Ultrastructural analysis of the effect of ascorbic acid on secretion and assembly of extracellular matrix by cultured chick embryo chondrocytes

Homogeneous hydroxyapatite/alginate composite hydrogel promotes calcified cartilage matrix deposition with potential for three-dimensional bioprinting

Calcified cartilage regeneration plays an important role in successful osteochondral repair, since it provides a biological and mechanical transition from the unmineralized cartilage at the articulating surface to the underlying mineralized bone. To biomimic native calcified cartilage in engineered constructs, here we test the hypothesis that hydroxyapatite (HAP) stimulates chondrocytes to secrete the characteristic matrix of calcified cartilage. Sodium citrate (SC) was added as a dispersant of HAP within alginate (ALG), and homogeneous dispersal of HAP within ALG hydrogel was confirmed using sedimentation tests, electron microscopy, and energy dispersive spectroscopy. To examine the biological performance of ALG/HAP composites, chondrocyte survival and proliferation, extracellular matrix production, and mineralization potential were evaluated in the presence or absence of the HAP phase. Chondrocytes in ALG/HAP constructs survived well and proliferated, but also expressed higher levels of calcified cartilage markers compared to controls, including Collagen type X secretion, alkaline phosphatase (ALP) activity, and mineral deposition. Compared to controls, ALG/HAP constructs also showed an elevated level of mineralized matrix in vivo when implanted subcutaneously in mice. The printability of ALG/HAP composite hydrogel precursors was verified by 3D printing of ALG/HAP hydrogel scaffolds with a porous structure. In summary, these results confirm the hypothesis that HAP in ALG hydrogel stimulates chondrocytes to secrete calcified matrix in vitro and in vivo and reveal that ALG/HAP composites have the potential for 3D bioprinting and osteochondral regeneration.

Analyzing Biological Performance of 3D-Printed, Cell-Impregnated Hybrid Constructs for Cartilage Tissue Engineering

Three-dimensional (3D) bioprinting of hybrid constructs is a promising biofabrication method for cartilage tissue engineering because a synthetic polymer framework and cell-impregnated hydrogel provide structural and biological features of cartilage, respectively. During bioprinting, impregnated cells may be subjected to high temperatures (caused by the adjacent melted polymer) and process-induced mechanical forces, potentially compromising cell function. This study addresses these biofabrication issues, evaluating the heat distribution of printed polycaprolactone (PCL) strands and the rheological property and structural stability of alginate hydrogels at various temperatures and concentrations. The biocompatibility of parameters from these studies was tested by culturing 3D hybrid constructs bioprinted with primary cells from embryonic chick cartilage. During initial two-dimensional culture expansion of these primary cells, two morphologically and molecularly distinct cell populations ("rounded" and "fibroblastic") were isolated. The biological performance of each population was evaluated in 3D hybrid constructs separately. The cell viability, proliferation, and cartilage differentiation were observed at high levels in hybrid constructs of both cell populations, confirming the validity of these 3D bioprinting parameters for effective cartilage tissue engineering. Statistically significant performance variations were observed, however, between the rounded and fibroblastic cell populations. Molecular and morphological data support the notion that such performance differences may be attributed to the relative differentiation state of rounded versus fibroblastic cells (i.e., differentiated chondrocytes vs. chondroprogenitors, respectively), which is a relevant issue for cell-based tissue engineering strategies. Taken together, our study demonstrates that bioprinting 3D hybrid constructs of PCL and cell-impregnated alginate hydrogel is a promising approach for cartilage tissue engineering.

The mechanism of ascorbic acid-induced differentiation of ATDC5 chondrogenic cells

