Abstract
Fibroblast-like (Type B) synoviocytes are cells in the synovial membrane that are responsible for production of both synovial fluid and the extracellular matrix in the synovial intima. Immunostaining of the horse synovial membrane for protein gene product (PGP) 9.5, which is a neuron-specific ubiquitin C-terminal hydrolase, demonstrated selective localization of the immunoreactivity in a synoviocyte population different from acid phosphatase-positive Type A synoviocytes. The immunoreactive cells were lined up in the synovial intima and extended dendritic processes towards the joint cavity to form a dense plexus on the surface. Electron microscopic examination clearly identified the PGP 9.5-immunoreactive cells as Type B synoviocytes characterized by developed rough endoplasmic reticulum and free ribosomes. Immunoreactivity for PGP 9.5 was diffusely distributed throughout the cytoplasm, including the tips of fine processes. Western and Northern blot analyses could not distinguish the corresponding protein and mRNA obtained from the brain and synovial membrane. The existence of the neuron-specific PGP 9.5 in Type B synoviocytes suggests a common mechanism regulating the protein metabolism between neurons and synoviocytes, and also provides a new cytochemical marker for identification of the cells.
S
Protein gene product (PGP) 9.5 was originally isolated from the human brain as a brain-specific protein with a molecular weight of approximately 27,000 (Jackson and Thompson 1981; Doran et al. 1983). PGP 9.5 is structurally related to ubiquitin C-terminal hydrolase and has been confirmed to exhibit its enzymatic activity (Wilkinson et al. 1989). Immunohistochemical studies revealed the predominant localization of PGP 9.5 in neural elements of the central and peripheral nervous systems (Thompson et al. 1983). PGP 9.5 is also shared by both sensory paraneurons, such as olfactory receptor cells, taste bud cells, and cutaneous Merkel cells, and endocrine paraneurons, such as pituitary endocrine cells, thyroid parafollicular cells, adrenal medullary cells, and pancreatic islet cells (Thompson et al. 1983; Wilson et al. 1988; Wang et al. 1990; Iwanaga et al. 1992). Because the cellular distribution of PGP 9.5 immunoreactivity other than neurons and paraneurons is restricted to some cells of the kidney, ovary, and testis (Wilson et al. 1988; Fraile et al. 1996), it is still believed that PGP 9.5 is the most reliable marker for neurons and para-neurons.
This article reports the selective localization of PGP 9.5 in Type B synoviocytes in the horse joint. Although we cannot present here the neuronal and paraneuronal nature of the horse synoviocytes, immunostaining for PGP 9.5 clearly demonstrated the detailed distribution and unique shape of Type B synoviocytes, which have not been previously documented.
Materials and Methods
Immunohistochemistry for PGP 9.5
Seven male thoroughbred horses without any abnormalities in the joint were used in this study. All experiments were performed under protocols following the Guidelines for Animal Experimentation, Graduate School of Veterinary Medicine, Hokkaido University. The animals were deeply anesthetized with pentobarbital and thiopental sodium and were sacrificed by bloodletting from the cervical artery. Joint capsules at the carpal joints on both sides were quickly removed and fixed in 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4) for 24 hr. They were dissected into small pieces (1 × 1 × 1 cm) and dipped in 30% sucrose solution overnight at 4C. Frozen sections about 20 μm thick were prepared in a cryostat (CM 3050; Leica, Nussloch, Germany) and mounted on poly-L-lysine-coated glass slides.
Immunostaining of PGP 9.5 was carried out according to the avidin-biotin complex (ABC) method. The sections were treated with 0.3% Triton X-100 in 0.1 M PBS, pH 7.4, for 1 hr and then with 0.3% H2O2 in methanol for inhibition of endogenous peroxidase activity. After treatment with a blocking serum for 30 min, the sections were incubated with a rabbit polyclonal antiserum against human PGP 9.5 (RA95101; Ultraclone, Isle of Wight, UK) diluted from 1000 to 48,000 overnight at room temperature (RT). The sections were then incubated with biotinylated goat anti-rabbit immunoglobulins and ABC (Histofine kit; Nichirei, Tokyo, Japan). The antigen-antibody reaction was visualized by incubation in 0.05 M Tris-HCl buffer (pH 7.6) containing 0.01% 3,3′-diaminobenzidine (DAB) and 0.002% H2O2. In some cases, DAB reactions were enhanced by adding 0.04% nickel ammonium sulfate.
