0306-4522/91 $3.00 + 0.00 Pergamon Press plc ,c 1991 IBRO

Neuroscience Vol. 44, No. 3, pp. 613 625, 1991 Printed in Great Britain

T R A N S F O R M I N G G R O W T H F A C T O R BETA ISOFORMS IN THE A D U L T RAT C E N T R A L A N D P E R I P H E R A L N E R V O U S SYSTEM K. UNSICKER,*K. C. FLANDERS,q"D. S. CISSEL,t R. LAFYATIS,tand M. B. SPORNt *Department of Anatomy and Cell Biology, University of Marburg, Robert-Koch-Str. 6, D-3550 Marburg, F.R.G. tLaboratory of Chemoprevention, National Cancer Institute, Bethesda, MD 20892, U.S.A. Abstract--The distribution of transforming growth factor-beta isoforms 1, 2 and 3 and transforming growth factor-beta 2 and 3 mRNAs in adult rat central and peripheral nervous system was examined using Northern blotting and isoform specific antibodies for immunocytochemistry. Transforming growth factor-beta 2 and 3 mRNA were present in all brain areas including cerebral cortex, hippocampus, striatum, cerebellum and brainstem. In sciatic nerve, transforming growth factor-beta 3 mRNA was highly expressed, but transforming growth factor-beta 2 mRNA was not detectable. Transforming growth factor-beta l-like immunoreactivity was confined to meninges and choroid plexus in the brain and connective tissue in peripheral ganglia and nerves. Transforming growth factor-beta 2 and 3 immunoreactivity entirely overlapped and, in general, were found in large multipolar neurons. Highest densities of irnmunoreactive neuronal perikarya were present in spinal cord and brainstem motor nuclei, hypothalamus, amygdaloid complex, hippocampus and cerebral cortical layers II, 1II and V. Most thalamic nuclei, superior colliculi, periaqueductal gray and striatum were almost devoid of transforming growth factor-beta 2- and 3-immunoreactive neurons. Fibrous astrocytes in white matter areas were intensely immunostained. Most dorsal root ganglionic neurons, their satellite cells and Schwann cells in peripheral nerves were also labeled. Transforming growth factor-beta 2- and 3-immunoreactive neurons were localized in brain regions that have been shown to contain neurons synthesizing and/or storing basic fibroblast growth factor suggesting possible opposing or synergistic effects of these peptide growth factors. However, the precise functions of local synthesis and storage of the transforming growth factor-beta isoforms in the nervous system are as yet unknown.

There is an increasing awareness in the field of neurosciences that the development, maintenance and repair of nervous system functions essentially requires the presence and subtle regulation of peptide growth factors. Nerve growth factor (NGF), the prototype of a growth factor, which governs development and differentiation of certain sets of neurons, 433 is only one example and, in terms of its neurotrophic actions, covers only a very small fraction of the diverse neuron and glial cell functions that may be under the control of growth factors. Fibroblast growth factors, "~,4°,41.58insulin and insulin-like growth factors, 17"25"47"54 transforming growth factor ( T G F ) alpha 3°6~ and platelet-derived growth factor 6'46'48 are now included in a rapidly expanding list of growth factors, which are found in the brain and exert important functions spanning from cell lineage specification, glial cell proliferation and neuronal differentiation to neuritogenesis, synapse formation, regulation of cell surface molecules and, reciprocally, growth factors. Interestingly, there are very few growth factors

Abbreviations: EGF, epidermal growth factor; FGFs, fibro-

blast growth factors; NGF, nerve growth factor; RA, retinoid acid; TGF. transforming growth factor. 613

which were not already known from other sources before they were discovered in the nervous system. Therefore, a pragmatic approach towards studying the presence and roles of peptide growth factors in the brain implies analyses for and functional assays on nerve tissues of such factors, which, so far, have only been known outside the nervous system. The family of T G F - b e t a s (TGF-//) at present comprises five distinct, yet highly homologous isoforms (for review see Ref. 50) acting as multifunctional regulators of cell proliferation and differentiation and performing critical roles during embryogenesis. 233738'44 Despite their high degree of structural conservation, TGF-fl isoforms seem to differ with regard to their selective expression, regulation and biological functions. Thus, TGF-fl 4 and 5 have been found in chick and frog, respectively, but not in mammals. 26''~ Moreover, TGF-fl 2, but not 1 and 5, induces mesoderm formation in amphibia. 5~ Several lines of evidence also suggest that the various TGF-/~ isoforms are selectively controlled by other growth and differentiation factors, such as glucocorticoids, retinoid acid (RA) and epidermal growth factor (EGF). Thus, RA and E G F have reciprocal effects on the levels of T G F - / / I and 2, expressed by A 549 human carcinoma cells (D. Danielpour, personal communication).

614

K. UNSICKERet al.

Considering these facts, studies on the possible differential distributions and roles of TGF-fl isoforms in the nervous system appear mandatory. Until very recently, it was not even clear whether TGF-fl occurred at all in neural tissues, since an immunocytochemical study had shown that TGF-fl 1 was absent from the CNS and confined to meningesY With the advent of more selective tools, i.e. TGF-fl isoformspecific antibodies ~3 16 and c D N A probes 9'18'59 it became possible to approach the issue of TGF-fl in neural tissues at a more refined level. In a previous paper 16 we have reported on the spatial and temporal development of TGF-fl in the nervous system of the mouse embryo. Immunoreactivities for TGF-fl 2 and 3, but not 1, were found in (radial) glial cells, neuronal perikarya and axons and, interestingly, appeared in (on) axons before showing in the cell bodies. Moreover, m R N A s for TGF-fl 1, 2 and 3 were found in the E l 5 mouse brain. EXPERIMENTAL PROCEDURES

