14. All semaphorins share a ‘sema domain’; a conserved 500 amino acid motif14,15.

Semas.jpg

Fig. . Schematic representation of semaphorins. The semaphoring family can be divided into eight classes. Classes 1 and 2 represent invertebrate semaphorins, classes 3-7 represent vertebrate semaphorins, and class 8, or V, depicts the viral semaphorins. Several conserved domains are present: an N-terminal sequence (SS), the semaphorin domain (sema), an immunoglobulin domain (Ig), a C-terminal basic domain (BD), thrombospondin repeats (TSP), and a glycosyl phosphatidyl inositol (GPI) anchor (adapted from De Wit & Verhaagen, 2003)14.

Presently, there are seven vertebrate secreted class 3 semaphorins known (termed Sema3A-G)16. Sema3s signal through neuronal receptor complexes that contain neuropilins (NRPs), a ligand-binding subunit, and plexins, a signal transducing component. Membrane-bound semaphorins do not require NRPs as co-receptors but bind directly to plexins17. Although the neuropilin family consists of two members (NRP1 and NRP2) and the plexin family of nine, neither neuropilins nor plexins are specific for individual secreted semaphorins14. NRP1 binds all Sema3s, whereas NRP2 binds to all but one Sema3, Sema3A. Moreover, NRPs also serve as receptors for members of the vascular endothelial growth factor (VEGF) family of angiogenesis18. Despite the assorted binding methods between the family members, Sema3-responses are thought to be mediated through specific combinations of NRP1 and NRP214.

Examination of neural scar tissue has revealed that meningeal fibroblasts invading the lesion core express repulsive Sema3A, Sema 3B, Sema3C, Sema3E, and Sema3F17,19. Following CNS injury, the two major descending motor pathways, the corticospinal tract (CST) and the rubrospinal tract (RST), which continuously express NRPs, fail to enter the semaphorin-positive portion of the neural scar, thus hampering regeneration. Moreover, ascending projections of dorsal root ganglion cells also fail to penetrate the region after CNS damage19. Since the semaphorin-neuropilin interaction seems to play a fundamental role in the inhibition of axon regeneration after spinal cord injury, this study aimed to neutralize the repulsive effect of Sema3s by introducing NRP-soluble receptor bodies into the scar-inducing tissue.

Primary fibroblasts were genetically modified with a lentiviral vector to express function blocking semaphorin-receptor bodies (NRP-bodies) for secreted Sema3s. It is hypothesized that these NRP-bodies modulate or dampen the repulsive action that occurs when NRPs, expressed in the CST and RST, come into contact with the semaphorin-positive lesion core by scavenging the secreted Sema3s.

Thus, in this study we aimed to neutralize the repulsive effects of neural scar-secreted Sema3s in vivo, via a new approach that utilizes NRP-soluble receptor bodies. Rats received dorsal partial cervical lesions, resulting mainly in the dysfunction of one forelimb, after which the injury site was injected with one of six NRP-body treatments. Furthermore, the red nucleus, a midbrain structure from where the RST axons emerge20, was stereotactically injected in order to trace the regenerating axons. Behavioral and histological analysis was used to assess regeneration over a 12-week observation period.

2. Materials and Methods

2.1 Animals and housing

For the purpose of these experiments, 60 female Fischer 344 rats (200 g; Harlan Laboratories, Inc) were used. The animals arrived one week prior to the start of the experiment to allow for acclimatization and handling. Throughout the experiment they were housed in group cages, in a temperature and humidity controlled room, under a 12:12 hour light/dark cycle with food and water ad libitum. The well-being of the animals was monitored daily (see supplementary Materials and Methods for a more detailed description of animal welfare monitoring) and documented by the responsible researcher. Eight animals were euthanized before the end of the experiment due to severe autotomy (see supplementary Materials and Methods), thus completing the experiment were n=9 for GFP, NRP2WT, NRP1Y297A, and NRP1VEGF, and n=8 for NRP1WT, and NRP1T316R. All experimental procedures were performed in accordance with the laws and regulations set by the European Union and were approved by the Animal Experimentation Committee KNAW (DEC# 2009.11).

2.2 Experimental set-up

2.2.1 In vivo application of NRP bodies in C4

All animals received a cervical (C4) unilateral RST lesion one week prior to the start of behavioral testing. Surgery was performed using ketamine (0.1 ml/100 g i.m.) and xylazine (0.05 ml/200 g i.m.) while body temperature was maintained at 37°C using a heating pad. After shaving the skin of the back, an incision was made in the cervical region and a laminectomy was performed at level C4. Subsequently, the dura mater was opened and a unilateral lesion of the RST was performed using a micro-knife. Immediately after the lesion was performed, the animals received 3 injections of primary fibroblasts. Three pressure injections of 1 μl (106 cells) of primary fibroblasts (harvested from a donor rat) were administered, genetically modified with a lentiviral vector to express one of six different NRP-bodies: a wild type NRP1 (NRP1WT), a wild type NRP2 (NRP2WT), a NRP1-body binding only to VEGF (NRP1VEGF), a NRP1-body binding only to Sema3s (NRP1Y297A), a double-mutant NRP1-body not capable of binding anything (NRP1T316R), and a control-group with green fluorescent protein (GFP). The primary fibroblasts were placed in the center of the lesion site and at 1 mm proximal and distal to the lesion core, using a glass capillary (tip diameter 60 μm; 0.2 μl/min). After infusion and gentle withdrawal of the glass capillary, muscle and skin were sutured using 5/o resorbable Vicryl sutures. After suturing, the animals received a 2 ml subcutaneous injection with saline in the flank to compensate for blood and general body fluid loss. Subsequently, the animals were placed in an incubator at 37°C and returned to their home cages following full awakening.

