The Embryological Development of the Form of the Trabeculae Bridging the Subaracnoid Space

In a modern textbook [1] one finds the comment “Although it is commonly stated that the brain “floats” in the cerebrospinal fluid (CSF), the brain is suspended in the CSF-filled subarachnoid space by the trabeculae”. Subarachnoid trabaculae are sheets or columns of collagen reinforced material that stretch between the arachnoid and pia membranes. They can be seen with light microscopes but they are too thin to be displayed by ultrasound. The arachnoid membrane was discovered by Gerardus Blasius (1626-1692), who named it “arachnoid” (Gr. Spider) from the “cobweb” of fine fibres attached to its underside [2]. Thus the existence of trabeculae has been known to anatomists for a very long time. Trabeculae constrain relative movement between skull and brain such as postulated in the Shaken Baby Syndrome hypothesis [3].


Introduction
In a modern textbook [1] one finds the comment "Although it is commonly stated that the brain "floats" in the cerebrospinal fluid (CSF), the brain is suspended in the CSF-filled subarachnoid space by the trabeculae". Subarachnoid trabaculae are sheets or columns of collagen reinforced material that stretch between the arachnoid and pia membranes. They can be seen with light microscopes but they are too thin to be displayed by ultrasound. The arachnoid membrane was discovered by Gerardus Blasius (1626-1692), who named it "arachnoid" (Gr. Spider) from the "cobweb" of fine fibres attached to its underside [2]. Thus the existence of trabeculae has been known to anatomists for a very long time. Trabeculae constrain relative movement between skull and brain such as postulated in the Shaken Baby Syndrome hypothesis [3].
The pattern of trabeculum structure is set in early embryonic development. This present paper is largely a based on the work of McLone and others in the 1970s.
Illustrations of trabeculae often show them as straight between arachnoid and pia membranes [1], but scanning electron microscopy shows a much more complex, and mechanically stronger honeycomb type of structure [4]. In Figure 1, (drawn from Figure 2 in that reference) elements appear to run in random directions, and be of random sizes. In such a structure there are no fault lines along which tears could run, and there are always some elements partially aligned with any imposed stress and hence able to resist stretch.
Subarachnoid trabeculae got the name trabeculae, "beams", from the fact that they appeared in sections as thin rods supporting the arachnoid above the pia. Scanning electron microscopy, (SEM), has revealed that many are actually sheets, or "curtains" that appear as rods when sectioned. Also, they are usually depicted as traversing straight across the cerebro-spinal fluid (CSF) filled subarachnoid space. Figure 1 is drawn from an electron microscope photograph of a section of a human optic nerve. Optic nerves are not actually nerves in the common use of that term. They are actually extensions of the forebrain (diencephalon) [5], carrying axons of retinal ganglion cells, shown in the yellow area. They are accompanied by extensions of the arachnoid and pial maters from the cerebrum surface, and enclosing a cerebro-spinal fluid (CSF) filled subarachnoid space. Their crosssection represents that of their brain. The trabeculae can be seen to have a random honey-comb like structure.