For immunostaining of whole-mount preparations, we used flat-surfaced synovial membrane and small pieces of the villous region, which contained approximately 10 villi. The staining procedure was essentially identical to the method used for frozen sections except for the constant use of Triton X-100-containing PBS for dilution of the antiserum and rinsing of sections. The tissues were always processed in a free-floating manner in glass bottles shaken by a rotator. After all the steps of immunostaining, the flat-surfaced synovial membrane and villi, which were cut off at the base of villi, were mounted on glass slides and sealed by coverslips with glycerin-gelatin.
Double Staining with Immunohistochemistry and Acid Phosphatase
For detection of acid phosphatase activity, frozen sections were stained according to Burnstone (1958). Control experiments for acid phosphatase reactions were simultaneously carried out by incubating with the medium containing 10 mM NaF, a potent inhibitor for this enzyme. The activity was completely inhibited by preincubation of the medium with NaF. After acid phosphatase histochemistry, the sections were immunostained using the PGP 9.5 antiserum as mentioned above.
Immunoelectron Microscopy
Frozen sections of the synovial membrane 20 μm thick were stained with the PGP 9.5 antiserum as described above except for omitting the treatment with Triton X-100 in PBS. After DAB reaction, stained sections were postfixed with 1% OsO4 for 30 min. They were dehydrated through a graded series of ethanol and embedded directly in Epon 812 on glass slides. After the embedded sections were detached from the glass slides, ultrathin sections were prepared and stained briefly with lead citrate for observation with a transmission electron microscope (H-7100; Hitachi, Tokyo, Japan).
Western Blotting
The horse synovial membrane, brain, jejunum, and liver were homogenized with 50 mM Tris-HCl buffer (pH 7.5) containing 10 mM EDTA and solubilized in the presence of 1 mM phenylmethylsulfonyl fluoride, 10 mM EDTA, and 1% Triton X-100 for 30 min on ice. The samples were reduced with 0.7 M 2-mercaptoethanol and subjected to 12% polyacrylamide electrophoresis. The proteins were electrophoretically transblotted onto a nitrocellulose membrane. The membrane was then blocked with 5% powdered milk in 0.01 M PBS containing 0.05% Tween 20 and incubated with the antiserum against PGP 9.5, followed by incubation with peroxidase-labeled goat anti-rabbit immunoglobulins (DAKO; Carpinteria, CA). The antigen-antibody complex was visualized with DAB solution containing H2O2.
Cloning and Sequencing of Horse PGP 9.5 cDNA
Total RNA was extracted from the brain and synovial membrane by the guanidine isothiocyanate method using TRIzol solution (Gibco BRL; Gaithersburg, MD), according to the manufacturer's directions. Poly (A)+ RNA was prepared by use of oligo-d(T) cellulose columns (Clontech; Palo Alto, CA), and used as a template for reverse transcription-polymerase chain reaction (RT-PCR). The forward (5′-tgctgctgctgtttcccctc-3′) and reverse (5′-aactggcgccatggttcacc-3′) PCR primers were designed on the basis of common sequences of human and rat PGP 9.5 cDNA (GenBank, X04741 and D10699, respectively) (Figure 6). The PCR product was electrophoresed in a 2.0% agarose gel, purified by Geneclean II Kit (Bio 101; Vista, CA) and ligated into a pCR II plasmid vector (Invitrogen; Carlsbad, KY). The sequences of the obtained cDNAs were determined using a dye terminator cycle sequencing kit (Perkin Elmer Applied Biosystems; Foster City, CA).
Whole-mount preparation of a villus immunostained with PGP 9.5 antiserum. Immunoreactive cells at the villous tip extend neuron-like processes towards the surface. Bar = 20 μm.
Northern Blot Analysis
Sixty μg of total RNA was separated on a 1% agarose/formaldehyde gel and transferred to a nylon membrane (Amersham; Poole, UK). A cDNA probe was labeled with [α-32 P]-dCTP using a multiprime DNA labeling kit (Amersham). Nylon membranes were hybridized with the labeled cDNA probe, washed, and exposed to X-ray film (X-OMAT AR; Kodak, Rochester, NY) for 1 week.
