Androgen action in fetal mouse spinal cord cultures: metabolic and morphologic aspects

62

Brain Re,search, 4()6 ( IqN7) 62- 72 Klscvicr

BRE 12416

Androgen action in fetal mouse spinal cord cultures: metabolic and morphologic aspects Kurt F. Hauser l, Neil J. MacLusky 3 and C. Dominique Toran-Allerand 1'2 t Center for Reproductive Sciences, 2Departments of Neurology and of Anatomy and Cell Biology, Columbia University, College of Physicians and Surgeons, New York, N Y 10032 (U.S.A.) and ~Department of Obstetrics and Gynecology, Yale Medical School, New Haven, CT06510 (U.S.A.)

(Accepted 29 July 1986) Key words: Androgen; Developing spinal cord; Organotypic culture; Testosterone metabolism; Aromatase; 5a-Reductase; Neurotrophic factor

Morphologic and metabolic aspects of androgen effects on the developing spinal cord were studied in organotypic cultures of the El3 (embryonic day 13) fetal mouse lumbosacral spinal cord, maintained as either hemisected, homologous explant pairs co-cultured with bulbocavernosus muscle (morphologic studies), or as whole cross-sectional segments without muscle in which the nutrient medium was supplemented with muscle extract (metabolic studies). Metabolic studies demonstrated the total absence of aromatase activity. 5a-Reductase activity, on the other hand, increased differentially in a segment-dependent manner in spinal cord explants from 0 to 35 days in culture, suggesting regional differences in the utilization of testosterone and its 5a-reduced metabolites. In all studies, spinal cord explants showed androgen-dependent increases in neurite outgrowth, although this was most pronounced in spinal cord-muscle co-cultures. These results indicate that androgens per se affect very early development throughout the entire lumbosacral spinal cord. and that this influence is not restricted to those segments reported to be sexually dimorphic in the adult.

INTRODUCTION Increasing evidence suggests that the spinal cord is a m a j o r site for androgen action within the central nervous system (CNS). Spinal reflexes associated with mating behavior in the rat are responsive to circulating testosterone levels in the adult 14. Both the target perineal muscles (medial and lateral bulbocavernosus, ischiocavernosus) responsible for copulatory behavior in the male 15'39 and the m o t o n e u r o n s innervating these muscles contain androgen receptors 3,8,1s,22. B r e e d l o v e and A r n o l d 3 described a motor nucleus in spinal cord segments L 5 and L 6 that is present in the adult male rat but not the female. They t e r m e d this v e n t r o m e d i a l l y located sexually dimorphic nucleus which innervates the medial and lateral bulbocavernosus muscles the 'spinal nucleus of the bulbocavernosus' (SNB). J o r d a n et al. Is r e p o r t e d a

similar dimorphism in the dorsal lateral m o t o r nucleus ( D L N ) of L 5 and L 6 which innervates the ischiocavernosus. K o j i m a and Sano 21 found sexually dimorphic patterns of serotonergic innervation, as well as increased size and n u m b e r of m o t o n e u r o n s in spinal segments La_ 2 which innervate the cremaster muscle in male rats. In the SNB and D L N of both sexes, m o t o n e u r o n size and n u m b e r are increased following testosterone (T) and 5a-dihydrotestosterone ( D H T ) administration perinatally or during the early postnatal period, but not by estradiol (Ez) 2-5. These and the additional findings of the absence of the SNB in (1) androgen receptor-deficient testicular feminized male (Tfm/y) mutants 3 and (2) normal male rats following p r e n a t a l anti-androgen (flutamide) administration 4, suggest that androgens per se may be responsible for changes occurring within the SNB during development. It is not known, however,

Correspondence: K.F. Hauser, Department of Anatomy, The M.S. Hershey Medical Center, The Pennsylvania State University, Hershey, PA 17033, U.S.A.

0006-8993/87/$03.50 (C~1987 Elsevier Science Publishers B.V. (Biomedical Division]

63 whether other androgen receptor-containing ventral cord regions are sensitive to androgen during development. Although androgen receptors have been demonstrated in the adult spinal cord of both sexes4°, their ontogenetic history is unknown. Similarly, while small numbers of estrogen-concentrating cells have been localized in the dorsal 2°'37, but not ventral 32, horns of the adult rat spinal cord (although some DLN motoneurons have been reported to show small amounts of nuclear accumulation of E 2 (ref. 5)), their presence during development is as yet unknown. The metabolic pathways of androgen in the developing spinal cord are unknown. In the brain, on the other hand, T either acts without conversion, or enters one of two metabolic pathways. (1) T can be converted to DHT by 5a-reductase, or (2) it can be converted to E 2 by aromatase (for review see MacLusky and Naftolin26). Both metabolic products have dramatic organizational effects in the brain during development (for reviews see MacLusky and Naftolin 26, Toran-Allerand46), as well as activational influences in the adult CNS (for reviews see McEwen29'3°). As part of an ongoing investigation into the developmental significance of androgen and estrogen receptors in the developing murine CNS, we have been studying androgen metabolism in individual segments of fetal spinal cord cultures. We have focused particularly on the lumbar spinal cord segments, because they include the androgen-responsive, sexually dimorphic regions associated with the rat SNB and DLN, which have also been documented at L 5 and L 6 in the male mouse, but not in the female and Tfm/y mutant (Hauser and Toran-Allerand, unpublished observations). In this study, we present novel evidence that T acts solely as an androgen without conversion to E2 to directly influence early spinal cord development at both sexually dimorphic and non-dimorphic lumbosacral levels. MATERIALSAND METHODS Spinal cord culture The entire spinal cord with attached dorsal root ganglia was removed aseptically from embryonic day 13 (El3) (in one case El4) mice obtained following cervical dislocation and rapid caesarian section 35. The mice were mated in our breeding colony for a

