Multiple cholinergic differentiation factors are present in footpad extracts: comparison with known cholinergic factors

While the sympathetic innervation of adult rat sweat glands is cholinergic, the axons that initially innervate the sweat gland are noradrenergic (Landis and Keefe, 1983; Leblanc and Landis, 1986; Stevens and Landis, 1987). Thus, there is a switch in the neurotransmitter phenotype of these sympathetic neurons. Cross-innervation studies in vivo indicate that this switch is target-directed. Neurons that normally innervate sweat glands fail to acquire cholinergic proper-ties when they are made to innervate the parotid gland, a target whose sympathetic innervation is ordinarily nora-drenergic (Schotzinger and Landis, 1990). Furthermore, normally noradrenergic sympathetic neurons that innervate hairy skin become cholinergic when they are made to innervate sweat glands (Schotzinger and Landis, 1988). Acqui-sition of immunoreactivity for vasoactive intestinal peptide (VIP), a neuropeptide present in mature sweat gland innervation, is also target-dependent (Landis et al., 1988; Schotzinger and Landis, 1988; Stevens and Landis, 1990 and unpublished observations). Although the evidence for target control of transmitter phenotype is compelling, the identity of the molecules involved is not known.

We and others previously found that extracts of rat foot-pads contain factor(s) that induce choline acetyltransferase (ChAT) and VIP, and reduce catecholamine and neuropep-tide Y (NPY) content in cultured sympathetic neurons (Rao and Landis, 1990; Rohrer, 1992). This activity was not detected in extracts of parotid gland or hairy skin, tissues that receive noradrenergic sympathetic innervation. The cholinergic differentiation activity is evident in developing footpads as early as postnatal day 5 (P5) and increases over a period of two weeks to reach adult levels. Therefore, it is present at the appropriate time in development to be responsible for the target-mediated noradrenergic-to-cholin-ergic switch.

Several purified proteins elicit the same noradrenergic-to-cholinergic switch in cultured sympathetic neurons and are therefore candidates for the differentiation signal in sweat glands. The cholinergic differentiation factor (CDF), originally purified from heart-cell-conditioned medium (Patterson and Chun, 1977; Fukada, 1985), is identical to leukemia inhibitory factor (LIF; Yamamori et al., 1989). Recombinant CDF/LIF also influences neuropeptide expression in cultured sympathetic neurons in a manner very similar to that of the footpad factor (Rao and Landis, 1990; Nawa et al., 1990). Studies using the reverse transcriptase/polymerase chain reaction (RT-PCR) revealed that CDF/LIF mRNA is selectively localized in postnatal rat footpads (Yamamori, 1991). Another candidate for the sweat gland cholinergic factor is ciliary neurotrophic factor (CNTF; Lin et al., 1989; Stockli et al., 1989). This protein regulates cholinergic, noradrenergic and neuropeptide properties in cultured sympathetic neurons in the same fashion as CDF/LIF (Ernsberger et al., 1989; Saadat et al., 1989; Rao et al., 1992a). A third cholinergic factor, MANS (mem-brane associated neurotransmitter stimulating factor), was partially purified from rat spinal cord (Wong and Kessler, 1987; Adler et al., 1989). MANS appears to be related to CNTF since it supports the survival of chick ciliary neurons (Rao et al., 1990) and antibodies to CNTF immuno-precipitate the cholinergic inducing activity of MANS and detect a band of 24×10³, the relative molecular mass of CNTF (Carnow et al., 1985; Manthorpe et al., 1986), on immunoblots of MANS preparations (Rao et al., 1992b). Finally, a cholinergic differentiation factor, which is bio-chemically distinct from CDF/LIF and CNTF, has been purified from skeletal muscle. It enhances the survival of motor neurons in vitro and in vivo and increases levels of ChAT in spinal cord cultures (McManaman et al., 1988, 1989, 1990). The ability of the skeletal muscle cholinergic differentiation factor to induce cholinergic properties in sympathetic neurons has not been tested. Since CDF/LIF and CNTF, two of the factors that induce cholinergic function in sympathetic neurons, affect motor neurons (Geiss and Weber, 1984; Martinou et al., 1989; 1992; Sendtner et al., 1990; Arakawa et al., 1990), it seemed plausible that the skeletal muscle cholinergic factor would also induce cholinergic function in sympathetic neurons.

To determine if any of the known cholinergic factors is present in sweat gland, we further characterized the cholinergic inducing activity from footpad extracts and compared its properties with CDF/LIF, CNTF and the cholinergic factor from skeletal muscle. We found that most of the differentiation activity present in footpads of young rats is associated with a molecule(s) that is immunoprecipitated by antibodies that recognize CNTF. We did not, however, detect either CNTF mRNA or protein in footpads in immunoblot, in northern blot or in situ hybridization analyses, suggesting that the major differentiation factor present in the extracts and recognized by CNTF antibodies is similar, but not identical, to CNTF. The second differentiation activity is as yet uncharacterized.

