A novel model of optic nerve injury established by microsurgery using the pterional approach in cats
Correspondence Address: Source of Support: National Natural Science Foundation of
China (No. 30271333), Conflict of Interest: None DOI: 10.4103/0028-3886.82726
Source of Support: National Natural Science Foundation of China (No. 30271333), Conflict of Interest: None
Background : Experimental studies have yielded a wealth of information related to the pathological mechanism of optic nerve injury. However, there is no suitable animal model to study intracranial optic nerve injury. Aim : To establish an experimental model of acute optic nerve injury. Material and Methods : We established an animal model of acute intracranial optic nerve injury using the classic pterional approach in cats and investigated electrophysiological and ultrastructural changes. We applied immunohistochemical staining to examine the expression of glial fibrillary acid protein, neurofilament protein, myelin basic protein pre- and post-injury. Results : We successfully established an animal model of acute intracranial optic nerve injury. Conclusion :The pathological processes of acute optic nerve injury may involve the following series of steps. Direct mechanical injury of the optic nerve leads to the death of oligodendrocytes in the optic nerve, which consequently results in optic nerve demyelination. Following optic nerve injury, the astrocytes in the injured area die and produce excitatory amino acids, which have an adverse effect on neurons, resulting in the proliferation and activation of astrocytes. The astrocytes can absorb the glutamic acid and transform it into atoxic glutamine. The glutamic acid can then injure retinal ganglion cells, resulting in the reduction of neurofilament proteins in the axons. We believe that the application of the pterional approach to establish optic nerve injury animal models has both a clinical and theoretical basis.
Keywords: Animal model, cat, optic nerve injury, pterional approach
The optic nerve is one of the best-studied central nervous system (CNS) fiber tracts because of its relative homogeneity, accessibility, and the ease with which the axons can be traced neuroanatomically. It is difficult to obtain injured human optic nerves. Thus, it is necessary to establish a suitable animal model of optic nerve injury for research purposes. The pterional approach is one of the most common surgical approaches in neurosurgery. It is typically used to decompress the optic canal to treat acute optic nerve injuries associated with craniocerebral traumas. In this study, we created an experimental model of optic nerve injury by using the pterional approach, and examined the related pathology.
In the 1990s, scientists developed optic nerve injury models. ,,, The common method for establishing models first involves incising the conjunctiva laterally to the cornea. After separating the retractor bulbi muscles, the optic nerve is identified and exposed near the eyeball. The mobilized nerve is crushed 2 mm distal to the eyeball with a noninvasive vascular clamp, pincette, or a specially made microscopic occluding forceps for several seconds. Alternatively, the optic nerve is cut off in part or as a whole to establish partial or whole optic nerve trans-section models, respectively. ,,,,, In 1994, Haciyakupoğlu and et al .  established an optic nerve crushing injury model using the above surgical method and by guiding a balloon behind the eyeball through a trochar and injecting physiological saline to cause optic nerve injury. This injury was similar to that caused by an intraorbital tumor. It is often difficult to compare the results obtained by different investigators on optic nerve compression injuries because of differences in the pressure levels and application methods used. In 2006, Sarikcioglu attempted to standardize the method to crush optic nerves.  He introduced a new method to crush the intraorbital optic nerve that used a Yasargil aneurysm clip with a closing force of 1.82 N.
The injuries in the animal models described above are all limited to the intraorbital optic nerve. Even today, research on intracranial optic nerve injuries is limited because the microenvironments of intraorbital optic nerves are different from those of intracranial optic nerves. Therefore, it is of practical importance to establish an animal model that imitates intracranial optic nerve injuries as closely as possible.
Thirty-six healthy adult cats, of both genders and weighing 3.0 to 3.5 kg, were provided by the Animal Experimental Center of Fudan University. The cats were randomly divided into the control group (n = 6) or the experimental group (n = 30). The experimental group was composed of five subgroups with six cats per subgroup, which were observed at 6 h and 1 d, 3 d, 7 d and 14 d after injury. The entire experimental procedure was designed in accordance with Regulations for the Administration of Affairs Concerning Experimental Animals formulated by the Ministry of Science and Technology of the People's Republic of China.
