- Email: [email protected]
Glial fibrillary acidic protein (GFAP)-like immunoreactivity in the rodent eye
Journal of Neuroimmunology, 8 (1985) 331-345 Elsevier
331
JN1 00271
Glial Fibrillary Acidic Protein (GFAP)-Like Immunoreactivity in the Rodent Eye Comparison Between Peripheral Glia of the Anterior Uvea and Central Glia of the Retina Hfakan Bjrrklund and Doris Dahl Department of Histology, Karolinska Institutet, Stockholm (Sweden), and Department of Neuropathology, Harvard Medical School and Spinal Cord Injury Research Laboratory, West Roxbury Veterans Administration Medical Center, Boston, MA (U.S.A.) • (Received 12 November, 1984) (Revised, received 2 January, 1985) (Accepted 2 January, 1985)
Summary Immunohistochemistry with antiserum raised against glial fibrillary acidic protein (GFAP) revealed a dense plexus of GFAP-positive fibers in normal rodent iris. These fibers were not stained with 2 monoclonal GFAP antibodies which readily stain astrocytes, suggesting that they contain a polypeptide closely related, but not identical, to CNS GFAP. The GFAP-positive iris fibers did not disappear after short-term intraocular grafting or culturing of irides; instead a conspicuous system of fluorescent, star-shaped cells appeared. In the retina Mialler glia were intensely fluorescent using GFAP antiserum whereas positive staining was observed with GFAP monoclonals only after injury to the retina. These antibodies, however, readily stained astrocytes in the inner layers of the normal retina. Taken together, these findings support the idea of GFA proteins as a group of closely related but not identical polypeptides. Key words: Astrocytes - E y e - Glial fibrillary acidic protein (GFAP) - I r i s - Lens epithelium- MUller g l i a - Retina
Supported by the Swedish Medical Research Council (14X-03185, 12P-7310) and Karolinska Institutets fonder. Dr. Dahl was supported by the Veterans Administration. Correspondence to: Dr. Hhkan Bj~Srklund at his present address: Department of Pharmacology, Pharmacia AB, S-75182 Uppsala, Sweden. 0165-5728/85/$03.30 © 1985 Elsevier Science Publishers B.V. (Biomedical Division)
332 Introduction
After the first antiserum was raised against glial fibrillary acidic protein (GFAP) almost 15 years ago (Bignami et al. 1972; Uyeda et al. 1972) it soon became clear that this protein is the major component of astrocytic intermediate filaments (see Bignami et al. 1980; Eng and DeArmond 1983). During the last few years it has, however, become clear that GFAP-like immunoreactivity is not confined to the CNS. Thus the presence of GFAP-Iike immunoreactivity has been demonstrated in enteric glial cells (Jessen and Mirsky 1980, 1983; Bj~brklund et al. 1984b; Jessen et al. 1984), the sciatic and other peripheral nerves, where G F A P immunoreactivity seems to be restricted to non-myelin-forming Schwann cells (Yen and Fields 1981; Dahl et al. 1982; Jessen et al. 1984), the olfactory nerve (Barber and Lindsay 1982) and several peripheral ganglia (Jessen et al. 1984). Furthermore, the presence of GFAPlike immunoreactivity in non-neuronal cells such as the lens epithelium (Hatfield et al. 1984) and Kupffer cells in the liver ( G a r d e t al. 1982) has recently been described. Whether the GFAP-Iike material outside the CNS is identical to astrocyte G F A P has been unclear (Davison and Jones 1981; Dahl et al. 1982). However, two independent laboratories have recently demonstrated that the GFAP-immunoreactive material in the PNS resides in a protein with a molecular weight of 49 000 and thus similar to CNS G F A P (Yen and Fields 1983; Jessen et al. 1984). Since Jessen et al. (1984) could only stain a small proportion of enteric glial cells and no Schwann or satellite cells using a monoclonal G F A P antibody which readily stained astrocytes, the molecular identity of peripheral G F A P is still unclear and they raised the possibility that G F A P is not one individual protein but a group of closely related polypeptides with differential distribution in various types of cells. In the present communication we have studied the presence and distribution of GFAP-positive cells and fibers in various tissues of the rodent eye using polyclonal as well as monoclonal G F A P antibodies. The eye is unusual in that both central type neurons nerve fibers and glia (in the retina) and peripheral nerve fibers and glia (in the iris, choroid membrane, cornea, etc.) are present within the same organ. Furthermore, since several of the tissues in the eye such as the iris, choroid membrane and to a certain extent the retina can be flat mounted, studies of fiber distribution are facilitated. Finally, because of the position of the eye surgical manipulations are easily performed. Some of these results have been reported previously (Bj6rklund 1984; Bj~brklund et al. 1984a, 1985).
Material and Methods Animals Albino rats Sprague-Dawley and mice (NMRI) were used (Alab, Stockholm, Sweden). Iris whole mounts were prepared from both adult rats and mice as well as from late embryonic and early postnatal rats on chromalum-coated slides (Falck 1962; Malmfors 1965). Short-term intraocular grafting of irides was performed as described by Olson and Malmfors (1970) using adult rats both as donors and
333 recipients. Culturing of adult rat irides for 24 h was performed as described by Ebendal et al. (1980). In one experiment 5 /~1 of a cell suspension prepared from cortex cerebri of 10-day-old rat pups was injected into the anterior eye chamber of adult rats (Bjrrklund 1984). Spread preparations of lens epithelium were prepared from adult mice. For studies of glial cells in rat retina the posterior part of the eye was either immediately frozen on dry-ice or fixed in a parabenzoquinone/ paraformaldehyde (PBQ) solution as described earlier (Bjrrklund et al. 1985). Both types of specimens were then sectioned at 8 or 14/tm on a cryostat. Alternatively the retinae were spread-prepared on chromalum-coated slides or occasionally divided into 2 or 3 pieces which were then smeared on a slide using a coverglass. In one experiment injury to the retina was performed by penetrating the posterior part of the eyebulb with a needle 4 days prior to killing.
lmmunofluorescence techniques All whole mounts, stretch preparations and retinal smears were air-dried from one to several hours before being further processed for immunohistochemistry. Sections of fresh-sectioned rat retinae were postfixed for 3 min in acetone at room temperature. Antiserum to G F A was raised in rabbits against degraded human antigen (Dahl and Bignami 1976) and was used diluted 1 : 100 in phosphate-buffered saline (PBS). Two monoclonal G F A antibodies raised against a cytoskeletal preparation from chicken brain were also used as culture supernatants diluted 1:100 in PBS. The preparation and characterization of these antibodies will be described in detail elsewhere (Dahl et al., in preparation). Antiserum to vimentin (R15) was a generous gift from Dr. Richard Hynes, Massachusetts Institute of Technology, Boston, MA, U.S.A. It was used diluted 1 : 5 0 in PBS. All slides were processed according to the indirect immunofluorescence technique (Coons 1958). After rinsing in PBS all specimens were incubated with the primary antibodies overnight at 4°C in a humid atmosphere. After rinsing in PBS for 3 x 10 min, the slides were incubated with fluorescein isothiocyanate- (FITC) or rhodamine-conjugated secondary antibodies for 1 h at room temperature in darkness. All antisera contained 0.3% (v/v) Triton X-100. After a final rinse (3 × 10 min), the slides were mounted in glycerine-phosphate buffer.