Preparation and characterization of antibodies Antibodies directed against peptide sequences of the mature forms of TGF-fl 1, 2 and 3, and the TGF-fl 2 and 3 precursors (so-called latency associated peptide) were raised in rabbits, affinity-purified and assayed for specificity including an evaluation in Western blots. T M Antibodies against TGF-fl 2 and 3 that have not been fully described in previously published work were raised against peptide sequence 50-75 of mature TGF-fl2, 50-60 of mature TGFfl3, 237-259 of pre-pro TGF-fl2 and 81-100 of pre-pro TGF-fl3, conjugated through terminal tyrosine residues to bovine serum albumin or ovalbumin with stoichiometric amounts of diazotized benzidine and injected into rabbits. Each antibody is specific for its respective isoform and maximal crossreactivity amounts to tess than 5%. Antibodies against the non-overlapping peptides of mature and pre-pro TGF-fls produced the same patterns of staining on tissue. lmmunohistochemistry Adult Sprague-Dawley rats (Hilltop Lab. Animals, Scottdale, PA) (150-200 g) were anesthesized with Nembutal and perfused via the heart with 200 ml of neutral buffered formalin. Brain, cervical spinal cord, lumbar dorsal root ganglia and several peripheral nerves (sciatic, sural, peroneal) were dissected and post-fixed at least 24 h in neutral buffered formalin. Tissues were embedded in paraffin and TGF-fl were localized in sections (brain: coronal sections) using the above antibodies (IgG concentration: 2-8 gg/ml) as reported by Flanders et al. :4 For controls, the primary antibody was incubated with a five-fold molar excess of immunizing peptide for 2 h before applying the mixture to the section. RNA extraction and Northern blot analysis RNAs from carefully and rapidly dissected cerebral cortex, hippocampus, striatum, brainstem, cerebellum and sciatic nerve were extracted by the method of Chirgwin et al. 7 using 4 M guanidinium isothiocynate buffer for extraction and centrifuging on a 5.7 M caesium chloride cushion. RNA electrophoresis of 10-#g samples and blotting were done essentially as described. 53 Northern blots were hybridized according to the method of Church and Gilbert s and simian TGF-fl 2, ~s and murine TGF-fl 39 probes were labeled by random priming using a kit from Boehringer Mannheim (FR,G.).

RESULTS Messenger R N A s for transforming growth .[actorbetas in brain and sciatic nerve Messenger R N A s for TGF-fl 2 and 3 were found in the adult rat brain (Fig. 1), as has previously been reported for the embryonic mouse brain} 6 In addition, the message for TGF-fl 3, but not TGF-fl 2, was found in sciatic nerve preparations. Because of the purification of the R N A from whole tissue leading to some degradation and due to multiple transcript sizes characteristic of TGF-fl 2, the individual bands are difficult to distinguish, however, are most clearly seen in lanes 2 and 3 of Fig. 1. Lanes 2 (hippocampus) and 3 (striatum) display relatively high levels of the two major transcripts and also express lower mol. wt transcripts. Lane 1 (cortex) expresses primarily the higher of the two major transcripts, while lanes 4 (brainstem) and 5 (cerebellum) express the lower mol. wt transcripts. There is no detectable expression of TGF-fl 2 m R N A in sciatic nerve. For TGF-fl 3, expression of the characteristic 3.9 kb band was highest in the sciatic nerve, with lesser expression in the various brain regions. TGF-fl 2 and 3 are known to be differentially expressed in various tissues. Similarly, different transcripts of TGF-fl 2 have been shown to be differentially expressed. 3v The TGF-fl 2 transcripts, in particular those being differentially expressed, have been shown to be transcribed using different promotors. 39

1 2 3 4 5 6

fJ2-18S

,e3

-18S

.Fig. 1. Expression of TGF-fl2 and TGF-fl3 mRNA in different regions of the nervous system. 10/zg of total RNA from various regions of the rat brain or from peripheral nerve was separated on a formaldehyde/agarose gel and blotted onto Nytran paper. These RNAs were then on parallel blots hybridized to TGF-fl 2, or TGF-fl 3 32p-labeled eDNA probes. The lanes contain RNA from the following regions: (1) cortex, (2) hippocampus, (3) striatum, (4) brainstem, (5) cerebellum and (6) sciatic nerve.

TGF-fls in the nervous system

615

Fig. 2. TGF-fl 2 (A) and 2 (B, C) immunoreactivity in the ventral part of the rat spinal cord. Transverse sections were processed as described in Experimental Procedures. Panel C is a higher magnification of the boxed area in panel A. In the gray matter large motoneurons (M) and their cell bodies are intensely stained. In white matter areas (W) fibrous astrocytes (arrowheads) and axons (arrows) are labeled. In D staining is abolished when anti-TGF-fl 2 is preincubated with a 50-fold molar excess of immunizing peptide. Absorption controls were also carried out with antibodies to TGF-fl 3 and its precursor using the respective peptides and resulted in an identical abolishment of staining (not shown).t6 Peroxidase, with Mavers hematoxylin counterstain. Scale bars for A, B and D, 100/~m and for C, 10/~m.

Immunohistochem&try General remarks. Using non-crossreactive antibodies, which are specific for the respective TGF-fl isoforms, we found virtually identical immunohistochemical localization of TGF-fl 2 and 3 and their respective precursors. In contrast, TGF-fl 1 immunoreactivity was confined to the meninges and choroid plexus (not shown), as has been reported previously. ~6'23 The only difference in staining with TGF-fl 2 and 3 antibodies concerned the intensity of staining (Fig. 2). This may relate to different affinities and/or avidities of the antibodies or reflect discrepant amounts of TGF-fl 2 and 3 in their locations. Due to

this overlap in TGF-fl 2 and 3 staining, only one representative photomicrograph will be shown. Immunoreactivities indicating the presence of TGF-fl 2 and 3 mature forms and precursor proteins were found both in neurons and in glial cells of the central and peripheral nervous system. In brain and spinal cord localization in neurons varied depending on the specific area and type of neuron. Generally, large multipolar neurons were more intensely immunoreactive than smaller neurons, and motor nuclei of the spinal cord and brainstem as well as pyramidal cell layers in the cerebral cortex and hippocampus were among those which displayed the densest ac-

