Neuron,

Vol. 4, 303-311,

February,

1990, Copyright

0 1990 by Cell Press

Translational Regulation in Cultured Sympathetic

of Somatostatin Neurons

Katharyn Spiegel: Vivien Wang,+ and John A. Kessler Department of Neurology Department of Neuroscience Albert Einstein College of Medicine Bronx, New York 10461

Summary Coculture of sympathetic neurons with ganglion nonneuronal cells elevated levels of preprosomatostatin mRNA but did not alter neuronal synthesis, content, or release of somatostatin. Treatment of sympathetic neurons with culture medium conditioned by exposure to ganglion nonneuronal cells similarly elevated preprosomatostatin mRNA. Treatment with conditioned medium elevated somatostatin levels in pure neuronal cultures, but not in neurons cocultured with nonneuronal cells. Conditioned medium also failed to increase peptide levels in neurons cultured on a substratum of killed nonneuronal cells, despite a large increase in preprosomatostatin mRNA. These observations suggest that contact of sympathetic neurons with nonneuronal cell membranes inhibits the increase in peptide synthesis, but not the increase in preprosomatostatin mRNA after treatment with conditioned medium. Thus neuronal interactions with nonneuronal cells regulate somatostatin metabolism at both the mRNA and peptide levels. Regulatory effects on the mRNA and the peptide are separable and do not necessarily occur in parallel, and translational controls may be the rate-limiting factors. Introduction A central issue for understanding development of the nervous system is how neurons express the appropriate neurotransmitters at proper levels. Considerable evidence suggests that neuronal interactions with nonneuronal cells play a critical role in regulating transmitter expression in cholinergic, catecholaminergic, and peptidergic neurons (Patterson and Chun, 1974; Bunge et al., 1978; Mudge, 1981; Kessler, 1985). The effects of nonneuronal cells may be mediated by the release of diffusible factors, by contact with the neuron, or by both. For example, sympathetic neurons, which in vivo display predominantly noradrenergic traits, become cholinergic either when they are exposed to soluble factors released by nonneuronal cells (Patterson and Chun, 1974; Bunge et al., 1978; Weber, 1981; Kessler, 1984) or when they contact certain nonneuronal cell membranes (Hawrot, 1980; Kess*Present address: Department of Pharmacology, Pharmaceutical Research Division, Warner-Lambert, Michigan. +Present rytown,

address: Regeneron, New York.

777 Old

Saw Mill

Parke-Davis, Ann Arbor, River

Road,

Tar-

ler et al., 1986; Wong and Kessler, 1987). The precise levels at which nonneuronal cells control transmitter metabolism are not fully known. At least some influences of nonneuronal cells are mediated at the mRNA level (Raynaud et al., 1987; Kessler et al., 1988, Sot. Neurosci., abstract). However, regulation of neurotransmitter metabolism may occur at the level of mRNA synthesis or stability, translational or posttranslational processing, catabolism, or release. A major focus of this study was to define intracellular mechanisms mediating effects of interactions between neurons and nonneuronal cells on transmitter expression. Sympathetic neurons may express a number of different transmitters including norepinephrine, acetylcholine, somatostatin, substance P, and others (Hijkfelt et al., 1977a, 197713; Patterson, 1978; Black, 1982; Kessler, 1984). The precise complement and levels of transmitters expressed by sympathetic neurons in vitro depend upon the culture environment. For example, synthesis of the peptide neurotransmitter, somatostatin, is stimulated by neuronal interactions with target tissues (Kessler, 1984). Coculture of sympathetic neurons with ganglion nonneuronal cells slightly decreases or has no effect on neuronal levels of somatostatin. However, coculture prevents the increase in somatostatin in neurons interacting with target tissues (Kessler, 1984). This study examines the effects of ganglion nonneuronal cells on neuronal levels of preprosomatostatin mRNA, somatostatin synthesis, and peptide release to define mechanisms underlying these complex cell-cell interactions. We find that nonneuronal cells regulate somatostatin metabolism both by the release of diffusible factors and by cell-cell contact and that controls are independently exerted on levels of the mRNA and on peptide synthesis. Results Effects of Coculture with Ganglion Nonneuronal Cells Cocultureof sympathetic neurons with ganglion nonneuronal cells stimulated neuronal expression of cholinergic traits and of substance P and decreased levels of noradrenergic traits (Patterson, 1978; Kessler, 1985). However, coculture exerted less effect on somatostatin levels, which were diminished or unchanged after coculture (Kessler, 1985). To define further the effects of nonneuronal cells on somatostatin expression, levels of preprosomatostatin mRNA were examined in neurons cultured in the presence and absence of ganglion nonneuronal cells. Cultured sympathetic neurons contained preprosomatostatin mRNA identical in size (0.65 kb) to mRNA isolated from hypothalamus. Coculture resulted in a more than 33-fold increase in preprosomatostatin mRNA (Figure 1). Despite the large increase in preprosoma-

A

A

10 T

O-O

NEURONS

0-O

NEURONS AND NONNEURONAL

ONLY

2

4

CELLS

8

TH

“b 6 z B

1 2

4

2

B

0

0

6

0

5

NEURONS ONLY WITH NON-NEURONAL

6

8

10

12

TIME (hours)

O-O

SOMATOSTATIN

e-0

PROSOMATOSTATiN

2.500

mRNA

PEPTIDE

Figure

1. Effects

of Coculture

with

Ganglion

Nonneuronal

Cells

Sympathetic neurons were cultured in the absence and presence of ganglion nonneuronal cells. After 10 days, the cultures were harvested and assayed for levels of somatostatin and preprosomatostatin mRNA. (A) Northern blot analysis of mRNA isolated from neurons cultured in pure neuronal culture (lane 1) or in the presence of nonneuronal cells (lane 2). The blot was hybridized to the somatostatin (SS) probe and then rehybridized to the tyrosine hydroxylase (TH) probe. mRNA from one culture dish of neurons was loaded onto each lane. Neuron numbers did not differ significantly between the pure neuronal cultures and the cocultu res. (B) Levels of somatostatin and preprosomatostatin mRNA obtained by densitometty of autoradiograms. Peptide is expressed as mean femtograms per neuron + SEM (n = 5). mRNA is expressed as mean arbitrary absorbance units k SEM. n = 4; 5 dishes per point, per experiment. *Differs from pure neuronal group at P < 0.01.

