There are 27 types of collagen molecules identified to date, referred to as types I to XXVII. These collagens are organized into either fibrils or sheetlike structures. In vitreous, nearly all of the collagen is in the form of fibrils that are thin, uniform, and heterotypic (of mixed composition) containing primarily collagen types II, IX, and the hybrid V/XI (Fig. 38-1). Gross¹ initially claimed that vitreous collagen fibrils are morphologically distinct from collagen in other connective tissues. However, Swann et al² demonstrated that the amino acid composition of the insoluble residue of vitreous is similar to that of cartilage collagen and later identified that it is most similar to cartilage collagen composed of alpha 1, type II chains.³ Comparisons of the arthrogenic and immunologic properties of collagens from bovine articular cartilage (type II) and vitreous showed that the two were indistinguishable by these assays.⁴ Subsequent studies⁵ demonstrated that, although vitreous collagen contains an alpha 1, type II chain similar to cartilage collagen, there is a lower alanine content. Furthermore, these studies found that vitreous collagen has additional peptides present as uncleaved extension chains containing an amino acid composition different from the alpha chain component. The investigators concluded, however, that the overall similarities in amino acid composition and in the types of cyanogen bromide cleavage peptides indicated that the fibers of the central and posterior peripheral regions of the vitreous are composed of a collagen that should be classified as type II. Linsenmayer et al⁶ measured in vivo synthesis of types I and II collagen in chick embryo vitreous by radioimmunoprecipitation after tritiated proline labeling and found that more than 90% of the labeled material in the vitreous was type II collagen. Snowden⁷ provided further physicochemical evidence in support of the similarities between vitreous and cartilage collagens. This may explain why certain clinical phenomena, such as inborn errors of type II collagen metabolism in arthro-ophthalmopathies, manifest phenotypic expression in each of these two tissues.

There are some distinct differences in the chemical composition of vitreous and cartilage collagens that are only partly owing to the presence of terminal peptide constituents in vitreous collagen.⁵ Swann and Sotman⁸ have demonstrated that the carbohydrate content of pepsin-solubilized vitreous alpha chains is significantly greater than that of cartilage alpha chains, indicating that the carbohydrate side chains of vitreous collagen are largely composed of disaccharide units similar to those found in basement membrane collagen. They proposed that these distinct chemical features are related to the special structure of the mature vitreous fibrils in vivo. Liang and Chakrabarti⁹ have shown that there are differences between bovine cartilage and vitreous with respect to collagen fibril growth, melting temperature, and fluorescence with a hydrophobic fluorescent probe. These investigators and others¹⁰ proposed that vitreous collagen should be considered a “special” type II collagen. Schmut et al¹¹ used differential salt precipitation of pepsin-solubilized collagen from bovine vitreous and found that type II collagen is the major component of native vitreous fibers.

Gloor¹² pointed out that collagen content is highest where the vitreous is a gel. Ayad and Weiss¹³ studied bovine vitreous collagen to determine whether the gel-like structure of vitreous could be explained on the basis of chemical composition. Their findings demonstrated that type II is the major vitreous collagen, but collagens composed of α1, α2, and α3 chains as well as C-PS disulfide-bonded collagen were present in concentrations similar to those in cartilage. In contrast to cartilage, however, vitreous type II collagen was significantly more hydroxylated in the lysine and proline residues. The α1, α2, and α3 collagen chains were interpreted by Van der Rest¹⁴ to represent type IX collagen, although Eyre et al¹⁵ felt that there was evidence to indicate the presence of type V collagen in vitreous. Furthermore, with respect to the disulfide-bonded collagen, vitreous had three times more C-PS1 and C-PS2 collagens than cartilage, although the molar ratio of C-PS 1 to C-PS2 in both was 1:1, suggesting that in both tissues these collagens are components of a larger molecule. Other studies¹⁴ demonstrated that these disulfide-linked triple-helix fragments were actually derivatives of type IX collagen. In this regard vitreous is again similar to cartilage, insofar as both contain type IX collagen,¹⁵ although the two tissues differ in the sizes of type IX collagen chains.¹⁶