single 45-min period (07.30-08.15 h) and checked for the presence of vaginal plugs (sperm positivity) immediately thereafter. The day a plug is found is considered day 0 of gestation. Whole cross-sections of spinal cord segments L 1, L 2, L 4 (in 1 case L3) , L 5, L6, and S 1were maintained with their attached dorsal root ganglia as organotypic explants 35 (Fig. 1). Since at this age embryos could not be readily sexed, it was assumed that the E13/E14 mouse spinal cord is sexually undifferentiated and bipotential. Each spinal cord level (one explant) was placed on a collagencoated coverslip, fed twice weekly with one drop (50 #1) of a complex biological nutrient medium standard for our laboratory and maintained in a Maximow double coverslip assembly45. This medium consisted of steroid-deficient (gelded horse) serum (18.6%); amodified Eagle's minimal essential medium (59.1%) supplemented with glutamine (292/~g/ml); glucose (9 mg/ml); sodium pyruvate (1 mM); transferrin (100 /zg/ml; Collaborative Research, Lexington, MA); selenium (4 ng/ml); putrescine (16.1 /~g/ml; Sigma Chemical Co., St. Louis, MO) and insulin (10/~g/ml, Collaborative Research). Spinal cord cultures were additionally supplemented with muscle extract (5%) 42, 9-day chick embryo extract (5%), nerve growth factor (100 ng/ml, Collaborative Research), estrone (El, 5 ng/ml; Sigma) and 17fl-estradiol (E2, 25 ng/ml; Sigma). Antibiotics are never used. The explants were cultured for 0 (short-term) or 32 days (long-term) prior to the administration of either 20 pmol of [7a-3H]T (0.5 /zCi, 25 Ci/mmol; New England Nuclear) or 8.7 pmol of [1,2-3H]19-OH androstenedione (0.5 pCi, 57.5 Ci/mmol; New England Nuclear) for 72 h. Both [3H]labeled androgens were extensively (3 x) repurified before use by chromatography on a column (10 cm x 1 cm) of Sephadex LH-20 in the system isooctane:methylene chloride:methanol, 50:50:1. Because of the dependence of the SNB and DLN on androgens, long-term cultures received additionally 20 ng/ml T (Sigma), but were subsequently deprived of T 24 h prior to the administration of the radiolabeled androgens. E 1 and E 2 were added as a trap for any labeled estrogens which might be formed by aromatization. In addition, cultures of the septum-preoptic area, regions which are known to aromatize androgens in vitro 46 as in vivo, were taken from newborn mice and maintained under identical conditions as the short-term

64

Fig. 1. Typical appearance of a living spinal cord explant after 23 h in culture. Note the attached dorsal root ganglia, symmetry between right and left halves, and distinct dorsal and ventral regions, x60.

spinal cord cultures to assure that aromatase activity was not lost due to an artifact of the culture conditions. Each culture with its spent medium was frozen individually on dry ice and stored at -70 °C prior to analysis. Controls consisted of samples of the starting medium, as well as blank collagen-coated coverslips and explants of El3 cartilage and fibroblasts maintained on collagen-coated coverslips for identical periods of time in the presence of radiolabeled androgens. Cultures were examined microscopically at daily intervals, which provided an opportunity to follow the development of the explant and outgrowth.

Analysis of androgen metabolites Analysis of the radiolabeled androgen metabolites was carried out essentially as described by MacLusky et al. 27. Unlabeled 'carrier' steroids 200/~g each of estrone (El), estradiol (E2), testosterone (T), androstenedione ( A E D I O N E ) , 5a-dihydrotestosterone (DHT), androsterone (AONE), androstandione ( A A D I O N E ) , 5a-androstan 3a-,17fl-diol (3aDIOL), 5a-androstan 3fl-,17fl-diol (3flDIOL), and [14C]labelled E2, T and A E D I O N E (2000 dpm each)

were added to the frozen cultures to facilitate separation and identification of the metabolites and correction for procedural losses. The cultures were extracted into 2 ml of ethanol. The ethanol extracts were evaporated to dryness under a stream of air and the androgen and estrogen metabolites present separated by phenolic partition. The non-phenolic fraction was further analyzed by thin-layer chromatography (TLC) in the systems described by Denef and coworkers 7. The estrogens in the phenolic extract were separated by TLC on silica gel GF254 plates (Brinkman, Westbury, NY) in the system described by Ruh 38, which resolves El and E 2 from T and its major non-phenolic metabolites. The separated EL and E 2 spots were eluted from the TLC plates with a small volume of methylene chloride and methanol, then evaporated to dryness and acetylated by overnight incubation at room temperature with 0.2 ml pyridine and 0.2 ml acetic anhydride. Residual acetic anhydride was neutralized by the addition of 1 ml methanol. The tubes were evaporated to dryness and the residual acetate fractions chromatographed once more on silica gel plates in the system methylene