I PULSE C

mRNA, neuronal levels of somatostatin were not significantly altered by coculture with nonneuronal ceils (Figure 1). By contrast, levels of tyrosine hydroxylase mRNA (1.8 kb; identical in size to adrenal tyrosine hydroxylase mRNA) were decreased by a mean of 50% in the same cocultures (Figure IA), and tyrosine hydroxylase activity was also slightly decreased (data not shown; see Kessler, 1985). Thus coculture of sympathetic neurons with ganglion nonneuronal cells significantly elevated levels of preprosomatostatin mRNA, but did not alter levels of somatostatin peptide. Neuronal somatostatin levels reflect release of the peptide as well as synthesis and catabolism. Neuronal release of somatostatin into the culture medium was therefore measured in the presence and absence of nonneuronal cells. Cultures were grown with peptidase inhibitors in the medium to prevent degradation

O-O e-0

2.500

Figure

TIME (hours)

3.000

PULSE

tostatin

I

CHASE

2. Effects

SOMATOSTATIN PROSOMATOSTATIN

I,

Cl&E

of Coculture

TIME

(hours)

on Somatostatin

Synthesis

Neurons cultured in the absence or the presence of ganglion nonneuronal cells for 2 weeks were changed to cysteine-free medium to which $S]cysteine was added (see Experimental Procedures). n = 2; 3 dishes per point, per experiment. (A) Cells cultured continuously in the presence of [%.]cysteine were examined by combined immunoprecipitation-HPLC for the incorporation of radioactivity into somatostatin. The data represent the combined counts of both the somatostatin and the prosomatostatin HPLC peaks. There was no difference in the prosomatostatin/somatostatin ratio in the presenceand absence of nonneuronal cells at any time point (see also [B] and [Cl). n = 2; 3 dishes per point, per experiment. (B and C) After 3 hr of culture in the presence of [%]cysteine (PULSE), some cultures were washed and chased with excess cold cysteine. Incorporation of radioactivity into prosomatostatin and somatostatin was examined in neurons cultured in the absence (B) and presence (C) of ganglion nonneuronal cells. n = 2; 3 dishes per point, per experiment.

Regulation 305

of Somatostatin

Expression

of somatostatin; addition of bacitracin (0.2 mg/ml) and captopril(O.5 mM) allowed recovery of more than 85% of the labeled somatostatin added to the medium (data not shown). Neurons cultured in the presence of nonneuronal cells released 2.8 + 1.1 fg of somatostatin per neuron per 24 hr, whereas neurons cultured in the absence of nonneuronal cells released 2.5 f 0.6 fg per neuron per 24 hr. Thus coculture did not significantly alter the rate of release of the peptide. To assess directly the effects of nonneuronal ceils on neuronal synthesis of somatostatin, [?S]cysteine incorporation into somatostatin was examined in pure neuronal cultures and in cultures also containing nonneuronal cells (see Experimental Procedures for description of the combined immunoprecipitation and HPLC techniques used). In the first series of studies, cells were cultured in cysteine-free medium containing 250 tXi/ml [35S]cysteine (1100 Ci/mmol), and the counts per minute incorporated into somatostatin were examined at varying times after addition of tracer (Figure 2A). The presence of nonneuronal cells did not significantly alter the counts per minute incorporated into somatostatin for periods of incubation of up to 12 hr. To help determine the rate of catabolism of somatostatin as well as the synthetic rate, cells were pulsed with the tracer for 3 hr, washed, and chased with a large excess of unlabeled cysteine (Figures 2B and 2C). Significant labeling of a peak comigrating with prosomatostatin was detectable within 1 hr of addition of tracer to either pure neuronal cultures (Figure 2B) or neuronal cultures also containing nonneuronal cells (Figure 2C). Two hours after addition of tracer, counts per minute in prosomatostatin reached a steady-state level and significant labeling of somatostatin itself was detectable. Two hours after the chase, labeling of prosomatostatin was no longer detectable and labeling of somatostatin reached its peak value. Thereafter, the number of counts per minute in somatostatin decreased continually over the subsequent 48 hr. The calculated rate of somatostatin synthesis in pure neuronal cultures (0.0023 fmol per neuron per hr) did not differ significantly from the synthetic rate in neurons cocultured with nonneuronal cells (0.0026 fmol per neuron per hr). The rate of decrease in labeling of somatostatin after the pulse (catabolism plus release) also did not differ significantly between pure neuronal cultures (half-life 23.5 hr) and cultures also containing nonneuronal cells (half-life 22.5 hr). Thus despite a large increase in preprosomatostatin mRNA, coculture of neurons with ganglion nonneuronal cells did not alter either the rate of synthesis or the rate of catabolism of somatostatin. Effects of Nonneuronal Cell Conditioned Medium The effects of coculture on preprosomatostatin mRNA metabolism could have reflected either release by nonneuronal cells of soluble regulatory molecules or direct contact of neurons with nonneuronal ceil membranes. To define the possible role of soluble fac-

A ss

TH

1234

B

20

1234

0 NEURONSONLY B%! WITH NON-NEURONAL CELLS

0 CONTROL GNCM

PEPTIDE Figure

3. Effects

CONTROL GNCM

0

mRNA

of GNCM

Neurons cultured in the absence or the presence of ganglion nonneuronal cells were treated with 30% GNCM. After 10 days of treatment, the cultures were examined for content of somatostatin and preprosomatostatin mRNA. (A) Northern blot hybridized to the somatostatin (SS) probe and rehybridized to the tyrosine hydroxylase fTH) probe. Lane 1, neurons only; lane 2, neurons + nonneuronal cells; lane 3, neurons + CNCM; lane 4, neurons + nonneuronal cells + GNCM. (B) Levels of somatostatin and preprosomatostatin mRNA. Peptide is expressed as mean femtograms per neuron k SEM. mRNA is expressed as mean arbitrary absorbance units + SEM. n = 3; 3 dishes per point, per experiment. *Differs from control (neurons only) at P < 0.005 by Anova. **Differs from control (neurons only) at P < 0.001 by Anova. * * *Differs from control (neurons only) at P < 0.002 by Anova.

