Expression of P450c17 in the Human Fetal Nervous System

Abstract

P450c17 catalyzes steroid 17α-hydroxylase and 17,20 lyase activities. P450c17 is expressed in human fetal and postnatal adrenals and gonads and in the developing mouse nervous system, but little is known about its expression in the human nervous system. We obtained portions of 9-, 10-, and 11-wk gestation human fetuses and delineated the pattern of expression of P450c17 in their peripheral nervous systems by immunocytochemistry using the P450c17 antiserum previously used to characterize P450c17 in the mouse brain. P450c17 was readily detected in the dorsal root ganglia (DRG) and spinal cord. Neural structures were identified with antisera to the cytoskeletal protein neural cell adhesion molecule; DRG were identified with antisera to the neuronal transcription factor BRN3A and neurotrophin receptor tropomyosin-receptor-kinase B. The identification of P450c17 was confirmed using commercial antisera directed against different domains of P450c17 and by using antisera immunodepleted with authentic human P450c17. We also found expression of the P450 cholesterol side-chain cleavage enzyme (P450scc) in the spinal cord and DRG. Expression of P450scc is limited to cell bodies; unlike P450c17, we never detected P450scc in fiber tracts. Catalysis by P450c17 requires electron donation from P450 oxidoreductase (POR). Dual-label immunohistochemistry detected P450c17 and POR colocalized in DRG bundles, but some fibers containing P450c17 lacked POR. These data suggest that neurosteroids synthesized via these two enzymes may act in the developing human nervous system. The expression of P450c17 in structures lacking POR means that P450c17 may not be steroidogenic in those locations, suggesting that P450c17 may have additional functions that do not require POR.

P450c17 is the single microsomal cytochrome P450 enzyme that catalyzes 17α-hydroxylase and 17,20 lyase activities and hence is required for the synthesis of cortisol, androgens, and estrogens (1). Human P450c17 exerts 17,20 lyase activity almost exclusively with Δ5 steroids (2), whereas rodent P450c17 preferentially catalyzes 17,20 lyase activity with Δ4 steroids (3); thus, most human sex steroid synthesis proceeds through the intermediacy of dehydroepiandrosterone (DHEA) and not through 17OH-progesterone, which is not efficiently converted to androstenedione by human P450c17 (2, 4). The activities of P450c17, like that of all microsomal cytochrome P450 enzymes, require the flavoprotein P450 oxidoreductase (POR) (for review, see Ref. 5). The 17,20 lyase activity of human P450c17 is increased by P450c17 phosphorylation (6, 7) and is enhanced by the allosteric action of cytochrome b5 (2, 8). Human P450c17 mRNA is encoded by the CYP17A1 gene (9) on chromosome 10 (10) and is abundant in the human fetal adrenal and testis but is not found in the human fetal ovary or placenta (11, 12).

P450c17 has been found in the brains of rodents (13–20), birds (21–24, reviewed in Ref. 25), guinea pig (26), fish (27–30), amphibians (31, 32), and reptiles (33). In rodents, P450c17 expression in the nervous system appears mainly to be found during embryonic development (13) but may be expressed in adult hippocampus (18). Physiological concentrations of DHEA and DHEA sulfate (DHEAS) normally found in the fetal mouse brain differentially stimulated outgrowth of neurites from primary cultures of embryonic d 16 fetal mouse neurons in vitro: DHEA stimulated axonal growth, whereas DHEAS stimulated dendritic growth (34). DHEA also increased axonal outgrowth from cultured sensory neurons and from primary cultures of dorsal root ganglion (DRG) neurons (35). Recent studies have demonstrated that DHEA is a ligand for the neurotrophin receptor tropomyosin-receptor-kinase (Trk) A (36), thereby establishing a new mechanism for DHEA action and suggesting this may be the mechanism through which DHEA stimulates axonal growth. DHEA and DHEAS also accelerate spinal cord recovery after traumatic injury in rodents (37, 38). However, genetic proof for an essential role for P450c17 in the brain is absent because knockout of the mouse cyp17 gene causes preimplantation embryonic lethality (39), whereas absent human P450c17 activity causes a complex endocrine syndrome that is compatible with adult life (40, 41). Thus, it appears that P450c17 is likely to serve a permissive or complementary role, rather than being a uniquely required factor. We show that P450c17 is expressed in human fetal nervous system at 9–11 wk and that expression is robust in the fetal spinal cord, DRG, and trigeminal ganglia.