65 c h l o r i d e : e t h y l a c e t a t e (98:2). R a d i o c h e m i c a l purity

c a v e r n o s u s muscles with s o m e s u r r o u n d i n g c o n n e c -

of t h e final s e p a r a t e d a n d r o g e n and e s t r o g e n m e t a -

tive tissue and pieces o f a d j o i n i n g p e r i n e a l muscles

bolites was c o n f i r m e d by crystallization to c o n s t a n t

( l e v a t o r ani, i s c h i o c a v e r n o s u s ) , w e r e r e m o v e d f r o m

specific activity f r o m a c e t o n e and h e x a n e . F o r the

m a l e m i c e (in s o m e cases f e m a l e p e r i n e a l muscles

cultures i n c u b a t e d with 1 9 - O H A E D I O N E ,

w e r e used with essentially similar results), and co-

o n l y the

p h e n o l i c f r a c t i o n was p r o c e s s e d , as d e s c r i b e d a b o v e .

c u l t u r e d with spinal c o r d e x p l a n t s using p r e v i o u s l y

T h e n o n - p h e n o l i c m e t a b o l i t e s f r o m t h e s e cultures

d e s c r i b e d m e t h o d s 35'36. E a r l i e r stages w e r e not at-

w e r e discarded.

t e m p t e d , b e c a u s e of the difficulty in identifying t h o s e

Co-culture o f spinal cord explants with muscle

w e r e n o t i n c l u d e d in the m e d i u m in this p a r a d i g m to

muscles in fetal mice. M u s c l e and e m b r y o extracts T o e v a l u a t e w h e t h e r o r n o t a n d r o g e n s had an ef-

m a x i m i z e any t r o p h i c i n f l u e n c e s p r o v i d e d by the

fect on the m o r p h o l o g i c a l d e v e l o p m e n t o f spinal c o r d

muscle explants p e r se. O n e h e m i s e c t e d spinal c o r d

cultures, h o m o l o g o u s ( m i r r o r ) pairs 45 f r o m the right

s e g m e n t was t r e a t e d with 20 ng/ml T and 20 ng/ml

and left halves of l u m b a r spinal c o r d s e g m e n t s 1 - 6

D H T , while the c o n t r o l , h o m o l o g o u s half, was de-

were explanted

In s o m e in-

p r i v e d of a n d r o g e n s . A f t e r 27 days in vitro, h o m o l o -

stances, strips o f n e w b o r n m e d i a l and lateral bulbo-

gous e x p l a n t pairs w e r e fixed and stained for n e u r o f i -

as d e s c r i b e d a b o v e .

TABLE I

Androgen metabolites present in short- and long-term spinal cord cultures, and short-term cultures of the septum-preoptic areafollowing exposure to [3H]testosteronefor 72 h Values are presented as mean (pmoi) + S.E.M. Abbreviations: T, testosterone; AEDIONE, androstenedione; DHT, 5a-dihydrotestosterone; 3aDIOL, 5a-androstan-3a-17fl-diol; 3flDIOL, 5a-androstan-3fl-17fl-diol; AONE, androsterone; AADIONE, 5a-androstan-3,17-dione. (S.E.M. are not reported for the septum-preoptic area since only two observations were made.)

T

AEDIONE

DHT

3aDIOL

3flDIOL

AONE

AADIONE

16.70 + 0.713 10.78 + 1.07

0.061 + 0.048 0.302 + 0.026

0.437 + 0.061 0.389 + 0.045

1.047 + 0.132 5.226 + 0.695

0.083 + 0.006 0.463 + 0.079

0.115 + 0.035 1.47 + 0.284

0.060 + 0.034 0.577 _+ 0.157

16.12 + 1.28 11.57 + 1.18

0.072 + 0.044 0.229 + 0.053

0.399 + 0.056 0.604 + 0.222

0.747 + 0.108 4.11 + 0.436

0.056 + 0.021 0.372 + 0.047

0.119 + 0.013 1.18 + 0.149

0.044 + 0.028 0.479 + 0.158

Short Long

15.34 + 1.47 12.09 + 0.534

0.195 + 0.067 0.279 + 0.055

0.461 + 0.038 0.405 + 0.044

0.704 + 0.081 4.15 + 0.709

0.085 + 0.021 0.362 + 0.063

0.127 + 0.030 1.08 + 0.118

0.118 _.+0.058 0.490 + 0.128

L5 Short Long

16.74 + 0.847 13.26 + 0.985

0.163 + 0.100 0.312 + 0.047

0.498 + 0.092 0.416 + 0.085

0.587 + 0.165 3.39 + 0.674

0.051 + 0.007 0.202 + 0.034

0.065 + 0.014 1.04 + 0.259

0.155 + 0.089 0.461 + 0.136

Short Long

16.71 + 0.786 14.13 + 0.955

0.141 + 0.071 0.315 + 0.046

0.628 + 0.082 0.271 + 0.066

0.545 + 0.059 2.48 + 0.530

0.070 + 0.020 0.249 + 0.102

0.066 + 0.015 0.852 + 0.193

0.171 + 0.121 0.392 + 0.121

Short Long

16.23 + 0.648 12.67 + 0.925

0.364 + 0.234 0.304 + 0.032

0.523 + 0.086 0.494 + 0.074

0.741 + 0.117 3.72 + 0.592

0.046 + 0.014 0.284 + 0.039

0.080 + 0.018 1.12 ___0.113

0.280 + 0.235 0.475 + 0.094

L1

Short Long L2 Short Long L3_4

L6

Sl

Septum/ preoptic area

14.28

0.05

0.61

3.72

0.43

0.40

0.11

Preoptic area

12.65 + 1.35

0.02 __+0.01

0.98 + 0.12

3.93 + 1.95

0.24 + 0.06

0.64 + 0.14

0.16 + 0.05

66 brils (neurofilament protein bundles) by a modified Holmes' reduced silver-impregnation method (Toran-Allerand, unpublished).