616

K. UNSICKERet al.

cumulation of immunoreactive neuronal perikarya. Fibrous astrocytes populating several white matter areas of the central nervous system were generally intensely stained, whereas astrocytes in gray matter displayed weak or no immunoreactivity. In peripheral ganglia, most neurons and Schwann cells contained TGF-fl 2 and 3 immunoreactivities. Central nervous system Spinal cord. Neuronal perikarya and dendrites emanating from those that stained intensely with antibodies to TGF-fl 2 and 3 were found in all laminae of the gray matter with the exception of laminae 1 and 2 (Figs 2 and 6A). All large neuronal cell bodies in the motor nuclei displayed strong staining providing the overall impression of a massive accumulation of immunoreactivity in the ventral part of the gray matter (Fig. 2C). Laminae 3-7 contained fewer, and laminae I and 2 virtually no immunoreactive cells (Fig. 6A). Neurites in the spinal cord gray were stained with varying intensities. Ependymal cells of the central canal also reacted with the TGF-fl 2 and 3 antibodies. In the white matter intense staining was associated with fibrous astrocytes and axons (Fig. 2C), whose axoplasm appeared intensely labeled. It was not possible to resolve whether oligodendrocytes were immunoreactive. Brainstem and cerebellum. A schematic presentation on the distribution of neuronal cell bodies immunoreactive for TGF-fl 2 and 3 in the rat brainstem is provided in Figs 6B-E. The highest concentration of immunopositive neurons was found in the nuclei of motor cranial nerves, in the inferior olivary nuclei, in the nucleus raphe pallidus and in the pontine nuclei. Large numbers of intensely labeled perikarya were also present in the cochlearis and vestibular nuclei. The spinal and mesencephalic trigeminal (Figs 3C and D) and most nuclei in the reticular formation, in particular the gigantocellular, contained intensely stained large neurons singly or in clusters that were interspersed among heterogeneously stained neuronal perikarya. In the gracile and cuneate nuclei as well as in the neuronal cell layers of the superior colliculus labeled neurons were very scarce (Fig. 3B). The neuronal cell bodies in the central gray surrounding the aqueduct were unstained. Ependymal cells of the fourth ventricle and aqueduct reacted with TGF-fl 2 and 3 antibodies, while the epithelium of the choroid plexus contained all three TGF-fl isoforms. Fibrous astrocytes in white matter tracts, such as pyramidal, trigeminal and lemniscal, were intensely stained. In the upper brainstem (Fig. 6E) immunoreactive perikarya were found at medium density in the substantia nigra (Fig. 3F), pars compacta and reticularis, in the red nucleus (Fig. 3E) and the group of interpenduncular nuclei. In the cerebellum the TGF-fl isoforms 2 and 3 occurred both in the cortex and in the medulla including cerebellar nuclei. Among the Purkinje cell bodies about 5% were intensely stained (Fig. 3A), while most of the remaining ones were unstained, few being

weakly stained. Immunoreactive Purkinje cells had no preferential location within the cerebellum. Staining of granule and Golgi neurons was variable. The molecular layer exhibited homogenous staining at medium intensity associated with the neuritic networks and the neuronal perikarya. This pattern did not suggest a preferential labeling of Bergmann glial cells. However, astrocytes in the cerebellar white matter were clearly stained. In all cerebellar nuclei, large and medium-sized neurons were TGF-flimmunoreactive. Diencephalon. (i) Habenula and subfornical organ. A moderate number of intensely stained neuronal celt bodies were found in the habenular nuclei, in greater density in the medial as compared to the lateral habenular nuclei (Fig. 4A). In the subfornical organ cells immunoreactive for the TGF-fl 2 and 3 isoforms were present in very large numbers. (ii) Thalamus. The subnuclei of the thalamus were entirely devoid of immunoreactive perikarya (Fig. 6F), with few notable exceptions. Large numbers of TGF-fl 2 and 3-positive neurons occurred in the reticular and in the dorsolateral thalamic nuclei. A small population of immunoreactive cells was present in the lateral geniculate nucleus. (iii) Subthalamus. A moderate number of immunopositive neurons was seen in the subthalamic nucleus and zona incerta (Fig. 6F), (iv) Hypothalamus. Of all diencephalic areas the hypothalamus displayed the greatest density of immunostained cell bodies (Fig. 6F). Numerous positive neurons were found in the arcuate, ventromedial hypothalamic, tuberal, suprachiasmatic and supraoptic nuclei. In the supraoptic nucleus all magnocellular neurons and their processes were strongly immunostained. The medial preoptic area and tuber cinereum contained a moderate amount of immunoreactive neuronal perikarya, and their numbers were low in the dorsomedial, magnocellular lateral and other hypothalamic nuclei. Tanycytes including their processes lining the floor of the third ventricle were intensely stained (Fig. 4B). In the optic nerve, presumed astrocytes with small cell bodies and ramified processes were heavily stained, and so were axons in the optic nerve. (v) Retina. Intense labeling was seen in the outer and inner plexiform synaptic layers and in cell bodies and axons of retinal ganglion cells. Inner segments of photoreceptor cells and internal end feet of Mfiller glial cells were also stained (Fig. 5A). Telencephalon. (i) Cerebral cortex. All areas and layers of the cortex contained neurons that were stained with antibodies to TGF-fl 2 and 3 (Fig. 6F). The densest concentration of immunoreactive perikarya was found in layers II, III and V (Figs 6F and 4C). In particular, the large neuronal somata in layer V and their apical dendrites were very intensely labeled (Fig. 4C and D). (ii) Hippocampus. TGF-fl 2 and 3-like immunoreactivities were very prominent and found in pyra-

!

Fig. 3. Localization of TGF-/3 2 and 3 immunoreactivity in selected areas of cerebellum and brainstem of the adult rat. Cerebellum (A), nucleus gracilis (B), nucleus of the spinal trigeminal tract (C), nucleus of the mesencephalic trigmenial tract (D), red nucleus (E), substantia nigra (F). In the cerebellum, few Purkinje cells (P), cells in the molecular (m) and granular layers and fiber and synaptic areas in these layers displayed immunostaining. The brainstem nuclei shown in panel (B) through (F) contain various proportions of TGF-/3 2- and 3-immunoreactive neuronal perikarya. Peroxidase, with Mayers hematoxylin (A, B, D F) or methyl green (C) counterstains. Scale bars, 10 l~m. 617

w

Fig. 4. Localization of TGF-fl 2 and 3 immunoreactivity in selected areas of the diencephalon and telencephalon habenular nuclei (A), tanycytes in the floor of the 3rd ventricle (B), parietal cerebral cortex (C and D), hippocampus (E), medial septal nucleus (F) and striatum (G). In the habenular complex some neurons located near the ependymal cell (e) lining of the ventricle are labeled. Tanycytes and their processes are strongly immunoreactive. In the parietal cerebral cortex, layers 2, 3 and 5 have the densest accumulation of immunoreactive perikarya. P, hippocampal pyramidal neurons. Arrows in G mark intensely stained neurons in the striatum. Scale bars in A, B and D, lO~m and C, E~G, 1 0 0 # m 618