Materials and Methods

Human tissue

Human fetal tissue was obtained at San Francisco General Hospital from elective abortions under a protocol approved by the University of California, San Francisco, Committee on Human Research. Permission to collect fetal tissues was obtained from the women undergoing elective abortion. The gestational age of first-trimester specimens was primarily estimated by ultrasonographic measurements of crown-rump length and further confirmed by foot-length measurement. Specimens were harvested within 30 min of the abortion procedure, placed in Hanks' balanced salt solution containing calcium and magnesium and supplemented with 50 μg/ml gentamycin (Mediatech, Inc., Manassas, VA) and Fungizone (2.5 μg of amphotericin B and 2.05 μg of sodium deoxycholate per milliliter; Invitrogen Corp., Carlsbad, CA), and held on ice for transport to the laboratory. Elapsed time between the surgical procedure and freezing was less than 3 h. The samples were removed from buffer and stored semidry on a strip of foil at −80 C until processed.

Histology

Human fetal specimens that contained spinal cord and DRG were placed flat on the dorsal side to establish a left/right axis; frontal sections dividing the sample along the dorsal/ventral axis were obtained and prepared for analysis. Sections of 14–18 μm were collected from an HM 505 E Microm cryostat onto Superfrost Plus slides (Fisher Scientific, Pittsburgh, PA). Prepared slides with adjacent sections were air dried and stored at −80 C. For immunohistochemistry, the prepared slides were fixed in 4% paraformaldehyde for 5 min, rinsed in PBS 0.1% Triton X-100 to remove fixative, and placed in blocking solution for 2–4 h at room temperature. Blocking solution consisted of PBS, with 0.1% Triton X-100, 0.1% BSA, and 10% serum (either goat or donkey) the same as the secondary antibody. Incubation with the primary antibody occurred in a humidified chamber overnight at 4 C. Samples were incubated with either donkey or goat secondary antibodies after four 15-min rinses with PBS to remove the primary antibody. Secondary antibodies were applied to the samples for 2–4 h at room temperature in a humidified chamber, followed by two 15-min washes with PBS to remove the secondary antibody. A third rinse contained the nuclear stain 4′,6-diamidino-2-phenylindole (DAPI; a fluorescent stain that binds to AT-rich regions of DNA) for an additional 15 min followed by a final rinse of PBS to remove excess DAPI before applying coverslips in mounting solution (crystal mount; Biomeda, Foster City, CA). Images were collected with the ×10 and ×20 objectives from a Leica DMRB upright fluorescent microscope (Buffalo Grove, IL) or at ×10, ×20, or ×63 (oil objective) with an upright Zeiss Axioimager M2 with Apotome fluorescent microscope (New York, NY). Images were captured with Leica Firecam imaging software for the Macintosh computer platform and Zeiss Axiovision imaging software for the personal computer platform (Thornwood, NY). Postprocessing of images and figure construction were done with Adobe Photoshop CS5 (San Jose, CA) and/or Google Picasa software for the personal computer platform (Mountain View, CA).

Antisera

The polyclonal rabbit antihuman P450c17 antiserum (42) and its use for immunohistochemical analysis of mouse P450c17 (13) have been described; we used this antibody diluted 1:1000. Goat polyclonal antibodies against human P450c17 from Santa Cruz Biotechnology (Santa Cruz, CA) (catalog no. N17:SC-46084, lot no. g0306; catalog no. C12:SC-46080, lot no.I1407; and catalog no. N18:SC-46085, lot no. b1307) were used at 1:500 dilution, and control rabbit IgG (catalog no. H48:SC-66849, lot no. A0108; Santa Cruz) was used at 1:500 dilution. The rabbit polyclonal antiserum against human P450 cholesterol side-chain cleavage enzyme (P450scc) has been described (43), and rabbit polyclonal antiserum against rat POR was from Stressgen (Ann Arbor, MI); both were used diluted 1:500.