10

0 ,'-" a.

5

! 'i,

RESULTS 0

[3H]Androgen metabolism The distribution of androgen metabolites recovered from the short- and long-term spinal cord cultures after incubation with [3H]T is shown in Table I. In all cases, the major metabolite recovered was 3aDIOL, followed by D H T , A O N E , 3fl-DIOL, AAD I O N E and A E D I O N E . In the short-term cultures (4 observations/segment), metabolic activity was relatively low; the majority of substrate remaining unchanged at the end of the 72-h incubation. In the long-term cultures (6-8 observations/segment), on the other hand, there was an increase in the total quantity of 5a-reduced products obtained. There was also an apparent increase in 3a, 3fl and 17fl hydroxysteroid dehydrogenase activities, as evidenced by significantly increased quantities of AONE, 3flDIOL, A A D I O N E and A E D I O N E recovered from the majority of the long term cultures (ANOVA, P < 0.01) and the increased 3 a D I O L to D H T ratio. The total quantities of 5a-reduced metabolites (DHT, A O N E , 3 a D I O L , 3flDIOL and A A D I O N E ) produced by the cultures are presented in Fig. 2. In the short-term cultures, no differences were observed between the amounts of 5a-reduced metabolites produced by cultures from the different spinal cord levels. After long-term culture, however, there were greater differences in the total 5a-reduced products from different segments than were present in short-term cultures, although these differences were not quite significant, using A N O V A (P = 0.053). However, when comparing short and longterm cultures, there were disproportionate increases in the metabolic activities of cultures especially from L I - L 4 and $1, compared to those from L5 and L 6. In fact, there was no significant change after long-term culture in the quantities of 5a-reduced metabolites produced by the L 6 explants (P < 0.05).

Aromatase activity Incubation of short- or long-term cultures with either [3H]T or [3H]19-OH A E D I O N E yielded no

L1

L2

L4

L5

L6

$1

SPINAL CORD SEGMENT

Fig. 2. Total 5a-reduced products (mean pmol + S.E.M.) produced by individual short-term (black bars) and long-term cultures (white bars) from different lumbosacral spinal cord levels 72 h following administration of 20 pmol [3H]testosterone. While 5a-reductase activity was evident in short-term cultures, no segment differences were apparent. In contrast, long-term cultures showed a significant segment-dependent increase in 5a-reductase activity when compared to short-term explants in all segments except L 6.

detectable levels of E l or E2 formation above the blank of the assay procedure (30-50 cpm; approximately 10-3 pmol) (at least 4 observations/labeled precursor/segment). In contrast, incubation of 2 explants each from the septum-preoptic area with either [3H]T or [3H]19-OH A E D I O N E yielded large quantities of both E1 (x = 0.008 and 0.13 pmol, respectively) and E2 (x = 0.13 and 1.01 pmoi, respectively). The preoptic area alone yielded 0.08 + 0.01 pmol of E 1 and 1.52 ___0.44 pmol of E2 in the presence of [3HIT (3 observations each), while incubation with [3H]19-OH A E D I O N E yielded 0.52 pmol E l and 2.14 pmol E 2 (2 observations each). In the case of the preoptic area, cultures receiving 19-OH A E D I O N E converted more than 30% of the added substrate to estrogen.

Androgen-increased outgrowth Androgen administration resulted both in an increase in the area and apparent density of the neurite outgrowth from silver-impregnated homologous explant pairs from all spinal cord segments after 27 days in vitro. In the case of spinal cord segments which had been co-cultured with bulbocavernosus muscles there was survival of muscle fragments, as well as an even greater increase in the amount of neurite outgrowth compared to explants grown in the presence of muscle and embryo extracts alone (Fig. 3). Examination of silver-impregnated cultures revealed the heterogeneous nature of the neurites; consisting of a mixture of the processes of dorsal root ganglia and

67

Fig. 3. Appearance of homologous-paired right and left spinal cord segments co-cultured with bulbocavernosus muscles in the absence (A) and presence (B) of androgens after 27 days in vitro. Note the increased numbers of neurites emanating from the androgentreated explant. Darkfield micrographs of Holmes' silver-impregnated cultures from segment L5. ×50.

those emanating from numerous regions of the spinal cord explant. The ventral roots, when they could be clearly identified, were composed of greater numbers of neurites in the androgen-treated cultures. In

muscle co-cultured explants, the ventral root neurites were better organized than in those receiving muscle and embryo extracts alone, and often projected directly to the explanted muscle fragment. Af-