2.2.2 Stereotactical neuroanatomical tracing of the RN

All animals underwent tracing surgery following completion of the behavioral tests (exactly 11 weeks after receiving the C4 lesion). General anesthesia was induced by isoflurane inhalation (5% isofluorane mixed in pure oxygen) in a Plexiglas induction chamber. Anesthesia was maintained with an airflow-system, using a 2-3% isoflurane mixture that was supplied to the animals through a breathing mask fitted over the nose, while body temperature was maintained at 37°C using a heating pad. Afterwards, their scalp was shaved and their skin opened to expose the skull. A small hole was drilled unilaterally in the skull at 5.4 mm posterior and 0.7 mm lateral to Bregma, while the animal was held in a stereotactical frame in horizontal position. Subsequently, 0.5 μl of a 10% solution of biothinylated dextran amine (BDA, 10 000 MW) was pressure injected using a glass capillary (tip diameter 60 μm; 0.2 μl/min) into the red nucleus (RN) at 6.6 mm depth from the dorsal surface of the brain. After withdrawal of the capillary, the skin of the skull was sutured using 5/o resorbable Vicryl sutures and the rats were allowed to recover in a separate cage.

2.3 Behavioral testing

2.3.1 BBB open field locomotor test

A modification of the Basso, Beattie, and Bresnahan (BBB) rating scale21 was used to assess hindlimb function in an open field (50x60 cm). The BBB locomotor test was developed to evaluate the severity of bilateral contusion injuries at the thoracic level. In our lesion model, the rats received unilateral cervical lesions affecting only one side. The test was thus modified to assess the BBB score only on the affected (left) side. The rats were observed for 3 minutes by two trained individuals and scored from 0 (no observable movement) to 21 (normal locomotion). The animals were observed three days post-surgery, then weekly for six weeks, and bi-weekly the six weeks thereafter. Parameters include: weight support, plantar/dorsal stepping, coordination, toe clearance, predominant paw position during initial contact/lift off, stability, and tail position (for a more detailed description, see Basso et al., 1995)21.

Fig. 2. Example of a rat’s paw prints during one runway crossing on the CatWalk. Prints are labeled automatically. Note: RF = right forelimb, LF = left forelimb, RH = right hindlimb, LH = left hindlimb.

Catwalk example.jpg

2.3.2 CatWalk gait analysis

The rats were subjected to gait assessment with the CatWalk 10.0 XT (Noldus Information Technology) automated gait analysis runway. First, pre-operative baseline measurements were obtained. Subsequently, the rats were tested weekly from one to six weeks post-surgery, and bi-weekly for the next six weeks (in all cases, 3 complete runs per animal were obtained). In a darkened room, the animals were individually placed on the left side of a runway consisting of a glass surface (130x25 cm) and black plastic walls (spaced 8 cm apart). The rats used in this study were pre-trained to cross the walkway during their week of acclimatization and quickly learned the task without external motivation.

The technology of the CatWalk is such that only at those points where the animal makes contact with the glass plate, light exits the floor and scatters, illuminating the points of contact only and recording these with a high speed color camera (fig. 2). The intensity of the emitted signal depends on the degree of contact between the paw and the glass plate and increases with applied pressure. The signal is digitized and transferred to the CatWalk computer program where many parameters of the recorded gaits can be studied. The following parameters are of most interest in our C4 lesion model:

Intensity, the mean pressure exerted by one individual paw during floor contact, during the whole crossing of the walkway. Expressed in arbitrary units.

Print area, the total floor area contacted by the paw during stance phase. Expressed in cm2.

Swing, the time interval between two consecutive paw placements of the same paw. Expressed in seconds.

Stride length, the distance between two consecutive paw placements of the same paw. Expressed in pixels.

Stance duration, the absolute duration of the stance phase that depends on the animal’s walking speed. Expressed in seconds.

Duty cycle, the ratio between stance duration and step cycle duration. Expressed in %.

Step pattern, there are a total of six different step sequences that a rat can use as it places its paws one after another (table 1).

Table 1

Limb sequences in regular step patterns

Category Sequence

Cruciate RF-LF-RH-LH or LF-RF-LH-RH

Alternate RF-RH-LF-LH or LF-RH-RF-LH

Rotary RF-LF-LH-RH or LF-RF-RH-LH

Note: RF = right forelimb, LF = left forelimb,

RH = right hindlimb, LH = left hindlimb (adapted

from Vrinten and Hamers, 2003).

