The subcellular distribution of prolyl hydroxylase determined immunohistochemically

Prolyl hydroxylase catalyses the biosynthesis of hydroxyprolyl residues in collagen. In the presence of oxygen and various cofactors it hydroxylates prolyl residues in collagen and model substrates which occur in sequences of the general structure –X–Pro–Gly–, where X is alanine or a variety of other amino acids but not glycine (McGee, Rhoads & Udenfriend, 1971). Although it is established that proline must be present in peptide linkage before enzymic hydroxylation, it is not clear in which intracellular compartment this mainly occurs. Available evidence suggests that hydroxylation takes place either in the ‘ground’ cytoplasm or within the cisternae of the rough endoplasmic reticulum (Grant & Prockop, 1972). The point at which hydroxylation takes place in the process of collagen chain biosynthesis is also not resolved. One view holds that hydroxylation proceedsparipassu with growth of nascent chains (Miller & Udenfriend, 1970), while the other claims that hydroxylation takes place after the completion of chain synthesis but before its secretion from the cell (Prockop, 1970). The present investigation was undertaken in an attempt to resolve some of these problems. This paper reports the subcellular distribution of prolyl hydroxylase in rat fibroblasts and chondrocytes as determined by immunohisto-chemistry, and by ultracentrifugation analysis of fibroblast homogenates. A preliminary report of this work has been presented elsewhere (Al-Adnani, Patrick & McGee, 1973).

Horseradish peroxidase (type VI) and 3,3’-diaminobenzidine tetrachloride were purchased from Sigma Chemical Co. (London). Rabbit anti-goat IgG (IgG fraction) and goat IgG were obtained from Cappel Laboratories (Downington, Pennsylvania) and 4,4’-difluoro-3,3’-dinitrodiphenylsulphone (FNPS) from K & K Laboratories (Los Angeles). The specific anti serum to rat prolyl hydroxylase used throughout this study was prepared as described elsewhere (McGee, Langness & Udenfriend 1971; Roberts, McGee & Udenfriend, 1973). All tissue culture reagents were obtained from Biocult Laboratories (Paisley, Scotland).

Rabbit anti-goat IgG (IgG fraction) was conjugated with horseradish peroxidase essentially as described by Nakane & Pierce (1967). Horseradish peroxidase, 50 mg, and rabbit anti-goat IgG, 50 mg, were dissolved in 2 ml of 0·5 M carbonate buffer, pH 10·0, and 0·25 ml of FNPS (°’S% in acetone) was added slowly with constant stirring on ice. This solution was stirred gently for 6 h at 4 °C and then dialysed overnight against 0·05 M phosphate buffer, pH 7·5, containing 0·15 M NaCl (PBS). The precipitate was removed by ccntrifugation and an equal volume of saturated ammonium sulphate added to the supernatant. The precipitated protein recovered by centrifugation was redissolved in 2 ml of PBS and the IgG peroxidase conjugate separated from unreacted IgG and peroxidase by molecular sieve chromatography on a column of G200-Sephadex (90 × 1·5 cm) equilibrated and eluted with PBS. The immunoreactive IgG peroxidase conjugate was present in the descending limb of the first protein peak eluted; this was pooled and frozen in aliquots of 1 ml at – 20 °C until required.

Fibroblast cultures were prepared from rat embryonic skin (15 – 18 days old). The skin was disaggregated by trypsin (025% in 0·15 M NaCl) and the detached cells grown in glass culture flasks in an air atmosphere at 37 °C. The culture medium consisted of minimum essential medium buffered with HEPES containing 10% foetal calf serum, 2·00 mM glutamine, 0·25 mM ascorbate, 0·25 × 10⁻⁶M ferric nitrate together with 50 units penicillin and 50 units streptomycin per ml. Skin fibroblasts were grown also on glass slides placed in Petri dishes under the same conditions.

For electron microscopy monolayers were washed briefly in 0·15 M NaCl and the cells fixed for 30 min at 4 °C in 4% paraformaldchyde in 0·01 M phosphate, pH 7·4, containing 5% sucrose. The latter buffer was used for all subsequent washing procedures. After fixation the monolayers were washed and the cells scraped from the flask with a rubber policeman into the same buffer. The cells were pelleted at 600 g for 10 min and resuspended in 1 ml of a 1 in 10 dilution of specific antiserum to rat prolyl hydroxylase for 60 min. The cells were washed for 10 min and resuspended in 1 ml of anti-goat IgG peroxidase conjugate for 2 h. After a further 15-min wash they were fixed for 60 min in 2% glutaraldehyde in 0·06 M phosphate buffer, pH 7·4. The bound cellular conjugate was localized by its peroxidase activity following incubation for 15 min in a reaction mixture containing 0 ·075% 3,3’-diaminobenzidine and 0·001% hydrogen peroxide in 0·05 M Tris buffer, pH 76 (Graham & Karnovsky, 1966). The cells were washed and osmicated for 60 min in 1% osmium tetroxide and then processed for electron microscopy by routine methods. Ultrathin sections were viewed in a Philips 200 or 301 electron microscope at 60 kV.