Results
Light Microscopy
Immunostaining of frozen sections using the antiserum against PGP 9.5 demonstrated cells scattered in the synovial intima that covered the inner surface of the fibrous capsule (Figure 1). The PGP 9.5-immunoreactive cells were small and round, extending a few thin, long cytoplasmic processes. Typical fibroblasts in the subintimal collagenous layer were free of immunoreactivity. In the synovial villi, most of the cells gathering along the surface were intensely immunolabeled (Figures 2 and 3), sometimes displaying an epithelium-like arrangement. The immunoreactive cells in the villi possessed a round cell body and extended branching processes towards the joint cavity. The processes on the surface were regularly and densely arranged to form a superficial plexus, a kind of lamina limitans (Figure 3). Here, again, slender fibroblasts intervening in the subintimal dense connective tissue remained unreactive. At villous tips, subintimal spaces lacked collagen bundles and were occupied by a clear amorphous material, including blood vessels. The non-fibrous core contained PGP 9.5-immunoreactive dendritic or spindle-shaped cells, which appeared to be continuous with intimal immunoreactive cells (Figure 3). Other intense immunoreactivity for PGP 9.5 occurred in nerve fibers dispersed mainly in the villous region, but these were few. The synovial membrane lacked PGP 9.5-immunoreactive neuronal somata. The synoviocytes and nerves showed the same staining for the antiserum diluted in different concentrations. The most intense immunoreactivity was obtained in 1:12,000-20,000. When the antiserum was more diluted, the immunoreactivity in the synoviocytes and nerve elements gradually decreased in intensity. The positive staining produced by the rabbit anti-PGP 9.5 serum could be completely eliminated by substitution of the immune serum with a nonimmune rabbit serum.
Immunostaining of the horse synovial membrane by use of PGP 9.5 antiserum. (Figure 1) Round cells with a few thin processes are positively stained in the synovial intima. Immunoreactive synoviocytes gather in the intima of the villous region and extend dendritic processes towards the joint cavity (Figure 2), frequently exhibiting a densely arranged plexus on the surface, like a lamina limitans (Figure 3, arrows). In the tip of a villus (Figure 3), immunoreactive cells also appear at the villous core. (Figure 4) A whole-mount preparation of the synovial membrane lining the joint capsule shows dispersed immunoreactive cells possessing thin and long processes which run roughly in the same direction. Double staining for acid phosphatase and PGP 9.5 (Figure 5) discriminates several round cells (red) from dendritic PGP 9.5-immunoreactive cells (brown). Bars = 20 μm.
Whole-mount preparations of the flat-surfaced synovial membrane and villi demonstrated the entire shape of the immunoreactive synoviocytes, especially extensions of the processes. The cells dispersed in the intima of flat synovial membrane extended uni- or bipolar processes which ran in roughly the same direction as those of adjacent cells and which exhibited terminal branching (Figure 4). When approaching a villous region, these processes became more complicated and branched in a dendritic fashion. All immunoreactive cells in the villi extended branching processes towards the joint cavity to form a plexus of processes on the synovial surface (Figure 6).
Histochemistry for acid phosphatase stained round cells scattered on and near the surface of the synovial membrane. Double staining with the PGP 9.5 antiserum showed a complementary distribution of PGP 9.5 immunoreactivity and acid phosphatase activity (Figure 5). No cells were convincingly positive with both markers. Thus, different cells labeled by each method were intermingled in the synovial intima, although acid phosphatase-positive cells were less numerous than PGP 9.5-immunoreactive cells.
Electron Microscopy
By electron microscopy, immunoreactive cells were identified as cells labeled with electron-dense materials (Figures 7 and 8). The immunoreactivity occurred diffusely and homogenously throughout the cytoplasm along the entire extension, including the tips of fine processes. Immunoreactive cells gathered to form a rough cell line close to the surface of the synovial membrane, but there were no specializations for cell-to-cell contacts. These cell bodies were characterized by free ribosomes and well-developed rough endoplasmic reticulum with broad cisternae (Figure 8). Filamentous structures were rich in the cytoplasm, especially in the processes. There were a few lysosomes but no secretory granules, which were reported in synoviocytes of some rodents (Linck and Porte 1978). These ultrastructural characteristics corresponded to those of Type B synoviocytes (Barland et al. 1962; Graabaek 1984). The immunoreactive cells extended irregular-shaped processes directed towards the joint cavity (Figure 7). The surface of the synovial membrane was fragmentally covered by PGP 9.5-immunoreactive processes of different thicknesses. In some places, the amorphous interstitium of the synovial intima was directly exposed to the joint cavity (Figure 8). Type A synoviocytes, which were characterized by the rich existence of lysosomes and vacuoles, were negative in reaction.
Electron micrographs showing PGP 9.5-immunoreactive Type B cells (B) in the synovial intima. The cells extend irregular-shaped processes which reach the joint cavity (Figure 7). Arrows indicate parts of nonimmunoreactive Type A cells. Thin processes extending from an immunoreactive cell body, which is characterized by the rich existence of expanded endoplasmic reticulum, partially cover the synovial surface (Figure 8). Bars = 2 μm.