Results
GFAP-Iike immunoreactivity in rodent irides Using polyclonal G F A P antibodies a dense system of fluorescent fibers was seen in the dilator plate of whole-mounted adult rat irides (Figs. l a and b). The GFAP-positive fibers formed a system of thicker bundles associated with a finemeshed network of thin fibers. A high number of positive fibers were observed in the sphincter area where they could frequently be seen reaching the sphincter margin. Thin, delicate fibers were seen surrounding large blood vessels with a distribution similar to the sympathetic innervation of these structures (Fig. lb). The GFAP-positive fibers in the mouse iris had a slightly more regular organization with a lower
334 number of fiber bundles radiating towards the sphincter (Fig. 2). These bundles were surrounded by a comparatively sparse network of thinner fluorescent fibers. In several respects the GFAP-positive fiber plexus resembled the distribution of autonomic nerve fibers in the mouse iris. In both species cell body-like areas (i.e., negative swellings surrounded by GFAP-like immunoreactivity) were observed both in fiber bundles and at branching sites of thinner fibers (Figs. la and c). Although the fluorescence intensity of the GFAP-positive fibers in the iris was relatively high they did not show the same intensity as astrocytes. This was clearly observed on rat irides where cell suspensions from cortex cerebri of 10-day-old rat pups had been injected into the anterior eye chamber of adult rats. When examined 10 days later individual as well as groups of GFAP-positive astrocytes were seen distributed unevenly over the iris. These cells had a higher fluorescence intensity than the intrinsic GFAP-positive iris fibers (Fig. 3). Furthermore, such astrocytes were readily demonstrated with the monoclonal GFAP-antibodies whereas none of these G F A P antibodies revealed any intrinsic fibers in the rat iris. The ontogeny of the GFAP-positive iris fibers was studied in the rat. Already at embryonic day 18, the earliest time point when the iris could be reliably stretch-prepared, fluorescent fibers were observed forming a sparse system of strongly fluorescent fibers (Fig. 4a). A gradual increase in fiber density was observed with increasing age so that at postnatal day 6 a system of GFAP-positive fibers resembling the adult organization was observed although the density of fluorescent fibers had still not reached the adult level (Fig. 4b).
GFAP-like immunoreactivity in grafted, host and cultured irides To ensure that no axons in the iris were stained with G F A P antiserum adult rat irides were grafted to anterior eye chamber of adult recipients in order to cause complete degeneration of all nerve fibers. The grafts were then studied after 1, 2 and 6 days in oculo. Similarly in one experiment irides were cultured for 24 h. Neither in grafted nor in cultured irides were any degenerating GFAP-positive fibers observed. However, both types of irides, and to a lesser extent host irides, became swollen and more difficult to stretch-prepare. This made it difficult to reliably assess the density of GFAP-positive fibers. Interestingly, in all 3 types of irides, as well as in other types of disturbed irides (i.e., as an intrinsic response to intraocular injections of cell suspensions) a high number of strongly fluorescent spider-like cells were observed (Fig. 5). These cells usually had 5-8 extensively branching processes. In contrast to astrocytes and GFAP-positive iris fibers where a filamentous immunoreactivity was seen the fluorescence in these cells had a grainy appearance. The cellular origin of these cells is unclear; however, they do not seem to be part of the normal
Fig. 1. Distribution of GFAP-like immunoreactivityin adult stretch-prepared rat (a and b) and mouse (c) iris. In a fiber bundles are seen meanderingover the dilator plate intermingled with a network of thin, smooth fibers, b: A large blood vessel with thin GFAP-positivefibers is seen traversing the dilator. In c cell body-like structures are seen at junctional points in the GFAP-positive plexus (arrows). Similar structures although less obvious are seen also in a. All specimens reacted with GFAP antiserum. (a) ×135, (b and c) ×330. (a and b from BjOrklund et al. 1984a.)
335
336
Fig. 2. GFAP-positive fibers in adult mouse iris. The sphincter is seen in the lower part, the ciliary body and part of the choroid membrane in the upper part of the figure. The bundles show a more radial orientation than in the rat iris (cf. Fig. la), whereas the thinner fibers form a well-organized network against a negative background. Positive fibers are observed also in the sphincter, ciliary body and choroid membrane, lmmunohistochemistry using GFAP antiserum. × 135.
337
Fig. 3. GFAP-positive cells on the iris of a rat injected 10 days earlier with a cell suspension prepared from 10-day-old rat pups. Note the obvious difference in fluorescence intensity between the intrinsic GFAP-positive fiber network and the astrocvtic cells. GFAP antiserum. × 330.
/
/~
/r,'iJ¸i ¸
.
a
,x',
•
\
Fig. 4. GFAP-like immunoreactive fibers in developing rat iris. a: Embryonic day 18; b: postnatal day 6. G F A P antiserum. × 330.
338
• -
,I
.,.
4
-
i
_. .
,.~- ~ .
.
.
.
' 7)
...':;
.
>~ ~.:
~
:_
1
a
~ .._ i-
-
..
~
,
/
h\l il" ~
~
i|"
_.,,,,~.-
•
Fig. 5. GFAP-Iike cells in ( a ) host iris 24 h after intraocular grafting of an iris graft and (b) a cultured iris after 24 h in vitro. Normal intrinsic GFAP-positive fibers are seen intermingled with brightly fluorescent cells. G F A P antiserum. ( a ) x 330, (b) X 135,
t
'-
~O
!'o
~a.
,¢?
,O
.... %
,(-
-~"
Fig. 6. Spread-prepared lens epithelium from an adult mouse. The epithelial cells are clearly GFAP-positire. G F A P antiserum, x 190.
339
GFAP-positive network. Interestingly, they are readily observed even 24 h after grafting or culturing although more cells are observed in iris grafts and host irides 6 days after grafting. Furthermore, we have observed that they tend to disappear after
f
k \~
--
z..',
. -. o's.j; -
T
.
I
.
- ~ --i-" (_~
i~ ' ~ C't
,
,.¥
-,,¢
/
"
Fig. 7. a, b and c: Miiiler gila in rat retina, a: GFAP-positive M~ller glia in a spread preparation of unfixed retina, b: Sectioned PBQ-fixed retina showing a high number of GFAP-positive Mi~ller glia spanning the retina, c: Vimentin-positive Mi~ller glia in a similar preparation, d: GFAP-positive fibers surrounding a blood vessel in a spread-prepared retina. G F A P and vimentin antisera. (a-d) x 135.
340 longer time periods in irides from eyes injected with cell suspensions. Thus, one might speculate that whatever their cellular origin (see below), these cells appear rapidly after various disturbances and then disappear with time. GFAP-Iike immunoreactivity in mouse lens epithelium Using smear preparations, adult mouse lens epithelial cells were found to be strongly fluorescent when stained with G F A P antiserum. The cells had a typical epithelial morphology with a small, rounded cytoplasm where an obvious perinuclear fluorescence was observed (Fig. 6). No GFAP-Iike immunoreactivity was observed in the nuclei. GFAP- and oimentin-like immunoreactioity in rat retina The mammalian retina contains at least 2 types of macroglial cells, the Mialler glia spanning most of the width of the retina and retinal astrocytes in the ganglion cell and nerve fiber layers. Using G F A P antiserum intensely fluorescent Mi~ller glia
Fig. 8. Typical star-shaped retinal astrocytes in unfixed spread preparations of rat retina immunohistochemicallyvisualized using a monoclonalGFAP antibody. (a) × 135, (b) x 330.