TGF-/~s in the nervous system

619

Fig. 5. TGF-/3 2 and 3 immunoreactivity in the retina (A), dorsal root ganglion (B) and sciatic nerve (C). In the retina all cellular and synaptic layers, except for photoreceptor outer segments (asterisk) are stained. Arrows mark inner and feet of Miiller glial cells. In the dorsal root ganglion most neuronal cell bodies, axons and ganglionic satellite cells (arrow) react with anti-TGF-/~ 2 and 3 antibodies. Cross-sections of peripheral nerves such as sciatic nerve display immunoreactive axons (arrowheads) and Schwann cells (arrows). Scale bars in A, 100/~m and B and C, 10pm.

midal cell perikarya, dendrites and in granule cells of the dentate gyrus (Fig. 6F). In addition, fiber and synaptic areas apical of pyramidal cell bodies were intensely and homogeneously labeled (Fig. 4E). There were no pronounced differences concerning the labeling pattern of the different areas of the cornu ammonis. Large neurons in the hilus of the dentate gyrus were strongly immunoreactive. (iii) Amygdala and piriform cortex. Moderate to large numbers of immunoreactive neuronal cell bodies were found throughout the amygdaloid complex and in the piriform cortex. Highest densities of stained neurons occurred in the medial amygdaloid nucleus and in the cortical amygdaloid group. The band of piriform cortical neurons exhibited very intense staining in virtually all perikarya (Fig. 6F). (iv) Septal area. Magnocellular neurons in the medial septal and diagonal band nuclei displayed intense TGF-/3 2 and 3-like immunoreactivities (Figs 4F and 6F). A similarly high proportion of immunoreactive neuronal cell bodies was seen in the dorsal part of the lateral septal nucleus. The septohippocampal nucleus contained a small number of positive cell bodies. (v) Caudate-putamen. At low magnificactions (Fig. 4G), the entire caudate-putamen appeared intensely stained except for areas containing the fiber tracts of the internal capsule. At closer inspection it was apparent that only very few neuronal perikarya (approximately 5%) were immunoreactive (Fig. 5G) and that most of the staining was associated with extraperikaryal fiber and synaptic areas. (vi) Commissures. In the corpus callosum, fornix and anterior commissure small, process-bearing cells,

presumably astrocytes, spreading out between fiber bundles were labeled with antibodies to TGF-/~ 2 and 3. Peripheral nervous system Dorsal root ganglia. Neuronal perikarya and processes were, in the great majority, immunoreactive for both TGF-~ 2 and 3. However, intensities of staining varied considerably, and a few neuronal cell bodies seemed to be almost devoid of immunoreactivity (Fig. 5B). In the latter, very intense labeling of the satellite cells was consistently observed. Whether all satellite cells were reactive could not be resolved. Peripheral nerves. The axoplasm of most axons appeared intensely labeled with antibodies to TGF-/~ 2 and 3. Some faint staining was often associated with a narrow rim that surrounded the myelin sheath suggesting that myelinating Schwann cells were immunoreactive (Fig. 5C). DISCUSSION

Northern blotting and immunohistochemistry were used in this study to reveal areas of synthesis and location and of the translation products of TGF-/3 1, 2 and 3. In confirmation of previous results 16"23TGFfl 1 was absent from the neurons and glial cells and confined to meninges and choroid plexus. The only (para-) neuronal cell found so far to contain TGF-/3 1 is the adrenal chromaffin cell, 56 (Flanders, unpublished observation) and its tumor counterpart, the PC12 rat pheochromocytoma cell. 28 Nonetheless, TGF-fl 1 is active on glial cells as has been shown by its mitogenic effect on cultured Schwann cells. 49 However, since the TGF-/3 isoforms can replace each

,

~ • ~.

...........

~

~op

-

\

y,~\

"

-

, '

GrC

bas c lnteraural - 2.30ram tera

/ ~ ' ~ - ~ / -

Bregma - 11.30mm

Cervical 5 Intemural - 3.80mm

Bregnm - 12.80ram

:i:~?i:~:i:!::~:i:i?iii!:

":':::if:.,. Y

2~;

csc

\ CGL "~:ili~:!:..

" '.'.:.::::::::.

'"'

CGD

}co~ ~iiii:;~ii!ii::: ~f

CGLV

nl: xsc

Interaural 3.20mm d

Intcraural 1.20mm

Bregum - 7.80ram

B ~ g m a - 5.80ram

G

F Fr2

.~ = ~^1

'd I \~"~

.~ LSI

'

~..'N ~N.~\~ x~ \'A

"

::":':":':':'::':'~:;:~ii~':~:'i;~i

"

" ::::::::::::::::::::::::::::::::::::

lntemural 19.20ram lntefaural 5.86mm

Bregma - 3.14mm

Bregma 0.20ram

F-fls in the nervous system other in several bioassay systems, 2°'s° TGF-fl 1 may not be a physiologically relevant isoform for peripheral and central nervous system neuronal and glial cells.

Synthesis and storage of transforming growth factorbetas 2 and 3 TGF-fl 2 and 3 m R N A s and protein have a wide distribution in the peripheral and central nervous system. Their m R N A s were found in five representative regions of the brain, including cerebral cortex, hippocampus, striatum, cerebellum and brainstem. Presence of the TGF-fl 3 message in peripheral nerve suggests synthesis of the TGF-fl 3 in Schwann cells, which also stain for the TGF-fl 3 protein. The positive immunoreaction for TGF-fl 2 in the sciatic nerve is hard to reconcile with the very low or lacking signal in the Northern blot, but clearly reflects a discrepancy in the amount of transcript and translation product. In situ hybridization on chick embryos (Marascalco and Unsicker, unpublished observation) revealed the TGF-fl 3 message unequivocally over white matter tracts of the spinal cord, suggesting TGF-fl synthesis in glial cells. Given the strong immunoreactivities for TGF-fl 2 and 3 in fibrous astrocytes of white matter, we propose that these cells rather than oligodendrocytes are the sites of synthesis. The functional roles of white matter astrocytes have not been clearly determined. In situ hybridization in the chick embryo also revealed the m R N A for TGF-fi 2 and 3 to be present in retinal ganglion cells. Together, these data suggest that both glial cells and neurons in the CNS can synthesize TGF-fl.