Prolyl hydroxylase was localized also in rat embryonic cartilage essentially as described for monolayers with the following modifications. Whole tibias from 1-week-old rat embryos were fixed in 4% buffered paraformaldehyde for 4 h then washed overnight in buffer at 4 °C. Sections were cut at 50 μ m, processed through the immunohistochemical procedure and embedded in Epon as described above.

Prolyl hydroxylase activity of fibroblast cultures was measured by the tritium release method (Hutton, Tappel & Udenfriend, 1966) after cell sonication in conditions similar to those described elsewhere (McGee, Langness & Udenfriend, 1971).

The cell cultures used in this investigation had a fibroblast morphology and produced abundant collagen fibres as determined by reticulin stains and electron microscopy. Prolyl hydroxylase activity in these cells increased several fold as the cultures progressed from log to stationary phase of growth (Fig. 1); this is similar to results obtained using established fibroblast lines (Comstock & Udenfriend, 1970). The immunoreactivity of the conjugate was established by immunodiffusion against goat IgG; the peroxidase activity of the conjugate was detected by its reaction with diaminobenzidine and hydrogen peroxide directly on the immunodiffusion plate (Fig. 2A, B).

The conditions for detecting the enzyme immunohistochemically were determined originally at the light-microscope level using fibroblasts growing on glass slides. These cells were always in late log phase when prolyl hydroxylase activity was maximal. Among several fixatives employed (commercial formalin, paraformaldehyde, glutaraldehyde, osmium tetroxide, acetone and ethanol), 4% paraformaldehyde produced the most acceptable degree of cell preservation without abolishing the antigenicity of the enzyme. Its localization at the light-microscope level is shown in Fig. 3; the peroxidase reaction product is seen as discrete granules scattered throughout the cell, usually in highest concentration around the nucleus. These fibroblasts did not contain any endogenous peroxidase activity. The reaction illustrated in Fig. 3 A was not observed when normal goat serum was substituted for antiserum (Fig. 3B). Similarly the reaction was abolished by first absorbing the antiserum with rat skin prolyl hydroxylase.

All of the ultrastructural observations were carried out on late log-phase cultures or on rat embryonic cartilage. The electron-microscopic appearances of cultured cells treated with normal goat serum and anti-goat IgG peroxidase conjugate is shown in Fig. 4. No peroxidase reaction product is seen in these preparations or in cartilage controls. It should be noted that in control cells not subsequently stained with heavy metal salts the ribosomes on the rough endoplasmic reticulum (RER) are extremely difficult to visualize. This seems to be a general featuie of tissue fixed in paraformaldehyde since we have recently observed the same phenomenon with rough endoplasmic reticulum of hepatocytes preserved in this fixative. The RER, however, is easily identified by the parallel arrangement of its membranes and the frequent presence of an electron-dense material within its channels (Figs. 5, 6) which is a normal feature of such cells (Ross & Benditt, 1961). The ultrastructural localization of prolyl hydroxylase is illustrated in Figs. 5–7. The enzyme, identified by the electron-dense peroxidase reaction product, is distributed along the membranes of the RER. On some membranes it has a linear distribution (Figs. 5, 7), but on others it appears focal (Fig. 6), tentatively suggesting that prolyl hydroxylase is spatially related to the ribosomes. A reaction product was never observed within RER cisternae nor in the Golgi apparatus. Of particular note, no reaction was ever found on the nuclear envelope of any cell examined (Fig. 7).

Centrifugation analysis of the subcellular distribution of collagen proline hydroxylase in late log-phase rat fibroblasts is shown in Table 1. About 60% of the enzyme activity recoverable from a cell-free homogenate was present in the 100000 g pellet and about 40% in the 10000g supernatant. If each centrifugal fraction after its isolation was treated with 0·1% Triton X-100 before assay the amount of enzyme activity detectable in the 600 g supernatant and the 100000 g pellet increased 2·4- and 2·7-fold, respectively, while the 100000g supernatant showed no change; on this basis about 80% of the enzyme recoverable from a cell-free homogenate is present in the microsomal fraction and about 20% in the 100 000 g supernatant. It is probable that Triton treatment reveals enzyme ‘buried’ in membranes which is not accessible to substrate under normal assay conditions. It is conceivable of course that Triton may also release enzyme from within the cisternae of the RER but considering the immunohistochcmical results this seems less likely. It is important to note that these experiments were performed in the absence of EDTA and dithiothreitol, both of which ‘stabilize’ the activity of prolyl hydroxylase. When fibroblasts are ruptured in the presence of the latter reagents 90% of this enzyme is found in the 100000 g supernatant (Roberts et al. 1973).