341 TABLE 1 A S U M M A R Y O F S T R U C T U R E S T H A T STAIN W I T H T H E P O L Y C L O N A L A N D M O N O C L O N A L G F A P ANTIBODIES, RESPECTIVELY
Normal iris fibers Reactive cells in disturbed irides Lens epithelial cells Normal Miiller glia Mtiller glia in injured retina Retinal astrocytes
Polyclonal G F A P
Monoclonal G F A P
Yes Yes Yes Yes Yes Yes
No No Not tested No Yes Yes
forming typical parallel arrays of fibers were stained in spread-prepared and smeared retinae (Fig. 7a). Individual Mi~ller-like fibers with a conspicuous thickening at one end were also observed in large numbers. In the brain and spinal cord we have found that astrocytes are better stained in fresh-frozen, acetone-fixed, sections than in sections of formalin-fixed tissues. Surprisingly, GFAP-positive Mialler glia were readily observed in sections of PBQ-fixed retina (Fig. 7b) whereas only a few faintly stained fibers were seen in fresh-frozen, acetone-fixed, sections. Furthermore, using monoclonal GFAP antibodies, fluorescent Mi~ller glia were only observed in injured retinae. In agreement with earlier studies (Shaw and Weber 1984), vimentin antiserum revealed many Mi~ller glia in spread preparations/smears and in acetone-fixed sections whereas only weakly fluorescent fibers were seen in sections of PBQ-fixed retinae (Fig. 7c). GFAP-positive fibers were also observed surrounding blood vessels (Fig. 7d). Star-shaped GFAP-positive retinal astrocytes with approximately 8-12 thin, smooth, strongly fluorescent processes were abundant in spread-prepared and smeared normal retina using both GFAP antiserum and monoclonal antibodies (Fig. 8). In sections star-shaped cells were observed in the ganglion cell and nerve fiber layers with both GFAP-antiserum and monoclonal GFAP antibodies.
Discussion
The present results demonstrate that GFAP-positive fibers are abundant in several tissues of the rodent eye (see Table 1 for a summary). Apart from the iris and retina we have observed GFAP-positive fibers in the choroid membrane. Supporting the findings of Hatfield et al. (1984) we have also seen GFAP immunoreactivity in the lens epithelium. The ceils responsible for the GFAP immunoreactivity in these various tissues are clearly of different types. In the retina both Mialler glia and retinal astrocytes are considered to be part of the normal astrocytic lineage, whereas the cells stained in the lens epithelium are of epithelial origin. Schwann cells are the only type of macroglial cell considered to be present in the iris. Several laboratories have also shown that Schwann cells in several peripheral nerves are GFAP-positive (Yen and Fields 1981, 1983; Dahl et al. 1982; Jessen et al. 1984). Interestingly,
342 GFAP-Iike immunoreactivity seems to be restricted to non-myelinating Schwann cells. It might therefore be speculated that the high density of unmyelinated nerve fibers in the iris (for a review see Tervo et al. 1982) is the reason why this tissue has such a dense GFAP-positive fiber network. It is noteworthy that thin GFAP-positive fibers were observed surrounding large blood vessels in the iris with a distribution similar to the sympathetic, non-myelinated, innervation of these vessels. In contrast such fibers are not observed with neurofilament antisera (Seiger et al. 1984). Instead neurofilament antiserum seems to preferentially stain nerve fibers with a large diameter and thus to a large extent myelinated (Seiger et ai. 1984). Several lines of evidence, however, indicate that the GFAP-like protein in the iris is not identical to CNS GFAP. Thus, the fluorescence intensity was clearly lower in GFAP-positive iris fibers as compared to CNS astrocytes growing on the iris. Furthermore, neither of 2 GFAP monoclonals that readily stain astrocytes were able to stain fibers in the iris. Similarly we have not observed staining of enteric glial cells using these antisera (Bjrrklund and Dahl, unpublished observations). Interestingly, Jessen et al. (1984) have recently reported that rat peripheral G F A P immunoreactivity resides in a 49 kDa protein and thus is identical in size to rat CNS GFAP. However, they could not observe any peripheral GFAP-immunoreactive structures, apart from a small proportion of enteric glial cells, using a monoclonal G F A P antibody. Taken together these results suggest that G F A P might not be one individual polypeptide but a group of two or more closely related proteins. Using G F A P antiserum we have been able to demonstrate that normal albino rat Mialler glia are GFAP-positive. Surprisingly, our monoclonal G F A P antibodies failed to demonstrate Mialler glia in normal retina, whereas retinal astrocytes were strongly stained with both G F A P monoclonals. Only after injury to the eye could staining of Mi~ller glia be observed. A possible explanation for these findings is that the G F A P molecules in normal Miailer glia and retinal astrocytes are not identical, with the latter cell type containing G F A P similar to that in brain astrocytes." It is of course also conceivable that a higher concentration of G F A P is necessary for the monoclonal G F A P antibodies to stain Mtiller glia and therefore positive staining is seen only in injured retina where the content of G F A P in all probability has increased within individual cells. Previously G F A P has not been observed in normal mammalian Mialler glia (Bignami and Dahl 1979; Bromberg and Schachner 1978; Dixon and Eng 1981; Shaw and Weber 1983, 1984) but has been detected in rats as a response to traumatic injury (Bignami and Dahl 1979) or retinal degeneration (Shaw and Weber 1983) and in some human retinae obtained at autopsy (O'Dowd and Eng 1979). GFAP-positive Mi~ller glia were, however, recently demonstrated by Bignami (1984) in goldfish retina. Since Mtiller glia are strongly vimentin-positive in normal rat (Dahl et al. 1981; Schnitzer et al. 1981; Shaw and Weber 1983) Bignami raised the possibility that Mialler glia contain intermediate filaments that are GFAP-vimentin heteropolymers (Quinlan and Franke 1983) with vimentin being the normally dominating protein in mammalian retina while the amount of G F A P increases in response to retinal injury. An intriguing finding is the appearance of a large number of star-shaped, GFAP-positive cells in various types of disturbed irides using G F A P antiserum. The
343 nature of these cells is intriguing; their distribution suggests that they are probably not completely corresponding to the cell body-like structures observed in normal irides (see above). However, they could represent normally GFAP-negative Schwann cells that undergo morphological a n d / o r biochemical changes as a response to disturbances of various types. It is well known that the amount of G F A P immunoreactivity increases in sciatic nerves during Wallerian degeneration (Yen and Fields 1981; Dahl et al. 1982). Alternatively these cells do not belong to the neuroectoderm. Since positive cells are detected in cultured irides, they have to be present normally in the irides and cannot have migrated from other organs. Interestingly, Kreutzberg and Gr~iber (1984) have reported that microglia in the facial or hypoglossal central motor nuclei can be stained with both polyclonal and monoclonal G F A P antibodies following cutting of the appropriate motor nerves. However, ultrastructurally such cells did not contain intermediate filaments of the astrocytic type (i.e., classic fibrous astrocyte fiber bundles). Similarly the reactive GFAP-positive cells in disturbed retinae do not have a clearly filamentous G F A P immunoreactivity. Thus an interesting possibility is that these cells belong to the reticuloendothelial system. Such an hypothesis is further supported by the finding of G a r d et al. (1982) that Kupffer cells, which are macrophage-type cells, are G F A P positive. That non-macroglial cells can express G F A P immunoreactivity is also demonstrated by the staining of lens epithelial cells. In conclusion, a rich plexus of GFAP-positive fibers is present in normal rodent iris. Since 2 monoclonal G F A P antibodies failed to stain these fibers, they probably contain a polypeptide not identical to CNS GFAP. In the retina Mtiller glia are intensely stained with G F A P antiserum but not with monoclonal G F A P antibodies which readily stained retinal astrocytes. Thus these studies lend further support to the idea of G F A polypeptides as a group of two or more closely related proteins that share some common antigenic determinants and have a similar molecular weight.
Acknowledgements For support in m a n y ways we are grateful to Drs. Lars Olson, ~,ke Seiger and Amico Bignami. We thank Lena Hultgren, Anna Hultghrdh, Barbro Standwerth and Ingrid Str6mberg for technical assistance and Ida Engqvist for secretarial help.