The pterional approach was used to create acute optic nerve injury. The cats were anesthetized with an intraperitoneal injection of 20 g/L of pentobarbital sodium (30 mg/kg). The limbs were fixed in a lateral, recumbent position with the head towards one side. The access for intravenous infusion was established, and 800,000 IU of penicillin were added to 250 mL of saline. The skin was prepared and disinfected with tamed iodine, and an operative incision of approximately 3 cm was made along the line between the lateral canthus and the tragus [Figure 1]a. The scalp and aponeurosis were incised. Then the temporal muscle was pulled away with a retractor, and the periosteum was detached. The skull was drilled with a dental drill to enlarge the bone window to a size of 2 × 2 cm. The dura mater was electrically cauterized following thorough hemostasis. It was then incised in a star shape to the bone margin under microscope to drain the cerebrospinal fluid (CSF). The brain tissue was slowly elevated, and the white optic nerve could be viewed deep along the skull base [Figure 1]b. We examined whether there was vascular attachment, and the approximately 3-mm optic nerve was mobilized. The optic nerve was squeezed with noninvasive vascular clip [catalog no. W40130, Shanghai Medical Instruments (Group) Corp., Ltd., Surgical Instruments Factory; closing force, 20 g] for 20 s, and then the clip was removed. The optic nerve was washed with saline, and the skull was closed following hemostasis. Postoperatively, the pupillary size and the direct and indirect light reflexes were observed. As shown in [Figure 1]a-b, the procedure established a successful experimental model with a larger ipsilateral pupil and a loss of light reflexes. [Figure 1]c-d reveals that the operative incision via the pterional approach and exposure of the optic nerve which are commonly used in neurosurgery, are similar to those applied in animal studies.
Visual evoked potential recordings
The flash visual evoked potential (F-VEP) was tested in all of the cats before and after the experiment. Both the latency and amplitude were recorded. The cats were anesthetized with an intraperitoneal injection of 20 g/L of pentobarbital sodium (30 mg/kg). Then the limbs were fixed in a prone position with a scaffold made specifically for this purpose. The skin was prepared and disinfected, and a median incision of approximately 4 cm was made in the posterior occipital scalp. A vertical incision was made to the border between the anterior border roots of both ears at its midpoint. The skull was drilled at approximately 5 cm posterior to the midpoint on the vertical line (the location of the visual cortex), and a bone window of about 2×2 cm crossed over the midline. The dura mater was kept open, so the superior sagittal sinus could be viewed in the center of the dura mater. The silver recording electrode was kept in contact with the dura surface without causing damage, whereas the ground electrode was placed subcutaneously at the operated site. The upper and lower eyelids were propped open with an eye speculum. The recorded eye was positioned toward the light source at 45°, and the other eye was covered with black cloth. A current of 0.1 to 300 Hz was passed through the recording system tape, recorded for 250 ms, and overlapped 100 times. The brightness was 5 cd/(s·m 2 ), the brightness of background was 30 cd /(s· m 2 ), and the stimulus interval was 1 s.
Collection of samples
Cats in both the control and experimental groups were anesthetized with overdose of pentobarbital sodium (20 g/L), and an aortic cannulation via the left ventricle was performed after thoracotomy. The superior vena cava was opened to drain the perfusate, and perfused with 500 mL of saline followed by 25 g/L of glutaral phosphate buffer solution (PBS, 250 mL). The optic nerve sample was then removed immediately and placed in fixative solution. Optic nerves samples were prepared at different time points before the model was established and at 6 h, 1 d, 3 d, 7 d and 14 d after the model was established. The sections were stained and observed under electron and light microscopes.
Transmission electron microscopy
The optic nerve specimens were fixed in situ with a 4% paraformaldehyde-0.5% glutaraldehyde solution in PBS for 2 h at 4°C. After rinsing the specimens in PBS, some cells were post-fixed for 30 min with 1% osmium tetroxide in PBS. They were then dehydrated through a graded ethanol series, en bloc stained during dehydration with a saturated solution of uranyl acetate in 70% ethanol, and embedded in Araldite to perform ultrastructural studies of the cells. Serial ultra-thin sections were cut with a diamond knife and collected on nickel grids. All the specimens were examined with a Zeiss EM 900 electron microscope operating at 80 kV.
After being de-waxed and hydrated, the sections were placed in 10g/L H 2 O 2 for 10 to 15 min to remove the endogenous peroxydase and then incubated in confining liquid at room temperature for 5 min. Subsequently, the sections were incubated in a 37°C incubator for 2 h, with different antibodies against the glial fibrillary acid protein (GFAP, l:400, NeoMarkers), neurofilament protein (NF, l:400, NeoMarkers), or myelin basic protein (MBP, 1:800, Serotec). Following addition of the antibodies, the sections were incubated at room temperature for 10 min, and then the corresponding biotinylated secondary antibodies were added. After 10 min, the streptavidin-biotin-peroxidase complex was added. Following each step, the sections were rinsed with PBS. 3,3'-Diaminobenzidine (DAB) was added to observe the staining. At the same time, hematoxylin was added for clear observation.
The size of the bilateral pupils, and the direct and indirect light reflexes were observed postoperatively. The established models were suitable because the operated pupil was larger than the contralateral pupil, the direct light reflex was absent and the animals exhibited an indirect light reflex.
Flash visual evoked potential
The VEP latencies and amplitudes were obviously different between the control group [Figure 2]a and the model group at each time point. In the recorded VEPs, the P1 wave latency was prolonged, and the amplitude was reduced [Figure 2]b and c.