Regularity index (RI), calculated from the number of normal step sequences and the total number of steps according to the following formula:

Base of support (BOS), the distance between either the two front paws or the two hind paws.

2.3.3 Forelimb locomotor assessment scale

To assess forelimb use during locomotion in rats injured at the cervical level, the forelimb locomotor assessment scale (FLAS) was developed22. The rats were individually placed on the right side of a clear alleyway (plexiglass; 90x10 cm) and video recordings of the animals traversing the alleyway were made with an HD digital camera (3 videos of at least 4 consecutive step cycles per animal per session). The animals were later graded on a scale that measures movements of shoulder, elbow, and wrist joints, forepaw position and digit placement, forelimb-hindlimb coordination, compensatory behaviors adopted while walking, and balance. Scores range from 0-64. The rats were tested weekly for the first six weeks post-surgery and bi-weekly thereafter.

2.4 Histological processing

2.4.1 Transcardial perfusion and tissue preparation

To allow for histological processing, the animals were sacrificed. After deep anaesthetization with sodium pentobarbital (Nembutal) 0.11 ml/100 g i.p. (30 mg/kg body weight), the animals were transcardially perfused with saline followed by 4% paraformaldehyde in 0.1 M phosphate buffer pH 7.4 (PFA). Brain and spinal cord were dissected and post-fixed overnight in 4% PFA at 4°C. Subsequently, the tissue was treated overnight (at 4°C) with 0.25 M EDTA and 30% sucrose in TBS to enhance tissue penetration and for cryoprotection. The spinal cords were then cut into four sections: cervical C1, C2-C6, C7, and the remainder. The cervical sections were rapidly frozen in dry-ice-cooled isopentane and stored in -80°C until use. The brains and remaining spinal cord sections were stored in TBS at 4°C and have not been used in this experiment.

2.4.2 Cryosectioning

The cervical spinal cord sections C1 and C7 were cut into 20 μm thick transverse sections using a cryostat (Leica CM 3050S) at -21°C. The sections were then thaw-mounted on Superfrost Plus microscope slides (Menzel Gläser, Germany), dried at room temperature overnight and stored in -80°C until use. Cervical sections C2-C6 were not used in this experiment.

2.4.3 Immunohistochemistry

The microscope slides containing cervical sections C1 and C7 were removed from the -80°C freezer and thawed at 37°C for at least one hour. Subsequently, the sections were rehydrated in PBS and post-fixed in 4% PFA for 10 minutes. The slides were then washed 10 minutes twice in 0.2% Triton X-100 in PBS. After washing, the sections were blocked for one hour in 0.4% Triton X-100 and 10% goat or calf serum in PBS (blocking mix). Afterwards, the sections were incubated overnight at 4°C with the primary antibody, rabbit-anti-GFAP (anti-glial fibrillary acidic protein; marker for mature astrocytes; Dako), in a 1:500 dilution in blocking mix. The next day, the sections were washed 10 minutes twice in 0.2% Triton X-100 in PBS and once in PBS-only. After washing, the sections were incubated with the secondary antibodies anti-Streptavidin-Cy3 (reacts with BDA; 1:700), anti-rabbit IgG-DyLight488 (1:700), and Hoechst 33258 (stains DNA; 1:1000), for 1.5 hours at room temperature. Lastly, the sections were washed three times 5 minutes in PBS, coverslipped with Mowiol and left to solidify at 4°C.

2.5 Image analysis

Three C1 and C7 images of complete sections (10x magnification) were captured for each animal with a fluorescent microscope (Axiovert 25 CFL Inverted Microscope, Zeiss). Subsequently, the area (mm2) of each hemisphere (contralateral and ipsilateral to the lesion site) was quantified using software written by institute personnel.

2.6 Statistics

Two-way ANOVAs with Bonferroni post-hoc analyses were performed when comparing the BBB mean scores in individual weeks. BBB score trajectories, BBB subscore, all Catwalk trajectories and the FLAS scores were analyzed using One-way ANOVAs with Bonferonni multiple comparisons post-hoc tests. The general area differences between C1 and C7 hemisections were analyzed using an unpaired Student’t t-test, and the area differences and tissue reduction scores were compared using One-way ANOVAs with Bonferroni multiple comparison and Bartlett’s test for equal variances.

3. Results

3.1 BBB

The BBB procedure was modified by solely observing and scoring limb function ipsilateral to the lesion site. All rats had BBB scores of 21 prior to injury. Three days after injury, the mean BBB score decreased to 12.9 ± 0.6 (mean ± s.e.m.) for the GFP-group, 13.6 ± 0.4 for the NRPVEGF-group, 15.0 ± 0.9 for the NRPY297A-group, 14.0 ± 0.5 for the NRPT316R-group, 15.0 ± 0.9 for the NRP1WT-group, and 15.8 ± 1.5 for the NRP2WT-group. The mean BBB-score did not improve throughout the experiment, but decreased further for each group to 11.3 ± 0.2 for GFP, 13.6 ± 1.2 for NRPVEGF, 12.1 ± 0.3 for NRPY297A, 11.5 ± 0.3 for NRPT315R, 12.6 ± 0.7 for NRP1WT, and 13.0 ± 1.0 for NRP2WT (see figure 3a). Although by week 12 the mean BBB scores of the NRP1VEGF-, the NRP1Y297A-, and the NRP1WT-groups are significantly higher than the score of the GFP-group (p<0.05), there exists little qualitative difference between these scores; BBB scores between 11 and 13 indicate frequent to consistent weight supported plantar steps, but with deficits in forelimb-hindlimb coordination21. These results are in line with a persistent deficiency in forelimb functioning. However, Koopmans and colleagues23 note that the correct assessment of forelimb-hindlimb coordination by observers is a serious limitation within the BBB locomotor rating scale.