The ATDC5 cell line exhibits a multistep process of chondrogenic differentiation analogous to that observed during endochondral bone formation. Previous investigators have induced ATDC5 cells to differentiate by exposing them to insulin at high concentrations. We have observed spontaneous differentiation of ATDC5 cells maintained in ascorbic acid-containing alpha-MEM. A comparison of the differentiation events in response to high-dose insulin vs. ascorbic acid showed similar expression patterns of key genes, including collagen II, Runx2, Sox9, Indian hedgehog, and collagen X. We took advantage of the action of ascorbic acid to examine signaling events associated with differentiation. In contrast to high-dose insulin, which downregulates both IGF-I and insulin receptors, there were only minimal changes in the abundance of these receptors during ascorbic acid-induced differentiation. Furthermore, ascorbic acid exposure was associated with ERK activation, and ERK inhibition attenuated ascorbic acid-induced differentiation. This was in contrast to the inhibitory effect of ERK activation during IGF-I-induced differentiation. Inhibition of collagen formation with a proline analog markedly attenuated the differentiating effect of ascorbic acid on ATDC5 cells. When plates were conditioned with ATDC5 cells exposed to ascorbic acid, ATDC5 cells were able to differentiate in the absence of ascorbic acid. Our results indicate that matrix formation early in the differentiation process is essential for ascorbic acid-induced ATDC5 differentiation. We conclude that ascorbic acid can promote the differentiation of ATDC5 cells by promoting the formation of collagenous matrix and that matrix formation mediates activation of the ERK signaling pathway, which promotes the differentiation program.

Collagen/annexin V interactions regulate chondrocyte mineralization

Physiological mineralization in growth plate cartilage is highly regulated and restricted to terminally differentiated chondrocytes. Because mineralization occurs in the extracellular matrix, we asked whether major extracellular matrix components (collagens) of growth plate cartilage are directly involved in regulating the mineralization process. Our findings show that types II and X collagen interacted with cell surface-expressed annexin V. These interactions led to a stimulation of annexin V-mediated Ca(2+) influx resulting in an increased intracellular Ca(2+) concentration, [Ca(2+)](i), and ultimately increased alkaline phosphatase activity and mineralization of growth plate chondrocytes. Consequently, stimulation of these interactions (ascorbate to stimulate collagen synthesis, culturing cells on type II collagen-coated dishes, or overexpression of full-length annexin V) resulted in increase of [Ca(2+)](i), alkaline phosphatase activity, and mineralization of growth plate chondrocytes, whereas inhibition of these interactions (3,4-dehydro-l-proline to inhibit collagen secretion, K-201, a specific annexin channel blocker, overexpression of N terminus-deleted mutant annexin V that does not bind to type II collagen and shows reduced Ca(2+) channel activities) decreased [Ca(2+)](i), alkaline phosphatase activity, and mineralization. In conclusion, the interactions between collagen and annexin V regulate mineralization of growth plate cartilage. Because annexin V is up-regulated during pathological mineralization events of articular cartilage, it is possible that these interactions also regulate pathological mineralization.

Response of engineered cartilage tissue to biochemical agents as studied by proton magnetic resonance microscopy

OBJECTIVE

To test the hypothesis that magnetic resonance imaging (MRI) results correlate with the biochemical composition of cartilage matrix and can therefore be used to evaluate natural tissue development and the effects of biologic interventions.

METHODS

Chondrocytes harvested from day-16 chick embryo sterna were inoculated into an MRI-compatible hollow-fiber bioreactor. The tissue that formed over a period of 2-4 weeks was studied biochemically, histologically, and with MRI. Besides natural development, the response of the tissue to administration of retinoic acid, interleukin-1beta (IL-1beta), and daily dosing with ascorbic acid was studied.

RESULTS

Tissue wet and dry weight, glycosaminoglycan (GAG) content, and collagen content all increased with development time, while tissue hydration decreased. The administration of retinoic acid resulted in a significant reduction in tissue wet weight, proteoglycan content, and cell number and an increase in hydration as compared with controls. Daily dosing with ascorbic acid increased tissue collagen content significantly compared with controls, while the administration of IL-1beta resulted in increased proteoglycan content. The water proton longitudinal and transverse relaxation rates correlated well with GAG and collagen concentrations of the matrix as well as with tissue hydration. In contrast, the magnetization transfer value for the tissue correlated only with total collagen. Finally, the self-diffusion coefficient of water correlated with tissue hydration.

CONCLUSION

Parameters derived from MR images obtained noninvasively can be used to quantitatively assess the composition of cartilage tissue generated in a bioreactor. We conclude that MRI is a promising modality for the assessment of certain biochemical properties of cartilage in a wide variety of settings.