68 ter a week or more in vitro, innervated muscle fragments displayed synchronized contractions, indicating innervation of the muscle by motoneurons within the explant. In the absence of androgen, in contrast, both the density and length of the neurites from cocultured spinal cord explants as well as the number of myocytes appeared reduced. Most commonly, much of the cultured muscle underwent fatty degenerative changes after several weeks in vitro. Additional studies concerning these experiments, including co-culture of spinal cord with non-perineal (thigh), androgen receptor-containing 22 striated muscles, or muscle extract alone will be described in a subsequent paper to evaluate whether or not the responses are unique to co-culture with bulbocavernosus muscles. DISCUSSION Testosterone is readily metabolized in E13 spinal cord explants cultured in the absence of muscle targets, suggesting that androgens may be physiologically active in the spinal cord during its early development. The importance of the target in regulating the motoneuron number in general is well recognized, as is the unique dependence of the perineal muscles on androgen. Thus the inclusion of muscle and embryo extracts in the medium was designed to provide a modicum of trophic sustenance in order to evaluate the role of androgens in spinal cord cultures directly without the potentially modifying influences of its androgen-sensitive and androgen-metabolizing target. Since the short- and long-term cultures are of approximately similar size and volume, the increase in androgen metabolism in the long-term cultures is consistent with the process of differentiation and development in vitro, although further studies will be required to determine whether there is a change in the number of cells in the explants during culture. The difference between L5 and L 6 and the other spinal segments studied ( L I - L 4, $1) suggests that there are lower levels of 5ct-reductase activity in those segments containing the sexually dimorphic nuclei. In striking contrast, no aromatase activity was observed in the spinal cord explants. Even in the presence of 19-OH-AEDIONE, which is the preferred substrate for aromatase, no detectable amounts of estrogen were formed in either the short- or longterm cultures. It should be noted that high levels of

aromatase activity are present under identical culture conditions in cultures from other regions of the newborn CNS. These include the hypothalamus, preoptic area (reported in the present study), as well as the cerebral cortex, but not the cerebellum (MacLusky and Toran-Allerand, in preparation). Thus, any influence estrogens might have on spinal cord development would have to be due to circulating estrogens directly. Differences have been found in the amounts of 5areductase in different CNS regions in rodents 7'23'41"47. The presence of 5a-reductase activity has been inversely correlated with the presence of estrogen-receptor containing cells23; with the reduction in 5a-reductase activity perhaps resulting from the flux of T through the aromatase pathway to form estrogen 23. However, in the present study no such alternative pathway exists. Moreover, estrogen receptor concentrations are very low throughout the adult spinal cord 20'32'37, including the region of the SNB 5, and their presence during development is unknown. While aromatase might compete with 5ct-reductase for T in identical or adjoining regions of the brain, the importance of 5ct-reductase mediating the actions of androgens per se in the developing spinal cord should not be overlooked, since aromatase activity is not present. Why 5a-reductase is present in reduced amounts in L 5 to L 6 during development is uncertain. If the reduction in 5ct-reductase activity is the result of differences associated with the small numbers (approximately 300 each) of SNB and/or DLN cells, then one would expect these differences to be greatly attenuated when examining entire cross-sectional explants. Since this is not the case, the findings suggest that 5ct-reductase activity is not limited to those sexually dimorphic motoneurons, but to other neurons or glial cells within the explant, as has been suggested in vivo 23. Moreover, it is not known (1) whether 5ctreductase-containing subpopulations of cells differ, or (2) whether differences in the content of the enzyme reside within the individual cells themselves in regionally dependent manners. In adult rats, a significantly greater percentage of DLN motoneurons accumulate [3H]T in males than in females, but there is no sex difference in [3H]DHT labeling, suggesting differences can be present within individual cells 5. Additional variability might be due to the possible

69 variable anatomical location of spinal motor nuclei. Motor nuclei innervating specific muscles in both the rat 33 and mouse 31 can differ considerably in their rostral-caudal position between individuals. In the adult rat both T and DHT, but not E2, enhance penile reflexes 13 which are associated with the SNB and DLN TM. The above evidence suggests that 5a-reductase is important and produces sufficient amounts of DHT in adult male rats. In the frog, Xenopus laevis, there is an increase in 5a-reductase activity in the spinal cord segments associated with the clasp reflex involved in male mating behavior 9'19. In adult rats, T enhances hypoglossal nerve regeneration, while D H T does not have a pronounced effect 52. If the products of the aromatization of T are not responsible for this effect, then 5a-reductase activity would be important in determining the availability of each respective androgen. During development DHT, but not T, appears to be differentially effective in the expression of male external sexual characteristics 51. In cases where a deficiency or abnormality in the enzyme 5a-reductase exist, the external genitalia are female in appearance, despite the production of T by functioning testes 17'4s. Thus, 5a-reductase is critical for the masculinization of external male genitalia. It is possible that T and DHT also act differently towards targets in the developing CNS and that varying amounts of 5a-reductase activity in the target tissue might differentially affect this response. Breedlove 2, using HRPtracing, found that in female rats treated with T during development, the SNB motoneurons correctly innervate the retained bulbocavernosus muscles which are normally only present in the male. However, in DHT-treated females the motoneurons innervating these muscles were located outside their normal position, and the SNB itself was reduced in size. The reason(s) for this difference is unclear, yet it indicates that T and DHT can influence developmental events differently. The distribution of 5a-reductase activity might normally serve to direct the correct positioning and subsequent innervation of SNB motoneurons with their appropriate target muscles in the male. At the SNB, the availability of T, therefore, appears to be increased, while that of D H T is decreased by these apparent regional differences in 5ct-reductase activity. However, the opposite is true with respect to the muscle targets which produce DHT. Regional