A late log-phase monolayer culture of rat fibroblasts was washed twice with 0·06 M phosphate buffer (pH 7·4) containing 5% sucrose and the cells scraped off the flask with a rubber police-man into the same buffer. (This buffer was identical to that used for the preparation of cells for immunohistochemistry.) The cells were recovered by centrifugation, resuspended in the same buffer and homogenized in a loose-fitting all glass homogenizer. The homogenate was centrifuged at 600 g for 5 min. An aliquot of this supernatant was kept on ice for enzyme assay and the remainder centrifuged at 100000 g for 90 min. The 100000g pellet was resuspended in the same sucrose solution used for homogenization.

Prolyl hydroxylase activities of all fractions were measured together at the end of the centrifugation procedure. The results recorded are the total enzyme activity of each fraction.

The immunohistochemical observations illustrated in this paper indicate that prolyl hydroxylase antigen is present exclusively on or in the membranes of the rough endoplasmic reticulum in fibroblasts and ehondroeytes. The focal distribution of the antigen on the RER in some cells suggests that it may be spatially associated with ribosomes but not necessarily an integral part of their structure. Since late log-phase fibroblasts contain only enzyme in the active form (McGee & Udenfriend, 1972) the distribution of the peroxidase reaction product reflects the intracellular localization of active enzyme. These observations, together with the ultracentrifugation data demonstrating that 80% of the activity of prolyl hydroxylase is membrane-bound, indicate that the bulk of this enzyme is intimately associated with the site of protein synthesis and suggests therefore that hydroxylation and collagen chain synthesis occur at the same intracellular site. This evidence contradicts earlier reports that hydroxylation may take place in the cell cytoplasm or within the cisternae of the RER (Grant & Pròckop, 1972). The latter suggestions were based on the observation that, in purification schemes for prolyl hydroxylase, 90% of the enzyme activity in collagen-synthesizing tissue is recoverable in the soluble protein fraction (Roberts et al. 1973). However, in these experiments the tissue was subjected to severe homogenization procedures in the presence of reducing and chelating reagents which might easily dislodge enzyme from membranes. More recently, however, other groups have reported similar data indicating that a significant amount of prolyl hydroxylase is released from microsomal fractions by Triton treatment (Guzman & Cutroneo, 1973).

The current concept of piotein synthesis in mammalian cells visualizes the nascent chain growing through the ribosome into the lumen of the endoplasmic reticulum (Sabatini & Blobel, 1970) and the completed protein subsequently secreted. If collagen polypeptides were hydroxylated after completion of translation (Prockop, 1970) it would be expected that prolyl hydroxylase would be found distal to the site of synthesis and in the lumen of the RER or elsewhere. Since enzyme was never found in the latter locations but only in close association with the membranes of the protein-synthesizing machinery it could be postulated that the immunohisto-chemical evidence is more in keeping with the view that proline hydroxylation takes place as collagen chains are being assembled (Miller & Udenfriend, 1970), rather than after the completion of polypeptide synthesis (Prockop, 1970). However, the inherent limits in resolution obtainable with the peroxidase-labelled antibody procedure are such that it is not possible to exclude that enzyme exists on the cisternal side of the membranes of the RER closely adjacent to the point where completed polypeptides emerge from the ribosome. If the enzyme existed in this situation hydroxylation could occur after completion of translation.

The observation that prolyl hydroxylase is absent from the nuclear membrane is interesting. Although it is not possible to exclude entirely artifactual loss of antigen during the preparative procedures, it is tempting to speculate that this observation represents functional differences between the ribosome-covered nuclear envelope and the membranes of the RER within fibroblasts.

After completion of this work Olsen, Berg, Kishiday & Prockop (1973) reported on the location of prolyl hydroxylase in a non-mammalian system. Using a direct ferritin-labelled antibody procedure, it was shown that the enzyme existed within the cisternae of the RER. However, this result was obtained by examining fragments of formaldehyde-fixed cells which had been disrupted by brisk homogenization and subjected to centrifugation at 20000 g. It is not impossible that these preparative procedures could displace prolyl hydroxylase from membranes of the RER into the cisternae as the authors themselves concede. Moreover, in one electron micrograph illustrating their paper, ferritin particles can be seen unassociated with any organellar structure, suggesting that the enzyme could have been dislocated from the membranes of the RER not only into the cisternae but also diffusely into the cell sap. In intact cells we never found reaction product inside the lumen of the RER. Furthermore, recent data based on ultracentrifugation analysis of collagen-synthesizing systems have also led others to the conclusion that prolyl hydroxylase is attached to the membranes of the rough endoplasmic reticulum (Diegelman, Bernstein & Peterkofsky, 1973).

This work has been carried out in part with the aid of a grant from the Medical Research Council to one of us (R.S.P.). We wish to acknowledge also support from the Distillers Company Ltd. One of us (M.S.A.) is supported by a scholarship from the Ministry of Higher

Education of Iraq. We have received valuable technical assistance from Mr S. Kane and Mrs M. Rodger.