Introduction

The optic nerve, on its way to the retina, is enclosed in dural, arachnoid and pial sheaths. The arachnoid cells form the subarachnoid space (SAS) surrounding the nerve. Furthermore, they build cell layers, which are the basis of a fluid barrier that should prevent the outflow of cerebrospinal fluid (CSF) from the SAS [6]. However, it has been shown that CSF, and even particulate matter, may drain into the orbit and may actually reach the cervical lymphatics [7, 11, 16]. This phenomenon was especially observed under conditions of increased intracranial pressure (ICP) [4, 11, 16, 17].

The morphological basis remains unclear. Previous investigators have described the specific morphology in the distal portion of the optic nerve sheath (ONS) [8, 18]. It was shown that the meningeal architecture is different in the proximal and distal portions of the ONS: in the distal portion it was thought that tortuous channels and microchannels connect the SAS with the dura mater. It was suggested that the arachnoid barrier layer (ABL) is absent in this region, thus enabling CSF outflow [8]. In contrast, others described an ABL in the optic nerve in the region where CSF outflow was observed [7]. Our previous investigations demonstrated a different morphological behaviour of the ABL in the proximal portion of the nerve at a normal ICP in comparison to increased ICP. Similar to findings in the cortical meninges [6], the ABL may disrupt when there is a strong increase in ICP, as is the case during aneurysmal rupture. During high-pressure cisternography this disruption allows a massive outflow of CSF into the orbit [4].

Extending our previous studies, we now investigate, under the condition of only a mild increase in ICP, the ABL at the termination of the optic nerve. With regard to the afore-mentioned conflicting ultrastructural results, we specifically asked whether the observed outflow of CSF at the termination of the optic nerve occurs, even in the presence of a continuous ABL.

Materials and methods

General procedures

Ten male cats, weighing between 3.6 kg and 4.2 kg, were used for the experiments. Five cats were infused with contrast medium and the other five were used for ultrastructure control examinations. The experiments were performed with institutional approval TS 504-42502-90/401 of the local government (Bezirksregierung Hannover).

Anesthesia was induced by intramuscular injection of 30 mg/kg body weight of ketamine and supplemented by a small dose of barbiturate. The animals were intubated endotracheally and ventilated by a small-animal respirator. The femoral blood vessels were catheterised for the continuous recording of arterial blood pressure and the continuous administration of drugs. Physiological body temperature was maintained by a heating blanket. Anesthesia was continued with the infusion of ketamine at a rate of 30 mg/h combined with pancuronium at 0.8 mg/h. Arterial blood gas parameters were checked routinely and kept within normal limits.

A microcatheter was inserted through the atlanto-occipital membrane into the cisterna magna. A pressure transducer was connected to the cisternal catheter by a three-way stopcock.

Cisternal infusion

Using an adjustable syringe pump, 0.1 ml/min of X-ray contrast medium Isovist 300 (Schering, Berlin, Germany) was infused over 60 min. The infusion was interrupted when the ICP reached more than 20 mmHg. X-ray studies were performed 1, 10, 20, 30 and 60 min after starting the infusion. Immediately after the cisternal infusion, the animals were fixated.

Fixation and electron microscopic examination

Arterial perfusion-fixation was performed through a catheter introduced into the aortic arc after clamping the descendent aorta. The cava vein was opened by slit incision. Following perfusion with 100 ml heparinised saline, 200 ml aldehyde fixative (2.0% formaldehyde, 2.5% glutaraldehyde, 0.1 M sodium cacodylate and 0.025% CaCl2 at pH 7.4 [12]) were infused over 10 min.

After the perfusion, the eyeballs were removed and the optic nerves were exposed by microsurgical techniques. The specimens were immersed in the aldehyde fixative for 24 h. Samples were postfixed in 2% osmium-tetroxide in cacodylate buffer for 2 h and dehydrated in increasing concentrations of alcohol.

For scanning electron microscopic (SEM) examination, specimens of the optic nerve were dried in a critical point dryer with CO2 and mounted on suitable stubs. The specimens were coated with gold and observed and photographed with a Philips 505 scanning electron microscope. For transmission electron microscopic (TEM) examination, 3–5 mm specimens of the optic nerve were embedded in epon. Ultra-thin sections were stained with 5% uranyl acetate and 0.4% lead citrate and examined using a Zeiss 10 CR electron microscope.