differences in spinal cord 5a-reductase activity might somehow be correlated with regional differences in the innervation patterns with respect to the sexually dimorphic muscles. One speculative possibility is that DHT promotes the secretion of trophic factors from both the muscles and the spinal cord. If so then it might be appropriate for those segments of the spinal cord containing the SNB motoneurons to have lower 5a-reductase activity during development. Such a pattern might prevent the motoneurons from being 'attracted' back into the cord. In any event, the present study indicates that variations in 5a-reductase activity is one developmental mechanism by which the differential effect of T and DHT on androgen responsive targets within the spinal cord may be mediated. Short-term cultures (E13) are capable of metabolizing T very early during development. For this to be physiologically significant in vivo, however, there must be a source of androgen so early during development. By E17 (ref. 50) in the rat and even earlier in the rabbit 51, plasma T has been reported to be present in male fetuses and in adjacent female siblings. More recent evidence, furthermore, has shown that the rodent placenta also produces both T and A E D I O N E in significant amounts as early as El0 (refs. 43,44), or E12 (ref. 28), suggesting that the placenta may serve as an important source of androgen during early neural development. Androgen-treated, silver-impregnated spinal cord/muscle co-cultures showed a dramatic increase in neurite outgrowth surrounding the explant when compared with cultures deprived of androgen for 27 days in vitro. This neurite-promoting effect of androgen was observed in all lumbosacral spinal cord segments as long as bulbocavernosus muscle was present. However, we cannot determine from these observations whether androgen is acting directly on the spinal cord and/or muscle to elicit this response, since in the absence of androgen neither the muscles nor the neurites thrive (muscle or embryo extracts were not added in co-culture experiments) which is similar to what is observed postnatally in vivo 6. It was also noted that androgens increased neurite outgrowth from all lumbar spinal cord segments co-cultured with bulbocavernosus muscle, indicating that motoneurons throughout the lumbar spinal cord which do not normally innervate these muscles can do so in response to androgen. It is interesting to speculate that

70 the unique androgen-dependence of these muscles could be responsible for the formation and survival of an SNB-like nucleus at any lumbar level, and that spatiotemporal factors align those axons from the motoneurons positioned at the location of the SNB with the bulbocavernosus muscles during development. In cultures receiving muscle and embryo extracts alone, there was a less dramatic albeit definite androgen-dependent increase in neurite outgrowth, compared to explants co-cultured with muscle. Assuming that androgen is only increasing the growth of a subpopulation of the neurites comprising the total outgrowth (presumably the ventral roots), then the less defined arrangement of neurites, due to the absence of the organizing influence of muscle targets, might obscure changes occurring in only a portion of those neurites. In addition, muscle and embryo extracts may be inadequate in providing the same trophic sustenence as actual muscle. Nevertheless, the morphologic and biochemical findings utilizing muscle and embryo extracts suggest a direct action of androgen on the developing spinal cord. The recent findings of Fishman and Breedlove 1° imply that the SNB is dependent on a direct androgen action on the target bulbocavernosus muscles. In our co-culture studies, therefore, the primary target for androgen would be the bulbocavernosus muscles. While we do not dispute the above interpretation, evidence for a direct neuronotrophic effect of androgens on motoneurons throughout the lumbar spinal cord has been found in explants that have been cultured in the presence of muscle extract alone 16, and is supported by the above-mentioned results of the present study. This prenatal effect of androgen on motoneuron survival, which is independent of synaptic contact with muscle, differs from the highly specific, largely postnatal action of androgen on bulbocavernosus muscle survival. Combined with the present metabolic results, this suggests that androgens act directly on the spinal cord by a separate mechanism in addition to any indirect action via the muscle. Thus the observed trophic effect of androgen indicates that non-aromatized androgens directly influence the developmental interaction between motoneurons and muscle.

Our observations of neuronotrophic and myotrophic effects of androgen in spinal cord-muscle co-cultures during development may have clinical implications for the adult by contributing to an understanding of the pathophysiology of amyotrophic lateral sclerosis (ALS) a sexually differentiated disease. Appel 1 proposed the lack of a 'disorder-specific neurotrophic hormone' as responsible for the loss of motoneurons in ALS. Denervated muscle releases a 56kDa protein which promotes terminal motor axon sprouting ~1 and against which the serum from some ALS patients is immunoreactive j2. Androgens or perhaps estrogens formed by aromatization 2~'25 might influence the very availability of the 56-kDa protein reported to cause motor axon sprouting. Weiner 49 has proposed that the loss of androgen receptors might have a role in the etiology of ALS based, in part, on a 1.5-2.5-fold higher incidence of the disease in men, and the fact that the only motoneurons which are spared happen to be in cranial nerve nuclei that do not bind appreciable amounts of androgen 49. Since Onuf's nucleus 34 is also spared in advanced ALS 49, Weiner speculated that Onuf's nucleus also might not bind significant amounts of androgen. However, a different interpretation is necessary, since Onuf's nucleus in humans is believed to be analogous to the rat SNB which shows high levels of androgen binding 5. The presence of androgen receptors in adult spinal cord and muscle suggests that androgen has a continued influence on nerve-muscle interactions throughout life. ACKNOWLEDGEMENTS The skilled technical assistance of Dr. Emil Lev and Mr. Cliff Hurlburt and the typing assistance of Mrs. Barbara Hauser are gratefully acknowledged. This research was supported in part by the National Institutes of Health (HD-08364), the Whitehall Foundation, and a National Institute of Mental Health Research Scientist Development Award (MH-00192) to C.D.T.-A. ; Institutional funds from a Mellon Foundation Grant to The Center for Reproductive Sciences, and by the National Institutes of Health (NS-19610) to N.J.M.