tors, the effects of nonneuronal cell conditioned medium were examined. Ganglion nonneuronal cells (principally Schwann cells with a few fibroblasts [Kessler, 19851) were grown to confluence, and culture medium withdrawn after 2 days of exposure to the nonneuronal cells was used to treat cultures of sympathetic neurons. Treatment of pure neuronal cultures with ganglion nonneuronal cell conditioned medium (GNCM) elevated levels of preprosomatostatin mRNA more than 21-fold and increased peptide levels more than 3-fold (Figure 3). However, levels of somatostatin in cocultures treated with CNCM were not higher than those in untreated, pure neuronal cultures, despite much larger (42-fold higher) levels of preprosomatostatin mRNA. These observations indicate that ganglion nonneuronal cells released a soluble factor(s) which stimulated levels of preprosomatostatin mRNA. The factor(s) also increased levels of somatostatin in neurons grown alone, but had no effect on peptide in neurons cocultured with nonneuronal cells. By contrast, treatment with GNCM significantly decreased levels of both tyrosine hydroxylase mRNA (Figure 3) and tyrosine hydroxylase activity

NEWKXl 306

O-O l - l

ZO-

PEPTIDE mRNA

-aI

Table 1, Effects Levels of NGF

:=

D! 1 0

10

20

30

50

Effects

of Treatment

PERCENTRFCM

Figure RFCM

4. Dose-Response

Curve

for

in the Presence

of Saturating

NGF

- RFCM

+ RFCM

100 nglml

3.4 + 0.03 3.9 + 0.03 3.7 + 0.05

9.4 + 0.07 9.9 + 0.05 9.3 f 0.07

1 &ml 5 &ml

Sympathetic neurons were cultured with NGF in either the absence (- RFCM) or 30% RFCM. After IO days, the cultures of somatostatin, which is expressed ron + SEM (n = 4).

ra 50

40

of RFCM

various concentrations of the presence (+RFCM) of were assayed for content as femtograms per neu-

with

Pure neuronal cultures were treated for 10 days with different concentrations of RFCM and were examined for content of somatostatin (open circles) and preprosomatostatin mRNA (closed circles). Peptide is expressed as mean femtograms per neuron + SEM. mRNA is expressed as mean arbitrary absorbance units k SEM. n = 3; 5 dishes per point for somatostatin and 3 dishes per point for preprosomatostatin mRNA.

in these cultures. Results obtained using medium conditioned by exposure to purified ganglion Schwann cells (n = 3; 5 cultures per group per experiment) or purified sciatic nerve Schwann cells (n = 1; 5 cultures per group) were similar to the effects of GNCM (data not shown). Isolation and cultureof ganglion nonneuronal cells or ganglion Schwann cells in the absence of neurons is time-consuming and yields relatively small amounts of tissue. To find a more convenient source of cells that produce factors which stimulate levels of preprosomatostatin mRNA, we examined the effects of media conditioned by exposure to various primary and secondary cell lines. RN22 schwannoma cell conditioned medium stimulated preprosomatostatin mRNA comparably to GNCM (data not shown). Medium conditioned by exposure to neonatal rat skin fibroblasts (RFCM) was even more effective in stimulating levels of preprosomatostatin mRNA in sympathetic neurons (see Figure 4). In view of existing prior studies of effects of RFCM on sympathetic neurons (Patterson and Chun, 1977; Smith and Kessler, 1988) and the potent effects on levels of preprosomatostatin mRNA, subsequent studies used RFCM as a tool for studying the regulation of somatostatin expression. A dose-response study of the effects of RFCM on pure neuronal cultures was performed. Treatment with as little as 10% RFCM elevated preprosomatostatin mRNA levels more than IO-fold (Figure 4). Effects on the mRNA saturated at 20% RFCM (IS-fold increase), i.e., higher doses did not further increase the mRNA level. By contrast, the effect of 20% RFCM on peptide levels was less than half the maximal effect, which was achieved at 40% RFCM (3-fold increase). These very different dose-response curves suggested that different factors in RFCM may regulate preprosomatostatin mRNA and peptide metabolism.

Since fibroblasts and Schwann cells in culture can produce nerve growth factor (NGF; see Lindholm et al., 1987), which increases somatostatin levels in sympathetic neurons (Kessler, 1985), the effects of RFCM were examined in the presence of saturating concentrations of NGF, i.e., levels at which additional NGF had no further effect. Increasing the dose of NGF in the culture medium from 100 rig/ml to 1 @ml only slightly increased somatostatin levels (14% increase), and further increases in NGF had no effect (Table 1). RFCM treatment in the presence of 1 @ml NGF produced effects on somatostatin almost identical to those observed in the presence of the lower NGF concentration (Table 1); consequently the effects of RFCM cannot be due to NGF. Effects of Neuronal Contact with Nonneuronal Cell Membranes GNCM and RFCM stimulated levels of preprosomatostatin mRNA in sympathetic neurons regardless of whether the neurons were grown in isolation or in the presence of ganglion nonneuronal cells. CNCM and RFCM also stimulated peptide levels in pure neuronal cultures. However, conditioned medium treatment of neurons cocultured with ganglion nonneuronal cells did not alter somatostatin levels. These observations suggested that the nonneuronal cells inhibited stimulation of somatostatin, but not stimulation of preprosomatostatin mRNA. To test the hypothesis that direct contact of neurons with nonneuronal cell membranes inhibited stimulation of peptide levels, neurons were cultured on a substrate of nonneuronal cells, Schwann cells, or skin fibroblasts that had been killed by exposure to 10% trichloroacetic acid (Hawrot, 1980; Kessler et al., 1984). Culture of sympathetic neurons on a substrate of killed nonneuronal cells increased preprosomatostatin mRNA levels approximately 4-fold compared with neurons cultured on the usual collagen substratum (Figure 5). However, culture on the substrate of killed nonneuronal cells had no effect on somatostatin levels. Treatment of neurons cultured on killed nonneuronal cells with RFCM further increased levels of preprosomatostatin mRNA (II-fold increase), but peptide levels were again unchanged. By contrast, treatment with RFCM elevated levels of both preprosoma-