Results

Cisternography

Radiological examination demonstrated leakage of contrast medium into the orbit, which was most pronounced at the distal portion in comparison to the proximal portion of the optic nerve. This finding is regularly demonstrable after 20 min of contrast infusion (Figs. 1, 2).

Fig. 1
figure 1

Inferior posterior view of a radiograph of the skull after 30 min of cisternal contrast infusion. The leakage of contrast medium (asterisk) into the orbital tissue is confined to the distal portion of the optic nerve (O). N nasal contrast deposit

Full size image
Fig. 2
figure 2

Lateral view of a radiograph of the skull after 60 min of cisternal contrast infusion. The orbital tissue (O) is filled with contrast medium. Note the cisternal catheter (arrowhead) with no leakage at the site of introduction. A cervical lymph node with its sinus (long arrow) and the efferent lymphatic vessel (short arrows) are contrasted

Full size image

Ultrastructure of the proximal portion of the optic nerve sheath

Beneath the pial and inner arachnoid covering of the optic nerve, the proximal portion of the ONS consists of the outer arachnoid layer (Fig. 3), a compact neurothelial layer and the dura mater. The outer arachnoid layer, which forms the outer mesothelial covering of the SAS, consists of four to eight cell layers. Occasionally, the covering is not continuous and excavations can be found. Thus, the arachnoid cell layers form three-dimensional membranous curtains, which are permeable through excavations of the SAS, as identified in SEM examination. Intercellular collagen is frequently observed between the arachnoid cells (Fig. 4).

Fig. 3
figure 3

Scanning electron microscopic (SEM) view of the distal optic nerve sheath (ONS). The optic nerve (ON) is covered by pial and inner arachnoid (AIL) cell layers. The subarachnoid space (SAS) is surrounded by the inner arachnoid and outer arachnoid (AOL) cell layers. Note the excavation (EC) of the SAS

Full size image
Fig. 4
figure 4

Transmission electron microscopic (TEM) view of the outer optical nerve sheath in the proximal portion of the optical nerve. The SAS is lined by the AOL. The outer arachnoid cell layers are characterised by the occurrence of intercellular collagen (C) and excavations of the SAS (E) that are partly filled with membranous bodies (arrowheads). The neurothelial cell layers (N) are separated from the AOL by a narrow intercellular cleft. Collagen and excavations are absent in the neurothelial cell layers. The dura mater (D) consists of collagenous fibres and few cells (approximately ×6,400 magnification)

Full size image

Neurothelial cell layers are separated from the outermost arachnoid cell layer by a narrow intercellular cleft (Fig. 4). No electron dense material, basal lamina, or filaments were seen in this cleft. In contrast to the outermost arachnoid layer bordering the SAS, the neurothelial layer is characterised by an absence of intercellular collagen fibres and excavations. Four to eight neurothelial cell layers form three-dimensional membranous curtains, which, in contrast to the arachnoid cell layers, are not disrupted by open junctions.

The dura mater consists of many collagenous fibres and few cells: these do not form curtains like the neurothelial and arachnoid cells (Fig. 4).

This meningeal architecture can be found along the whole orbital route of the optic nerve, except for a small region where the dura mater joins the sclera.

Ultrastructure of the distal portion of the optic nerve sheath

At the termination of the perioptic SAS, near the ocular bulb, the dura mater joins the sclera and the outer arachnoid joins the inner arachnoid and the pia mater of the optic nerve. Here, the ONS shows significant morphological features as a morphological correlate for CSF outflow.

The window area consists of an arachnoid trabecular network, i.e. collagenous fibres covered by a mesothelial lining of arachnoid cells. Arachnoid trabeculae reach the neurothelial layer. This network lies in the excavations of the SAS. Excavations are irregular-shaped extensions of the SAS into the outer arachnoid layer. These excavations of the SAS reach the neurothelial layers (Fig. 5a, b). Thus, the arachnoid mesothelial lining of the SAS is not continuous.

Fig. 5a, b
figure 5

Scanning electron microscopic views of the distal ONS in the area of the window. a Two excavations (EC) of the SAS with an arachnoid trabeculum (T) and the dura mater (D) are shown. b The view from inside the subarachnoid space in the direction of the neurothelial (N) layer shows the continuity of the excavations with the subarachnoid space. Arachnoid trabecula and arachnoid cell bridges (B) are visible within these excavations

Full size image

The excavations of the SAS do not show collagenous fibres in contrast to the trabecula (Fig. 6a, b). The arachnoid cells of the window area form a meshwork of bridges and trabecula connected by tight, open and desmosomal junctions (Fig. 6b). All arachnoid cells contain numerous intracellular filaments and anchoring filaments in the mesothelial lining cells (Fig. 6c).