References Barber, P.C. and R.M. Lindsay, Schwann cells of the olfactory nerves contain glial fibrillary acidic protein and resemble astrocytes, Neuroscience, 7 (1982) 3077-3090. Bignami, A., Glial fibrillary acidic (GFA) protein in Miiller glia -- Immunofluorescencestudy of the goldfish retina, Brain Res., 300 (1984) 175-178. Bignami, A. and D. Dahl, The radial glia of Mialler in the rat retina and their response to injury -- An immunofluorescence study with antibodies to the glial fibrillary acidic protein, Exp. Eye Res., 28 (1979) 63-69. Bignami, A., L.F. Eng, D. Dahi and C.T. Uyeda, Localization of the glial fibrillary acidic protein in astrocytes by immunofluorescence,Brain Res., 43 (1972) 429-435.
344 Bignami, A., D. Dahl and C. Rueger, Glial fibrillary acidic (GFA) protein in normal neural cells and in pathological conditions. In: S. Fedoroff and L. Hertz (Eds.), Advances in Cellular Neurobiology, Vol. 1, Academic Press, New York, 1980, pp. 285-319. Bjrrklund, H., Astrocytes grown in oculo - - Expression of cell morphologies on the iris as revealed by GFA-immunohistochemistry, Int. J. Develop. Neurosci., 2 (1984) 377-386. Bjrrklund, H., D. Dahl, L. Olson and ~,. Seiger, Glial fibrillary acidic protein-like immunoreactivity in rodent iris - - Normal distribution and reactive changes following transplantation, J. Neurosci., 4 (1984a) 978-988. Bj~Srklund, H., D. Dahl and ~,. Seiger, Neurofilament and glial fibrillary acidic protein-related immunoreactivity in rodent enteric nervous system, Neuroscience, 12 (1984b) 277-287. Bjrrkhind, H., A. Bignami and D. Dahl, Immunohistochemical demonstration of glial fibrillary acidic protein in normal rat Mialler glia and retinal astrocytes, Neurosci. Lett., 54 (1985) 363-368. Bromberg, J.S. and M. Schachner, Localization of nervous system antigens in retina by immunohistology, Invest. Ophthal. Visual Sci., 17 (1978) 920-924. Coons, A.H., Fluorescent antibody methods. In: J.F. Danielli (Ed.), General Cytochemical Methods, Academic Press, New York, 1958, pp. 399-422. Dahl, D. and A. Bignami, lmmunogenic properties of the glial fibrillary acidic protein, Brain Res., 116 (1976) 150-157. Dahl, D., A. Bignami, K. Weber and M. Osborn, Filament proteins in rat optic nerves undergoing Wallerian degeneration - - Localization of vimentin, the fibroblastic 100 ~, filament protein in normal and reactive astrocytes, Exp. Neurol., 73 (1981) 496-506. Dahl, D., N.H. Chi, L.E. Miles, B.T. Nguyen and A. Bignami, Glial fibrillary acidic (GFA) protein in Schwann cells - - Fact or artifact, J. Histochem. Cytochem., 30 (1982) 912-918. Davison, P.F. and R.N. Jones, Filament proteins in central, cranial and peripheral mammalian nerves, J. Cell Biol., 88 (1981) 67-72. Dixon, R.G. and L.F. Eng, Glial fibrillary acidic protein in the retina of the developing albino rat - - An immunoperoxidase study of paraffin-embedded tissue, J. Comp. Neurol., 195 (1981) 305-322. Ebendal, T., L. Olson, ~,. Seiger and K.-O. Hedlund, Nerve growth factors in the rat iris, Nature (Lond.), 286 (1980) 25-28. Eng, L.F. and S.J. DeArmond, Immunochemistry of the glial fibrillary acidic protein. In: H.M. Zimmerman (Ed.), Progress in Neuropathology, Vol. 5, Raven Press, New York, 1983, pp. 19-39. Falck, B., Observations on the possibilities of the cellular localization of monoamines by a fluorescence method, Acta Physiol. Scand., 56 (Suppl. 197) (1962) 1-25. Gard, A.L., F.P. White and G.R. Dutton, G F A P in rat liver in Kupffer cells?, Soc. Neurosci. Abstr., 8 (1982) 240. Hatfield, J.S., R.P. Skoff, H. Maisel and L.F. Eng, Glial fibrillary acidic protein is localized in the lens epithelium, J. Cell Biol., 98 (1984) 1895-1898. Jessen, K.R. and R. Mirsky, Glial cells in the enteric nervous system contain glial fibrillary acidic protein, Nature (Lond.), 286 (1980) 736-737. Jessen, K.R. and R. Mirsky, Astrocyte-like gila in the peripheral nervous system - - An immunohistochemical study of enteric glia, J. Neurosci., 3 (1983) 2206-2218. Jessen, K.R., R. Thorpe and R. Mirsky, Molecular identity, distribution and heterogeneity of glial fibrillary acidic protein - - An immunoblotting and immunohistological study of Schwann cells, satellite cells, enteric glia and astrocytes, J. Neurocytol., 13 (1984) 187-200. Kreutzberg, G.W. and M. Gr~iber, Molecular cytology of satellite glial cells during regeneration of motoneurons, Soc. Neurosci. Abstr., 10 (1984) 1085. Malmfors, T., Studies on adrenergic nerves - - The use of rat and mouse iris for direct observations on their physiology and pharmacology at cellular and subcellular levels, Acta Physiol. Scand., (Suppl. 248) (1965) 1-93. O'Dowd, D.K. and L.F. Eng, lmmunocytochemical localization of the glial fibrillary acidic (GFA) protein in the M~ller cell of the human retina, Soc. Neurosci. Abstr., 5 (1979) 431. Olson, L. and T. Malmfors, Growth characteristics of adrenergic nerves in the adult rat. Fluorescence histochemical and 3H-noradrenaline uptake studies using tissue transplantation to the anterior chamber of the eye, Acta Physiol. Scand., (Suppl.) 348 (1970) 1-112.
345 Quinlan, R..A. and W.W. Franke, Molecular interactions in intermediate-sized filaments revealed by chemical cross-linking - - Heteropolymers of vimentin and giial filament protein in cultured human glioma cells, Europ. J. Biochem., 132 (1983) 477-484. Schnitzer, J., W.W. Franke and M. Schachner, Immunocytochemical demonstration of vimentin in astrocytes and ependymal cells of developing and adult mouse nervous system, J. Cell Biol., 90 (1981) 435-447. Seiger, ,~., D. Dahl, C. Ayer-LeLievre and H. Bj~rklund, Appearance and distribution of neurofilament immunoreactivity in iris nerves, J. Comp. Neurol., 223 (1984) 457-470. Shaw, G. and K. Weber, The structure and development of the rat retina - - An immunofluorescence microscopical study using antibodies specific of intermediate filament proteins, Europ. J. Cell Biol., 30 (1983) 219-232. Shaw, G. and K. Weber, The intermediate filament complement of the retina - - A comparison between different mammalian species, Europ. J. Cell Biol., 33 (1984) 95-104. Tervo, T., K. Tervo and L. Er~nk6, Ocular neuropeptides, Med. Biol., 60 (1982) 53-60. Uyeda, C.T., L.F. Eng and A. Bignami, Immunological study of the glial fibrillary acidic protein, Brain Res., 37 (1972) 81-89. Yen, S.-H. and K.L. Fields, Antibodies to neurofilament, glial filament and fibroblast intermediate filament proteins bind to different cell types of the nervous system, J. Cell Biol., 88 (1981) 115-126. Yen, S.-H. and K.L. Fields, Schwann cells contain a protein similar to the CNS astroglial filament protein, Soc. Neurosci. Abstr., 9 (1983) 235.