Proteoglycan synthesis by cultured human chondrocytes

Iliac crest biopsies are important in the detection of human skeletal dysplasias. Therefore, culture of these cells may serve as a valuable method for studying proteoglycan metabolism in chondrocytes of individuals with skeletal abnormalities. Morphological and biochemical studies were performed on human iliac crest chondrocytes grown in monolayer and in agarose gels. Two proteoglycan populations of different hydrodynamic size and glycosaminoglycan composition were synthesized by cells grown in monolayer. Chondrocytes cultured in an agarose gel for 2 weeks synthesized proteoglycans identical to those of the native tissue with respect to hydrodynamic size and glycosaminoglycan chain length. However, the ratio of chondroitin-6-sulfate to chondroitin-4-sulfate was higher than in the native tissue. This ratio was not influenced by different sulfate concentrations in the medium. Moreover, treatment with ascorbic acid did not influence proteoglycan synthesis; however, there was a pericellular accumulation of proteoglycans.

Heterogeneous distribution of the precursor of type I and type III collagen and fibronectin in the rough endoplasmic reticulum of palatal mesenchymal cells of the mouse embryo cultured in ascorbate-depleted medium

In order to examine the intracellular distribution of precursors of type I and type III collagen and fibronectin in the palatal mesenchymal (MEPM) cells of the mouse embryo cultured under ascorbate-deficient conditions, immuno-electron-microscopic studies were carried out by use of affinity purified antibodies for these proteins. MEPM cells were obtained from the palatal shelves of 14-day-old mouse fetuses and cultured for 3-7 days in medium, either with or without 50 ng/dish/day ascorbic acid. Results obtained were as follows: (1) Although the rough endoplasmic reticulum (rER) of MEPM cells cultured for 5 days in ascorbate-supplemented medium was flattened, that in cells cultured in ascorbate-deficient medium had a distended or vesicular appearance. (2) Vesicular or distended rER showed heterogeneous staining for both type I and type III collagen, namely, some parts of rER showed positive staining for both types of collagen, while others showed negative staining. (3) Both type I and type III collagen showed codistribution in the same vesicular rER. (4) Vesicular rER showed negative or very faint labelling for fibronectin. These results may suggest regional differences in the function of rER.

The role of ascorbic acid in mesenchymal differentiation

Survival of all higher vertebrates requires that they either synthesize vitamin C (ascorbic acid) or obtain it from their diet. The role of ascorbic acid as a reductant for the iron prosthetic group of hydroxylase enzymes involved in collagen biosynthesis is well established. In contrast, the relationship between the biochemical functions of ascorbic acid and the broad defects in connective tissue formation associated with vitamin C deficiency is less obvious. This review will develop the hypothesis that vitamin C is required for the differentiation of mesenchyme-derived connective tissues such as muscle, cartilage, and bone. It is proposed that the collagen matrix produced by ascorbic acid-treated cells provides a permissive environment for tissue-specific gene expression.

Independent secretion of proteoglycans and collagens in chick chondrocyte cultures during acute ascorbic acid treatment

The mechanisms regulating the secretion of proteoglycans and collagens in chondrocytes, in particular those operating at the level of the rough endoplasmic reticulum (RER), are largely unknown. To examine these mechanisms, I studied the effects of acute ascorbate treatment on the secretion of two collagen types (types II and IX) and two proteoglycan types (PG-H and PG-Lb, the major keratan sulphate/chondroitin sulphate proteoglycan and the minor chondroitin sulphate proteoglycan respectively in cartilage) in scorbutic cultures of chick vertebral chondrocytes. I found that the scorbutic chondrocytes synthesized underhydroxylated precursors of types II and IX collagen that were secreted very slowly and accumulated in the RER. When the cultures were treated acutely with ascorbate, both macromolecules underwent hydroxylation within 1-1.5 h of treatment, and began to be secreted at normal high rates starting at about 2 h. Proteoglycan synthesis and secretion, however, remained largely unaffected by ascorbate treatment. Both the half-time of newly synthesized PG-H core protein in the RER and its conversion into completed proteoglycan were unchanged during treatment. Similarly, the overall rates of synthesis and secretion of both PG-H and PG-Lb remained at control levels during treatment. The data indicate that secretion of types II and IX collagen is regulated independently of secretion of PG-H and PG-Lb. This may be mediated by the ability of the RER of the chondrocyte to discriminate between procollagens and proteoglycan core proteins.