Hong and Davison¹⁷ identified a procollagen in the soluble fraction of rabbit vitreous that was identified as type II by segment long-spacing banding patterns. Detection of a propeptide extension only at the N terminus prompted these investigators to conclude that this was a novel type II procollagen. More recent studies by Bishop et al¹⁸ confirmed the presence of this procollagen. These findings as well as the identification of immature cross-links in adult bovine vitreous¹⁹ and morphologic evidence of synthesis of vitreous collagen by the peripheral retina in the adult eye²⁰ all suggest that postnatal collagen synthesis does occur, although probably only at a relatively low level. This question is relevant to both the origin of vitreous macromolecules as well as the question of vitreous physiology following vitrectomy. Concerning the former, studies by Newsome et al²¹ in the chick embryo implicated retinal Müller cells as the cells responsible for vitreous collagen synthesis during early development, whereas hyalocytes assume that responsibility later in life. Previous²⁰ and recent in vitro²² studies confirmed that Müller cells can synthesize collagen. It does not necessarily follow, however, that these are responsible for vitreous collagen synthesis. Rather, another hypothesis proposes that the ciliary body is the site of synthesis of the macromolecules of both the vitreous body and the internal limiting lamina of the retina.^23,24 What is not clear, however, is whether these macromolecules, particularly collagen, are resynthesized throughout life.

Balazs claimed that following vitreous removal there is no resynthesis of vitreous collagen, only hyaluronan. Recent studies, however, suggest that following vitrectomy a type II procollagen is secreted in humans.²⁵ What is not clear, however, is whether this is de novo synthesis that is induced by the act of vitreous removal, or whether there is ongoing synthesis of vitreous collagen in the adult that is up-regulated after vitrectomy. Clinical experience would suggest that there is little or no up-regulation of collagen synthesis following vitrectomy, since reoperations reveal that the gel state of vitreous is not re-established, at least not in the first weeks or months following vitrectomy, the usual time frame in most clinical scenarios. Whether this is owing to a lack of collagen resynthesis or some other molecular component of vitreous that is responsible for gel formation is open to conjecture and serves as fertile ground for future research.

Glycosaminoglycans are polysaccharides of repeating disaccharide units, each consisting of hexosamine (usually N-acetyl glucosamine or N-acetyl galactosamine) glycosidically linked to either uronic (glucuronic or iduronic) acid or galactose. The nature of the predominant repeating unit is characteristic for each glycosaminoglycan, and the relative amount, molecular size, and type of glycosaminoglycan are said to be tissue specific.⁴² Glycosaminoglycans do not normally occur as free polymers in vivo but are covalently linked to a protein core; the ensemble is called a proteoglycan. A sulfated group is attached to oxygen or nitrogen in all glycosaminoglycans except HA. Balazs⁴³ first documented the presence of sulfated galactosamine-containing glycosaminoglycans in bovine vitreous (<5% of total vitreous glycosaminoglycans), and others^44,45 identified these as chondroitin-4-sulfate and undersulfated heparan sulfate. Studies in the rabbit⁴⁶ found a total vitreous glycosaminoglycans content of 58 ng with 13% chondroitin sulfate and 0.5% heparan sulfate.

As described by Mayne,⁸¹ vitreous is organized in a dilute meshwork of collagen fibrils interspersed with extensive arrays of long HA molecules. The collagen fibrils provide a scaffold-like structure that is inflated by the hydrophilic HA. If collagen is removed, the remaining HA forms a viscous solution; if HA is removed, the gel shrinks⁵³ but is not destroyed. Early physiologic observations⁸² suggested the existence of an interaction between HA and collagen that stabilizes collagen. Biomechanical studies⁸³ of vitreous viscoelasticity suggested similar effects on HA. Early investigators⁸⁴ observed reversible complexes of an electrostatic nature between solubilized collagen and various glycosaminoglycans and suggested that collagen-HA interaction occurs on a physicochemical rather than chemical level. Others⁸⁵ later demonstrated that the sulfate group of a glycosaminoglycan is largely responsible for such interactions with the guanidino groups of arginine and epsilon-amino groups of lysine in collagen. Comper and Laurent⁵³ proposed that in vitreous electrostatic binding occurs between negatively charged polysaccharides and positively charged proteins. These authors extensively reviewed the existing data characterizing the electrostatic properties of glycosaminoglycans and the factors influencing their electrostatic interactions with different ions and molecules.