To overcome this limitation, they suggest to assess coordination based on the regularity index (RI), achieved through implementing the CatWalk. In their study, animals with a RI of 100% in all three CatWalk crossing were considered as having consistent coordination. Animals with a RI of 100% in two out of three crossings were considered as having frequent coordination. Animals with a RI of 100% in one out of three crossings were considered as having occasional coordination. If none of the three crossings had a RI of 100%, absence of coordination was concluded. However in our sample, some animals did not display consistent coordination during their baseline measurement when assessed using the RI method. The lowest RI during baseline measurements was 92.3%. Therefore, we used this as our cut-off point to assign a new BBB coordination score as follows: RI of >92.3% in all three crossings is considered consistent, RI of >92.3% in two out of three crossings is considered frequent, RI of >92.3% in one out of three crossings is considered occasional, no RI >92.3% in all three crossings is considered absence of coordination.

all BBB.jpg

Fig. 3. Schematic representation of BBB scores. (a) BBB scores of individual animals grouped per treatment on day 3, week 6, and week 12 post-surgery. By week 12, NRP1VEGF, NRP1Y297A, and NRP1WT score higher than GFP (all p<0.05). (b) BBB score trajectory during the course of the experiment using the regular observation21 method and a new method implementing the regularity index (RI) as suggested by Koopmans and colleagues23. Before RI implementation, NRP1VEGF, NRP1Y297A, NRP1WT, and NRP2WT score higher than GFP (all P<0.01). (c) BBB Subscore trajectory during the course of the experiment. NRP2WT scores higher than GFP (p<0.01).

Using the regular method of observing coordination, it was found that the NRPVEGF-, NRPY297A-, NRP1WT-, and the NRP2WT-group, but not the NRP1T316R-group, show improved left forelimb function over time compared to the GFP-group (p<0.01) (see figure). However, using the more reliable RI-method to assign a coordination score, no significant differences in BBB score over time could be found (see figure 3b).

Lastly, the BBB subscores of the left forelimb were analyzed. The BBB subscore is a 7-point scale within the BBB that evaluates toe clearance, paw position, stability, and tail position, with 7 representing normal function24. It allows the evaluation of recovery in animals with inconsistent forelimb-hindlimb coordination. These scores show improved function in the NRP2WT-, compared to the GFP-group (p<0.01) (see figure 3c).

3.2 Catwalk gait analysis

3.2.1 Intensity of the left forelimb

The intensity measure reflects the mean pressure exerted by one paw during floor contact. Mean baseline intensity of the left forelimb was between 69.25 ± 1.2 (mean ± s.e.m.) (a.u.) and 71.2 ± 0.6. Six weeks post-surgery, intensity was reduced to between 27.8 ± 0.5 and 39.9 ± 1.9 and did not improve throughout the experiment; at week 12, mean intensity was between 28.5 ± 0.9 and 32.5 ± 2.1. The NRP1VEGF-group exerted more pressure on their left forelimb than the GFP-group during the course of the experiment (p<0.05; see figure 4).

LF Intensity (all).jpg

Fig. 4. Graph depicting the intensity (expressed in a.u.) of the left forelimb (LF) during the course of the experiment. NRPVEGF showed improved intensity over time (p<0.05), but by week 12 this effect is diminished.

3.2.2 Print area of the left forelimb

Print area indicates the total floor area contacted by the paw during stance phase. Mean baseline print area of the left forelimb was between 1.4 ± 0.2 (mean ± s.e.m.) (cm2) and 1.7 ± 0.1. Six weeks post-surgery, print area was reduced to between 1.0 ± 0.5 and 1.1 ± 0.4. Thereafter, mean print area increased to between 1.5 ± 0.2 and 2.8 ± 0.4 at week 12, a mean contacted floor area much greater even than baseline. However, this surge in print area seems to be due to software difficulties concerning the CatWalk program (see discussion). During the course of the experiment, no group shows improved print area of the left forelimb compared to GFP (data not shown).

3.2.3 Swing of the left forelimb

The swing measure reflects the time interval between two consecutive paw placements of the same paw. Mean swing of the left forelimb was 0.11 ± 0.00 (mean ± s.e.m.) (s) during the baseline measurements. At six weeks post-surgery, the mean swing had increased to between 0.13 ± 0.01 and 0.21 ± 0.02, but was restored to between 0.08 ± 0.01 and 0.19 ± 0.03 at week 12. During the course of the experiment, all groups, excluding NRP1WT, show improved swing compared to the GFP-group (p<0.01) (see figure 5). In week 12 individually, all groups show improved swing compared to GFP (p<0.01), with NRP1T316R, NRP1WT, and NRP2WT seemingly returning back to baseline-level.