Western and Northern Blottings
Western blotting yielded a predominant immunoreactive band, at approximately 27 kD, in the extracts of brain, jejunum, liver, and synovial membrane (Figure 9A). The immunoreaction in the synovial membrane was significantly intense, although it was weaker than that in the brain and intestine, which exhibited large numbers of PGP 9.5-containing nerve elements.
Western and Northern blot analyses. (A) blotting of the equine brain (Lane 1), jejunum (Lane 2), synovial membrane (Lane 3), and liver (Lane 4) using anti-PGP 9.5 serum. A predominant protein band for PGP 9.5 is seen around 27 kD in all samples. (B) Northern blotting of PGP 9.5 using the brain (Lane 1) and synovial membrane (Lane 2) of the horse. In the synovial membrane, a small but significant amount of PGP 9.5 mRNA is expressed at the same position, with the most intense band in the brain.
The expression of PGP 9.5 mRNA was examined in the equine brain and synovial membrane. RT-PCR using two conserved regions between human and rat PGP 9.5 cDNA as primers detected a clear band at 404-489 base pairs (BP) in both samples. Cloning of the PCR products from the equine brain and synovial membrane demonstrated the identical sequence with 420 BP (registered to GenBank AB013344). Its homology with the corresponding region of human and rat PGP 9.5 was 98% and 97% in amino acid sequence (not shown) and 92% and 89% in cDNA sequence (Figure 10). In Northern blot analysis in which a cDNA fragment of horse brain PGP 9.5 was used as a probe, a significant band of the same size as the brain PGP 9.5 was detected in the synovial membrane at a condition of total RNA 60 μg, although it was less intense than that of the brain (Figure 9B).
Alignments of partial nucleotide sequences of horse, human, and rat PGP 9.5 cDNA. These sequences, 420 BP in length, correspond to position 144-563 of human PGP 9.5 cDNA (X04741). Clear spaces in the human and rat alignments indicate nucleotides identical with the horse molecule. Twenty base pairs (open squares) at 5′- and 3′-terminals were used as primers for RT-PCR.
Discussion
The existence of PGP 9.5-like immunoreactivity in Type B synoviocytes is surprising from the general view of selective localization of PGP 9.5 in neurons and some ontogenetically and functionally related cells (Doran et al. 1983; Wilson et al. 1988; Iwanaga et al. 1992). Although checking for the specificity of the immunoreaction lacked a preincubation test of the antiserum with a corresponding antigen, synoviocytes and nerve fibers distributed in the synovial membrane showed the same staining for to the PGP 9.5 antiserum diluted at different concentrations. Western blotting using the same antiserum revealed an identical molecular weight for immunoreactive substances, equal to that of human brain PGP 9.5 (Doran et al. 1983), between the synovial membrane and the brain of the horse. Moreover, cDNA cloning and subsequent Northern blot analysis using the horse synovial membrane demonstrated a significant expression of mRNA which is identical in partial sequence and molecular size to mRNA for brain PGP 9.5. Because the synovial membrane contains only a few nerve fibers, it is reasonable to consider that the synovial PGP 9.5-like immunoreactivity and PGP 9.5 mRNA are derived exclusively from synoviocytes. These findings strongly suggest that PGP 9.5 itself or closely related molecules are contained in the synovial membrane of the horse.
Identification of the PGP 9.5-immunoreactive cells as Type B synoviocytes is supported by the present double staining combined with acid phosphatase activity. Histochemistry for acid phosphatase, a representative lysosomal enzyme, can label macrophagic cells with a developed lysosomal system in various tissues. Graabaek (1985) showed the predominant localization of acid phosphatase activity in the lysosomes of synoviocytes and also distinctive differences in the development of the lysosomal system between synovial Type A and B cells in the rat. In the horse synovium, acid phosphatase histochemistry also discriminated Type A cells from PGP 9.5-immunoreactive Type B cells and other cellular elements. The double staining of PGP 9.5 and acid phosphatase activity appears to be one practical way of demonstrating fibroblast-like and macrophage populations.