Ascorbic acid stimulates production of glycosaminoglycans in cultured fibroblasts

The effect of ascorbic acid on collagen synthesis is well characterized. Proteoglycans and their attached glycosaminoglycans are components of the extracellular matrix closely associated with collagen fibers. We examined whether ascorbic acid also plays a role in glycosaminoglycan production. Synthesis and deposition of glycosaminoglycans into the extracellular matrix and secretion into the media were followed in human skin fibroblasts cultured in the presence and absence of ascorbic acid. Specific glycosaminoglycans were identified and quantitated by differential enzyme digestion, ion-exchange column chromatography, and cellulose-acetate electrophoresis. No major qualitative changes in glycosaminoglycans were observed. However, quantitatively, synthesis of glycosaminoglycans increased 30 to 90%, and deposition into the extracellular matrix increased 80% in the presence of ascorbic acid. This effect was only in part secondary to decreased levels of collagen, and the diminished capacity of underhydroxylated collagen to bind proteoglycans. The effect of ascorbic acid on extracellular macromolecules is thus more pervasive than previously assumed.

Proteoglycan core protein and type II collagen gene expressions are not correlated with cell shape changes during low density chondrocyte cultures

Chondrocytes isolated from chicken embryo sterna were cultivated in low density monolayer cultures to induce their dedifferentiation. At different stages of the long-term cultures, changes in expression of a cartilage-specific sulfated proteoglycan and cartilage-characteristic type II collagen have been examined and related to the shape change of cells using in situ hybridization and immunocytochemistry. At the beginning of the culture, all cells exhibit a round shape and express the cartilage phenotype. Then, during the course of the culture, chondrocytes flatten and become fibroblast-like, but this morphological modification does not start for all the cells at the same time. Interestingly, the loss of cartilage proteoglycan or type II collagen expression did not occur for all polygonal or fibroblast-like cells. Moreover, we observed a variability in the steady state levels of RNA or protein accumulation among chondrocytes exhibiting a similar shape, as judged by the intensity of hybridization signal or immunofluorescence over the cells. These observations support the hypothesis that the shape change does not have a causative role in the chondrocyte phenotype expression, but is rather a secondary effect of the dedifferentiation process. Furthermore, the disappearance of hybridizable core protein or type II collagen mRNA during the dedifferentiation process was coincident with the disappearance of the proteins for which they code as detected by immunohistochemical staining. This suggest that core protein and type II collagen gene expressions are controlled primarily at the transcriptional level in long-term chondrocyte cultures.

Ultrastructural localization of the major proteoglycan and type II procollagen in organelles and extracellular matrix of cultured chondroblasts

The mechanisms of synthesis and intracellular routing of the various cartilage matrix macromolecules are still unclear. We have studied this problem in cultured chondroblasts at the ultrastructural level using monospecific antibodies against the core protein of the keratan sulfate/chondroitin sulfate-rich cartilage proteoglycan (KS:CS-PG) or Type II procollagen, and cuprolinic blue, a cationic dye that binds to the glycosaminoglycan chains of proteoglycans. Intracellularly, the proteoglycan antibodies localized KS:CS-PG and its precursors primarily in the Golgi complex and secretory vesicles. In contrast, the bulk of Type II procollagen was found within the rough endoplasmic reticulum (ER). While devoid of collagen, the extracellular matrix was rich in KS:CS-PG molecules some of which studded the chondroblast plasmalemma. Cuprolinic blue staining indicated that the proteoglycans present in the Golgi complex fell into a predominant class of large proteoglycans, probably representing KS:CS-PG, and a minor class of smaller proteoglycans. Groups of these divergent proteoglycans often occupied distinct Golgi subcompartments; moreover, single large proteoglycans appeared to align along the luminal surface of Golgi cisternae and secretory vesicles. These results suggest that in cultured chondroblasts KS:CS-PG and Type II procollagen are differentially distributed both in organelles and in the extracellular matrix, and that different proteoglycan types may occupy distinct subcompartments in trans Golgi.

Effect of ascorbic acid on secreted proteoglycans from rabbit articular chondrocytes

Addition of ascorbic acid (25, 50 100 micrograms/ml) to cultures of rabbit articular chondrocytes did not change the total amount of proteoglycans produced. However, it induced an increased retention of these macromolecules in the pericellular fraction. The size of the proteoglycan subunits and the length of glycosaminoglycan chains, released in the medium, were not modified on exposure to ascorbic acid (25 micrograms/ml). On the other hand, the rate of non-sulfated chondroitin was increased 2.5-fold, whereas chondroitin-4-sulfate was depressed 1.5-fold.