Regulation 307

of Somatostatin

Expression

A

A

ss

TH

1234

1234

TH

lB15

1234

B 3DT

0 m

ON COLLAGEN ON KILLED NON-NEURONAL CELLS

B 0 CONTROL tZ%YMANS

PEPTIDE Figure 5. Effects of Culture Nonneuronal Cells

1

1’”

mRNA on a Substratum

of Killed

Ganglion

Neurons were cultured either on the normal collagen substratum or on ganglion nonneuronal cells that had been killed by 10% trichloroacetic acid. Some of the cultures were treated with 30% RFCM. After 10 days of treatment, the cultures were assayed for content of somatostatin and preprosomatostatin mRNA. n = 3; 3 dishes per point, per experiment. (A) Northern blot hybridized to the somatostatin (SS) probe, rehybridized to the tyrosine hydroxylase (TH) probe, and then the cyclophilin (lB15) probe. Lane 1, control on collagen; lane 2, control on killed cells; lane 3, RFCM treatment of neurons on collagen; lane 4, RFCM treatment of neurons on killed cells. (B) Levels of somatostatin and preprosomatostatin mRNA. Peptide is expressed as mean femtograms per neuron + SEM. mRNA is expressed as mean arbitrary absorbance units f SEM. n = 3; 5 dishes per point, per experiment. * Differs from control neurons on collagen at P < 0.001 and from control neurons on killed cells at P < 0.025. **Differs from all other groups at P < 0.01. * **Differs from control neurons on collagen at P < 0.025. * ***Differs from control neurons on collagen at P < 0.001 and from control neurons on killed cells at P < 0.01.

tostatin mRNA (22-fold increase) and peptide (5-fold increase) in neurons cultured on collagen. Levels of tyrosine hydroxylase mRNA and activity were both decreased by RFCM treatment. Levels of mRNA for cyclophilin (lB15), a protein that appears to be constitutively expressed in most tissues (Danielson et al., 1988), were the same (+12%) for all groups, which suggests that equal amounts of mRNA were loaded onto each lane. These observations indicate that culture on the substrate of killed nonneuronal cells inhibited stimulation of somatostatin by RFCM. Virtually identical results were obtained after culture on a layer of purified killed ganglion Schwann cells (n = 4; 3 cultures per group per experiment). By contrast, culture on a layer of killed skin fibroblasts did not stimulate neuronal levels of preprosomatostatin mRNA, but did in-

mRNA

PEPTIDE

Figure

6. Effects

of Treatment

with

MANS

Factor

Pure neuronal cultures were treated with MANS factor for 7days and examined for content of somatostatin and preprosomatostatin mRNA. (A) Northern blot hybridized to the somatostatin (SS) probe and rehybridized to the tyrosine hydroxylase (TH) probe. Lane 1, control; lane 2, MANS factor treatment. (B) Levels of somatostatin and preprosomatostatin mRNA. Peptide is expressed as mean femtograms per neuron + SEM. mRNA is expressed as mean arbitrary absorbance units f SEM. n = 3; 3 dishes per point, per experiment. *Differs from control at P < 0.005 **Differs from control at P < 0.01.

hibit the RFCM.

Molecules

increase

in somatostatin

Mediating

after

Effects of Ceil-Cell

treatment

with

Contact

Previous studies have shown that contact of sympathetic neurons with nonneuronal cell membranes stimulates expression of cholinergic traits and of preprotachykinin mRNA and substance P (Hawrot, 1980; Kessler et al., 1986). The molecule that appears to mediate this interaction (membrane-associated neurotransmitter stimulating factor; MANS factor) has been partially purified (Wong and Kessler, 1987; Adler et al., 1989). To determine whether this membrane molecule mediated effects of cell-cell contact on somatostatin expression, cultured sympathetic neurons were treated with the factor. MANS factor treatment significantly elevated levels of both preprosomatostatin mRNA and somatostatin (Figure 6). Thus MANS factor may mediate the stimulatory effect of cell-cell contact on preprosomatostatin mRNA, but clearly does not cause the contact-mediated inhibition of increases in peptide.

Neuron 308

Discussion

Sympathetic neuron interactions with nonneuronal cells exerted complex effects on somatostatin metabolism. Coculture significantly elevated levels of preprosomatostatin mRNA without altering neuronal synthesis, content, or release of somatostatin. Furthermore, coculture prevented the increase in peptide levels, but not the increase in preprosomatostatin mRNA levels, after treatment with conditioned medium. Thus coculture exerted rate-limiting effects on neuronal somatostatin biosynthesis at a translational level. Some of the effects of coculture were mediated by diffusible factors released from the nonneuronal cells, and some resulted from cell-cell contact. The mechanisms underlying these two types of regulatory influences differed significantly. Diffusible

Factors

Released

by Nonneuronal

Cells

In pure neuronal cultures, treatment with nonneuronal cell conditioned medium elevated levels of somatostatin as well as its mRNA. The dose-response curve differed significantly for effects on somatostatin compared with effects on mRNA levels, which suggests that different factors regulate peptide and mRNA metabolism. The stimulatory effects on somatostatin contrasted with the effects on tyrosine hydroxylase mRNA and tyrosine hydroxylase enzyme activity, both of which were decreased by treatment. Furthermore, treatment did not alter neuronal survival, total protein, or levels of the constitutively expressed mRNA for cyclophilin (1815). Thus nonneuronal cells released diffusible factors that produced relatively specific changes in transmitter metabolism. Prior studies have shown that nonneuronal cell conditioned medium also stimulates cholinergic and substance P development. Effects on these different transmitter systems can be separated by chromatographic techniques (Nawaand Patterson, 1988, Sot. Neurosci., abstract); thus multiple regulatory molecules are apparently released by the nonneuronal cells. Membrane-Associated