For cytochemical detection of Type B synoviocytes, the following antisera and markers have been reported to be available. MAb 67 was able to stain the cell surface of synoviocytes, but also reacted with connective tissue fibers and vascular elements in humans (Stevens et al. 1990). The antiserum against prolyl hydroxylase could label Type B synovial lining cells but could not distinguish these from subintimal fibroblasts (Wilkinson et al. 1992). UDPGD, which is a rate-limiting enzyme for hyaluronan synthesis, appears to be more specific to Type B cells than MAb 67 antigen and prolyl hydroxylase (Wilkinson et al. 1992). However, histochemistry for UDPGD could not stain the entire extension of cells because of less intense reactivity and limited intracellular localization (Edwards 1994). On the other hand, PGP 9.5 is a cytosolic protein diffusely distributed throughout the cytoplasm and may exist richly in a cell, as suggested by the fact that PGP 9.5 is a major protein component of the neuronal cytoplasm (Doran et al. 1983). Furthermore, its antigenicity remains strong even in tissue sections processed by the conventional procedure. Intense immunoreactivity was obtained in tissues that were fixed in formalin for several months and embedded in paraffin (unpublished data). Therefore, the present immunohistochemistry for PGP 9.5 revealed their entire shapes, including characteristic dendritic processes, at both light and electron microscopic levels.
Previous cytochemical and ultrastructural studies have suggested that Type B synoviocytes possess broad, branching processes that extend towards the joint cavity (Linck and Porte 1981; Graabaek 1985; Wilkinson et al. 1992). Moreover, immunostaining of the human synovium with an anti-VCAM-1 MAb exhibited a meshlike aggregation of cytoplasmic processes on the surface of the synovial membrane (Wilkinson et al. 1993). These findings may correspond to an ultrastructural observation by Barland et al. (1962) describing how the processes overlap and intertwine to form a loose network in the human synovial membrane, although the authors did not ascertain the origin of the processes. The conspicuously branching processes of Type B synoviocytes, being reminiscent of neuronal dendrites, were more clearly demonstrated in the horse synovial membrane by the present PGP 9.5 immunostaining. The tips of the dendritic processes extended on the synovial surface to exhibit a kind of lamina limitans. Although the functional significance of the processes facing the joint cavity is unknown, exposure of the processes to the cavity is reasonable if they secrete some elements directly into synovial fluid. Moreover, the regularly arranged processes on the synovial surface may function as a receptor site under certain mechanical and chemical conditions, such as pressure, viscosity, and change of chemical composition.
The existence of a neuron-specific substance in Type B synoviocytes can not be explained by ontogenetic events. Earlier ultrastructural observations, which focused on the rich existence of secretory granules, have indicated the endocrine nature of Type B synoviocytes (Linck and Porte 1978; Okada et al. 1981; Graabaek 1984). Mouse and rat synoviocytes in particular contain many membrane-bounded, electron-dense granules, which resemble those of pituitary and pancreatic endocrine cells. The cluster formation of Type B cells close to fenestrated vessels and their rapid degranulation in response to stimuli (Linck and Porte 1981) suggest a function as recepto-secretory cells, a characteristic shared by paraneuron members (Fujita et al. 1988). In the horse synovium, however, Type B cells lacked the cytoplasmic granules and were comparable to agranular synoviocytes found in other mammals, such as guinea pig, rabbit, and human (Linck and Porte 1978). PGP 9.5 has been recognized as a ubiquitin C-terminal hydrolase showing brain-specific expression. Ubiquitin is a normal component of most eukaryotic cells, and it is assumed that the covalent attachment of ubiquitin to proteins (ubiquitination) plays a role in metabolic processes of intracellular proteins (Fried et al. 1987). The ubiquitin C-terminal hydrolase is a key molecule which controls the ubiquitination by the cleavage of amide bonds at the C-terminus of ubiquitin (Hershko 1991). With the dual distribution of PGP 9.5 in synoviocytes and neurons, we can postulate a common mechanism for regulation of cell metabolism, including degradation or modulation of some cytoplasmic and nuclear proteins.
PGP 9.5 is a useful marker for cytochemical detection of Type B synoviocytes in the horse. At present, however, this does not hold true for other mammals. Several immunohistochemical studies have used PGP 9.5 antisera for investigation of the innervation in the joint of rat, sheep, and human but did not describe any positive staining in synoviocytes (Mapp et al. 1990,1996; Shimizu et al. 1996; Tahmasebi-Sarvestani et al. 1997). The amino acid sequence of PGP 9.5 is highly conserved throughout evolution, as suggested by the fact that PGP 9.5 antisera can detect nerve elements in all mammalian and some submammalian species (Thompson and Day 1988; Mann et al. 1996). Further studies will be required to determine whether the existence of PGP 9.5 in synoviocytes is unique to the horse or whether related substances, which can not be detected by the antiserum used, are generally present in the synoviocytes of mammals.
Footnotes
Acknowledgements
We thank Dr Hiroyuki Taniyama (Rakuno Gakuen University) for cooperation with the sampling.