71 REFERENCES 1 Appel, S.H., A unifying hypothesis for the cause of amyotrophic lateral sclerosis, parkinsonism, and Alzheimer disease, Ann. Neurol., 10 (1981) 499-505. 2 Breedlove, S.M., Hormonal control of the anatomical specificity of motoneuron-to-muscle innervation in rats, Science, 227 (1985) 1357-1359. 3 Breedlove, S.M. and Arnold, A.P., Hormone accumulation in a sexually dimorphic motor nucleus in the rat spinal cord, Science, 210 (1980) 564-566. 4 Breedlove, S.M. and Arnold, A.P., Hormonal control of a developing neuromuscular system. 1. Complete demasculinization of the male rat spinal nucleus of the bulbocavernosus using the anti-androgen flutimide, J. Neurosci., 3 (1983) 417-423. 5 Breedlove, S.M. and Arnold, A.P., Sex differences in the pattern of steroid accumulation by motoneurons of the rat lumbar spinal cord, J. Comp. Neurol., 215 (1983) 211-216. 6 (~ih~ik, R., Gutman, E. and HanzlNovfi, V., Involution and hormone-induced persistence of the M. sphincter (levator) ani in female rats, J. Anat., 106 (1970) 93-110. 7 Denef, C., Magnus, C. and McEwen, B.S., Sex differences and hormonal control of testosterone metabolism in rat pituitary and brain, J. Endocrinol., (1973) 605-621. 8 Dionne, F.T., Dube, J.Y., Lesage, R.L. and Tremblay, R.R., In vivo androgen binding in rat skeletal and perineal muscles, Acta Endocrinol., 91 (1979) 362-372. 9 Erulkar, S.D., Kelley, D.B., Jurman, M.E., Zemlan, F.P., Schneider, G.T. and Krieger, N.R., Modulation of the neural control of the clasp reflex in male Xenopus laevis by androgens: a multidisciplinary study, Proc. Natl. Acad. Sci. U.S.A., 78 (1981) 5876-5880. 10 Fishman, R.B. and Breedlove, S.M., Neonatal androgen maintains sexually dimorphic perineal muscles in the absence ofinnervation, Soc. Neurosci. Abstr., 11 (1985) 530. 11 Gurney, M.E., Supression of sprouting at the neuromuscular junction by immune sera, Nature (London), 307 (1984) 546-548. 12 Gurney, M.E., Inhibition of terminal axonal sprouting by serum from patients with amyotrophic lateral sclerosis, New Engl. J. Med., 311 (1984) 933-939. 13 Hart, B.L., Activation of sexual reflexes of male rats by dihydrotestosterone but not estrogen, Physiol. Behav., 23 (1979) 107-109. 14 Hart, B.L., Role of testosterone secretion and penile reflexes in sexual behavior and sperm competition in male rats: a theoretical contribution, Physiol. Behav., 31 (1983) 823-827. 15 Hart, B.L. and Melese-d'Hospital, P.Y., Penile mechanisms and the role of the striated penile muscles in penile reflexes, Physiol. Behav., 31 (1983) 807-813. 16 Hauser, K.F. and Toran-Allerand, C.D., Testosterone increases cell numbers in organotypic cultures of fetal mouse spinal cord, Soc. Neurosci. Abstr., 11 (1985) 898. 17 Imperato-McGinley, J., Guerrero, L., Gautier, T. and Peterson, R.E., Steroid 5a-reductase deficiency in man: an inherited form of male pseudohermaphroditism, Science, 186 (1974) 1213-1215. 18 Jordan, C.L., Breedlove, S.M. and Arnold, A.P., Sexual dimorphism and the influence of neonatal androgen in the dorsolateral motor nucleus of the rat lumbar spinal cord, Brain Research, 249 (1982) 309-314. 19 Jurman, M.E., Erulkar, S.D. and Krieger, N.R., Testos-