Balazs⁴³ hypothesized that the hydroxylysine amino acids of collagen mediate polysaccharide binding to the collagen chain via O-glycosidic linkages. These polar amino acids are present in clusters along the collagen molecule, which may explain why proteoglycans attach to collagen in a periodic pattern.³² Hong and Davison¹⁷ identified a type II procollagen in the soluble fraction of rabbit vitreous and proposed a possible role for this molecule in mediating collagen-HA interaction. In cartilage, “link glycoproteins” have been identified that interact with proteoglycans⁸⁶ and HA.⁸⁷ Supramolecular complexes of these glycoproteins are believed to occupy the interfibrillar spaces. In Asakura’s³² studies of bovine vitreous stained with ruthenium red there are amorphous structures located on collagen fibrils at 55- to 60-nm intervals along the fibrils and filaments connecting these amorphous masses to adjacent collagen fibrils. These filaments may represent link structures of either a glycoprotein or proteoglycan nature. HA is known to interact with link proteins, as well as with an HA-binding glycoprotein called hyaluronectin.⁸⁸ In the cornea, chondroitin sulfate and keratan sulfate bridge the interfibrillar spaces and keep the fibrils at specified distances to achieve transparency.⁸⁹ The protein cores of these proteoglycans could be the linkage sites to collagen fibrils.

Bishop³⁵ has proposed that an understanding of how vitreous gel is organized and stabilized requires an understanding of what prevents collagen fibrils from aggregating and by what means the collagen fibrils are connected to maintain a stable gel structure. Studies⁹⁰ have shown that the chondroitin sulfate chains of type IX collagen bridge adjacent collagen fibrils in a ladder-like configuration spacing them apart. This arrangement might account for vitreous transparency, because keeping vitreous collagen fibrils separated by at least one wavelength of incident light would minimize light scattering, allowing unhindered transmission of light to the retina for photoreception. However, depolymerizing with chondroitinases does not destroy the gel, suggesting that chondroitin sulphate side chains are not essential for vitreous collagen spacing. Complexed with HA, however, the chondroitin sulfate side chains might space the collagen fibrils^33,89,90 apart, but Bishop believes that this form of collagen-HA interaction is very weak. Instead, he proposes that the LRR protein opticin is the predominant structural protein in short-range spacing of collagen fibrils. Concerning long-range spacing, Scott³³ and Mayne et al⁹¹ have claimed that HA plays a pivotal role in stabilizing the vitreous gel, supporting the early beliefs of Balazs. However, studies⁹² using lyase to digest vitreous HA demonstrated that the gel structure was not destroyed, suggesting that HA is not essential for the maintenance of vitreous gel stability and leading to the proposal that collagen alone is responsible for the gel state of vitreous.³⁵

Recent studies⁹³ have provided evidence to support the hypothesis that versican plays a crucial role in the supramolecular organization of vitreous by dint of versican’s ability to bind hyaluronan via its N terminal and other binding partners via its C-terminal region, resulting in structural stability of the vitreous. The binding of versican to HA is stabilized by a glycoprotein called link protein, which has also been identified in vitreous.⁵² Mutations that alter the splicing of the central chondroitin sulphate–bearing domains of versican have been implicated in the vitreoretinopathy Wagner syndrome, a condition with excess vitreous liquefaction.⁹⁴

Duke-Elder⁹⁵ claimed that the first theories of vitreous structure proposed that the vitreous is composed of “loose and delicate filaments surrounded by fluid,” as conceptualized in 1741 by Demours,⁹⁶ who formulated the alveolar theory. In 1780 Zinn⁹⁷ proposed that the vitreous is arranged in a concentric, lamellar configuration similar to the layers of an onion. The dissections and histologic preparations of Von Pappenheim⁹⁸ and Brucke⁹⁹ provided evidence for the lamellar theory. The radial sector theory was proposed by Hannover in 1845.¹⁰⁰ Studying coronal sections at the equator, he described a multitude of sectors approximately radially oriented around the central anteroposterior core that contains the Cloquet canal. Hannover likened this structure to the appearance of a cut orange. In 1848 Bowman¹⁰¹ introduced the fibrillar theory. Using microscopy, he described fine fibrils that form bundles that Retzius¹⁰² described as fibrous structures arising in the peripheral anterior vitreous that assume an undulating pattern similar to a horse’s tail in the central vitreous but maintain a concentric configuration at the periphery. The elegant studies of Szent-Gyorgi¹⁰³ in 1917 supported the descriptions of Retzius and introduced the concept that vitreous structure changes with age. Eisner¹⁰⁴ studied dissections of human vitreous and found “membranelles,” which he described as funnels packed into one another diverging outward and anteriorly from the prepapillary vitreous. Worst¹⁰⁵ has also studied preparations of dissected human vitreous and described that the tracts of Eisner constitute the walls of cisterns within the vitreous body. In Worst’s studies these cisterns are visualized by filling with white India ink. Worst has also studied the premacular vitreous in great detail and has proposed the existence of a “bursa premacularis,” which he described as a pear-shaped space that is connected to the cisternal system in front of the ciliary body. Kishi and Shimizu¹⁰⁶ recently described a “posterior vitreous pocket” similar to what Worst has reported, which they claim to be an anatomic structure. However, because more than 95% of their study population was age 65 years or older, this is likely to be an age-related phenomenon.¹⁰⁷