Cell culture reagents were obtained from Gibco (Grand Island, NY). 24- and 96-well culture plates were purchased from Corn-ing (Corning, NY). Affi-gel 10 was obtained from Bio-rad (Rich-mond, CA.). The Centricon filters were purchased from Amicon (Danvers, MA). [³H]acetyl-CoA was purchased from New Eng-land Nuclear (Wilmington, DE). T4 polynucleotide kinase and 3′ end terminal transferase were obtained from (BRL), gamma ³²P from Amersham (Arlington Heights, IL) and ³⁵S-dATP from NEN. The VIP RIA kit was obtained from INCSTAR (Min-neapolis, MN) and the NPY RIA kit and rat NPY were obtained from Peninsula (Belmont, CA). Dispase was purchased from Boehringer-Mannheim (Indianapolis, IN) and collagenase from Worthington Biochemicals (Freehold, NJ). Genescreen mem-branes were obtained from NEN. Vectastain ABC immunoperox-idase kits were obtained from Vector Laboratories (Burlingame, CA) and diaminobenzidine from Polysciences (Warrington, PA). The rabbit antiserum against CGRP and the avidin-Texas red con-jugates were obtained from Amersham. Other chemicals were purchased from Sigma Chemical Co. (St Louis, MO). Nerve Growth Factor (NGF) was provided by Dr Ken Neet (Department of Bio-chemistry, Chicago Medical School). An antiserum that blocks the activity of CDF/LIF was obtained from R & D Systems (St Paul, MN) and an antiserum generated against recombinant rat CNTF was the kind gift of Dr Morrissey at Regeneron Pharmaceuticals (Tarrytown, NY). Animals were obtained from Zivic Miller (Zelienople, PA).

Recombinant rat CNTF, produced in E. coli, was obtained from Regeneron Pharmaceuticals (Tarrytown, NY). Recombinant human LIF was obtained from AMGEN (Thousand Oaks, CA). CDF/LIF purified from heart-cell-conditioned medium was pro-vided by Dr Keiko Fukada (SUNY Downstate, New York, NY; Fukada et al., 1991). The cholinergic differentiation factor from skeletal muscle was obtained from Dr J. McManaman (Baylor College of Medicine, Houston, TX); the HAP7 fraction was used for all experiments (McManaman et al., 1989). Footpad extracts were prepared as described previously with the exception that the whole footpad was used rather than the gland-rich core (Rao and Landis, 1990). Unless otherwise stated, the DEAE fraction of extracts from 21-day-old rats was used. Sciatic nerve extracts were prepared from adult rats as described previously (Rao et al., 1990). The 0.25 M NaCl eluate of the DEAE column was used for all the experiments.

The soluble fractions of homogenates of footpads of 21-day-old rats were concentrated (tenfold) with a centriprep microconcen-trator (Amicon), and the final sample buffer contained 10 mM phosphate, pH 7.0. 10 mg protein was chromatographed on a MonoQ column (Pharmacia) using a Whatman HPLC. Fractions of 0.5 ml were collected while eluting with a linear gradient to 0.1 M NaCl (Fig. 1), and assayed for cholinergic activity on cultured sympathetic neurons. Active fractions from several runs of the MonoQ column were pooled, concentrated and chro-matographed on a MonoP HR 5/5 column (Pharmacia) that had been equilibrated with 25 mM bis-Tris, pH 7.1. Elution was with a polybuffer at pH 4.0 containing saturating levels of iminodi-acetic acid. The flow-through and eluted fractions were adjusted to neutral pH and stored at −70°C. Control experiments showed that the polybuffer did not affect the growth or differentiation of neuronal cultures. Active fractions (pH 4.9 –5.2) were pooled, con-centrated and chromatographed on a Sepharose 12 column (Phar-macia) equilibrated in 0.1 M NaCl, 10 mM phosphate, pH 7.0. Fractions of 0.5 ml were collected and assayed for biological activity. Protein standards were chromatographed under identical conditions to determine the approximate sizes of proteins collected in the various fractions.

The active fractions from the MonoP column were pooled, con-centrated ten fold and subjected to preparative polyacrylamide gel electrophoresis in sodium dodecyl sulfate (SDS-PAGE). A 15% acrylamide gel was used in a Biorad apparatus. Regions of the gel of appropriate relative molecular mass were cut out and loaded on a Biorad electroeluter. Elution was at 10 mA/tube for 4 hours in a buffer containing 50 mM ammonium bicarbonate and 0.1% SDS. Samples were concentrated and SDS precipitated by adding KCl to a final concentration of 0.1 M, as described by Suzuki and Terada (1988). Bovine serum albumin (1%; BSA) was added to the samples, which were then dialyzed on a Centricon apparatus (10×10³M_r size cutoff) to remove excess KCl. Samples were stored at −70°C until used.

Cultures of rat sympathetic neurons were prepared as described by Hawrot and Patterson (1979). Neurons were dissociated enzy-matically with Dispase (5 mg/ml) and collagenase (1 mg/ml) and plated in 96-well polylysine-laminin-coated plates. 1000 –2000 neurons were plated per well. Neurons were grown in L15-CO₂ medium with 7S NGF (100 ng/ml), 10 μM cytosine arabinoside and 5% adult rat serum, and the medium changed every second day. The cholinergic factors and column fractions were diluted in growth medium, sterilized by passage through a 0.2 μm filter and added to the neurons on the third day of culture. Neurons were harvested for assay between days 9 and 14 of culture.