Anatomical aspects Despite their wide distribution in the nervous system, TGF-fl 2 and 3 could be localized to discrete anatomical regions, and seemed to be abundant in some, but scarce or even absent from others. As a general feature, large neuronal cell bodies, in contrast to smaller ones, used to be intensely immunoreactive. There were, however, notable exceptions to a seemingly positive correlation of neuron size and intensity of immunoreactivity. For example, most Purkinje cells did not stain, and a certain number of magnocellular neurons in the red nucleus was also immunonegative. It was also apparent that areas whose functions, in very general terms, are related to receiving, transmitting and modulating sensory information, had only small populations of TGF-fl 2- and 3-immunoreactive neurons. Moreover, most thalamic subnuclei

were devoid of positive neurons. On the other hand, "endocrine" regions of the brain, like hypothalamus, and limbic structures were rich in TGF-fl linking their putative functions to the interface of hormonal and neuronal circuitries. Even so, the significance of this differential neuron- and area-specific pattern to the distribution of TGF-fl is not readily apparent. It may seem appropriate that long-range projecting neurons, in order to maintain these elaborate neuritic networks, must store larger amounts of these proteins than smaller neurons. To clarify this, studies including application of tracers, Golgi and ultrastructural techniques have to be designed using, for example, approximately 5% TGF-fl-immunoreactive striatal neurons, in order to investigate whether they represent the medium-sized densely spined neurons, which are the main source of striatal efferents. -~2 Likewise, it may turn out that the proportion of immunoreactive Purkinje cells represent those that project beyond the deep cerebellar nuclei into the brainstem. Although a possible co-distribution of TGF-fl 2 and 3 with certain neurotransmitters or modulators was not addressed in this study, it is probably safe to assume that TGF-fls do not co-localize with one particular neurotransmitter. For example, lamina V in the parietal cerebral cortex, where most neuronal perikarya are intensely stained, is very heterogenous with regard to its neurotransmitters. Glutamate, somatostatin, and vasoactive intestinal polypeptide have been reported to occur in these neuronsJ 2

Co-localization of transJorming growth factor betas with other growth .factors and their receptors Although there is only very limited information available with regard to a distributional mapping of growth factors in the nervous system, comparisons with our results reveal a few interesting overlaps. Several members of the family of fibroblast growth factors (FGFs) including acidic and basic F G F (bFGF) as well as higher molecular weight forms occur in neural tissues. ~ Northern blotting and in situ hybridization data indicate that, e.g. the hypothalamus and the hippocampus lm'~9 are prominent sites of b F G F synthesis, as opposed to thalamus and striaturn (Unsicker, unpublished observation). Astroglial cells can synthesize b F G F in tqtro. 12 Whether they also do it in t,ivo (Persson and Olson, personal communication) or whether neurons are the key site of synthesis m,~9 of b F G F in the brain is still debated. The distribution of b F G F protein in the rat brainstem and in some areas of the forebrain has been studied by immunocytochemistry; 2~ (Walicke, per-

Fig. 6(A G). Distribution of TGF-fl2 and 3 immunoreactive neuronal perikarya in selected coronal sections of the rat brain. Sections with respective stereotaxic coordinates are taken from the atlas by Paxinos and Watson, 43 which should also be consulted for abbreviations. The distribution density of TGF-fl immunoreactive neuronal perikarya along different coronal plains is coded as follows: [] less than 5%, [] 5-25%, [] 25-80%, ~ more than 80% immunostained perikarya, of 200 neurons evaluated within a given area. Sections were cut at 8-/~m thickness. A total of four brains was examined. N S C 44,3

I

621

622

K, UNSICKERet al.

sonal communication). There seems to be some significant overlap with the mapping of TGF-fl (this study) in that mostly large neuronal perikarya, as e.g. in the cerebral cortex, hippocampus and brainstem motor nuclei are bFGF-immunoreactive. Interestingly, the nuclei gracilis and cuneatus, which we found to be virtually devoid of TGF-flimmunoreactive perikarya, also lacked neurons staining with b F G F in the adult rat brain. The same nuclei, however, and several other brainstem nuclei, revealed a larger proportion of bFGF-positive neurons two weeks after birth. 2~ Although the developmental aspect was not addressed in the present study, we can conclude, based on a previous analysis of the developing mouse brain, 16 that TGF-fl staining is more ubiquitous in the developing than in the adult brain, covering areas like the thalamus which have only very few positive neurons in the adult. However, staining patterns for b F G F and TGF-fl do not entirely overlap. Probably the most important difference is that b F G F immunoreactivity has not been detected in glial cells in vii30. 21"45 Some co-distribution also exists for TGF-fl, the insulin receptor and insulin, respectively.5's7 Particularly high insulin receptor density has been reported for the hippocampus, medial amygdala, piriform cortex and the hypothalamus. In contrast, sites of TGF-~ synthesis as visualized by in situ hybridization are confined to the dentate gyrus, caudate nucleus and olfactory areas, 61 while TGF-~ immunoreactivity is exclusively found in a subpopulation of astrocytes. ~ Wilcox and Dernyck 61 failed to detect specific hybridizations for TGF-fl 1 in the brain. N G F , which acts on cholinergic forebrain neurons 6° has a more restricted distribution in the rat brain than the above peptide growth factors, and the same holds true for interleukin-l, which regulates N G F synthesis; 34 (personal communication). Principal sites of N G F synthesis are the hippocampus and cerebral cortex, 2'32 from where the protein is transported by retrograde axonal flow to the medial septal and basal nucleus of Meynert. To what extent these different growth factors interact and regulate each other in the brain remains to be studied. F G F and TGF-fl can act synergistically and antagonistically in many cellular systems modulating the activities of proteases, protease inhibitors and affecting extracellular matrix production? 4'52 Conceivably, they may perform similar functions in neural tissues. N G F has been shown to induce TGF-fl 1 m R N A and protein in the paraneuronal PC12 cell line, 28 and TGF-fl 3 m R N A in cultured dorsal root ganglionic neurons (Unsicker, Flanders and Marascalo, unpublished observation) suggesting functional relationships between these peptide growth factors. Functional aspects and conclusions