Under the electron microscope, the normal optic nerve myelin sheath had a complete and organized structure. The tramal plates were clear and tightly arranged, the axolemma was complete, the axonal density was not enhanced or decreased, and the microfilaments and microtubules were well distributed in the homogeneous axoplasm [Figure 3]a. Following optic nerve injury, the endoneurium, myelin sheath, tramal plates, axolemma, and axon were in disorder. Most of the myelin sheaths were arranged loosely. Some myelin sheaths folded inwards excessively to invade into the axoplasm, and others were evaginated and convoluted. The myelin sheaths were thickened, the tramal plate was separated, and the gaps between the layers were enlarged. The mitochondria in the axoplasm swelled, and the gaps between the cristae widened. The cristae were unclear or completely destroyed, and some collapsed or even disappeared, which led to mitochondria rarefaction and the formation of vacuole-like structures [Figure 3]b and c. Nerve fibers with normal structures were rarely observed around the damaged sites.
In non-injured optic nerves, the NF staining was observed as brown stripes that were arranged in a regular, parallel manner along the neuraxis [Figure 4]a. Following injury, the NF staining appeared irregular with bead-like brown staining next to the injured site [Figure 4]b. This staining gradually faded with time post-injury [Figure 4]c. In non-injured optic nerves, the MBP staining appeared as an even brownish-yellow pattern, which was arranged regularly along the axons [Figure 4]d. Following injury, the arrangement was still observed, but less staining was visible [Figure 4]e. Additionally, after injury, the GFAP staining was faded or absent at the injury site [Figure 4]f. The staining intensity at the proximate end of the injured site gradually increased and achieved peak levels on day 14 [Figure 4]g. [Figure 4]h shows normal GFAP staining.
The pterional approach is one of the most common surgical approaches in neurosurgery. In 1942, Dandy first reported the approach to manage anterior communicating artery aneurysms. Subsequently, after several modifications by Yasargil, the approach became the classic neurosurgical approach for treating intracranial aneurysms. The pterional approach is often used to perform optic nerve decompression surgery to treat acute optic nerve injuries that occur in association with craniocerebral trauma. The approach has the following advantages. (1) The operation can treat intracranial and optic nerve injuries at the same time. (2) The decompression of the optic nerve is performed under direct vision with a large surgical visual field, so the operation is convenient with a low possibility of optic nerve injury. (3) It is easy to locate and determine the intensity of the injury. Since the optic nerve is fully in the visual field, it is easier to open the optic nerve sheath using this approach than with other approaches. (4) There is no risk of injuring other nerves. (5) There is no disfigurement, since the incision is within the hairline. Therefore, the use of the pterional approach to establish optic nerve injury animal models has both a clinical and theoretical basis.
There are many similarities between the surgical approach used in this experiment and the clinical pterional approach. (1) In terms of body position, the head is turned on one side so that zygomatic process of the frontal bone is at the highest point. (2) The incision is made on the connecting line from the front of tragus from the superior border of the zygomatic arch to the lateral canthus, which is similar to the skin incision in the clinical pterional approach. (3) The operation uses the pterional approach to incise the scalp and aponeurosis, bluntly dissect the temporalis, and then retract, dissect the periosteum, grind off the skull bone, extend the bone window, and cut the bone off toward the direction of skull base as far as possible. (4) The operation is similar to the pterional approach, since it involves draining the cerebrospinal fluid, slowly elevating the pyriform lobe, and exploring the optic nerve along the skull base.
There are a number of differences between the two methods. The anatomic part of the cat's brain equivalent to the human brain's temporal lobe is pyriform lobe. The cat's skull base is usually flat, and it is easier to reach the sellar area and the optic foramen by slightly elevating the pyriform lobe, whereas in clinical operations, the frontal lobe would generally be elevated.
In this study, the cat was selected as the animal model. The main reason for this selection is that cats have a high tolerance for surgical procedures, and the gyrencephalic features are similar to the folded cerebral cortex in humans. Moreover, cat pupils react to light similarly to human pupils.
The optic nerve is a collection of retinal ganglion cell (RGC) axons that are composed of oligodendrocytes, astrocytes, and microglia. Its myelin is formed by oligodendrocytes. In this experiment, the axon marker NF, astrocyte marker GFAP, and oligodendrocyte marker MBP were selected to perform immunohistochemical staining. NF staining of the proximate end of the injured site showed no significant changes at 6 h. On day 1, the staining was slightly enhanced, and the staining gradually faded from day 3 onward to day 14. The GFAP staining of the directly injured site faded gradually, and at the proximate end of the injured site, the staining showed no significant change during the early stage (days 1-3), but enhanced staining was observed after 7 d and achieved its peak on day 14 after injury. After optic nerve injury, MBP staining was absent, suggesting there was a significant demyelination phenomenon. MBP indirectly reflects the function of oligodendrocytes. Previous studies on Browman-Wyse (BW) rats (visual function loss, optic nerve dysmyelination) and Jimpy rats reported that significant decreases of oligodendrocytes led to dysmyelination. , The areas affected by the primary injury showed morphological disruption, loss of beta-III tubulin axonal staining, and reduced myelinated axon density. 