331
JN1 00271
Glial Fibrillary Acidic Protein (GFAP)-Like Immunoreactivity in the Rodent Eye Comparison Between Peripheral Glia of the Anterior Uvea and Central Glia of the Retina Hfakan Bjrrklund and Doris Dahl Department of Histology, Karolinska Institutet, Stockholm (Sweden), and Department of Neuropathology, Harvard Medical School and Spinal Cord Injury Research Laboratory, West Roxbury Veterans Administration Medical Center, Boston, MA (U.S.A.) • (Received 12 November, 1984) (Revised, received 2 January, 1985) (Accepted 2 January, 1985)
Summary Immunohistochemistry with antiserum raised against glial fibrillary acidic protein (GFAP) revealed a dense plexus of GFAP-positive fibers in normal rodent iris. These fibers were not stained with 2 monoclonal GFAP antibodies which readily stain astrocytes, suggesting that they contain a polypeptide closely related, but not identical, to CNS GFAP. The GFAP-positive iris fibers did not disappear after short-term intraocular grafting or culturing of irides; instead a conspicuous system of fluorescent, star-shaped cells appeared. In the retina Mialler glia were intensely fluorescent using GFAP antiserum whereas positive staining was observed with GFAP monoclonals only after injury to the retina. These antibodies, however, readily stained astrocytes in the inner layers of the normal retina. Taken together, these findings support the idea of GFA proteins as a group of closely related but not identical polypeptides. Key words: Astrocytes - E y e - Glial fibrillary acidic protein (GFAP) - I r i s - Lens epithelium- MUller g l i a - Retina
Supported by the Swedish Medical Research Council (14X-03185, 12P-7310) and Karolinska Institutets fonder. Dr. Dahl was supported by the Veterans Administration. Correspondence to: Dr. Hhkan Bj~Srklund at his present address: Department of Pharmacology, Pharmacia AB, S-75182 Uppsala, Sweden. 0165-5728/85/$03.30 © 1985 Elsevier Science Publishers B.V. (Biomedical Division)
332 Introduction
After the first antiserum was raised against glial fibrillary acidic protein (GFAP) almost 15 years ago (Bignami et al. 1972; Uyeda et al. 1972) it soon became clear that this protein is the major component of astrocytic intermediate filaments (see Bignami et al. 1980; Eng and DeArmond 1983). During the last few years it has, however, become clear that GFAP-like immunoreactivity is not confined to the CNS. Thus the presence of GFAP-Iike immunoreactivity has been demonstrated in enteric glial cells (Jessen and Mirsky 1980, 1983; Bj~brklund et al. 1984b; Jessen et al. 1984), the sciatic and other peripheral nerves, where G F A P immunoreactivity seems to be restricted to non-myelin-forming Schwann cells (Yen and Fields 1981; Dahl et al. 1982; Jessen et al. 1984), the olfactory nerve (Barber and Lindsay 1982) and several peripheral ganglia (Jessen et al. 1984). Furthermore, the presence of GFAPlike immunoreactivity in non-neuronal cells such as the lens epithelium (Hatfield et al. 1984) and Kupffer cells in the liver ( G a r d e t al. 1982) has recently been described. Whether the GFAP-Iike material outside the CNS is identical to astrocyte G F A P has been unclear (Davison and Jones 1981; Dahl et al. 1982). However, two independent laboratories have recently demonstrated that the GFAP-immunoreactive material in the PNS resides in a protein with a molecular weight of 49 000 and thus similar to CNS G F A P (Yen and Fields 1983; Jessen et al. 1984). Since Jessen et al. (1984) could only stain a small proportion of enteric glial cells and no Schwann or satellite cells using a monoclonal G F A P antibody which readily stained astrocytes, the molecular identity of peripheral G F A P is still unclear and they raised the possibility that G F A P is not one individual protein but a group of closely related polypeptides with differential distribution in various types of cells. In the present communication we have studied the presence and distribution of GFAP-positive cells and fibers in various tissues of the rodent eye using polyclonal as well as monoclonal G F A P antibodies. The eye is unusual in that both central type neurons nerve fibers and glia (in the retina) and peripheral nerve fibers and glia (in the iris, choroid membrane, cornea, etc.) are present within the same organ. Furthermore, since several of the tissues in the eye such as the iris, choroid membrane and to a certain extent the retina can be flat mounted, studies of fiber distribution are facilitated. Finally, because of the position of the eye surgical manipulations are easily performed. Some of these results have been reported previously (Bj6rklund 1984; Bj~brklund et al. 1984a, 1985).
Material and Methods Animals Albino rats Sprague-Dawley and mice (NMRI) were used (Alab, Stockholm, Sweden). Iris whole mounts were prepared from both adult rats and mice as well as from late embryonic and early postnatal rats on chromalum-coated slides (Falck 1962; Malmfors 1965). Short-term intraocular grafting of irides was performed as described by Olson and Malmfors (1970) using adult rats both as donors and
333 recipients. Culturing of adult rat irides for 24 h was performed as described by Ebendal et al. (1980). In one experiment 5 /~1 of a cell suspension prepared from cortex cerebri of 10-day-old rat pups was injected into the anterior eye chamber of adult rats (Bjrrklund 1984). Spread preparations of lens epithelium were prepared from adult mice. For studies of glial cells in rat retina the posterior part of the eye was either immediately frozen on dry-ice or fixed in a parabenzoquinone/ paraformaldehyde (PBQ) solution as described earlier (Bjrrklund et al. 1985). Both types of specimens were then sectioned at 8 or 14/tm on a cryostat. Alternatively the retinae were spread-prepared on chromalum-coated slides or occasionally divided into 2 or 3 pieces which were then smeared on a slide using a coverglass. In one experiment injury to the retina was performed by penetrating the posterior part of the eyebulb with a needle 4 days prior to killing.
lmmunofluorescence techniques All whole mounts, stretch preparations and retinal smears were air-dried from one to several hours before being further processed for immunohistochemistry. Sections of fresh-sectioned rat retinae were postfixed for 3 min in acetone at room temperature. Antiserum to G F A was raised in rabbits against degraded human antigen (Dahl and Bignami 1976) and was used diluted 1 : 100 in phosphate-buffered saline (PBS). Two monoclonal G F A antibodies raised against a cytoskeletal preparation from chicken brain were also used as culture supernatants diluted 1:100 in PBS. The preparation and characterization of these antibodies will be described in detail elsewhere (Dahl et al., in preparation). Antiserum to vimentin (R15) was a generous gift from Dr. Richard Hynes, Massachusetts Institute of Technology, Boston, MA, U.S.A. It was used diluted 1 : 5 0 in PBS. All slides were processed according to the indirect immunofluorescence technique (Coons 1958). After rinsing in PBS all specimens were incubated with the primary antibodies overnight at 4°C in a humid atmosphere. After rinsing in PBS for 3 x 10 min, the slides were incubated with fluorescein isothiocyanate- (FITC) or rhodamine-conjugated secondary antibodies for 1 h at room temperature in darkness. All antisera contained 0.3% (v/v) Triton X-100. After a final rinse (3 × 10 min), the slides were mounted in glycerine-phosphate buffer.