Factors

Coculture of neurons with ganglion nonneuronal cells inhibited an increase in peptide levels despite elevated levels of preprosomatostatin mRNA. The inhibitoryeffect on peptide was clearly not mediated by a diffusible factor, since GNCM treatment actually increased peptide levels. This observation suggests that peptide levels are inhibited by contact of neurons with ganglion nonneuronal cells. Culture of neurons on a substratum of killed nonneuronal cells similarly inhibited stimulation of peptide levels by RFCM, despite a large increase in preprosomatostatin mRNA levels; these findings strongly support the conclusion that contact of neurons with nonneuronal cell membranes inhibits an increase in peptide synthesis. However, culture of neurons on the substratum of killed cells paradoxically stimulated levels of preprosomatostatin mRNA; thus cell-cell contact independently

regulates somatostatin metabolism at both the mRNA and the protein levels. Presumably these effects are mediated by at least two membrane molecules. Since treatment of sympathetic neurons with MANS factor, which mimicks the effects of nonneuronal cell membranes on cholinergic and substance P development (Wong and Kessler, 1987), increased preprosomatostatin mRNA, this molecule may mediate some or all of the stimulatory effects of cell-cell contact on preprosomatostatin mRNA. The molecule(s) mediating inhibitory effects on peptide synthesis is not yet known. Previous studies have indicated the importance of cell-cell contact in regulating neurotransmitter development. For example, culture of sympathetic neurons on a substrate of fixed heart cells results in an increase of cholinergic traits (Hawrot, 1980). Similarly, the development of mesencephalic dopaminergic neurons is stimulated by contact with striatal cell membranes (Prochiantz et al., 1981). Plasma membranes prepared from neurons and Schwann cells strikingly increase choline acetyltransferase activity and induce substance P expression in pure sympathetic neuron cultures (Kessler et ai., 1986). Furthermore, membranes isolated from adrenal chromaffin cells greatly increase tyrosine hydroxylase levels in the same cell type in vitro (Saadat and Thoenen, 1986). These observations suggest that neuronal interactions with cell surface and matrix molecules may play a decisive role in neuronal differentiation and transmitter expression. The present study indicates that somatostatin development may also be regulated by cell-cell contact. Contact-mediated stimulation of tyrosine hydroxylase levels and of substance P reflects increases in the respective species of mRNA (Saadat et al., 1987; Kremer et al., 1987, Sot. Neurosci., abstract). In this respect the stimulatory effects of cellcell contact on levels of preprosomatostatin mRNA are comparable. However, the paradoxical contactmediated inhibition of increases in peptide levels does not have a clear cut parallel in previous studies of transmitter expression. Levels

of Regulation

of Somatostatin

Metabolism

A number of lines of evidence suggest that neuronal interactions with nonneuronal cells regulate somatostatin metabolism by several independent mechanisms and at multiple metabolic levels. First, preprosomatostatin mRNA was stimulated by coculture without an increase in peptide synthesis. Second, there were different dose-response curves for the effects of RFCM treatment on peptide and mRNA levels. Finally, coculture further increased preprosomatostatin mRNA levels despite saturating levels of RFCM, an observation suggesting that at least two different molecules which stimulated preprosomatostatin mRNA are produced by nonneuronal cells. The effects on mRNA levels could reflect increases in either rate of transcription, stability of the mRNA, or both; experiments to distinguish among these possibilities are in progress. In any case, the increase in mRNA did not lead to an increase

Regulation 309

of Somatostatin

Expression

in peptide synthesis in cocultures because of the inhibitory effect of cell-cell contact. This indicates that, in some circumstances, transcriptional controls are not the rate-limiting factors for somatostatin biosynthesis.

Cell Type(s) Producing

the Regulatory

Molecules

Schwann cell conditioned medium stimulated levels of preprosomatostatin mRNA and somatostatin in pure neuronal cultures. Culture of neurons on killed Schwann cells also increased preprosomatostatin mRNA levels but inhibited an increase in peptide levels. Thus this single cell type produces the stimulatory and inhibitory molecules. By contrast, culture on a layer of killed fibroblasts did not stimulate neuronal levels of preprosomatostatin mRNA, although it did inhibit the increase in somatostatin after treatment with RFCM. Thus fibroblasts produce the inhibitory factor and the soluble stimulatory factor, but fail to produce the membrane-associated stimulatory factor. Since MANS factor appears to mediate the stimulatory effect of cell-cell contact on preprosomatostatin mRNA, these observations suggest that Schwann cells, but not fibroblasts, produce MANS factor.

General

Significance

of Observations

These findings indicate that posttranscriptional mechanisms may regulate peptide levels despite increases in levels of mRNA. These observations may help explain, in part, the often noted discrepancies between the distribution and levels of some peptides and their mRNAs in the nervous system. Thus neurons that express an mRNA may contain little of the appropriate peptide if the neurons contact other cells that exert inhibitory influences on peptide synthesis. There are numerous circumstances under which dual control of transmitter expression by nonneuronal cells at the translational as well as the transcriptional level might be physiologically important. First, the dual mechanism would allow the same nonneuronal cell to exert different regulatory influences on different neurons within the same structure. Thus, a neuron in contact with the nonneuronal cell would be limited with respect to somatostatin expression by the translational controls, whereas a neuron in proximity but not contact would be influenced only by the stimulatory effects of the diffusible factor. Such a mechanism might be particularly important in the event of injury, since proliferating nonneuronal cells would limit transmitter synthesis in the area of injury while stimulating a compensatory increase in surrounding neurons. Second, during development, transmitter expression could be initiated at the transcriptional level without expression of high levels of transmitter, which might exert inappropriate effects on surrounding cells. Finally, the translational controls might provide a cap to prevent synthesis of inappropriate levels of transmitter while still permitting transcriptional regulation below this level; stated otherwise, in the event of convergent stimulatory influences on somato-

statin transcription, the translational controls would provide a ceiling for the maximal rate of transmitter synthesis, while the stimulatory effect on transcription would help maintain a floor for the minimal rate of the synthesis. In summary, the present study indicates that complex interactions between sympathetic neurons and ganglion nonneuronal cells influence somatostatin development. Effects are mediated by several distinct mechanisms, and transcriptional and posttranscriptional controls may be exerted independently.