LF Swing (all).jpg

Fig. 5. Graph depicting the swing (expressed in seconds) of the left forelimb (LF) during the course of the experiment. All groups, excluding NRP1WT, show improved swing over time compared to GFP (p<0.01), and by week 12, all groups show this effect (p<0.01).

3.2.4 Stride length of the left forelimb

Stride length represents the distance between two consecutive paw placements of the same paw. At baseline, mean stride length of the left forelimb was between 14.7 ± 0.4 and 15.5 ± 0.3 (mean ± s.e.m.) (cm). Six weeks post-surgery, mean stride length decreased to between 8.1 ± 0.6 and 10.5 ± 1.1, and had decreased even further by week 12 to between 7.7 ± 0.7 and 10.4 ± 0.9. During the course of the experiment, no group shows improved stride length of the left forelimb compared to GFP (data not shown).

3.2.5 Stance duration of the left forelimb

Stance duration reflects the absolute duration of the stance phase that depends on the animal’s walking speed. The mean stance duration during baseline measurements was between 0.14 ± 0.01 and 0.16 ± 0.01 (mean ± s.e.m.) (s). Six weeks post-surgery, mean stance duration had decreased to between 0.08 ± 0.01 and 0.10 ± 0.01, but restored to between 0.11 ± 0.01 and 0.13 ± 0.01 by week 12. All groups show a similar trajectory of stance duration during the course of the experiment, but only NRP2WT shows significant improvement compared to GFP (p<0.001) (see figure 6), and none turned back to baseline-level.

Fig. 6. Graph depicting the stance duration measurement (expressed in seconds) of the left forelimb (LF). NRP2WT shows improved stance duration over time compared to GFP (p<0.001).LF Stance duration (all).jpg

3.2.6 Duty cycle of the left forelimb

The duty cycle measurement expresses the ratio between stance duration and step cycle duration. Thus, it expresses the percentage of time that a paw is in contact with the floor during one step cycle. During baseline measurements, the mean duty cycle was between 53.4 ± 1.1 and 56.0 ± 1.5 (mean ± s.e.m.) (%). Six weeks post-surgery, mean duty cycle was decreased to between 28.9 ± 2.7 and 41.0 ± 2.3, but restored to between 38.2 ± 2.8 and 57.2 ± 2.7 by week 12. NRP1Y297A, NRP1WT, and NRP2WT show improved duty cycle compared to GFP (p<0.01, p<0.05, and p<0.001, respectively), with NRP1Y297A and NRP1WT returning back to baseline-level (see figure 7).

LF Duty cycle (all).jpg

Fig. 7. Graph depicting the duty cycle (expressed in %) of the left forelimb (LF). NRP1Y297A, NRP1WT, and NRP2WT show improved duty cycle compared to GFP (p<0.01, p<0.05, and p<0.001, respectively).

3.2.7 Step pattern

There are a total of six different step patterns a rat can use (see table 1) as it places its paws during locomotion. At baseline, the animals predominantly use an alternate pattern, but as the experiment progressed, all patterns were used with no clear preferance (see figure 8).

step patterns.jpg

Fig. 8. Graphs depicting the step patterns used by the animals at baseline and during week 8. At baseline, the dominant stepping pattern is AB with little variation, whereas in week 8 all patterns are used with no clear preference. Note: CA = cruciate pattern A, CB = cruciate B, AA = alternate A, AB = alternate B, RA = rotary A, and RB = rotary B.

3.2.8 Regularity index (RI)

The RI is calculated from the number of normal step sequences and the total number of steps (see section 2.3.2). At baseline, regularity index was between 98.5 ± 0.8 and 100 ± 0.0 (mean ± s.e.m.) (%). Six weeks post-surgery, RI was decreased to between 68.9 ± 4.3 and 79.8 ± 2.5, and remained on this level throughout the experiment. All groups, excluding NRP1T316R, showed an improved RI compared to GFP (between p<0.05 and p<0.001), but no group returned to baseline-level (see figure 9).

Fig. 9. Graph depicting the regularity index over time. During the course of the experiment all groups, excluding NRP1T316R, show an improved regularity index compared to GFP (between p<0.05 and p<0.001).Regularity Index (all).jpg

3.2.9 Base of support

The BOS is the distance between either the two front paws (FL BOS) or the two hind paws (HL BOS). At baseline, FL BOS was between 1.73 ± 0.08 and 1.96 ± 0.06 (mean ± s.e.m.) (cm) and HL BOS between 2.45 ± 0.09 and 2.75 ± 0.07. Six weeks post-surgery, FL BOS had dispersed to between 1.66 ± 0.13 and 2.12 ± 0.11 and HL BOS to between 2.37 ± 0.11 and 2.95 ± 0.09. By the end of the experiment, FL BOS remained on a similar level, between 1.56 ± 0.11 and 2.20 ± 0.12, while HL BOS increased slightly to between 2.44 ± 0.10 and 3.23 ± 0.11. Throughout the course of the experiment, NRP1VEGF, NRP1Y297A, and NRP1T316R show an increase FL BOS compared to GFP (p<0.01, p<0.05, and p<0.001, respectively) (see figure 10a). NRP1Y297A, and NRP1T316R show a similar increase in HL BOS compared to GFP (p<0.001, and p<0.05, respectively), as does NRP2WT (p<0.05) (see figure 10b).