Ciliary ganglia were dissected from embryonic day 8 (E8) chicks, dissociated and plated in DMEM with 10% fetal calf serum as described by Varon et al. (1979). 1000-2000 neurons were plated directly into medium in 96-well culture plates coated sequentially with polylysine and laminin. The cultures were incu-bated for 24 hours at 37°C, and fixed with 2% glutaraldehyde as described by Barbin et al. (1984). The number of surviving neu-rons was determined by counting phase bright cells with neurites.

The induction of cholinergic function was determined by assay-ing ChAT activity in homogenates essentially according to the method of Fonnum (1969). To increase the sensitivity of the assay, an incubation period of an hour was used. All activity was inhibitable by 500 μM naphthylvinyl pyridine, a specific inhibitor of ChAT activity (Smith et al., 1967; White and Wu, 1973).

Tyrosine hydroxylase (TH) content was assayed by immunoblot. Aliquots of cultured neuron extracts (10⁴ neurons) were subjected to SDS-PAGE on a 15% minigel and the proteins transferred to nitrocellulose overnight at 30 mA. The blots were blocked (5% defatted milk in Tris-buffered saline, pH 7.2) and then incubated for two hours with a monoclonal antibody that rec-ognizes tyrosine hydroxylase (TH; 1:1000 dilution, the kind gift of Dr Ann Acheson, Edmonton, Alberta, Canada; Rohrer et al., 1986). The blots were stained and developed as described for the CNTF immunoblots.

Neuropeptide levels were determined by radioimmunoassay. Neuronal cultures (10⁴ neurons) were rinsed once with PBS and homogenized in 100 μl 2 M acetic acid. After boiling for five minutes, samples were centrifuged for 1 minute in an Eppendorf microfuge. Supernatants were dried under vacuum and stored at −70°C for subsequent assay. VIP was assayed using primary anti-bodies previously demonstrated to show minimal crossreactivity with other peptides. NPY was assayed by the delayed tracer method as specified in the kit using antibodies, standards and labeled tracer from Amersham. The standard curve was produced with rat NPY.

For the immunoprecipitation experiments in which the biological activity of the factors was tested, aliquots of CDF/LIF and extracts of sciatic nerve from adult rats or footpads from 21-day-old animals sufficient for a cell culture assay were added to buffer (phosphate-buffered saline [PBS], pH 7.3 with 2% BSA, 0.2% Triton X-100, and 0.02% polyethylene glycol 6000). A polyclonal anti-serum generated against recombinant rat CNTF (obtained from Dr Donna Morissey, Regeneron Pharmaceuticals) was added to each vial to a final concentration of 10 μM. After an overnight incubation, the antigen-antibody complex was absorbed to 10 μl of protein A Sepharose (in the same buffer) for a further two hours at room temperature. The bound complexes were separated by centrifugation and the supernatant was diluted into L15-CO₂ medium and used for cell culture assays. To ensure that the loss of activity was due to a specific effect of the antibody, aliquots of the cholinergic factors were incubated without the antibody and treated as described above.

A polyclonal antiserum generated in rabbit against CDF/LIF puri-fied from heart-cell-conditioned medium (the kind gift of Dr Keiko Fukada, SUNY Downstate, New York, NY; Fukada and Towle, 1992) was used for antibody blocking experiments. Sympathetic neurons were grown in Optimem defined medium with 100 ng/ml NGF. On the second day after plating, extract from postnatal day 8 rat pups or CDF/LIF purified from heart cell conditioned medium (20 ng/ml) were added to the cultures. Antiserum was added to a final dilution of 1:15 which had been previously deter-mined to block ChAT induction in response to treatment with purified CDF/LIF. Cultures were fed every fourth day and harvested on the ninth day after initiating antibody treatment. Samples were assayed for ChAT activity as described above. Two controls were performed to examine the specificity of the antiserum. Cultures were grown in the same dilution of antiserum but in the absence of CDF/LIF or in the presence of antiserum and sciatic nerve extract (25 μg/ml).

Aliquots of extracts were run on a 15% SDS-PAGE minigel (Biorad) and the proteins were blotted onto nitrocellulose (overnight, 30 mA). The blots were blocked (5% defatted milk in Tris-buffered saline, pH 7.2) and then incubated for two hours with anti-CNTF antiserum (1:1000 dilution) or with the antibody preincubated with 10 μM CNTF. The blots were then sequentially incubated with a biotinylated secondary antibody (1 hour at room temperature) and avidin conjugated to alkaline phosphatase (0.5 hour). The bound enzyme was detected with NBT (nitroblue tetra-zolium) and BCIP (bromo-chloro-indoyl phosphate) in 10 mM bicarbonate buffer, pH 9.5. After color development, the reaction was stopped by rinsing in distilled water.

Total RNA was prepared from liver, foot pad and sciatic nerve using the single step guanidinium-isothiocyanate method (Chom-czynski and Sacchi, 1987). The RNA was denatured on a 1.5% agarose gel containing 12.2 M formaldehyde and blotted onto Genescreen Plus membranes. Blots were probed with a 45 base pair antisense oligonucleotide probe for rat CNTF (bases 99-144). The probe was labeled at the 5′ end using T4 polynucleotide kinase and gamma ³²P. Hybridization was performed at 42°C for 48 hours in buffer consisting of 1 M NaCl, 30% deionized formamide, and 10% dextran sulfate. Blots were sequentially washed at 50 °C in 0.1×SSC and exposed to Kodak XAR-5 film for 3 –5 days.