Results
GFAP-Iike immunoreactivity in rodent irides Using polyclonal G F A P antibodies a dense system of fluorescent fibers was seen in the dilator plate of whole-mounted adult rat irides (Figs. l a and b). The GFAP-positive fibers formed a system of thicker bundles associated with a finemeshed network of thin fibers. A high number of positive fibers were observed in the sphincter area where they could frequently be seen reaching the sphincter margin. Thin, delicate fibers were seen surrounding large blood vessels with a distribution similar to the sympathetic innervation of these structures (Fig. lb). The GFAP-positive fibers in the mouse iris had a slightly more regular organization with a lower
334 number of fiber bundles radiating towards the sphincter (Fig. 2). These bundles were surrounded by a comparatively sparse network of thinner fluorescent fibers. In several respects the GFAP-positive fiber plexus resembled the distribution of autonomic nerve fibers in the mouse iris. In both species cell body-like areas (i.e., negative swellings surrounded by GFAP-like immunoreactivity) were observed both in fiber bundles and at branching sites of thinner fibers (Figs. la and c). Although the fluorescence intensity of the GFAP-positive fibers in the iris was relatively high they did not show the same intensity as astrocytes. This was clearly observed on rat irides where cell suspensions from cortex cerebri of 10-day-old rat pups had been injected into the anterior eye chamber of adult rats. When examined 10 days later individual as well as groups of GFAP-positive astrocytes were seen distributed unevenly over the iris. These cells had a higher fluorescence intensity than the intrinsic GFAP-positive iris fibers (Fig. 3). Furthermore, such astrocytes were readily demonstrated with the monoclonal GFAP-antibodies whereas none of these G F A P antibodies revealed any intrinsic fibers in the rat iris. The ontogeny of the GFAP-positive iris fibers was studied in the rat. Already at embryonic day 18, the earliest time point when the iris could be reliably stretch-prepared, fluorescent fibers were observed forming a sparse system of strongly fluorescent fibers (Fig. 4a). A gradual increase in fiber density was observed with increasing age so that at postnatal day 6 a system of GFAP-positive fibers resembling the adult organization was observed although the density of fluorescent fibers had still not reached the adult level (Fig. 4b).
GFAP-like immunoreactivity in grafted, host and cultured irides To ensure that no axons in the iris were stained with G F A P antiserum adult rat irides were grafted to anterior eye chamber of adult recipients in order to cause complete degeneration of all nerve fibers. The grafts were then studied after 1, 2 and 6 days in oculo. Similarly in one experiment irides were cultured for 24 h. Neither in grafted nor in cultured irides were any degenerating GFAP-positive fibers observed. However, both types of irides, and to a lesser extent host irides, became swollen and more difficult to stretch-prepare. This made it difficult to reliably assess the density of GFAP-positive fibers. Interestingly, in all 3 types of irides, as well as in other types of disturbed irides (i.e., as an intrinsic response to intraocular injections of cell suspensions) a high number of strongly fluorescent spider-like cells were observed (Fig. 5). These cells usually had 5-8 extensively branching processes. In contrast to astrocytes and GFAP-positive iris fibers where a filamentous immunoreactivity was seen the fluorescence in these cells had a grainy appearance. The cellular origin of these cells is unclear; however, they do not seem to be part of the normal
Fig. 1. Distribution of GFAP-like immunoreactivityin adult stretch-prepared rat (a and b) and mouse (c) iris. In a fiber bundles are seen meanderingover the dilator plate intermingled with a network of thin, smooth fibers, b: A large blood vessel with thin GFAP-positivefibers is seen traversing the dilator. In c cell body-like structures are seen at junctional points in the GFAP-positive plexus (arrows). Similar structures although less obvious are seen also in a. All specimens reacted with GFAP antiserum. (a) ×135, (b and c) ×330. (a and b from BjOrklund et al. 1984a.)
335
336
Fig. 2. GFAP-positive fibers in adult mouse iris. The sphincter is seen in the lower part, the ciliary body and part of the choroid membrane in the upper part of the figure. The bundles show a more radial orientation than in the rat iris (cf. Fig. la), whereas the thinner fibers form a well-organized network against a negative background. Positive fibers are observed also in the sphincter, ciliary body and choroid membrane, lmmunohistochemistry using GFAP antiserum. × 135.
337
Fig. 3. GFAP-positive cells on the iris of a rat injected 10 days earlier with a cell suspension prepared from 10-day-old rat pups. Note the obvious difference in fluorescence intensity between the intrinsic GFAP-positive fiber network and the astrocvtic cells. GFAP antiserum. × 330.
/
/~
/r,'iJ¸i ¸
.
a
,x',
•
\
Fig. 4. GFAP-like immunoreactive fibers in developing rat iris. a: Embryonic day 18; b: postnatal day 6. G F A P antiserum. × 330.
338
• -
,I
.,.
4
-
i
_. .
,.~- ~ .
.
.
.
' 7)
...':;
.
>~ ~.:
~
:_
1
a
~ .._ i-
-
..
~
,
/
h\l il" ~
~
i|"
_.,,,,~.-
•
Fig. 5. GFAP-Iike cells in ( a ) host iris 24 h after intraocular grafting of an iris graft and (b) a cultured iris after 24 h in vitro. Normal intrinsic GFAP-positive fibers are seen intermingled with brightly fluorescent cells. G F A P antiserum. ( a ) x 330, (b) X 135,
t
'-
~O
!'o
~a.
,¢?
,O
.... %
,(-
-~"
Fig. 6. Spread-prepared lens epithelium from an adult mouse. The epithelial cells are clearly GFAP-positire. G F A P antiserum, x 190.
339
GFAP-positive network. Interestingly, they are readily observed even 24 h after grafting or culturing although more cells are observed in iris grafts and host irides 6 days after grafting. Furthermore, we have observed that they tend to disappear after
f
k \~
--
z..',
. -. o's.j; -
T
.
I
.
- ~ --i-" (_~
i~ ' ~ C't
,
,.¥
-,,¢
/
"
Fig. 7. a, b and c: Miiiler gila in rat retina, a: GFAP-positive M~ller glia in a spread preparation of unfixed retina, b: Sectioned PBQ-fixed retina showing a high number of GFAP-positive Mi~ller glia spanning the retina, c: Vimentin-positive Mi~ller glia in a similar preparation, d: GFAP-positive fibers surrounding a blood vessel in a spread-prepared retina. G F A P and vimentin antisera. (a-d) x 135.
340 longer time periods in irides from eyes injected with cell suspensions. Thus, one might speculate that whatever their cellular origin (see below), these cells appear rapidly after various disturbances and then disappear with time. GFAP-Iike immunoreactivity in mouse lens epithelium Using smear preparations, adult mouse lens epithelial cells were found to be strongly fluorescent when stained with G F A P antiserum. The cells had a typical epithelial morphology with a small, rounded cytoplasm where an obvious perinuclear fluorescence was observed (Fig. 6). No GFAP-Iike immunoreactivity was observed in the nuclei. GFAP- and oimentin-like immunoreactioity in rat retina The mammalian retina contains at least 2 types of macroglial cells, the Mialler glia spanning most of the width of the retina and retinal astrocytes in the ganglion cell and nerve fiber layers. Using G F A P antiserum intensely fluorescent Mi~ller glia
Fig. 8. Typical star-shaped retinal astrocytes in unfixed spread preparations of rat retina immunohistochemicallyvisualized using a monoclonalGFAP antibody. (a) × 135, (b) x 330.