Fig. 10. Graphs depicting the BOS of both fore- and hindlimbs. (a) BOS of the FL during the course of the experiment. NRP1VEGF, NRP1Y297A, and NRP1T316R increased BOS compared to GFP (p<0.01, p<0.05, and p<0.001, respectively). (b) BOS of the HL during the course of the experiment. NRP1Y297A, NRP1T316R, and NRP2WT increased BOS compared to GFP (p<0.001, and p<0.05 for the latter two).BOS.jpg

3.3 FLAS

Per animal, three video recorded runs were scored. During the course of the experiment, all goups, excluding NRP1T316R, show improved forelimb function compared to GFP (between p<0.05 and p<0.001) (see figure 11).

Fig. 11. Graph depicting the mean FLAS scores over time. During the course of the experiment, NRP1VEGF (p<0.001), NRP1Y297A, NRP2WT (both p<0.01), and NRP1WT (p<0.05) show improved LF locomotion compared to GFP.FLAS score (all).jpg

3.4 Image analysis: area quantifications

Three sections each of C1 (see figure 12) and C7 (see figure 13) were quantified for area of the hemisection contralateral and ipsilateral to the lesion site. The general area (mm2) of the hemisections contralateral to the lesion site were larger in both C1 and C7 (both p<0.001) (data not shown). There were no between group differences in either area or reduction score (percentage of ipsilateral tissue loss, compared to contralateral tissue loss; see figure 14) of C1 or C7, but in the C7 sections the trend appears that all groups have a lower reduction score (and thus more tissue sparing) than GFP (see figure 14).

Fig. 12. Sample of images taken from the stained C1 sections (10x magnification) of each group. GFAP is in green, and BDA-binding SA-Cy3 is in red. The slightly enlarged image at the bottom is a blow-up of the NRP1T316R section, showing the ipsilateral and contralateral outlines that were done in order to quantify area of the respective hemisections.C1 10x.jpg

Fig. 13. Sample of images taken from the stained C7 sections (10x magnification) of each group. GFAP is in green, and BDA-binding SA-Cy3 is in red. The slightly enlarged image at the bottom is a blow-up of the NRP1VEGF section, showing the ipsilateral and contralateral outlines that were done in order to quantify area of the respective hemisections.C7 10x.jpg

Fig. 14. Graphs depicting the percentage of ipsilateral tissue reduction compared to contralateral tissue reduction. There are no statistical differences between groups, but the trend that GFP has more ipsilateral tissue loss compared to all the other groups appears in the C7-figure.reductyion scores.jpg

4. Discussion

This study aimed to investigate whether the repulsive action of Sema3s in spinal cord regeneration can be reduced by utilizing a strategy that neutralizes the different Semas with NRP1 and NRP2 soluble receptor bodies. The consistent tendency of treatment-groups to show an increased performance on the BBB, FLAS, and CatWalk, compared to the control group (GFP), indicates that treatment of the injured spinal cord with soluble NRP-bodies results in improved behavioral outcome, but the lack of (an) individual treatment group(s) dominating in improved performance suggests there is no clear selectivity among the different isoforms. Furthermore, the histological analyses indicate that the Sema-NRP interaction effect is not exclusive to axon regeneration, as there seem to be many qualitative differences between groups that cannot be explained by axon regeneration alone.

The BBB results obtained by observation alone show that throughout the experiment, all groups excluding NRP1T316R, score higher than GFP. Higher BBB scores generally indicate improved locomotion of the hind limb. However, this difference in score, though significant, lies between 11 and 13, indicating frequent to consistent weight supported stepping with deficient coordination. In other words, coordination is the determining factor in mean score. Nothing much changes after the BBB coordination score is substituted with the CatWalk RI. Though the differences are no longer significant, they still lie in the range of 11 to 13. Most studies utilizing a cervical hemisection lesion obtain similar scores25, but similar lesions at different heights may lead to full recovery of hindlimb function26. In any case, the most notable difference between the mean BBB scores with and without RI implementation is the trajectory. Without RI, the mean score seems to diminish a little more every week until it reaches its final range, whereas the RI graph shows a drop to between 11 and 13 from week 1 on. Substituting coordination with RI makes the scoring more reliable21, but in the end it all comes down to a coordination deficiency in our animals. Since coordination is largely dependent on the use of all limbs27, it could indicate that the left forelimb is still non-functional.

The BBB is not an optimal test for our lesion model, as it is more sensitive to severe lesions and hind limb dysfunction while our lesion aimed to target one forelimb. However, the hind limb ipsilateral to the lesion is also affected, though to a much lesser extent; immediately after surgery and throughout the observation period the animals showed a varied pattern of hind limb deficiency, most evident in inefficient toe clearance and intra- or extrarotated paw placement. Improved axon regeneration would thus not only improve forelimb function, but that of the hind limb as well. Another reason to include this test is as a means of comparison; the CatWalk and FLAS are both relatively new methods, whereas the BBB is a well established and widely used locomotion test in the field of spinal cord injury.