The present study, while descriptive, clearly identifies the TGF-fl isoforms 2 and 3 in distinct areas and

cellular structures of the nervous system. It was undertaken with the aim of providing a necessary data base upon which to design and interpret future experiments and possibly obtain functional cues for a family of peptide growth factors that until very recently did not seem to be present or have functions in the nervous system. A mitogenic effect of TGF-fl 1 and 2 on cultured Schwann cells has now been documented. 49 Moreover, we have found that TGF-fl ! modulates the proliferative effect of basic F G F on cultured astrocytes (Liidecke, personal communication) and plasminogen activator release in cultured neuroblastoma cells (Hamm et al., personal communication). Such effects are certainly consistent with the distribution of TGF-fl found in the present study, although many details, such as modes of access of the factors to their target cells and specificities for certain cellular subpopulations remain to be investigated. For example, both stimulation by an autocrine mechanism (frequently found with TGF-fl-responsive cells) 18'27and through presentation by the neuron or axon may constitute routes, by which TGF-//s act on Schwann and astroglial cells. With regard to astrocytes there was an apparent heterogeneity of TGF-// immunoreactivity. While astrocytes in the spinal cord, brainstem and cerebeltar white matter as well as in the optic nerve reacted intensely, astrogtial cells in the cerebral cortex were unstained. There is increasing evidence for astrocytes being far more heterogenous than has previously been accounted for by distinguishing type 1 and type 2 astrocytes, 36'62It will be important to correlate these cellular diversities with differences in the TGF-fl contents and, possibly, responses to TGF-fl. While astroglial cells in the gray matter of the central nervous system were either unstained or very weakly reactive, their counterparts in the peripheral nervous system, e.g. dorsal root ganglia, were often, but not consistently, immunoreactive. This variability may reflect a complex pattern of regulation of TGF-fl rather than a methodological artifact. Localization of a particular antigen in myelinating glial cells surrounding axons, at a light microscopic level, is often difficult. Immunoelectromicroscopy is likely to settle the issue of T G F - / / i n oligodendrocytes and myelinating Schwann cells. At an ultrastructural level, it will also be possible to determine the precise localization of the intense TGF-fl immunoreactivities in areas with synaptic densities, as e.g. cerebellar glomeruli, and cerebellar and hippocampal molecular layers. Functions of TGF-fl at synaptic sites may include regulation of cell surface components, protease and protease inhibitor activities and other mechanisms involved in the stabilization and constant renewal of synapses. Although TGF-fl 1 has been reported to exert a neurotrophic effect on cultured motoneurons, 35 and activin, a member of TGF-fl superfamily, induces neuronal differentiation and survival of P19 teratoma cells, 5-~ we have been unable so far to find any survival-promoting effects of

TGF-/~s in the nervous system TGF-/J 1, 2 a n d 3 for cultured e m b r y o n i c chick ciliary, dorsal root ganglionic a n d spinal cord n e u r o n s (Liidecke et al., u n p u b l i s h e d observation). A m o n g the challenges set by the new discovery o f the presence of a m u l t i f u n c t i o n a l family o f g r o w t h factors in the nervous system is also the enigma of the co-localization of TGF-/3 2 a n d 3. These two isoforms, despite their high degree o f homology, differ considerably in their p r o m o t e r regions 3~39

623

suggesting t h a t they can be differentially regulated. T h e i r possibly differential functions can now be studied, since sufficient a m o u n t s of r e c o m b i n a n t purified TGF-/3 3 have become available (Roberts, personal c o m m u n i c a t i o n ) . U. was supported by grants from the Fulbright Commission and German Research Foundation. Acknowledgements--K.