Podhajsky et al observed that when the optic nerve was crushed, the distal neuraxon of the ganglion cells developed Wallerian degeneration, resulting in a healing reaction at the mechanically injured site. GFAP and NF staining was absent at the injured site. GFAP staining of the proximate end of the injured site gradually enhanced and reached 1.7 times the normal staining by week 3, while the NF staining intensity gradually decreased within 6 weeks. The number of oligodendrocytes decreased at the injured site, but was nearly normal at the proximate end. Cohen et al. also observed in fishes that astrocytes disappeared at the injured site during the initial 2 d after injury, and macrophages were noticeably proliferating at the injured site. However, in fishes, the astrocytes were absent for a very short time, and they grew and proliferated at the directly injured site within 2 weeks. The enhanced GFAP staining at the proximate end of the injured site suggests that there is an increased number of astrocytes, resulting from migration or division.
A study by McPhilemy  confirmed that there is an increase in GFAP mRNA synthesis after optic nerve transection, and this the up-regulation of GFAP expression occurs in reactive glial cells. Scar formation, or the lack of nutritional factors in glial scars, blocked axonal regeneration. This indicates that the disappearance of astrocytes from the directly injured site may be because regenerated axons fail to grow through this area. The area of astrocytosis, as defined by GFAP immunostaining, usually exceeds the extent of axonal loss. Proteolipid protein (PLP) mRNA expression was reduced in oligodendrocytes in the denervated area of the lesioned nerve in comparison to the innervated zones of the lesioned nerve and the contralateral intact nerve. This down-regulation correlated with the axonal loss rather than the area of astrocytosis. These data support the contention that axons are necessary for oligodendrocytes to maintain full expression of their major myelin protein genes.  Furthermore, researchers have demonstrated that the RGCs in directly injured axons developed acute necrosis and released glutamic acid. ,,, Glutamic acid is an excitatory amino acid, which is released from neurons under a state of stress, affecting ganglionic cells through N-methyl-D-aspartic acid (NMDA) receptors. Meanwhile, astrocytes can absorb glutamic acid and transform it into atoxic glutamine. The results of our experiments are partially consistent with those of previous studies.
Taken together with the results of previous studies, our results suggest a pathological process of acute optic nerve injuries. During the early stage (6 h to 3 d) after compression of the optic nerve, astrocytes in the directly injured site die, which is reflected by the decreased or absent GFAP staining. In the early stage, NF staining of the proximate end is slightly enhanced, possibly as a result of the limited regeneration of axons, of which the growing tip can reach the proximate end of the injured site. After optic nerve injury, the excitatory amino acid is produced, which has a toxic effect on neurons, resulting in the proliferation and activation of astrocytes. The astrocytes can absorb glutamic acid and transform it into atoxic glutamine, which is reflected in the enhanced GFAP staining at the proximate end of the injured site after 7 to 14 d. However, glutamic acid-injured RGCs further result in a decrease of NF within axons. Therefore, direct mechanical injury of the optic nerve leads to the death of oligodendrocytes, and the decrease of oligodendrocytes then leads to demyelination.
This acute optic nerve injury model has the following features: (1) the ability to select the intracranial optic nerve to be injured to closely mimic clinical injuries, (2) low surgical trauma and a low mortality rate in the experimental animals, (3) the degree of injury is definite and controllable with good reproducibility, and (4) the pathological process is predictable and clinically similar to the human eye. Therefore, this model is suitable for research on the pathological changes and regenerative repair of the optic nerve after injury.
To conclude, this study proposes that the pathological process of acute optic nerve injury involves the following series of steps. Direct mechanical injury of the optic nerve leads to the death of oligodendrocytes in the optic nerve, which consequently results in optic nerve demyelination. Following optic nerve injury, the astrocytes in the injured area die and produce excitatory amino acids, which have an adverse effect on neurons, resulting in the proliferation and activation of astrocytes. The astrocytes can absorb the glutamic acid and transform it into atoxic glutamine. The glutamic acid can then injure retinal ganglion cells, resulting in the reduction of neurofilament proteins in the axons. We believe that the application of the pterional approach to establish optic nerve injury animal models has both a clinical and theoretical basis.
The authors would like to thank Dr. Gang Sui for his excellent technical assistance. This study is supported by the National Natural Science Foundation of China (No. 30271333).
[Figure 1], [Figure 2], [Figure 3], [Figure 4]