Immunofluorescence studies on cartilage matrix synthesis. The synthesis of link protein, chondroitin sulfate proteoglycan monomer and type II collagen

A comparison of the synthesis and deposition of fibrous type II collagen and the constituents of chondroitin sulfate proteoglycan (CSPG) aggregates, CSPG monomer and link protein, was made for chicken sternal chondrocytes in culture, using simultaneous double immunofluorescence and lectin localization. Chondrocytes deposited only CSPG constituents--and not type II collagen--into the extracellular matrix (ECM). Intracellular precursors of CSPG monomer were localized primarily in perinuclear regions, but were observed in other cytoplasmic vesicles as well. Link protein antibodies stained the same intracellular structures, but stained the perinuclear cytoplasm less intensely. In contrast, type II procollagen was distributed in vesicles throughout the cytoplasm and was clearly absent from the distinctive, CSPG precursor-containing vesicles. Fluorescence-labelled lectins were used to further identify intracellular membrane compartments. Wheat germ agglutinin (WGA) and Ricinus lectins (which recognize carbohydrates added in the Golgi) stained the perinuclear cytoplasm, while concanavalin A (conA) (which recognizes mannose-rich oligosaccharides added co-translationally) stained vesicles throughout the rest of the cytoplasm and not the perinuclear cytoplasm. The distinctive CSPG-containing vesicles were not stained with WGA or Ricinus agglutinins. Data presented elsewhere demonstrate that the vesicles do not react with monoclonal antibodies which recognize chondroitin sulfate (CS) or keratan sulfate (KS) determinants. Thus, we conclude that the vesicles accumulate CSPG precursors which have not been modified by Golgi-mediated processes. The data indicate that matrix molecules may be segregated selectively prior to transit through the Golgi complex. The co-distribution of link protein and CSPG monomer precursors in vesicles prior to further, Golgi-mediated modification may reflect an as yet undetermined function of these vesicles in the processing or assembly of CSPG.

Synthesis of cartilage matrix by mammalian chondrocytes in vitro. III. Effects of ascorbate

Chondrocytes isolated from bovine articular cartilage were plated at high density and grown in the presence or absence of ascorbate. Collagen and proteoglycans, the major matrix macromolecules synthesized by these cells, were isolated at times during the course of the culture period and characterized. In both control and ascorbate-treated cultures, type II collagen and cartilage proteoglycans accumulated in the cell-associated matrix. Control cells secreted proteoglycans and type II collagen into the medium, whereas with time in culture, ascorbate-treated cells secreted an increasing proportion of types I and III collagens into the medium. The ascorbate-treated cells did not incorporate type I collagen into the cell-associated matrix, but continued to accumulate type II collagen in this compartment. Upon removal of ascorbate, the cells ceased to synthesize type I collagen. Morphological examination of ascorbate-treated and control chondrocyte culture revealed that both collagen and proteoglycans were deposited into the extracellular matrix. The ascorbate-treated cells accumulated a more extensive matrix that was rich in collagen fibrils and ruthenium red-positive proteoglycans. This study demonstrated that although ascorbate facilitates the formation of an extracellular matrix in chondrocyte cultures, it can also cause a reversible alteration in the phenotypic expression of those cells in vitro.

Immunofluorescence studies of chondroitin sulfate proteoglycan biosynthesis: the use of monoclonal antibodies

Several monoclonal antibodies which recognize different antigenic determinants of chondroitin sulfate proteoglycan were used to study chondroitin sulfate proteoglycan biosynthesis in chicken chondrocyte cultures. The intracellular sites of synthesis and processing and extracellular deposition in matrix were localized by double immunofluorescence reactions. One rat monoclonal antibody, S103L , which recognizes an antigenic determinant of the core protein of the chicken cartilage chondroitin sulfate proteoglycan monomer, was used to identify both extracellular chondroitin sulfate proteoglycan and intracellular compartments containing chondroitin sulfate proteoglycan precursors. Intracellular staining with S103L was localized to perinuclear regions, and, in some chondrocytes, to a few other cytoplasmic vesicles as well. When chondrocytes were not fed for several days, intracellular chondroitin sulfate proteoglycan precursors were accumulated in larger compartments distributed throughout the cytoplasm. Polyclonal chondroitin sulfate proteoglycan antibodies displayed similar staining characteristics. In contrast, several of the monoclonal antibodies, including the rat monoclonals S11D and P100D , and the mouse monoclonals 1-B-5, 3-B-3 and 9-A-2, did not recognize native chondroitin sulfate proteoglycan, but reacted only with chondroitinase ABC-digested (and/or hyaluronidase-digested) chondroitin sulfate proteoglycan. These antibodies were particularly useful in the demonstration of the extracellular codistribution of chondroitin sulfate proteoglycan with either type II collagen or fibronectin. In other experiments, the monoclonal antibodies to chondroitin sulfate proteoglycan served to demonstrate that the perinuclear subset of intracellular compartments is uniquely involved in the addition of chondroitin sulfate oligosaccharides to the chondroitin sulfate proteoglycan core protein. Lastly, using the mouse monoclonal 5-D-4, which recognizes keratan sulfate determinants, the perinuclear region was identified as the site for keratan sulfate addition. Results suggest heterogeneity of keratan sulfate synthesis at the level of individual chondrocytes, even for cells apparently containing equivalent amounts of intracellular chondroitin sulfate proteoglycan.