REFERENCES

I. Abraham J. A., Mergia A., Whang J. L., Tumolo A., Friedman J., Hjerrild K. A., Gospordarowicz D. and Fiddes J. C. (1986) Nucleotide sequence of bovine clone encoding fi~r the angiogenic protein, basic fibroblast growth factor. Science 233, 545 548. 2. Ayer-LeLi~vre Ch., Olson L., Ebendal T., Seiger A. and Persson H. (1988) Expression of the B-nerve growth factor gene in hippocampal neurons. Science 240, 1339-1341. 3. Baird A. and B6hlen P. (1990) Fibroblast growth factors. In Handbook o f Experimental Pharmacology (eds Sporn M. B. and Roberts A. B.), Vol. 95/I. Springer, Berlin. 4. Barde Y. A. (1989) Trophic factors andneuronal survival. Neuron 2, 1525 1534. 5. Baskin D. G., Woods S. C., West D. B., van Houten M., Posner B. I., Dorsa D. M. and Porte D. Jr (1983) Immunocytochemical detection of indulin in rat hypothalamus and its possible uptake from cerebrospinal fluid. Endocrinology 113, 1818--1825. 6. Besnard F., Perrand F., Sensenbrenner M. and Labourdette G. (1987) Platelet-derived growth factor is a mitogen for glial but not for neuronal brain ceils in vitro. Neurosci. Lett. 73, 287-292. 7. Chirgwin J. M., Przybyla A. E., MacDonald R. J. and Rutter W. J. (1979) Isolation of biologically active ribonucleic acid from sources enriched in ribonuclease. Biochemistry 18, 5294-5299. 8. Church G. M. and Gilbert W. (1984) Genomic sequencing. Proc. natn. Acad. Sci. USA 81, 1991 1995. 9. Denhez F., Lafyatis R., Kondaiah P., Roberts A. B. and Sporn M. B. (1991) Cloning by polymerase chain reaction of a new mouse TGF-/3, mTGF-/33. Growth Factors (in press). 10. Emoto N., Gonzalez A. M., Walicke P. A., Wada E., Simmons D. M., Shimasaki S. and Baird A. (1989) Basic fibroblast growth factor (FGF) in the central nervous system: idcntification of specific loci of basic FGF expression in the rat brain. Growth Factors 2, 21 29. 11. Fallon J. H., Annis C. M., Gentry L. E., Twardzik D. R. and Loughlin S. E. (1990) Localization of cells containing transforming growth factor precursor immunoreactivity in the basal ganglia of the adult rat brain. Growth Factors 2, 241 250. 12. Ferrera N., Ousley F. and Gospodarowicz D. (1988) Bovine brain astrocytes express basic fibroblast growth factor, a neurotropic and angiogenic mitogen. Brain Res. 462, 223-232. 13. Flanders K. C., Roberts A. B., Ling N., Fleurdelys B. E. and Sporn M. B. (1988) Antibodies to peptide determinants of transforming growth factor-/3 and their applications. Biochemistry 27, 739- 746. 14. Flanders K. C., Thompson N. L., Gissel D. S., van Obberghen-Schilling E., Baker C. C., Kass M. E., Ellingsworth L. R., Roberts A. B. and Sporn M. B. (1989) Transforming growth factor-~ 1: histochemical localization with antibodies to different epitopes. J. Cell Biol. 108, 653~660. 15. Flanders K. C., Cissel D. S., Mullen L. T., Danielpour D., Sporn M. B. and Roberts A. B. (1991a) Antibodies to transforming growth factor-/32 peptides: specific detection of TGF-/32 in immunoassays. Growth Factors (in press). 16. Flanders K. C., Cissel D. S., Roberts A. B., Watanabe S., Lafyatis R., Sporn M. B. and Unsicker K. (1991b) Immunohistochemical localization of transforming growth factor-/3s in the nervous system of the mouse embryo. Development (in press). 17. Gammeltoft S., Haselbacher G. K., Humbel R. E., Fehlmann M. and van Obberghen E. (1985) Two types of receptor for insulin-like growth factors in mammalian brain. Fur. molec. Biol. Org. J. 4, 3407 3412. 18. Glick A. B., Flanders K. C., Danielpour D., Yuspa S. H. and Sporn M. B. (1989) Retinoic acid induces transforming growth factor-/32 in cultured keratinocytes and mouse epidermis. Cell Regulation I, 87-97. 19. Gonzalez A. M., Emoto N., Walicke P., Shimasaki S. and Baird A. (1989) The distribution of basic fibroblast growth factor mRNA in the adult rat brain. Soc. Neurosci. Abstr. 15, 710. 20. Graycar J. L., Miller D. A., Arrick B. A., Lyons R. M., Moses H. L. and Derynck R. (1989) Human transforming growth factor /~'3: recombinant expression, purification and biological activities in comparison with transforming growth factors /31 and /32. Molec. Endocrinol. 3, 1977 1986 21. Grothe C., Zachmann K. and Unsicker K. (1991) Basic FGF-like immunoreactivity in the developing and adult rat brainstem. J. comp. Neurol. 305, 328 336. 22. Heimer L., Alheid G. F. and Zaborsky L. (1985) Basal ganglia. In The Rat Nervous System (ed. Paxinos G.), pp. 37 86. Academic Press, Sydney. 23. Heine U. I., Munoz E. F., Flanders K. C., Ellingsworth L. R., Lam H. Y. P., Thompson N. L., Roberts A. B. and Sporn M. B. (1987) Role of transforming growth factor-//in the development of the mouse embryo. J. Cell Biol. 105, 2861 2876. 24. Horton W. E., Higginbotham J. D. and Chandrasekhar S. (1989) Transforming growth factor-/3 and fibroblast growth factor act synergistically to inhibit collagen II synthesis through a mechanism involving regulatory DNA sequences. J. Cell Physiol. 141, 8 15. 25. lshii D. N. (1989) Relationship of insulin-like growth factor II gene expression in muscle to synaptogenesis. Proc. natn. Acad. Sci. USA 86, 2898 2902. 26. Jakowlew S. B., Dillard P. J., Kondaiah P., Sporn M. B. and Roberts A. B. (1988) Complementary deoxyribonucleic