Not relying on observation, the novel CatWalk method is an easy-to-use paradigm that generates an immense amount of information that would otherwise require the utilization of several different tests. The parameters of interest in this study concerned either forelimb-specific data or measures of all limbs concerning balance, weight distribution and coordination. Although our results showed that no group predominantly outperforms any others, NRP1VEGF, NRP1Y297A, and NRP2WT most often claim this position in individual parameters. This is surprising as NRP1WT and NRP2WT were expected to have the biggest effects.

The intensity, stance duration, print area and duty cycle parameters are closely linked. The former two depict a form of floor contact that is not determined by size or shape of the limb, the second-to-last reflects the actual contacted floor area, and the latter is a ratio derived from stance duration (per step cycle duration). In this study, intensity generally decreased with NRP1VEGF being the only group with a greater intensity than GFP. Stance duration, print area and duty cycle followed a U-shaped trajectory with NRP2WT having a longer stance duration and NRP1Y297A and NRP2WT having a higher duty cycle than GFP. A decrease in intensity, printarea, stance duration, and duty cycle could indicate loss of function of the forelimb, unequal weight distribution, or an increase in pain in the affected limb. Moreover, a decrease in print area can indicate deformity of the paw. For instance, most animals lost the ability to extend the digits on their paw, causing the paw to permanently form a fist. Since the stance duration and duty cycle restored somewhat, while the intensity continued to drop, it seems likely that the animals regained a degree of function in their forelimb, but with inefficient weight distribution. Similar results in terms of forepaw contact in overground locomotion were found in a study that utilized a hemicontusion model at C528 and in the first weeks after sciatic nerve injury29, although here, most parameters returned back to baseline.

The print area results command a separate discussion as an enormous surge can be observed in weeks 10 and 12 that does not occur anywhere else. It was recently discovered that the CatWalk remembers the calibration settings of the previous runs independently of the settings used in an experiment. In our case, colleagues had just begun testing the equipment for mice while we were nearing the end of our sessions. The calibration settings for mice were mistakenly transferred to our data set in weeks 10 and 12, causing a surge in print area to above baseline level. Although a similar effect cannot be seen in other graphs, the validity of weeks 10 and 12 remains questionable.

The swing and stride length parameters do not affect floor contact, but are indicators of improved shoulder-joint movement. The inability to use a limb often results in it becoming stiff and rigid. Improved swing and stride length measurements thus indicate improved shoulder joint movement, less rigidity, and increased function. Interestingly, all groups excluding NRP1WT showed decreased swing time, while no group showed increased stride length compared to GFP. In this study, stride length is generally decreased, while swing improves, indicating that the left-forelimb steps get shorter and faster, as if the animals is ‘tapping’. Although this suggests that the animals can move their forelimb, it is not a natural stepping motion and does not indicate recovery. Gensel and colleagues30 mention a decrease in stride length of the affected forelimb after cervical spinal cord injury, but they did not assess, or do not mention their swing speed results, nor do they elaborate on the notion of ‘stepping’, although they did find a trend towards an increase in number of steps.

Another important indicator for recovery is coordination, which is translated into the RI parameter in the CatWalk. RI is the ratio of the total number of step sequence patterns to the actual number of paw placements. Our results show that the animals regress from using predominantly an alternate step pattern to using all six patterns in a varied manner (an effect also found by Dunham and colleagues28, and Hamers and colleagues31), while all groups showed an improved RI compared to GFP. Although the RI trajectory suggests improved coordination, the random usage of step patterns indicates otherwise. This result fits with the BBB scores where the animals show improvement, but cannot recover from deficient forelimb-hind limb coordination. Hamers and colleagues31 have shown a similar decrease in RI and BBB scores in the first weeks after thoracic transaction injury. However, the animals in their study progress further to reach RI baseline levels and BBB scores >13. Thus, their animals regained frequent or consistent coordination whereas ours did not.

The final CatWalk parameter we examined is the BOS. BOS is measured between either the two front paws, or the two hind paws. Our results show that NRP1VEGF, NRP1Y297A, and NRP1T316R have increased BOS of the forelimbs compared to GFP, while NRP1Y297A, NRP1T316R, and NRP2WT have increased BOS of the hindlimbs compared to GFP. Ideally, the BOS of both front and hind paws should return to baseline to indicate recovery. Any increase or decrease in BOS suggests deficiency; decreased BOS suggests that the paws are placed closer together and is detrimental for stability, increased BOS suggests that the paws are spaced further apart, indicating compensation for lack of stability. Increased BOS of the front paws could also indicate sliding motions made with the affected front limbs, which sometimes happens when the limb very rigid, but the animal can still move the shoulder joint, causing the limb to extend out to the side in a sliding motion. Furthermore, increased BOS of the hindlimbs can also indicate autotomy (see supplementary Materials and Methods), or extrarotated paw placement. In any case, deviations from baseline are undesirable, thus groups differing from GFP in BOS measurement do not necessarily indicate improved recovery. Closer inspection of the graphs reveals that NRP1Y297A and NRP2WT return to baseline level in forelimb BOS, and the same is true for two groups plus NRP1WT in hind limb BOS. Chuang and colleagues32 have also found that BOS increases and then steadily returns back to baseline levels in recovering animals after traumatic injury to the CNS.