341 TABLE 1 A S U M M A R Y O F S T R U C T U R E S T H A T STAIN W I T H T H E P O L Y C L O N A L A N D M O N O C L O N A L G F A P ANTIBODIES, RESPECTIVELY
Normal iris fibers Reactive cells in disturbed irides Lens epithelial cells Normal Miiller glia Mtiller glia in injured retina Retinal astrocytes
Polyclonal G F A P
Monoclonal G F A P
Yes Yes Yes Yes Yes Yes
No No Not tested No Yes Yes
forming typical parallel arrays of fibers were stained in spread-prepared and smeared retinae (Fig. 7a). Individual Mi~ller-like fibers with a conspicuous thickening at one end were also observed in large numbers. In the brain and spinal cord we have found that astrocytes are better stained in fresh-frozen, acetone-fixed, sections than in sections of formalin-fixed tissues. Surprisingly, GFAP-positive Mialler glia were readily observed in sections of PBQ-fixed retina (Fig. 7b) whereas only a few faintly stained fibers were seen in fresh-frozen, acetone-fixed, sections. Furthermore, using monoclonal GFAP antibodies, fluorescent Mi~ller glia were only observed in injured retinae. In agreement with earlier studies (Shaw and Weber 1984), vimentin antiserum revealed many Mi~ller glia in spread preparations/smears and in acetone-fixed sections whereas only weakly fluorescent fibers were seen in sections of PBQ-fixed retinae (Fig. 7c). GFAP-positive fibers were also observed surrounding blood vessels (Fig. 7d). Star-shaped GFAP-positive retinal astrocytes with approximately 8-12 thin, smooth, strongly fluorescent processes were abundant in spread-prepared and smeared normal retina using both GFAP antiserum and monoclonal antibodies (Fig. 8). In sections star-shaped cells were observed in the ganglion cell and nerve fiber layers with both GFAP-antiserum and monoclonal GFAP antibodies.
Discussion
The present results demonstrate that GFAP-positive fibers are abundant in several tissues of the rodent eye (see Table 1 for a summary). Apart from the iris and retina we have observed GFAP-positive fibers in the choroid membrane. Supporting the findings of Hatfield et al. (1984) we have also seen GFAP immunoreactivity in the lens epithelium. The ceils responsible for the GFAP immunoreactivity in these various tissues are clearly of different types. In the retina both Mialler glia and retinal astrocytes are considered to be part of the normal astrocytic lineage, whereas the cells stained in the lens epithelium are of epithelial origin. Schwann cells are the only type of macroglial cell considered to be present in the iris. Several laboratories have also shown that Schwann cells in several peripheral nerves are GFAP-positive (Yen and Fields 1981, 1983; Dahl et al. 1982; Jessen et al. 1984). Interestingly,
342 GFAP-Iike immunoreactivity seems to be restricted to non-myelinating Schwann cells. It might therefore be speculated that the high density of unmyelinated nerve fibers in the iris (for a review see Tervo et al. 1982) is the reason why this tissue has such a dense GFAP-positive fiber network. It is noteworthy that thin GFAP-positive fibers were observed surrounding large blood vessels in the iris with a distribution similar to the sympathetic, non-myelinated, innervation of these vessels. In contrast such fibers are not observed with neurofilament antisera (Seiger et al. 1984). Instead neurofilament antiserum seems to preferentially stain nerve fibers with a large diameter and thus to a large extent myelinated (Seiger et ai. 1984). Several lines of evidence, however, indicate that the GFAP-like protein in the iris is not identical to CNS GFAP. Thus, the fluorescence intensity was clearly lower in GFAP-positive iris fibers as compared to CNS astrocytes growing on the iris. Furthermore, neither of 2 GFAP monoclonals that readily stain astrocytes were able to stain fibers in the iris. Similarly we have not observed staining of enteric glial cells using these antisera (Bjrrklund and Dahl, unpublished observations). Interestingly, Jessen et al. (1984) have recently reported that rat peripheral G F A P immunoreactivity resides in a 49 kDa protein and thus is identical in size to rat CNS GFAP. However, they could not observe any peripheral GFAP-immunoreactive structures, apart from a small proportion of enteric glial cells, using a monoclonal G F A P antibody. Taken together these results suggest that G F A P might not be one individual polypeptide but a group of two or more closely related proteins. Using G F A P antiserum we have been able to demonstrate that normal albino rat Mialler glia are GFAP-positive. Surprisingly, our monoclonal G F A P antibodies failed to demonstrate Mialler glia in normal retina, whereas retinal astrocytes were strongly stained with both G F A P monoclonals. Only after injury to the eye could staining of Mi~ller glia be observed. A possible explanation for these findings is that the G F A P molecules in normal Miailer glia and retinal astrocytes are not identical, with the latter cell type containing G F A P similar to that in brain astrocytes." It is of course also conceivable that a higher concentration of G F A P is necessary for the monoclonal G F A P antibodies to stain Mtiller glia and therefore positive staining is seen only in injured retina where the content of G F A P in all probability has increased within individual cells. Previously G F A P has not been observed in normal mammalian Mialler glia (Bignami and Dahl 1979; Bromberg and Schachner 1978; Dixon and Eng 1981; Shaw and Weber 1983, 1984) but has been detected in rats as a response to traumatic injury (Bignami and Dahl 1979) or retinal degeneration (Shaw and Weber 1983) and in some human retinae obtained at autopsy (O'Dowd and Eng 1979). GFAP-positive Mi~ller glia were, however, recently demonstrated by Bignami (1984) in goldfish retina. Since Mtiller glia are strongly vimentin-positive in normal rat (Dahl et al. 1981; Schnitzer et al. 1981; Shaw and Weber 1983) Bignami raised the possibility that Mialler glia contain intermediate filaments that are GFAP-vimentin heteropolymers (Quinlan and Franke 1983) with vimentin being the normally dominating protein in mammalian retina while the amount of G F A P increases in response to retinal injury. An intriguing finding is the appearance of a large number of star-shaped, GFAP-positive cells in various types of disturbed irides using G F A P antiserum. The
343 nature of these cells is intriguing; their distribution suggests that they are probably not completely corresponding to the cell body-like structures observed in normal irides (see above). However, they could represent normally GFAP-negative Schwann cells that undergo morphological a n d / o r biochemical changes as a response to disturbances of various types. It is well known that the amount of G F A P immunoreactivity increases in sciatic nerves during Wallerian degeneration (Yen and Fields 1981; Dahl et al. 1982). Alternatively these cells do not belong to the neuroectoderm. Since positive cells are detected in cultured irides, they have to be present normally in the irides and cannot have migrated from other organs. Interestingly, Kreutzberg and Gr~iber (1984) have reported that microglia in the facial or hypoglossal central motor nuclei can be stained with both polyclonal and monoclonal G F A P antibodies following cutting of the appropriate motor nerves. However, ultrastructurally such cells did not contain intermediate filaments of the astrocytic type (i.e., classic fibrous astrocyte fiber bundles). Similarly the reactive GFAP-positive cells in disturbed retinae do not have a clearly filamentous G F A P immunoreactivity. Thus an interesting possibility is that these cells belong to the reticuloendothelial system. Such an hypothesis is further supported by the finding of G a r d et al. (1982) that Kupffer cells, which are macrophage-type cells, are G F A P positive. That non-macroglial cells can express G F A P immunoreactivity is also demonstrated by the staining of lens epithelial cells. In conclusion, a rich plexus of GFAP-positive fibers is present in normal rodent iris. Since 2 monoclonal G F A P antibodies failed to stain these fibers, they probably contain a polypeptide not identical to CNS GFAP. In the retina Mtiller glia are intensely stained with G F A P antiserum but not with monoclonal G F A P antibodies which readily stained retinal astrocytes. Thus these studies lend further support to the idea of G F A polypeptides as a group of two or more closely related proteins that share some common antigenic determinants and have a similar molecular weight.
Acknowledgements For support in m a n y ways we are grateful to Drs. Lars Olson, ~,ke Seiger and Amico Bignami. We thank Lena Hultgren, Anna Hultghrdh, Barbro Standwerth and Ingrid Str6mberg for technical assistance and Ida Engqvist for secretarial help.