The probe used in the northern blot experiments was also used for in situ hybridization after labeling at the 3′ end using termi-nal transferase and ³⁵S-dATP. The sense probe to the same region was used as a control. The in situ hybridization method was as described by Siegel (1989). Briefly, fresh-frozen tissue sections were fixed in 4% formaldehyde for 5 minutes, rinsed three times in PBS, incubated in acetic anhydride/triethanolamine for 10 min-utes, rinsed two times briefly in 2×SSC (0.3 M NaCl, 0.03 M sodium citrate, pH 7.3) and finally dehydrated through a series of graded alcohols. The slides were air-dried and processed for hybridization. After a one hour prehybridization incubation (4×SSC, 50% formamide, 500 μg/ml sheared single-stranded DNA, 250 μg/ml yeast tRNA, 1 ×Denharts solution [0.02% Ficoll, 0.02% polyvinylpyrrolidone, 0.02% bovine serum albumin], and 10% dextran sulfate), the blots were rinsed with 2×SSC and radioactive probe, diluted in hybridization buffer, was applied at a concentration of 10⁶ cts/minute/slide. Parafilm coverslips were placed over the sections and hybridization was carried out in humid chambers overnight at room temperature. After washing, the sections were dipped in Kodak NTB-3 emulsion and exposed for 6 weeks. Sections were developed, fixed and counterstained with ethidium bromide or cresyl violet.

The distribution of CNTF immunoreactivity was examined in foot-pads and in hairy skin. Rats were killed with ether and perfused through the heart with 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.4, for 10 minutes. The tissue was fixed for an additional hour by immersion in the paraformaldehyde solution. After rinsing, the tissue was equilibrated with 30% sucrose in PBS. 15 μm cryostat sections were collected and thaw mounted onto gelatin-coated slides. The sections were incubated overnight at room temperature with a primary antiserum generated in rabbit against a synthetic peptide from CNTF (amino acids 181-200; Rende et al., 1992; the kind gift of Dr Manthorpe) diluted 1:2000 in dilution buffer (0.5 M NaCl, 0.01 M phosphate buffer, 3% bovine serum albumin, 0.1% sodium azide and 0.3% Triton X-100). The sections were rinsed, incubated in biotinylated sec-ondary antibody for 2 hours, rinsed and then incubated with the Vectastain avidin-horseradish peroxidase (HRP) complex for 2 hours. After rinsing, the bound antibody was visualized with 3,3′ diaminobenezidine (0.5 mg/ml) in 0.1 M Tris buffer, pH 7.2, containing 0.02% H₂O₂. Optimal reaction product developed within 15 seconds to 2 minutes. To control for antiserum specificity, sections were incubated without primary antibody or with primary antiserum preabsorbed with 20 μg/ml of recombinant CNTF. In a second set of experiments, double labeling for CNTF and calcitonin gene-related peptide (CGRP), a neuropeptide present in sensory fibers in footpad epidermis (Kruger et al., 1989), was per-formed. Sections were incubated with the CNTF antiserum, rinsed, incubated with biotinylated secondary antiserum and then with avidin conjugated to Texas red. A second series of incubations was then carried out to detect CGRP. The sections were incubated with a CGRP antiserum raised in rabbit, rinsed, incubated in goat anti-rabbit conjugated to fluorescein, rinsed and mounted in PBS/glycerin. The sequential use of two rabbit antisera could yield cross-reactivity between the secondary antisera. Control sections were incubated with the goat anti-rabbit fluorescein after the avidin-conjugated Texas red without the second primary antiserum and no specific fluorescein staining was evident.

Footpads from postnatal day 21 rats were homogenized and the extract was applied to a Mono Q anionic exchange column. This age was selected because previous studies had shown that footpads from 21-day-old rats had the greatest amount of cholinergic inducing activity on a per protein basis (Rao and Landis, 1990). As shown in Fig. 1A, the cholinergic activity bound to the column and was eluted as a single peak at 0.25 mM NaCl. This chromatographic pro-file is similar to that of CNTF and the cholinergic factor from skeletal muscle (Manthorpe et al., 1986; McManaman et al., 1988, 1989), but is different from that of the basic protein, CDF/LIF (Fukada, 1985). Similar results were obtained with extracts from rats 28 days and older (data not shown). The Mono Q peak fraction was then chro-matographed on a Mono P column. The cholinergic induc-ing activity eluted between pH 4.8 and 5.2 (Fig. 1B). The pH of the peak fraction, 5.0, is similar to the value reported for CNTF purified from rat sciatic nerve (Manthorpe et al., 1986). In each case, cholinergic inducing activity was detected in a single peak.