Early Development-the Potential Pia-Arachnoid Space (PPAS)
Much of the fundamental work on the initial development of the telencephalon was carried out by Weed, in 1917. In 1975 McLone [6,7] undertook a detailed electron microscopic study of this embryonic sequence in the mouse, and much of the following is based on his work.
The basic sequence of embryological development of the brain coverings (meninges) in mice and men is very similar. The significant differences relate to strengthening of the sub pia (neuroglia) in response to the greater brain weight [8].
The initial development of the subarachnoid space takes place in two phases during embryogenesis. The first phase is the development of a mesenchymal, space holding, layer. This comes forward from the future neck region and advances between the ectoderm and the neuroepithelium of the telencephalon [6]. It is composed of a gel-filled mesenchyme network, (ground substance) Figure 2. At this stage there are no arachnoid or pia membranes. The mesenchyme cells, are widely spaced, stellate cells, with long interconnecting cytoplasmid processes. The gel component is made up of chains of repeating disaccharides termed glycosaminoglycans [9]. Gases move through this gel by diffusion, there is no bulk movement. Osaka et al refer to this stage as meninx primitivea, or primitive subarachnoid space [10]. Initially there are no Pial or Arachnoid membranes in the PPAS.
During post conceptual days 10 to 13, liquid filled "holes" start to appear in the gel, randomly distributed throughout the PPAS (coloured blue in Figure 3). They gradually increase in size, appearing to push the mesenchyme cells and fibres before them. At 13-16 days, loss of gel at the upper and lower surfaces of the PPAS allows compaction of the mesenchyme to form membranes. Mesenchyme cells are pluripotential, they can convert to connective tissue, blood cells and vessels. They reinforce these new membranes which become the Pia and Arachnoid membranes [8]. The collagen content of the cerebral pial layer increases with progression to high mammals (i.e. mouse, rat, monkey, man) [7], indicating its importance in stabilising the brain-skull positioning.
Concentrations of fibrous material also appear lining the expanding liquid filled lacunae. McLone and Bondareff comment "Bundles of microfibrils and collagen are commonly associated with lacunae in the outer pia-arachnoid layer and may serve as struts to maintain an open subarachnoid pathway." Actually these trabecular "struts" are under tension, "guy ropes" might be a better analogy.
Eventually, where the resulting lacunae approach each other closely, the remaining mesenchymal cells and fibres become pressed into "curtains" stretched across the subarachnoid space. These curtains have holes in them through which fluids can pass ( Figure 4).
This establishes the possibility of cerebrospinal fluid flow. These remaining walls are the trabeculae. Osaka described the resultant fluid space as "essentially the 'cleared-out connective tissue space' which is formed in late embryonic life" [10].
By the 17 th day the framework of the subarachnoid space, consisting of the outer arachnoidal membrane, the trabeculae, and the inner pial layer has been established. It retains the random nature of the original liquid filled "holes" of early embryonic life.

Trabecular Attachments
Trabeculae may be strong but they will only be as useful as the strength of their attachments. In the arachnoid mater collagen fibres from the trabeculae stitch into the arachnoid mater which has formed in the top surface of the PPAS (Primitive Pia Arachnoid Space). The arachnoid mater becomes strongly reinforced with collagen and so can withstand relatively powerful forces.
The Pia Mater itself is only one cell thick so its strength depends on the structures below it. With electron microscopy, trabecular collagen fibres can traced through the Pia Mater to fibres in the Sub Pial space. In turn, fibres in the subpial space are attached to a basement membrane beneath which are a bed of astrocytes and oligodendocytes [11]. These  are described as Glia (Gr. Glue) because they act to hold neurons and capillaries together, though not with glue! They may be thought of as micro-octopuses. Instead of suction cups they wrap the end of each tentacle several times round the nearest axon, blood capillary, etc., a different one with each tentacle. There will be many more astrocytes than neurons so the net effect is that they are all drawn together. There are many astrocytes holding on to the basement membrane with one tentacle while holding axons etc. with the others. The result is that if a trabeculum pulls on the basement membrane from above, the local neurones get stretched below. When a boxer is temporarily knocked out various axons will have been stretched. This may act directly on the axon, or the stretch may have been transmitted to a local arteriole and induced temporary vasoconstriction. If the jolt is much stronger, as in Ommaya's experiments [12], vessels may be overstretched and bleed, showing as contusions. (The damage is below the Pia, at the brain surface, not above the arachnoid mater in the dura, as presumed in the Shaken Baby Syndrome hypothesis. This study has found that trabecular form and generation were well understood by anatomists in the 1970's and 1980's, but their significance for child protection was not recognised then, nor since. The subarachnoid space is not empty (Figure 4). It is full of fine fibres (trabeculae) which are too thin to show up on ultrasound machines. These trabeculae inherit their random direction, size, configuration etc. from the random positions of initiation of the substitution of fluid for ground substance gel in embryonic development. This is the explanation of the random appearance of the human optic nerve illustration in Figure 1.