624

K. IJNSICKER et al.

acid cloning of a novel transforming growth factor-/~ messenger ribonucleic acid from chick embryo chondrocytes. Molec. Endocrinol. 2, 747 755. 27. Kehrl J. H., Taylor A. S., Delsing G. A., Roberts A. B., Sporn M. B. and Fauci A. S. (1989) Further studies on the role of transforming growth factor /? in human B celt function. J. Immunol. 143, 1869~ 1874. 28. Kim S. J., Rudkin B. B., Kim K. Y., Danielpour D., Sporn M. B., Guroff G., Unsicker K. and Roberts A. B. (1991) Nerve growth factor induces transcription of transforming growth factor-/~ 1 in PC12 cells. Eur. mole('. Biol. Org. J. (submitted). 29. Kondaiah P., Sands M. J., Smith J. M., Fields A., Roberts A. B., Sporn M. B. and Melton D. A. (1990) Identification of a novel transforming growth factor-/~ (TGF-/~5) mRNA in Xenopus laevis. J. biol. Chem. 265, 1089-1093. 30. Kudlow J. E., Leung A, W. C., Kobrin M. S., Paterson A. J. and Asa S. L. (1989) Transforming growth factor-alpha in the mammalian brain: immunohistochemical detection in neurons and characterization of its mRNA. J. biol. Chem. 264, 3880- 3883. 31. Lafyatis R., Lechleider R., Kim S. J., Jakowlew S., Roberts A. B. and Sporn M. B. (1991) Structural and functional characterization of the transforming growth factor /~3 promoter: a cAMP responsive element regulates basal and induced transcription. J. biol. Chem. (in press). 32. Large T, H., Bodary S. C., Clegg D. O., Weskamp G., Otten U. and Reichardt L. F. (1986) Nerve growth factor gene expression in the developing rat brain. Science 234, 352-355. 33. Levi-Montalcini R. (1987) The nerve growth factor 35 years later. Science 237, 1154 1162. 34. Lindholm D., Heumann R., Meyer M. and Thoenen H. (1987) lnterleukin-I regulates synthesis of nerve growth factor in non-neuronal cells of rat sciatic nerve. Nature 330, 658~559. 35. Martinou J. C., Le Van Thai A., Valette A. and Weber M. J. (1990) Transforming growth factor [~1 is a potent survival factor for rat embryo motoneurons in culture. Devl Brain Res. 52, 175-181. 36. Miller R. H. and Raft M. C. (1984) Fibrous and protoplasmic astrocytes are biochemically and developmentally distinct. J. Neurosci. 4, 585-592. 37. Miller D. A., Lee A., Pelton R. W., Chen E. Y., Moses H. L. and Derynck R. (1989a) Murine translbrming growth factor-//2 cDNA sequence and expression in adult tissues and embryos. Molec. Endocrinol. 3, 1108--1114. 38. Miller D. A., Lee A., Matsui Y., Chert E. Y., Moses H. L. and Derynck R. (1989b) Complementary cDNA cloning of the murine transforming growth factor-r3 (TGF-fl3) precursor and the comparative expression of TGF-/i3 and TGF-//I messenger RNA in murine embryos and adult tissues. Molec. Endocrinol. 3, 1926-1934. 39. Noma T., Geiser S., Glick A. B., Miller I. E. and Sporn M. B. (1990) Molecular cloning and characterization of the promoter of the transforming growth factor-~2 gene. J. cell. Biochem. 14C, 307. 40. Otto D., Frotscher M. and Unsicker K. (1989) Basic fibroblast growth factor and nerve growth factor administered in gel foam rescue medial septal neurons after fimbria fornix transection. J. Neurosci. Res. 7,2, 83-91. 41. Otto D. and Unsicker K. (1990) Basic FGF reverses chemical and morphological deficits in the nigrostriatal system of MPTP-treated mice. J. Neurosci. 10, 1912-1921. 42. Parnavelas J. G. and McDonald J. K. (1983) The cerebral cortex. In Chemical Neuroanatomy, (ed. Emson P. C.). Raven Press, New York. 43. Paxinos G. and Watson Ch. (1986) The Rat Brain in Stereotaxic Coordinates. 2nd edn. Academic Press, Sydney. 44. Pelton R. W., Nomura S., Moses H. L. and Hogan B. L. M. (1989) Expression of transforming growth factor-/?2 RNA during murine embryogenesis. Development 106, 759-767. 45. Pettmann B., Labourdette G., Weibel M. and Sensenbrenner M. (1986) The brain fibroblast growth factor is localized in neurons. Neurosci. Lett. 68, 175-180. 46. Raivich G. and Kreutzberg G. (1987) Expression of growth factor receptors in injured nervous tissue 11. Induction of specific platelet-derived growth factor binding in the injured PNS is associated with a breakdown in blood-nerve barrier and endoneurial interstitial edema. J. Neurocytol. 16, 707-711. 47. Recio-Pinto E. and lshii D. N. (1988) Insulin and related growth factors: effects on the nervous system and mechanism for neurite growth and regeneration. Neurochem. Int. 12, 397~,14. 48. Richardson W. D., Pringle N., Mosley M. J., Westermark B. and Dubois-Dalcq M. (1988) A role for platelet-derived growth factor in normal gliogenesis in the central nervous system. Cell 53, 309-319. 49. Ridley A. J., David I. B,, Stroobant P. and Land H. (1989) Transforming growth factors-~q I and/~2 are mitogens for rat Schwann cells. J. Cell Biol. 1tl9, 3419 3424. 50. Roberts A. B. and Sporn M. B. (1990) The transforming growth factors-/L In Handbook o f Experimental Pharmacology, "Peptide Growth Factors and Their Receptors" (eds Sporn M. B. and Roberts A. B.), Vol. 95, pp. 419-472. Springer, Heidelberg. 51. Roberts A. B., Rosa F., Roche N. S., Coligan J. E., Garfield M., Rebbert M. L., Kondaiah P., Danielpour D., Kehrl J. H., Wahl S. W., Dawid I. B. and Sporn M. B. (1990) Isolation and characterization of TGF-/~2 and TGF-,q5 from media conditioned by Xenopus XTC cells. Growth Factors 2, 135-147. 52. Saksela O., Moscatelli D. and Rifkin D. B. (1987) The opposing effects of basic fibroblast growth factor and transforming growth factor beta on the regulation of plasminogen. J. Cell Biol. 105, 957 963. 53. Sambrook J., Fritsch E. F. and Maniatis T. (1989) Molecular Cloning, 2nd edn. Cold Spring Harbor Laboratory Press. 54. Sara V. R. and Carlsson-Schwirnt C. (1988) The role of insulin-like growth factors in the regulation of brain development. Prog. Brain Res. 73, 87 ~99. 55. Schubert D., Kimura H., LaCorbiere M., Vaughan J., Karr D. and Fischer W. H. (1990) Activin is a nerve cell survival molecule. Nature 344, 868-870. 56. Thompson N. L., Flanders K. C., Smith J. M., Ellingsworth L. R., Roberts A. B. and Sporn M. B. (1989) Expression of transforming growth factor-/~ I in specific cells and tissues of adult and neonatal mice. J. Cell Biol. 108, 661-669. 57. Unger J.. McNeill T. H., Moxley R. T., White M,, Moss A. and Livingston J. N. (1989) Distribution of insulin receptor-like immunoreactivity in the rat forebrain. Neuroscience 31, 143-157. 58. Unsicker K., Reichert-Preibsch H., Schmidt R., Pettmann B., Labourdette G. and Sensenbrenner M. (1987) Astroglial and fibroblast growth factor have neurotrophic function for cultured peripheral and central nervous system neurons. Proe. natn. Acad, Sei. USA 84, 5459-5463. 59. van Obberghen-Schilling E., Kondaiah P., Ludwig R. L., Sporn M. B. and Becker C. C. (1987) Complementary deoxyribonucleic acid cloning of bovine transforming growth factor-/~ 1. Molec. Endocrinol. 1, 693-~698.

TGF-/~s in the nervous system

625

60. Whittemore S. R. and Seiger A. (1987) The expression, localization and functional significance of/~-nerve growth factor in the central nervous system. Brain Res. Rev. 12, 439 464. 61. Wilcox J. N. and Derynck R. (1988) Localization of cells synthesizing transforming growth factor-alpha mRNA in the mouse brain. J. Neurosci. 8, 1901 1904. 62. Wilkin G. P., Marriott D. R. and Cholewinski A. J. (1990) Astrocyte heterogeneity. Trends Neurosci. 13, 43 46. (Accepted 22 April 1991)