terone 5a-reductase in spinal cord of Xenopus laevis, J. Neurochem., 38 (1982) 657-661. 20 Keefer, D.A., Stumpf, W.E. and Sar, M., Estrogen-topographical localization of estrogen-concentrating cells in the rat spinal cord following [3H]estradiol administration, Proc. Soc. Exp. Biol. Med., 143 (1973) 414-417. 21 Kojima, M. and Sano, Y., Sexual differences in the topographical distribution of serotonergic fibers in the anterior column of rat lumbar spinal cord, Anat. Embryol., 170 (1984) 117-121. 22 Krieg, M., Characterization of the androgen receptor in the skeletal muscle of the rat, Steroids, 28 (1976) 261-274. 23 Krieger, N.R., Scott, R.G. and Jurman, M.E., Testosterone 5a-reductase in rat brain, J. Neurochem., 40 (1983) 1460-1464. 24 Longcope, C., Widrich, W. and Sawin, C.T., Secretion of estrone and estradiol-17fl by human testis, Steroids, 20 (1972) 439-448. 25 MacDonald, B.S., Sykes, P.J., Adhikary, P.M. and Harkness, R.A., The identification of 17a-hydroxy-17'fl-methyl androsta-l,4-dien-3-one as a metabolite of 17'fl-hydroxy17a-methylandrosta-l,4-diene-3-one in man, Biochem. J., 122 (1971) P26. 26 MacLusky, N.J. and Naftolin, F., Sexual differentiation of the central nervous system Science, 211 (1981) 1294-1303. 27 MacLusky, N.J., Clark, C.R. and Naftolin, F., Is estrogen biosynthesis involved in androgen action in the spinal cord? Metabolism and binding of androgens in the spinal cord of the rat, Brain Research, submitted. 28 Matt, D.W. and Macdonald, G.J., In vitro progesterone and testosterone production by the rat placenta during pregnancy, Endocrinology, 115 (1984) 741-747. 29 McEwen, B.S., Neural gonadal steroid actions, Science, 211 (1981) 1303-1311. 30 McEwen, B.S., Steroid hormone action in the brain: cellular and behavioral effects. In F.O. Schmitt, S.J. Bird and F.E. Bloom (Eds.), Molecular Genetic Neuroscience, Raven Press, New York, 1982, pp. 265-275. 31 McHanwell, S. and Biscoe, T.J., The localization of motoneurons supplying hindlimb muscles of the mouse, Phil Trans. R. Soc. London Ser. B, 293 (1981) 477-498. 32 Morrell, J.I., Wolinsky, T.D., Krieger, M.S. and Pfaff, D.W., Autoradiographic identification of estradiol-concentrating cells in the spinal cord of the female rat, Exp. Brain Res., 45 (1982) 144-150. 33 Nicolopoulos-Stournaras, S. and Iles, J.F., Motor neuron columns in the lumbar spinal cord of the rat, J. Comp. Neurol., 217 (1983) 75-85. 34 Onufrowicz, B., Notes on the arrangement and function of the cell groups in the sacral region of the spinal cord, J. Nerv. Ment. Dis., 26 (1899) 498-504. 35 Peterson, E.R., Co-culture of adult skeletal muscle with fetal mouse spinal cord complex, TCA Manual, 4 (1978) 921-924. 36 Peterson, E.R. and Crain, S.M., Regeneration and innervation in cultures of adult skeletal muscle coupled with fetal rodent spinal cord, Exp. Neurol., 36 (1972) 136-159. 37 Pfaff, D.W. and Keiner, M., Atlas of estradiol-concentrating cells in the central nervous system of the female rat, J. Comp. Neurol., 151 (1973) 121-158. 38 Ruh, T.S., Simultaneous separation of estrogens and androgens using thin-layer chromatography, J. Chromatogr., 121 (1976) 82-84. 39 Sachs, B.D., Role of striated penile muscles in penile re-

72 flexes, copulation, and induction of pregnancy in the rat, J.

Reprod. Fertil., 66 (1982) 433-443. 40 Sar, M. and Stumpf, W.E., Androgen concentration in motor neurons of cranial nerves and spinal cord, Science. 197 (1977) 77-79. 41 Selmanoff, M.K., Brodkin, L.D., Weiner, R.I. and Siiteri, P.K., Aromatization and 5a-reduction of androgens in discrete hypothalamic and limbic regions of male and female rat, Endocrinology, 101 (1977) 841-848. 42 Smith, R.G. and Appel, S.H., Extracts of skeletal muscle increase neurite outgrowth and cholinergic activity of fetal rat spinal motor neurons, Science, 219 (1983) 1079-1081. 43 Soares, M.J. and Talamantes, F., Gestational effects on placental and serum androgen, progesterone, and prolactin-like activity in the mouse, J. Endocrinol., 95 (1982) 29-36. 44 Soares, M.J. and Talamantes, F., Midpregnancy elevation of serum androstenedione levels in the C3H/HeN mouse: placental origin, Endocrinology, 113 (1983) 1408-1412. 45 Toran-Allerand, C.D., Sex steroids and the development of the newborn mouse hypothalamus and preoptic area in vitro: implications for sexual differentiation, Brain Research, 106 (1976) 407-412. 46 Toran-Allerand, C.D., On the genesis of sexual differentia-

47

48

49 50

51

52

tion of the central nervous system: morphogenetic conscquences of steroidal exposure and possible role of ~t-fetoprotein, Prog. Brain Res., 61 (1984) 63-97. Verhoeven, G., Lamberigts, G. and De Moor, P., Nucleus associated steroid 5rx-reductase activity and androgen responsiveness. A study in various organs and brain regions ot' rats, J. Steroid Biochem.. 5 (1974) 93-1110. Walsh. P.C., Madden, J.D., Harrod, M.J.. Goldstein, J.[-., MacDonald, P.C. and Wilson, J D . , Familial incomplete male pseudohermaphroditism, type 2. Decreased dihydrotestosterone formation in pseudovaginal perineoscrotal hypospadias, N. Engl. J. Med., 291 (1974) 944-949. Weiner, L.P., Possible role of androgen receptors in amyotrophic lateral sclerosis, Arch. Neurol., 37 (1980) 129-131. Weisz, J. and Ward. I.L., Plasma testosterone and progesterone titers of pregnant rats, their male and female fetuses, and neonatal offspring, Endocrinology, 106 (1980) 306-316. Wilson, J.D., Goerge, F.V. and Griffin, J.E., The hormonal control of sexual development, Science, 211 (1981) 1278-1284. Yu, W.-H.A. and Srinivasan, R., Effect of testosterone and 5a-dihydrotestosterone on regeneration of the hypoglossal nerve in rats, Exp. Neurol., 71 ( 1981 ) 431-435.