To determine the approximate relative molecular masses of the cholinergic inducing activity, peak fractions from the Mono P column were chromatographed on a Sepharose 12 column. The activity eluted in a peak between the 16×10³M_r and 32 ×10³M_r size markers (Fig. 2). This relative mol-ecular mass is comparable to those of MANS (Wong and Kessler, 1987), CNTF (Manthorpe et al., 1986) and degly-cosylated CDF/LIF (Fukada, 1985). The relative molecular mass of the footpad cholinergic activity was determined more accurately by SDS-PAGE. Purified material from the Mono P column was subjected to SDS-PAGE. Slices of the gel corresponding to 22-26×10³ and 26-32×10³M_r were removed and the proteins eluted with an electroeluter. As shown in Fig. 3A, most of the cholinergic activity was recovered in the 22-26×10³M_r fraction.

The SDS-PAGE fractions were also tested for their effects on VIP, TH and NPY expression. As in the case of the cholinergic inducing activity, most of the TH- and NPY-suppressing and VIP-inducing activities were in the 22 – 26×10³M_r fraction (Fig. 3B-D).

Since the size and pI of the cholinergic differentiation activity in the footpad extract were very similar to those of CNTF, we examined the ability of SDS-PAGE-purified material to support the survival of E8 chick ciliary neurons in culture, the original assay for CNTF activity (Barbin et al., 1984; Manthorpe et al., 1986). Fig. 4 shows that the 22-26×10³M_r fraction purified from footpad extract contained a significant level of ciliary neurotrophic activity. Thus, this fraction contained not only cholinergic and peptidergic differentiating activity for cultured sympathetic neurons but also ciliary neurotrophic activity.

To determine whether the factor(s) present in the foot-pad extracts was antigenically related to CNTF, an anti-serum generated against recombinant rat CNTF was used to precipitate the cholinergic activity. Since sciatic nerve contains CNTF (Manthorpe et al., 1986; Lin et al., 1989; Stockli et al., 1989), extracts of sciatic nerve were used to optimize immunoprecipitation conditions. The CNTF anti-serum completely precipitated the cholinergic inducing activity from sciatic nerve extracts; the ChAT activity present in cultures treated with sciatic nerve extract from which the CNTF had been precipitated was equivalent to the activity present in control cultures (Fig. 5A). Further, previous studies have shown that the CNTF antiserum com-pletely immunoprecipitates the peptide-inducing activities present in sciatic nerve extract (Rao et al., 1992a). While the CNTF antiserum had no effect on the induction of ChAT by recombinant CDF/LIF, it precipitated approxi-mately 80% of the ChAT-inducing activity in the extract from footpads of 21-day-old rats (Fig. 5A). We also tested the ability of the CNTF antiserum to immunoprecipitate the activity in the footpad extract that induces VIP and sup-presses NPY levels (Fig. 5B,C). Approximately 50% of the VIP-inducing activity was precipitated, while only 20% of the NPY activity was removed. Thus, the CNTF antiserum cross-reacted with a factor(s) present in the footpad extract and immunoprecipitated the majority of the cholinergic inducing activity (see also Rohrer, 1992).

An immunoblot analysis of the extract was performed using the antiserum generated against recombinant rat CNTF to determine if the footpad extract contained authentic CNTF. Previous studies had demonstrated that extracts of sciatic nerve and footpads contain similar amounts of cholinergic inducing activity when compared on the basis of extracted protein (Rao and Landis, 1990). For these experiments, three different concentrations of footpad extract and sciatic nerve extract were tested to determine the amounts of each required to yield equivalent induction of ChAT activity: 10, 100 and 250 μg of sweat gland extract caused a 2.3-, 8.1- and 15.3-fold induction and 10, 100 and 250 μg of sciatic nerve extract caused a 3.5-, 12.6- and 23.5-fold change. Equal amounts of cholinergic inducing activity from extracts of sciatic nerve and footpad were sub-jected to SDS-PAGE, transferred to nitrocellulose and probed with the CNTF antiserum. This procedure yielded a strong signal for recombinant CNTF and a 24×10³M_r band in sciatic nerve (Fig. 6A,B). These bands were absent when the antiserum was preincubated with 10 μM recom-binant CNTF (Fig. 6C). No specific bands were evident in the lanes containing footpad extract which has cholinergic inducing activity (Fig. 6B, lanes 3,4) or hairy skin extract (lane 2) which lacks detectable cholinergic inducing activity (Rao and Landis, 1990). It is important to note that the lanes containing sciatic nerve and footpad extracts, when assayed on cultured sympathetic neurons, contained the same amount of cholinergic inducing activity. Further, we could still detect a band when ten-fold less sciatic nerve extract was loaded. These observations suggest that the major cholinergic activity present in the footpad extracts was not identical to CNTF.

We performed northern blot and in situ hybridization analyses to ascertain whether CNTF mRNA could be detected in footpad. When 30 μg of total RNA was probed, a band of 1.3 kb, the expected size of CNTF mRNA (Stockli et al., 1989), was detected in sciatic nerve and a faint band was detected in dorsal root ganglia (Fig. 7). In contrast, no specific binding was evident in lanes containing RNA from footpads or liver. When ten-fold less RNA from sciatic nerve was loaded than from footpad, CNTF message could still be detected in sciatic nerve (Fig. 7B). The same oligonuclotide probe was used for in situ hybridization studies of sciatic nerve and sweat gland (Fig. 8). Many Schwann cells in sciatic nerve exhibited a specific hybrid-ization signal with the antisense probe, while no such signal was apparent with sense control probe (data not shown). No signal was evident in sweat gland tissue with the antisense probe, over either the glands or the Schwann cells investing the sympathetic innervation of the sweat glands.