The FLAS scores, specifically designed to investigate forelimb locomotion, indicate that although the scores diminish drastically throughout the experiment, all groups excluding NRP1T316R have improved forelimb function compared to GFP. The FLAS, like the CatWalk, is a novel method which has not been used much yet. However, it is a forelimb specific assessment scale that is more sensitive to the lesion model used in this study than is the BBB. Even though it is more sensitive, it has become apparent that it is not the most adequate method to study this specific lesion. For example, there were instances where two rats displayed qualitative movement differences that were not picked up by the FLAS; one animal swiftly moved across the plexiglass alley while another stumbled through with a lot more difficulties, yet both animals received the same (lowest) FLAS score. In the future we hope to eradicate this problem by including other tests such as the cylinder test33, where the animal’s forelimb function is examined through exploratory behavior, and the pellet retrieval test34, where the animal has to grasp at and retrieve pellets while climbing stairs.

Although NRP2WT, NRP1Y297A, and NRP1VEGF seem to be the most regular groups to show increased recovery compared to GFP, no group clearly dominates any others in this respect. The behavioral results point in the direction of improved outcome for non-control treatment groups, but there seems to be no clear selectivity for the different isoforms. However, there are a few interesting outcomes. Firstly, NRP1T316R sometimes significantly differs from GFP. This could indicate that the double-mutant isoform, a control-group like GFP, is capable of interactions that were not expected. Secondly, the NRP-VEGF interaction seems to play a larger part than expected. VEGF is an important factor in traumatic injury to the spinal cord. However, implications of its role are contradictory. For example, Widenfalk and colleagues35 have shown that administration of VEGF to the lesion site immediately after injury improves recovery and increases tissue sparing. Herrera and colleagues36 on the contrary, have shown that administration of external VEGF after injury does not lead to the sparing of more neurons and that, like the scavenging action of our soluble NRPVEGF-body, a reduction in VEGF is desirable for neuroprotection.

Apart from studying the behavioral outcome, the spinal cords of the animals were subjected to histological analyses. The area of the contralateral and ipsilateral hemisections of C1 and C7 were quantified, resulting in data that showed no between-group differences in area or tissue reduction for either C1 or C7. Since C7 is distal to the lesion site (C1 is proximal), we expected to find a treatment-related effect and what can be observed is a trend towards less ipsilateral tissue loss at C7 in all treatment groups compared to GFP. The largest effect will probably be in the quantification of the RST fibers that can be seen (in red) in the 10x magnification images (see figures 12 and 13). There appear to be striking group differences in number of fibers. In the C1 sections, GFP shows almost no RST fibers whereas especially NRP1Y297A and NRP1WT display a lot, possibly indicating different degrees of anterograde degeneration. No fibers are found anywhere in the C7 sections, indicating failed regeneration of axons. Future investigations will reveal the quantitative and qualitative differences between the sections.

Although histological analyses are still in progress, the main indication following the histological results is the probability that the Sema-NRP interaction effect is not exclusive to axon regeneration. Due to the before-mentioned promiscuous binding activities of both the Sema and the NRP families, it is very likely that the introduction of soluble NRP receptor bodies allows for interactions with other cell types that may lead to unintended results such as different degrees of tissue sparing. NRPs are implicated in several different pathways that control cell migration37. Thus, tampering with NRP and Sema concentration may lead to the recruitment of astrocytes, oligodendrocyte precursors and other cell types that alter the molecular composition of the scar environment and affect yet other unknown pathways. Moreover, a distortion in Semas and NRPs may lead to modulation of the immune response by attracting or repelling as T and B cells or promoting invasion, leading to massive monocyte and macrophage recruitment38. In any case, soluble NRP bodies probably cause alterations in systems far beyond axon regeneration alone.

In summary, treatment of the injured spinal cord results in improved forelimb recovery with no clear selectivity among the different isoforms, although NRP2WT and NRP1VEGF deserve further investigation. Histological analysis, though still in progress, shows similar results with respect to tissue sparing. Future investigations focusing on RST fiber quantification and evaluation of GFAP scar volume and vascularization as well as Sema and immune response staining should elucidate the different effects of the isoforms and strengthen the behavioral results.

5. Acknowledgements

The author would, first and foremost, like to thank Vasil Mecollari for his extensive supervision and guidance during this project. Secondly, the author wishes to express her gratitude towards Paul Lucassen for acting as co-assessor and UvA-representative, and Ruben Eggers, and Barbara Hobo for all their help with the animals. Finally, the author would like to thanks Joost Verhaagen and all his colleagues of the Laboratory for Neuroregeneration at the Netherlands Institute for Neurosciences for allowing me to take part in your studies and for creating such a welcoming and motivating work environment.