Experimental

Procedures

Experimental Animals Pregnant Sprague-Dawley

rats (Taconic) were housed in clear plastic cages and were exposed to 650-800 Iux of cool-white fluorescent illumination from 5 a.m. to 7 p.m. daily. Ralston Purina lab chow and water were offered ad libitum. Neonates were routinely used within 24 hr after delivery. Tissue Culture

Techniques

The neonatal superior cervical ganglion was dissociated and grown as described (Kessler, 1984) in a medium consisting of Ham’s nutrient mixture F12 (GIBCO) with 10% fetal calf serum (Hyclone, Logan, UT), NGF fi (100 @ml), penicillin (50 U/ml; CIBCO), and streptomycin (50 pg/ml; GIBCO). Cultures were maintained at 37OC in a 95% air, 5% CO2 atmosphere at nearly 100% relative humidity. Cultures were fed 3 times per week in all experiments; treatments were begun on day 2 with the first feeding and continued for a total of 10 days before harvesting unless otherwise noted. Ganglion nonneuronal cells were eliminated by treatment on days 1 and 3 of culture with cytosine arabinofuranoside (15 PM). Virtually.100% of the remaining cells (8,000-11,000 per 35 mm dish) bound tetanus toxin and had the morphology of neurons (Kessler, 1985). In the case of coculture, cytosine arabinofuranoside treatment was omitted and the surviving cells consisted primarily of neurons and Schwann cells, with a small number (I%-5%) of fibroblasts, as determined by morphologic criteria and anti-RAN-l immunocytochemistry (Fields et al., 1978). Cell numbers were determined by counting as described (Kessler, 1985). In the experiments examining peptide release, bacitracin (0.2 mg/ml) and captopril (0.5 mM) were added 24 hr prior to harvesting. In the [35S]cysteine labeling experiments, neurons were grown in 16 mm wells instead of culture dishes.

Preparation of Crude of the MANS Factor

Membranes

and Partial

Purification

The procedures have been described previously (Wong and Kessler, 1987). Briefly, entire spinal cords of adult female rats were homogenized with a polytron, nuclei and cell debris were removed by low speed centrifugation, and the P2 membrane fraction was isolated by centrifugation at 100,000 x g for 1 hr. The resulting pellet was washed, and MANS factor was extracted by incubating the pellet in 4 M NaCI. To purify the factor partially, the membrane extract was filtered twice through a Sephadex G-75 ultrafine column, and the resulting active fraction was identified by testing for stimulation of choline acetyltransferase activity in superior cervical ganglion neurons and by SDS-PAGE (10% polyacrylamide). Preparation of CNCM and the Fixed Substratum of Ganglion Nonneuronal Cells The neonatal superior cervical ganglion was dissociated and grown as described above except that NGF was omitted from the medium and cultures were plated more densely (2-3 ganglia per 35 mm dish). The cultures were fed several times with ice-cold medium to ensure that no neurons survived. After IO-14 days, the cultures reached confluence. Therewere no detectable neu-

NWJr0tl 310

rons in these cultures, which consisted largely of Schwann cells with some fibroblasts, as determined by the above criteria. Medium collected from confluent cultures (CNCM) was filtered (0.45 PM; Millipore) and stored at -20°C prior to use. Some of these cultures were killed and lightly fixed with 10% trichloroacetic acid, as described previously (Hawrot, 1980; Kessler et al., 1984), for use as a substratum for neuronal cultures. Cultures of superior cervical ganglia on this substratum were treated with cytosinearabinofuranoside and handled identically to cultures plated onto collagen. There was no significant difference in neuron numbers in these cultures compared with cultures using the collagen substratum. Preparation of Schwann Cells and Schwann Cell Conditioned Medium Schwann cells were cultured from neonatal superior cervical ganglia or from neonatal rat sciatic nerve by the method of Brockes et al. (1979), except that the cells were dissociated and plated in Ham’s nutrient mixture F12 with 10% fetal calf serum and ganglion Schwann cell proliferation was initially stimulated by the method of Porter et al. (1986) using forskolin. Cells were allowed to grow to confluence, and medium removed from confluent cultures was filtered (0.45 pm; Millipore) and stored at -20°C prior to use. The fixed substratum of Schwann cells was prepared by the method described above for preparing the fixed substratum of ganglion nonneuronal cells. Preparation of RFCM Skin samples from embryonic rats (E17) were treated with trypsin, mechanically dissociated, and plated on tissue culture plastic in Eagle’s minimum essential medium containing 15% fetal bovine serum. Fibroblasts were allowed to grow to confluence in this medium, which minimized the survival of other cell types; at the first passage, the feeding medium was switched to F12 FCSIO so that it could be used to treat neuronal cultures. Medium collected from confluent cultures (RFCM) was filtered (045 pm; Millipore) and stored at -2OOC prior to use. Extraction of Total Cellular RNA Cultured neurons were scraped in ice-cold Puck’s saline C using a rubber policeman. The cells were centrifuged at 3000 x g for 5 min, and the cell pellets were homogenized in glass-glass homogenizers in 0.4 ml of lysis solution (IO mM Tris-HCI, 1 mM EDTA, 350 mM NaCI, 2% SDS, 7 M urea [pH 8.01). Homogenized samples were then extracted twice with an equal volume of phenol-chloroform-isoamyl alcohol (50/48/2, v/v/v) and once with 1 voi of chloroform-isoamyl alcohol (96/4, v/v). Total nucleic acid was precipitated with 0.3 M sodium acetate and 3 vol of ethanol and dissolved in H20. Northern Gel and Blot Techniques Denatured RNA samples (2-20 pg of total nucleic acid extracted from a single culture dish) were run on a 1% agarose gel containing 6% formaldehyde. RNA was transferred to Gene Screen Plus (New England Nuclear) by overnight capillary blotting in 10x SSC (lx SSC = 0.15 M NaCI, 0.015 M sodium citrate). The Northern blot was then baked in vacua at 80°C for 2 hr and stored under vacuum until needed. 32P-labeled RNA probes synthesized with SP6 RNA polymerase were used exclusively for hybridization detection of preprosomatostatin and tyrosine hydroxylase mRNAs. A rat somatostatin cDNA clone was generously provided by Dr. Richard Goodman (Goodman et al., 1983). The Xbal-Sau3A fragment encompassing the coding region was isolated and subcloned into the pGem2 vector (Promega) for preparation of RNA probes. A partial cDNA clone for rat tyrosine hydroxylase in the pSP65vector was a generous gift of Drs. Elaine Lewis and Dona Chikaraishi (Lewis et al., 1983). The cyclophilin (lB15) probe was generously provided by Dr. James Douglas (Danielson et al., 1988). Northern blots were prehybridized for 6-24 hr at 65°C in 50% formamide, 1 M NaCI, 1% SDS, 10% dextran sulfate. Blots were hybridized at 65°C for 6-24 hr in prehybridization solution to which 0.1 mg/ml denatured calf thymus DNA and 5 x IO6 dpm