Fig. 6a–c
figure 6

Transmission electron microscopic view of the periorbital neurothelial window. a The SAS is lined by the outer arachnoid cell layers, characterised by cell bridges (B) and trabecula (T) with numerous excavations (EC), partly filled with membranous bodies (arrows). Collagen (C) is surrounded by outer arachnoid cells, thus forming arachnoid trabecula. The neurothelial cell layer (N) consists partly of a single cell layer (arrowheada). The dura mater (D) shows abundant collagen fibres and few cells (approximately ×6,200 magnification). b At higher magnification, desmosomes (circles) are visible, connecting the arachnoid cells (approximately ×8,300 magnification). c The arachnoid cells are characterised by abundant intracellular fibres and anchoring filaments (AF) of an arachnoid trabecula bordering the SAS (approximately ×35,700 magnification)

Full size image

The neurothelial cells are layered in a parallel manner, similar to those in the proximal portion of the ONS, and they do not have filaments in their intercellular clefts. However, the number of cell layers is significantly reduced. Sometimes, the neurothel shows only one cell layer with pore-like openings (Fig. 7). The neurothepithelium is continuous in the distal portion of the ONS and ends spur-like in the junction between the dura and the sclera.

Fig. 7
figure 7

Scanning electron microscopic view of the distal ONS in the area of the window. After dissection of the outer arachnoid layers, pore-like structures (P) in the neurothelial layer are visible

Full size image

The main differences between the meninges in the distal and proximal portions of the ONS can be detailed as follows (Fig. 8):

  • In the distal portion of the ONS the neurothelial layers are significantly thinner and partly consist of only one layer. Open junctions between the neurothelial cells (Fig. 7) are found in the distal portion under physiological ICP

  • Excavations of the SAS are far more numerous in the distal portion of the ONS, thus creating the reticular meshwork of the periorbital arachnoid window (Fig. 5). The SAS excavations reach the neurothelial layer (Figs. 5b, 6a)

  • Intracellular filaments are more numerous in the distal portion of the ONS (Fig. 6c)

Fig. 8
figure 8

Schematic drawing of the ONS. In the proximal portion of the ONS (right side) the SAS is lined by the inner arachnoid layer (AIL) and the outer arachnoid layer (AOL) (see Fig. 3). The neurothel (NT) in the proximal portion of the ONS consists of neurothelial layers without intercellular collagenous fibres, open junctions or excavations (see Fig. 4). Before entering the ocular bulb (OB), the distal portion of the ONS terminates in the window area, where the dura mater (D) joins the sclera (S). It is characterised by numerous excavations (EC) of the SAS, separated by the trabeculae (T) of the AOL (see Fig. 6b). These excavations reach the NT. The NT in the window area, in contrast to the proximal portion of the ONS, consists of fewer neurothelial layers and open junctions (see Figs. 6a, 7)

Full size image

The dura mater in the distal portion of the optic nerve shows no significant differences from the proximal dura mater. Pre-existing microchannels or particularly loose collagenous tissue are not seen.

Ultrastructure of the conjunctival lymphatics

Following the cisternal infusion of X-ray contrast medium, it can be seen in the conjunctival lymphatics. Figure 9 shows the dilated initial lymph vessels in the conjunctiva with their specific morphological features, i.e. interendothelial openings and basal filaments. X-ray contrast medium can be observed in the interendothelial openings. Furthermore, contrast medium was deposited in the SAS of the ONS. It was not possible to demonstrate lymph vessels or contrast medium in the rest of the orbital tissue.