Since CNTF is known to be present in Schwann cells in sciatic nerve, and footpads from which the extracts were made contain branches of this nerve, we used a polyclonal antiserum generated against a peptide domain of CNTF (Rende et al., 1992) to localize CNTF-like immunoreactivity in footpad (Fig. 9). Sections of sciatic nerve, footpad and hairy skin were incubated with the polyclonal antiserum and specific staining was localized by the peroxidase-antiperoxidase method. As expected, strong staining was observed in a subset of Schwann cells in the sciatic nerve (data not shown; see Rende et al., 1992; Dobrea et al., 1992). In addition, CNTF-immunoreactivity was present in nerve fibers that course through the pad and under the epi-dermis. This distribution is similar to that of sensory fibers, including those that contain CGRP (Kruger et al., 1989). To determine whether the CNTF-immunoreactivity was, in fact, associated with sensory axons in footpads, double labeling with antisera against CGRP and CNTF was per-formed. In all cases, the staining patterns were overlapping. CNTF-immunoreactivity was also associated with nerve fibers associated with the epidermis and in the dermis of hairy skin (Fig. 9). No CNTF immunoreactivity was detectable in the sweat gland secretory coil or in the Schwann cells associated with the sympathetic innervation of the sweat glands. Thus, CNTF present in Schwann cells associated with sensory fibers could contribute to the CNTF-like activity present in extracts from footpads. Since cholinergic inducing activity is not present in extracts of hairy skin (Schotzinger and Landis, 1988; Rao and Landis, 1990) which, like footpads, contains nerve bundles with associated CNTF-immunoreactivity, it seems unlikely that the CNTF present in these nerve bundles is the source of the major fraction of the cholinergic inducing activity.

The size and pI values of the major cholinergic inducing activity in the footpad extract indicated that its biochemical properties were distinct from glycosylated CDF/LIF. It was possible, however, that the footpad factor is a post-translationally modified form of CDF/LIF that has biochemical and immunological properties distinct from those of the protein originally purified from heart-cell-conditioned medium. In addition, the finding that the anti-CNTF antibodies only partially immunoprecipitated the cholinergic inducing activity in footpad extracts suggested that there is more than one cholinergic factor in the footpad extract. To examine whether non-glycosylated CDF/LIF co-chro-matographs with the footpad activity, we added recombinant, non-glycosylated CDF/LIF to the footpad homogenate and, after an overnight incubation at 4°C, determined whether CDF/LIF was retained on a DEAE column. As expected (Fukada, 1985), CDF/LIF was not retained on the column, even in the presence of footpad extract (Fig. 10). The footpad factor was, however, retained and no cholin-ergic inducing activity was detectable in the flow-through fraction (Fig. 10). Thus, the major footpad activity does not behave like either glycosylated or non-glycosylated CDF/LIF.

To determine if any of the cholinergic inducing activity present in footpad extracts was immunologically related to CDF/LIF, we employed an antiserum generated against CDF/LIF purified from medium conditioned by rat heart cells (Fukada and Towle, 1992). This antiserum blocks the cholinergic inducing activity of the purified protein on cul-tured sympathetic neurons (Fig. 11). For these experiments, extracts were made from postnatal day 8 footpads since RT-PCR analysis indicated that the highest concentration of CDF/LIF mRNA is present at this age (Yamamori, 1991) and extracts from rat pups 5 days and older contain cholin-ergic inducing activity (Rao and Landis, 1990). As illus-trated in Fig. 11, the antiserum did not block the effects of the footpad extract on ChAT activity; it was not significantly different from that of control grown without CDF/LIF and the blocking antiserum. In addition, the presence of the antiserum had no effect on the low levels of ChAT activity normally present, or on the induction of ChAT by sciatic nerve extract. These observations are consistent with the failure of antibodies raised against the N-terminal region of CDF/LIF to immunoprecipitate the cholinergic inducing activity in extracts of footpads from 14-day-old and adult rats (Rao and Landis, 1990; our unpublished observations).

Skeletal muscle from young rats contains a protein that stimulates the development of ChAT activity in cultures of E14 rat spinal cord (McManaman et al., 1988). This factor has approximately the same relative molecular mass and pI as the major footpad factor. To assess further the similarity between the factors, we obtained a partially purified (Hap7) fraction of the skeletal muscle factor and tested its ability to alter the neurotransmitter properties of cultured sympathetic neurons. The factor did not alter neuronal survival. No cholinergic induction was observed at any of the concentrations tested and no reduction in TH was evident at the highest concentration tested (Fig. 12). The same fraction elicited a maximal induction of ChAT (150%) in spinal cord cultures (Dr J. McManaman, personal communica-tion). Therefore, it is unlikely that the footpad and skeletal muscle factor are similar.