of the probe bad been added. Blots were then extensively washed at 65OC and exposed to Kodak XAR-5 film at -9OOC. Autoradiographic signals were quantitated by densitometric analysis using a Hoefer Model GS 300 Transmittance/Reflectance Scanning Densitometer interfaced with a Hewlett-Packard 3392A Integrator. Biochemical Measurements For determinations of somatostatin content, neurons were scraped in ice-cold 2 M acetic acid using a rubber policeman. The samples were heated to 100°C for 5 min and centrifuged, and the supernatant was lyophilized for later analysis. Recovery of exogenous labeled somatostatin was 94%. Somatostatin was measured by radioimmunoassay as described previously (Arnold et al., 1982). Initial experiments were performed with an antibody that recognizes both somatostatin 28 and somatostatin 14, but recognizes prosomatostatin poorly. All results were then replicated with an antibody that recognizes the precursor as well as both somatostatin 14and somatostatin 28. Nodifferences were noted in results with the two antibodies. The antibodies cross-reacted less than 0.01% with any other peptide tested. Tyrosine hydroxylase was measured using a radioenzymatic assay as described previously(Kessler and Black, 1979). Protein was measured by the method of Lowry (Lowry et al., 1951). labeling with [%]Cysteine Sympathetic neurons were cultured in 16 mm wells (10,500 neurons per well; range 10,086-11,145) either with or without nonneuronal cells. After 2 weeks in vitro, the cultures were washed with cysteine-free F12 containing 10% dialyzed fetal calf serum and incubated in the cysteine-free medium containing 250 PCilml [a5S]cysteine (1100 Cilmmol). After varying periods of time, cultures were washed and harvested by scraping with a rubber policeman into 100 ul of 2 M acetic acid. The acid extract was heated to 100°C for 5 min and centrifuged, and the supernatant was lyophilized in preparation for immunoprecipitation. In some wells, the tracer was removed after a 3 hr incubation (pulse), the cells were washed 3 times with medium, and the medium was replaced with F12 FCSlo containing excess (40 pg/ml) cysteine (chase medium). At varying times thereafter, the cultures were harvested as described above. Immunoprecipitations and HPLC analyses were performed using the protocols of Staller and Shields (1988). Affinity-purified rabbit anti-somatostatin antiserum (10 ul) was used to immunoprecipitate each sample with constant mixing at 4OC for 24 hr. After isolation of somatostatin immunoreactive peptide from the protein A-Sepharose beads, excess cold somatostatin (5 ug) was added to each sample prior to HPLC analysis on an Altex Cl8 reverse phase column. The procedures used were identical to those of Staller and Shields (1988; gradient system 2), except that a Beckman model 332 HPLC system was used. The retention times for somatostatin and prosomatostatin were 11.5 min and 20.5 min, respectively. Recovery of labeled somatostatin added to some cultures at the time of harvesting was 71%. Acknowledgments We wish to thank Ms. Mary]ane Dougherty for excellent technical assistance and Mrs. Antoinette Barnecott for help in preparing this manuscript. This work was supported by National Institutes of Health grants NS20013 and NS20778 and by the Dysautonomia Foundation. Received

September

18, 1989;

revised

November

IO, 1989.

References Adler, ]., Schleifer, L., and Black, I. (1989). Partial purification and characterization of a membrane-derived factor regulating neurotransmitter phenotypic expression. Proc. Natl. Acad. Sci. USA 86, 1080-1083. Arnold, mann,

M. A., Reppert, S. M., Rorstad, H. T., Perlow, M. J., and Martin,

0. P., Sagar, S. M., KeutJ. 8. (1982). Temporal pat-

Regulation 311

of Somatostatin

Expression

terns of somatostatin immunoreactivity in the cerebrospinal fluid of the rhesus monkey: effect of environmental lighting. Neurosci. 2, 674-680. Black, I. B. (1982). tonomic neurons.

Stages of neurotransmitter Science 275, 1198-1204.

development

J.

in au-

Brockes, J. P., Fields, K. P, and Raff, M. C. (1979). Studies on cultured rat Schwann cells. I. Establishment of purified populations from cultures of peripheral nerve. Brain Res. 765, 105-108. Bunge, R. P., Johnson, M., and Ross, C. D. (1978). Nature and nurture in the development of the autonomic neuron. Science 799, 1409-1416. Danielson, P. E., Forss-Peter, S., Brow, Douglass, J., Mimer, R. J., and Sutcliffe, cDNA clone of the rat mRNA encoding 261-267.