References Barber, P.C. and R.M. Lindsay, Schwann cells of the olfactory nerves contain glial fibrillary acidic protein and resemble astrocytes, Neuroscience, 7 (1982) 3077-3090. Bignami, A., Glial fibrillary acidic (GFA) protein in Miiller glia -- Immunofluorescencestudy of the goldfish retina, Brain Res., 300 (1984) 175-178. Bignami, A. and D. Dahl, The radial glia of Mialler in the rat retina and their response to injury -- An immunofluorescence study with antibodies to the glial fibrillary acidic protein, Exp. Eye Res., 28 (1979) 63-69. Bignami, A., L.F. Eng, D. Dahi and C.T. Uyeda, Localization of the glial fibrillary acidic protein in astrocytes by immunofluorescence,Brain Res., 43 (1972) 429-435.
344 Bignami, A., D. Dahl and C. Rueger, Glial fibrillary acidic (GFA) protein in normal neural cells and in pathological conditions. In: S. Fedoroff and L. Hertz (Eds.), Advances in Cellular Neurobiology, Vol. 1, Academic Press, New York, 1980, pp. 285-319. Bjrrklund, H., Astrocytes grown in oculo - - Expression of cell morphologies on the iris as revealed by GFA-immunohistochemistry, Int. J. Develop. Neurosci., 2 (1984) 377-386. Bjrrklund, H., D. Dahl, L. Olson and ~,. Seiger, Glial fibrillary acidic protein-like immunoreactivity in rodent iris - - Normal distribution and reactive changes following transplantation, J. Neurosci., 4 (1984a) 978-988. Bj~Srklund, H., D. Dahl and ~,. Seiger, Neurofilament and glial fibrillary acidic protein-related immunoreactivity in rodent enteric nervous system, Neuroscience, 12 (1984b) 277-287. Bjrrkhind, H., A. Bignami and D. Dahl, Immunohistochemical demonstration of glial fibrillary acidic protein in normal rat Mialler glia and retinal astrocytes, Neurosci. Lett., 54 (1985) 363-368. Bromberg, J.S. and M. Schachner, Localization of nervous system antigens in retina by immunohistology, Invest. Ophthal. Visual Sci., 17 (1978) 920-924. Coons, A.H., Fluorescent antibody methods. In: J.F. Danielli (Ed.), General Cytochemical Methods, Academic Press, New York, 1958, pp. 399-422. Dahl, D. and A. Bignami, lmmunogenic properties of the glial fibrillary acidic protein, Brain Res., 116 (1976) 150-157. Dahl, D., A. Bignami, K. Weber and M. Osborn, Filament proteins in rat optic nerves undergoing Wallerian degeneration - - Localization of vimentin, the fibroblastic 100 ~, filament protein in normal and reactive astrocytes, Exp. Neurol., 73 (1981) 496-506. Dahl, D., N.H. Chi, L.E. Miles, B.T. Nguyen and A. Bignami, Glial fibrillary acidic (GFA) protein in Schwann cells - - Fact or artifact, J. Histochem. Cytochem., 30 (1982) 912-918. Davison, P.F. and R.N. Jones, Filament proteins in central, cranial and peripheral mammalian nerves, J. Cell Biol., 88 (1981) 67-72. Dixon, R.G. and L.F. Eng, Glial fibrillary acidic protein in the retina of the developing albino rat - - An immunoperoxidase study of paraffin-embedded tissue, J. Comp. Neurol., 195 (1981) 305-322. Ebendal, T., L. Olson, ~,. Seiger and K.-O. Hedlund, Nerve growth factors in the rat iris, Nature (Lond.), 286 (1980) 25-28. Eng, L.F. and S.J. DeArmond, Immunochemistry of the glial fibrillary acidic protein. In: H.M. Zimmerman (Ed.), Progress in Neuropathology, Vol. 5, Raven Press, New York, 1983, pp. 19-39. Falck, B., Observations on the possibilities of the cellular localization of monoamines by a fluorescence method, Acta Physiol. Scand., 56 (Suppl. 197) (1962) 1-25. Gard, A.L., F.P. White and G.R. Dutton, G F A P in rat liver in Kupffer cells?, Soc. Neurosci. Abstr., 8 (1982) 240. Hatfield, J.S., R.P. Skoff, H. Maisel and L.F. Eng, Glial fibrillary acidic protein is localized in the lens epithelium, J. Cell Biol., 98 (1984) 1895-1898. Jessen, K.R. and R. Mirsky, Glial cells in the enteric nervous system contain glial fibrillary acidic protein, Nature (Lond.), 286 (1980) 736-737. Jessen, K.R. and R. Mirsky, Astrocyte-like gila in the peripheral nervous system - - An immunohistochemical study of enteric glia, J. Neurosci., 3 (1983) 2206-2218. Jessen, K.R., R. Thorpe and R. Mirsky, Molecular identity, distribution and heterogeneity of glial fibrillary acidic protein - - An immunoblotting and immunohistological study of Schwann cells, satellite cells, enteric glia and astrocytes, J. Neurocytol., 13 (1984) 187-200. Kreutzberg, G.W. and M. Gr~iber, Molecular cytology of satellite glial cells during regeneration of motoneurons, Soc. Neurosci. Abstr., 10 (1984) 1085. Malmfors, T., Studies on adrenergic nerves - - The use of rat and mouse iris for direct observations on their physiology and pharmacology at cellular and subcellular levels, Acta Physiol. Scand., (Suppl. 248) (1965) 1-93. O'Dowd, D.K. and L.F. Eng, lmmunocytochemical localization of the glial fibrillary acidic (GFA) protein in the M~ller cell of the human retina, Soc. Neurosci. Abstr., 5 (1979) 431. Olson, L. and T. Malmfors, Growth characteristics of adrenergic nerves in the adult rat. Fluorescence histochemical and 3H-noradrenaline uptake studies using tissue transplantation to the anterior chamber of the eye, Acta Physiol. Scand., (Suppl.) 348 (1970) 1-112.
345 Quinlan, R..A. and W.W. Franke, Molecular interactions in intermediate-sized filaments revealed by chemical cross-linking - - Heteropolymers of vimentin and giial filament protein in cultured human glioma cells, Europ. J. Biochem., 132 (1983) 477-484. Schnitzer, J., W.W. Franke and M. Schachner, Immunocytochemical demonstration of vimentin in astrocytes and ependymal cells of developing and adult mouse nervous system, J. Cell Biol., 90 (1981) 435-447. Seiger, ,~., D. Dahl, C. Ayer-LeLievre and H. Bj~rklund, Appearance and distribution of neurofilament immunoreactivity in iris nerves, J. Comp. Neurol., 223 (1984) 457-470. Shaw, G. and K. Weber, The structure and development of the rat retina - - An immunofluorescence microscopical study using antibodies specific of intermediate filament proteins, Europ. J. Cell Biol., 30 (1983) 219-232. Shaw, G. and K. Weber, The intermediate filament complement of the retina - - A comparison between different mammalian species, Europ. J. Cell Biol., 33 (1984) 95-104. Tervo, T., K. Tervo and L. Er~nk6, Ocular neuropeptides, Med. Biol., 60 (1982) 53-60. Uyeda, C.T., L.F. Eng and A. Bignami, Immunological study of the glial fibrillary acidic protein, Brain Res., 37 (1972) 81-89. Yen, S.-H. and K.L. Fields, Antibodies to neurofilament, glial filament and fibroblast intermediate filament proteins bind to different cell types of the nervous system, J. Cell Biol., 88 (1981) 115-126. Yen, S.-H. and K.L. Fields, Schwann cells contain a protein similar to the CNS astroglial filament protein, Soc. Neurosci. Abstr., 9 (1983) 235.