Two distinct classes of prechondrogenic cell types in the embryonic limb bud

Differences are demonstrated in the chondrogenic potential of cells derived from the distal and proximal halves of chick wing buds from as early as stage 23, prior to the appearance of overt cartilage differentiation. In high cell density cultures, cells obtained from the distal portions of stage 23 or 24 limb buds are spontaneously chondrogenic in micromass cultures. Cells obtained from the proximal portions, however, become blocked in their differentiation as protodifferentiated cartilage cels, since these cells in micromass cultures make detectable type II collagen, but fail to synthesize significant levels of cartilage proteoglycan or to accumulate an extracellular matrix that will stain for sulfated glycosaminoglycans. Such cultures of proximal limb bud cells can be stimulated to form alcian blue staining nodules by the addition of 1 mM dbcAMP or 50 micrograms/ml ascorbate, or by mixing proximal cells with small numbers of distal cells (1 distal cell to 10 proximal cells). These results demonstrate the existence of two distinct stages among prechondrogenic mesenchyme cells. The earlier stage appears to be able to provide a chondrogenic stimulus to proximal cells.

Selective emergence of differentiated chondrocytes during serum-free culture of cells derived from fetal rat calvaria

Cells dispersed from the chondrocranial portions of fetal rat calvaria proliferated and performed specialized functions during primary culture in a chemically defined medium. Mature cultures were typified by multilayered clusters of redifferentiating cartilage cells. Flattened cells that lacked distinguishing features occupied areas between the clusters. Alkaline phosphate-enriched, ultrastructurally typical chondrocytes within the clusters were encased in a dense extracellular matrix that stained prominently for chondroitin sulfate proteoglycans. This matrix contained fibrils measuring 19 nm in diameter, which were associated with proteoglycan granules that preferentially bound ruthenium red. A progressive increase in the number of cells indicated the proliferation of certain elements in the primary culture. The cells in primary culture were biochemically as well as morphologically heterogeneous since they were found to synthesize type I and type II collagens. Homogeneous populations of redifferentiated chondrocytes were recovered as floating cells and were shown to express the chondrocyte phenotype in secondary culture. Subcultured cells synthesized type II collagen and its precursors almost exclusively and incorporated 35SO4 into proteoglycan monomer and aggregates to a greater degree than the cells in primary culture. The pattern of proteoglycan monomer and aggregate labeling resembled that of intact cartilage segments and bovine articular chondrocytes. Skin fibroblasts harvested from the same rat fetuses failed to proliferate when maintained under identical conditions. Hence, exogenous hormones, growth factors, and protein are not required for chondrocyte growth and maturation.

An ultrastructural study of fibroblasts derived from bovine ligamentum nuchae and their capacity for elastogenesis in culture

Fibroblast cultures were readily propagated from fetal bovine ligamentum nuchae. The ligament cells were easily cultured by standard techniques and were maintained in culture flasks for up to 57 days. During this time they accumulated an extensive extracellular matrix which contained the main structural elements of the parent tissue, namely collagen and elastic fibres. Elastogenesis was seen to proceed in two phases: the formation of parallel bundles of 10--12 nm wide microfibrils followed by the deposition within these bundles of amorphous elastin-like material. Elastic fibres were not produced in cultures that were supplemented with ascorbic acid either in the absence or presence of the lathyrogen BAPN.