There is heterogeneity in the distribution of collagen throughout the vitreous body. Chemical^108,109 and light-scattering^110,111 studies have shown that the highest density of collagen fibrils is present in the vitreous base, followed by the posterior vitreous cortex anterior to the retina, and then by the anterior vitreous cortex behind the posterior chamber and lens. The lowest density is found in the central vitreous and adjacent to the anterior cortical gel. HA molecules have a different distribution from collagen. They are most abundant in the posterior cortical gel with a gradient of decreasing concentration as one moves centrally and anteriorly.^112,113,114 Balazs¹¹⁴ has hypothesized that this is because vitreous HA is synthesized by hyalocytes in the posterior vitreous cortex and cannot traverse the ILL of the retina. HA leaves the vitreous to enter the posterior chamber by way of the annulus of the anterior vitreous cortex that is not adjacent to a basal lamina. Bound water (nonfreezable) has a distribution within the vitreous similar to that of HA,¹¹⁵ presumably resulting from binding by HA.

Laboratory investigations of vitreous structure have long been hampered by the absence of easily recognized landmarks within the vitreous body. Consequently, removal of the vitreous from the eye results in a loss of orientation. The transparency of the vitreous renders observation in conventional diffuse light unrewarding. Attempts to study vitreous structure with opaque dyes¹⁰⁵ do visualize the areas filled by dye but obscure the appearance of adjacent structures. The use of histologic contrast-enhancing techniques usually involves tissue fixation, which often includes dehydration of the tissue. Because vitreous is 98% water, dehydration induces profound alteration of the internal morphology. Consequently, any investigation of vitreous structure must overcome these difficulties.

The sclera, choroid, and retina can be dissected and the naked vitreous body can be maintained intact and attached to the anterior segment of the eye (Fig. 38-2A). This enables study of internal vitreous morphology without a loss of intraocular orientation. However, depending on the person’s age and consequently the degree of vitreous liquefaction,⁷² the dissected vitreous will remain solid and intact in young persons or will be flaccid and collapse in older individuals. Consequently, vitreous turgescence must be maintained to avoid distortion of intravitreal structure. Immersion of a dissected vitreous specimen that is still attached to the anterior segment into a physiologic solution maintains vitreous turgescence and avoids structural distortion (Fig. 38-2B). The limitations induced by the transparency of the vitreous were overcome by Goedbloed,¹¹⁶ Friedenwald and Stiehler,¹¹⁷ and Eisner¹⁰⁴ who employed darkfield slit illumination of the vitreous body to achieve visualization of intravitreal morphology. Illumination with a slitlamp beam directed into the vitreous body from the side and visualization of the illuminated portion from above produces an optical horizontal section of the vitreous body.¹¹⁸ The illumination/observation angle of 90° that is achieved by using this technique maximizes the Tyndall effect and, thus, overcomes the limitations induced by vitreous transparency. Furthermore, the avoidance of any tissue fixation eliminates the introduction of many of the artifacts that flawed earlier investigations. Recent studies have used these techniques to investigate human vitreous structure. Within the adult human vitreous there are fine, parallel fibers coursing in an anteroposterior direction^{39,118,119,120} (Figs. 38-3B and C and 38-4). The fibers arise from the vitreous base (Figs. 38-3H and 38-4) where they insert anterior and posterior to the ora serrata (Fig. 38-3H). As the peripheral fibers course posteriorly they are circumferential with the vitreous cortex, whereas central fibers undulate in a configuration parallel with the Cloquet canal.¹⁰² The fibers are continuous and do not branch. Posteriorly, these fibers insert into the vitreous cortex (Figs. 38-3E and F).