We have previously shown that extracts of footpads from both developing and adult rats contain cholinergic differentiation activity when assayed on cultures of neonatal rat sympathetic neurons (Rao and Landis, 1990). Characterization of the extracted factor(s) reveal that the major activity resides in either a single, or several closely related, soluble macromolecules of pI 5.0 and relative molecular masses of 22 –24×10³. We have compared the properties of this major activity to those of three well-characterized cholinergic factors, a cholinergic differentiation factor from skeletal muscle, CDF/LIF and CNTF. A fourth cholinergic factor, MANS, is not a likely candidate for the footpad factor. Unlike MANS (Adler et al., 1989; Wong and Kessler, 1987), the footpad activity is easily extracted in low salt conditions and no detectable activity is associated with membrane fractions from sweat glands (Rao and Landis, 1990; our unpublished data).

The cholinergic factor from skeletal muscle has a pI of 4.8 and a relative molecular mass of 22×10³M_r (McMana-man et al., 1988, 1989, 1990). While this factor had not been shown to induce cholinergic function in sympathetic neurons, CDF/LIF enhances survival of cultured motor neurons and increases their levels of ChAT (Martinou et al., 1989, 1992) and CNTF supports motorneuron survival in vitro and in vivo after nerve lesion (Sendtner et al., 1990; Arakawa et al., 1990). Thus, it seemed possible that the muscle-derived cholinergic factor would induce cholinergic function in cultured sympathetic neurons. When we tested a purified fraction of the skeletal muscle factor for ChAT induction under the same conditions in which footpad extract, CDF/LIF and CNTF cause a 10-to 20-fold induc-tion, the skeletal muscle factor did not alter either ChAT activity or TH content. These findings make it unlikely that this factor is responsible for the cholinergic effects of the footpad extract and, further, they suggest that the choliner-gic factor from skeletal muscle is distinct from CNTF.

Two soluble proteins, CDF/LIF and CNTF, exert very similar effects on the expression of neurotransmitter properties by cultured rat sympathetic neurons; they both increase ChAT and VIP expression and reduce TH and NPY expression (Ernsberger et al., 1989; Saadat et al., 1989; Yamamori et al., 1989; Nawa and Patterson, 1990; Nawa et al., 1990; Rao et al., 1992a). Each of these effects is also elicited by the purified footpad extract.

CDF/LIF is an attractive candidate for the differentiation factor in the footpad extract. Its effects on the neurotransmitter phenotype of cultured sympathetic neurons are indis-tinguishable from those of the purified footpad extract. Moreover, recent experiments using RT-PCR detected a selective expression of CDF/LIF mRNA in P8 footpads; in contrast, CDF/LIF mRNA was not detected in tissues of postnatal rats that are innervated by noradrenergic sympa-thetic neurons (Yamamori, 1991). Comparison of the biochemical properties of CDF/LIF with those of the partially purified footpad factor(s), however, reveals significant differences in size and charge. CDF/LIF purified from heart-cell-conditioned medium is a basic 45×10³M_r protein while the major cholinergic inducing activity present in the sweat gland extract is an acidic protein with a relative molecular mass in the 22 to 26×10³ range. In addition, antisera generated against an N-terminal peptide sequence from CDF/LIF failed to immunoprecipitate the cholinergic inducing activity from extracts of footpads from adult rats (Rao and Landis, 1990). The factor in footpad extracts could, however, represent a post-translationally modified form of CDF/LIF. Such modifications could account for the differences in physical properties, as well as the lack of reactivity with antibodies raised against the N-terminal region of CDF/LIF. A more stringent test for the presence of CDF/LIF is the use of an antiserum that directly blocks its activity. The binding of antibodies in such a serum to CDF/LIF would be less likely to be affected by post-translational modifications since the active site of even a modified CDF/LIF should be available for binding to its neuronal receptor. The blocking antiserum did not inhibit the cholinergic activity present in extracts of footpads from postnatal day-8 rat pups, the age at which CDF/LIF mRNA is most abundant (Yamamori, 1991). Thus, while message for CDF/LIF is present, we could not detect any CDF/LIF differentiation activity in the extract. It is possible that CDF/LIF is present in the footpad extracts at such a low level that a more stringent assay, for example, immuno-precipitation after separation of the distinct cholinergic inducing activities, is required to detect its presence. CDF/LIF present in footpads may also be lost during the extraction.

The biochemical properties of the major footpad cholin-ergic activity are strikingly similar to those reported for CNTF (Manthorpe et al., 1986; Lin et al., 1989; Stockli et al., 1989). The effects of CNTF and the footpad extract on the neurotransmitter properties of cultured sympathetic neu-rons are identical (Saadat et al., 1989; Rao and Landis, 1990; Rao et al., 1992a). Moreover, the footpad extract possesses ciliary neurotrophic activity, and this property co-purifies with the cholinergic-inducing activity. These findings, coupled with the immunoprecipitation of much of the cholinergic and neuropeptide-regulatory activities by anti-CNTF antibodies, provide evidence for the identity of CNTF as the major cholinergic factor in the footpad extract.