Fig. 9
figure 9

Lymph vessels in the conjunctiva. Upper section: SEM shows a dilated initial lymph vessel in the conjunctiva with its specific morphological features—interendothelial openings (O), endothelial covering (EC) and basal filaments (BF). Inset: X-ray contrast medium can be observed in an interendothelial opening. The SEM appearance is characteristic of macromolecules like contrast medium or staining markers like Berlin Blue. Lower section: Oak leaf-like endothelial cell (EC) connections (arrowheads) are a characteristic morphological feature on the luminal surface of lymph capillaries and helpful for identification; SEM micrograph

Full size image

Discussion

We examined the morphological differences between the distal and proximal ONS in order to establish the morphological basis for CSF outflow in the distal portion of the ONS at low ICP. We found the ABL in the distal portion of the ONS to be significantly thinner, but continuous, even at the end of the optic nerve. To explain the radiologically observed CSF outflow at low ICP, the continuous ABL is crossed via the open junctions and intracellular pores of the neurothelial cells.

Electron microscopic examination always carries the risk of misinterpreting perfusion or fixation artefacts. They were avoided using standardised procedures and physiological pressure during fixation. Only regularly demonstrable findings in material free from preservation artefacts were described.

Ultrastructure

In our study, the architecture of the distal ONS does not differ in principle from the meninges at other anatomical sites [13, 21]. There is an inner arachnoid layer covering the pia and an outer arachnoid layer bordering the neurothelial layer, thus the SAS is lined with arachnoid cells in each direction [14]. The ABL is seen to be continuous at the distal end of the optic nerve, but shrinks occasionally to a single layer. These results are consistent with the window area in the distal ONS, as has been described by others in the hamster and the rabbit [7, 8, 9, 18].

Comparing the ultrastructural description of the ONS in the literature with our results, the described meshwork channels [8, 18] correspond to the excavations of the SAS. They reach through the arachnoid to the neurothel and are most pronounced in the window area. With the SEM results presented here, it is possible to identify them as excavations of the SAS. They correspond to the cisterns of the arachnoid “reticular layer”, as described for the spinal meningeal funnels [21].

In our results there are no microchannels in the neurothelial layer. The described microchannels [8] in the neurothelial layer should be regarded as expansions of the extracellular space in the ABL, as described elsewhere [6].

CSF outflow

Our study demonstrates that CSF and particle transport into the orbit should follow the same principles as CSF outflow at the arachnoid villi. In the window area of the optic nerve, the excavations of the SAS terminate at a partly single-layered barrier that is exposed to a pressure gradient, i.e. ICP versus intraorbital pressure [15]. Thus, the pore-like openings observed can be considered an expression of the transarachnoid outflow of CSF, as has been described in arachnoid villi [19]. In contrast to the arachnoid villi no direct communication was observed between the vascular lumen and the SAS. Therefore, CSF drains into the orbital tissue.

In which direction does the CSF drain after entering the orbital tissue?

Gomez et al. proposed venous drainage [9], which we were not able to confirm like other authors [8, 18]. McComb et al. [16] demonstrated the drainage of dye from the optic nerve SAS into conjunctival lymphatics under increased ICP. In agreement with this observation, we found contrast medium in conjunctival lymph vessels after cisternal application at physiological ICP.

These observations support the idea of a lymphatic outflow of CSF [2, 5, 10, 16, 20]. The transfer between the SAS and the lymphatics in the orbita remains unclear under conditions of normal ICP. We would suggest that there is an interstitial route through the orbital tissue, because we did not find typical lymph vessels in the orbit, and this would correspond with the results of Bradbury and Cole and McComb et al. [2, 16]. We have described the orbital outflow of contrast medium under conditions of increased ICP [4]. In contrast, de la Motte described epidural lymphatics in this area [7], a finding we could not confirm.

In comparison to other routes, e.g. the olfactory or spinal nerves, the quantitative significance of CSF outflow along the optic nerve is probably low. However, the optic nerve may serve as an example for other perineural CSF outflow pathways that cannot be readily investigated. Our study provides further experimental evidence of the significance of the final lymphatic pathway of CSF absorption [2, 3], which may contribute to 50–60% of the total CSF absorption [1].

Conclusion

Our findings on the ultrastructural basis for the CSF outflow across the ONS meninges demonstrate a thinned but continuous neurothelial ABL in the distal ONS with pore-like openings. They are comparable with those of Tripathi et al., who studied the ABL in the region of the arachnoid villi. Therefore, we hypothesise that the ultrastructural basis for the outflow of CSF across the arachnoid villi and along the cranial nerves is uniform. CSF may cross the meninges in those regions where the ABL is thin and where the ABL is exposed to a pressure gradient. Thus, from the ultrastructural point of view, the mechanisms underlying the CSF outflow into the orbit and into the venous sinuses are essentially the same.