M. A., Calavetta, L., J. C. (1988). plB15: a cyclophilin. DNA 7

Fields, K. L., Brockes, J. F’., Mirsky, R., and Wendon, L. M. B. (1978). Cell surface markers for distinguishing types of rat dorsal root ganglion cells in culture. Cell 74, 43-51. Goodman, R. M., preprosomatostatin.

Aron, D. C., and Roos, B. A. J. Biol. Chem. 258, 5571-5573.

Hawrot, E. (1980). Cultured sympathetic neurons: derived and synthetic substrata on survival and Dev. Biol. 74, 136-151.

(1983).

Rat

effects of celldevelopment.

Hokfelt, T., Elfvin, L. G., Elde, R., Schultzberg, M., Goldstein, M., and Luft, R. (1977a). Occurrence of somatostatin-like immunoreactivity in some peripheral sympathetic noradrenergic neurons. Proc. Natl. Acad. Sci. USA 74, 3587-3591. Hokfelt, T, Elfvin, L. G., Schultzberg, M., Fuxe, K., Said, I., Mutt, V., and Goldstein, M. (1977b). lmmunohistochemical evidence of vasoactive intestinal polypeptide-containing neurons and nerve fibers in sympathetic ganglia. Neuroscience 2, 885-896. Kessler, J. A. (1984). Non-neuronal cell conditioned medium stimulates peptidergic expression in sympathetic and sensory neurons in vitro. Dev. Biol. 706, 61-69. Kessler, J. A. (1985). Differential regulation of peptide and cate cholamine characters in cultured sympathetic neurons. Neuroscience 75, 827-839. Kessler, J. A., and Black, I. B. (1979). The role of axonal transport in the regulation of enzyme activity in sympathetic ganglia of adult rats. Brain Res. 777, 415-424. Kessler, J. A., Adler, J. E., and Black, somatostatin regulate sympathetic ence 227, 1059-1061.

I. B. (1983). noradrenergic

Substance function.

P and Sci-

Kessler, J. A., Adler, J., Jonakait, G. M., and Black, I. B. (1984). Target organ regulation of substance P in sympathetic neurons in culture. Dev. Biol. 702, 417-425. Kessler, J. A., Conn, G., and Hatcher, membranes regulate neurotransmitter effects of a soluble brain cholinergic USA 83, 8726-8729.

V B. (1986). Isolated plasma expression and facilitate factor. Proc. Natl. Acad. Sci.

Lewis, E. J.,Tank, A. W., Weiner, N., and Chikaraishi, D. M. (1983). Regulation of tyrosine hydroxylase mRNA byglucocorticoid and cyclic AMP in a rat pheochromocytoma cell line. Isolation of a cDNA clone for tyrosine hydroxylase mRNA. J. Biol. Chem. 258, 14632-14637. Lindholm, D., Heumann, R., Meyer, M., and Thoenen, H. (1987). Interleukin-1 regulates synthesis of nerve growth factor in nonneuronal cells of rat sciatic nerve. Nature 330, 658-659, 1987. Lowry, 0. H., Rosebrough, (1951). Protein measurement Biol. Chem. 793, 265-275.

N. J., Farr, A. L., and Randall, R. J. with the folin phenol reagent. J.

Mudge, A. W. (1981). Effect of the chemical environment on levels of substance P and somatostatin in cultured neurones. Nature 292, 764-767. Patterson, P. H. (1978). Environment neurotransmitter function. Annu. Patterson, neuronal

determination Rev. Neurosci.

of autonomic 7, l-l%

P. H., and Chun, L. L. Y. (1974). The influence of noncells on catecholamine and acetylcholine synthesis

and accumulation rons. Proc. Natl.

in cultures of dissociated sympathetic Acad. Sci. USA 77, 3607-3610.

Patterson, P. H., and Chun, L. Y. (1977). The induction choline synthesis in primary cultures of dissociated pathetic neurons. I. Effects of conditioned medium. 56, 263-280.

neu-

of acetylrat symDev. Biol.

Porter, S., Clark, M. B., Glaser, L., and Bunge, R. D. (1986). Schwann cells stimulated to proliferate in the absence of neurons retained full functional capacity. J. Neurosci. 6, 3070-3078. Prochiantz, A., Dagnet, M.-C., (1981). Specific stimulation of cephalic dopaminergic neurones 293, 570-572.

Herbert,

A., and Glowinski, J. maturation of mesenby striatal membranes. Nature

in vitro

Raynaud, B., Faucon-Biguet, N., Vidal, S., Mallet, J., and Weber, M. J. (1987). The use of a tyrosine-hydroxylase cDNA probe to study the neurotransmitter plasticity of rat sympathetic neurons in culture. Dev. Biol. 779, 305-312. Saadat, S., and Thoenen, H. (1986). Selective induction of tyrosine hydroxylase by cell-cell contact in bovine adrenal chromaffin cells is mimicked by plasma membranes. J. Cell Biol. 703, 1991-1997. Saadat, S., Stechle, A., Lumouroux, A., Mallet, J., and Thoenen, H. (1987). Influence of cell-cell contact on levels of tyrosine hydroxylase in cultured bovine adrenal chromaffin cells. J. Biol. Chem. 262, 13007-13014. Smith, K., and Kessler, J. (1988). Non-neuronal cell-conditioned medium regulates muscarinic receptor expression in cultured sympathetic neurons. J. Neurosci. 8, 2406-2413. Staller, T J., and Shields, D. (1988). Retrovirus-mediated expression of preprosomatostatin: post-translational processing, intracellular storage, and secretion in GH3 pituitary cells. 1. Cell Biol. 707, 2087-2095. Weber, M. J. (1981). A diffusible factor responsible for the determination of cholinergic functions in cultured sympathetic neurons. J. Biol. Chem. 256, 3347-3453. Wong, V., and Kessler, J. A. (1987). Solubilization of a membrane factor that stimulates levels of substance P and choline acetyltransferase in sympathetic neurons. Proc. Natl. Acad. Sci. USA 84, 8726-8729.