Several lines of evidence, however, are inconsistent with such a conclusion. First, immunoblotting experiments with polyclonal antibodies against rat CNTF failed to reveal CNTF immunoreactivity in footpad extracts. Second, north-ern blot analysis failed to demonstrate detectable CNTF mRNA in footpads. Third, in immunocytochemical and in situ hybridization studies, neither sweat glands nor the Schwann cells associated with their sympathetic innervation had detectable CNTF immunoreactivity or mRNA. In each case, comparisons were made with sciatic nerve since extracts from sciatic nerve and sweat glands contain equivalent amounts of cholinergic inducing activity (Rao and Landis, 1990) and all the cholinergic inducing activity in sciatic nerve can be immunoprecipitated with CNTF anti-serum. Each of the methods yielded positive signals in analyses of sciatic nerve. This is particularly striking in the case of the immunoblot experiments in which equivalent amounts of cholinergic inducing activity from sweat gland were probed and a signal could be detected in tenfold less sciatic nerve extract. It is surprising that the CNTF anti-serum that precipitated cholinergic inducing activity from both sciatic nerve and sweat gland extracts recognized a band in the western blot of sciatic nerve extract but not sweat gland. The immunoprecipitation experiments, how-ever, were done with native protein while the immunoblot procedure could have denatured the relevant epitopes. Immunocytochemical studies demonstrate that at least some authentic CNTF should also be present in the extract; a sparse plexus of CNTF-immunoreactive nerve fibers is evi-dent in the footpads, coursing between the glands and beneath the epidermis. The contribution that the CNTF associated with these nerve fibers makes to the cholinergic inducing activity present in the footpad extracts remains to be defined. Since, however, a similar distribution of CNTF-immunoreactive fibers is present in hairy skin, extracts of which lack cholinergic inducing activity, it seems unlikely that the nerve fiber-associated CNTF represents the major source of the differentiation activity present in footpad extracts. Consistent with this interpretation is the observation that extracts of footpads of tabby mutant mice that lack sweat glands (Gruneberg, 1971) but have CNTF-immunoreactive fibers in their footpads contain significantly reduced levels of cholinergic inducing activity (our unpublished observations).

An alternative hypothesis in keeping with our results is that the major footpad factor is a CNTF-like molecule, which is recognized by some, but not all, of the probes used in our studies. Further characterization of the immunopre-cipitated material will clarify this issue. Evidence has recently been presented for the existence of such a CNTF homologue in chick (Eckenstein et al., 1990; Leung et al., 1992). This molecule, like mammalian CNTF, lacks a signal peptide but in contrast to mammalian CNTF is secreted into the medium by an unidentified mechanism (Lin et al., 1989; Stockli et al., 1989; Leung et al., 1992). In addition, the presence of an extracellular, soluble CNTF-like molecule has been reported in optic nerve (Hughes et al., 1988; Lil-lien et al., 1988) but it is not yet clear if this CNTF-like molecule is actually released into the medium (Lillien et al., 1988).

The CNTF antibodies only partially immunoprecipitated the neurotransmitter differentiation activities present in the footpad extract. About 80% of the cholinergic activity from footpads, and less of the neuropeptide-regulating activities was immunoprecipitated. It is unlikely that this is due to our immunoprecipitation procedures as identical precipitation experiments with sciatic nerve extract precipitated 100% of the cholinergic inducing and peptide inducing activity (Rao et al., 1992a). Further, independent immuno-precipitation experiments with a distinct antiserum to CNTF yielded a similar result (Rohrer, 1992). These observations suggest that there are two (or more) distinct cholinergic fac-tors present in the footpad extract and that the major component is CNTF-like. The presence of a second cholinergic factor, in principle, complicates the quantitative comparisons between sciatic nerve and footpad which form the basis for excluding authentic CNTF as the major differentiation activity. Even when 10-fold more RNA were probed from sweat glands than from sciatic nerve, however, we did not obtain evidence for the presence of CNTF in footpad.

In summary, we have found that most of the differentiation activity present in footpads of young rats is associated with a molecule(s) that is immunoprecipitated by anti-bodies that recognize CNTF. While CNTF message and protein were easily detected in northern blot, in situ hybridization and immunoblot analyses of sciatic nerve, we did not detect either in footpads. Some activity remains after immunoprecipitation, suggesting that the extract contains at least two distinct differentiation activities. If two components are indeed present in the extract the second, as yet uncharacterized, component, is unlikely to be CDF/LIF as immunoprecipitation and antibody blocking experiments have failed to detect its presence. Once the differentiation activities in the extract have been identified, their role in the directing the changes that occur in neurotransmitter phenotype in the developing sweat gland innervation in vivo can be defined.

The authors wish to thank Dr Ken Neet for providing NGF, Regeneron Pharmaceuticals for recombinant CNTF and anti-CNTF antiserum, Dr Keiko Fukada for purified CDF/LIF and anti-serum, Dr B. Samal of Amgen Pharmaceuticals for recombinant human LIF and Dr Chris Beattie for advice. We thank Drs Donna Morissey and Mark Furth of Regeneron Pharmaceuticals for the polyclonal antiserum to CNTF. This work was supported by NINDS Javits Neuroscience Investigator Awards to P.H.P. and S.C.L., a McKnight Foundation Neuroscience Research Award to P.H.P. and an AHA Grant-In-Aid to S.C.L. Dr Rao is the recipient of a postdoctoral fellowship from the Northeast Ohio Affiliate of the American Heart Association.