To identify the fibers of neural cells, we used mouse monoclonal antibody to Trk C and B receptors from Santa Cruz Biotechnology (catalog no. TRKB-3:SC-7268, lot no. b1108) and Abgent, Inc., (San Diego, CA) (catalog no. TRKC:am7688a, lot no. sg080924a), diluted 1:250. For other neural cells, the rabbit polyclonal antibody against brain-specific homeobox/POU domain protein 3A (BRN3A) was from Abcam (Cambridge, MA) (catalog no. 23579, lot no. 388042) diluted 1:500 and the monoclonal antibody for human neural cell adhesion molecule (NCAM; CD56) (catalog no. MAB2120z, lot no. pso1522887) diluted 1:250.

Primary antibody-antigen complexes were detected using fluorescent-labeled (Alexa Fluor 488 or 546) antibodies (Invitrogen) diluted 1:500. To detect rabbit primary antibodies, we used donkey antirabbit (catalog no. A21206), goat antirabbit (catalog no. A11070), and goat antirabbit (catalog no. A11034). To detect goat primary antibodies, we used donkey antigoat (catalog no. A11057 and A11055). The monoclonal antibodies were detected (catalog no. A11001 and A11018).

For immunodepletion studies, 1 μg of bacterially expressed human P450c17 protein purified to apparent homogeneity (44) was incubated with 1 ml of diluted rabbit or goat primary antibody overnight at 4 C. The incubated mixture was centrifuged at 13,500 × g to pellet any debris and used for immunohistochemistry. Before immunodepletion, the quality of the purified protein was assessed by Western blot transfer and detection.

Image analysis

In contrast to bright-field image acquisition, which works well with a wide range of objectives, capturing fluorescent images for analysis requires high magnification objectives to reveal detail within the specimen. Several figures in this study show detailed images captured with high magnification objectives, but some images show low-magnification views, which cannot be captured with low-magnification objectives. To overcome this obstacle, we used the Adobe Photoshop Photomerge function to compile 10–300 individual high-magnification images into a single picture. Some images were processed further with Adobe Photoshop to remove objects that reflect bright fluorescent light such as dust, glass particles, other debris, or bubbles. The main benefit of displaying low-magnification views is to provide anatomic perspective when viewing the image.

Results

Human fetal samples

Human fetal brains obtained from pregnancy termination at midgestation (∼20 wk) rarely survive intact and lack recognizable shape and landmarks, making ascertainment of specific brain regions impossible. This difficulty is especially true for the brain stem and thalamic regions that express P450c17 in the midgestation mouse brain, analogous to the midgestation developmental timeframe from which human tissue is often obtained. Hence, we sought to identify the expression of steroidogenic enzymes in fetal nervous system tissues that had identifiable structures and are known to express P450c17 in the rodent nervous system (13).

We obtained tissue from three human fetal samples at 9, 10, and 11 wk gestational age; all contained DRG and spinal cord. Fresh tissues that are intact from this gestational stage are extremely difficult to obtain from abortuses at our institute but are optimal for histological study because they lack calcified bone, making them easier to section than tissues from later gestational ages (e.g. 20 wk). Furthermore, the small size of the sample permits mounting a section representing a fairly large anatomic area on a single slide, facilitating the identification of specific structures.

Identification of neural structures in the 10wk human fetal sample

Staining for NCAM, a homophilic binding glycoprotein expressed on the surface of neurons and glia, identifies the spinal cord, which is derived from the neural tube, with the ventrally associated DRG. The DRG, derived from the neural crest, extend fiber tracts between the laterally located precartilaginous vertebral bodies (Fig. 1A). Although the precise orientation of the 10-wk specimen is unknown, the DRG are located repetitively along a quasi-anterior/posterior axis along the length of the spinal cord, which represents the midline of the sample. The tissue sections analyzed reflect a location within a quasi-dorsal/ventral axis within the specimen, close to the ventral region of the spinal cord in which DRG reside in close association with the spinal cord.

Fig. 1.

Identification of human fetal neural tissue. Characterization of the 10-wk gestation (gw10) sample by fluorescence immunohistochemistry identifies the spatial distribution of neural tissue within the sample. All figures display the sample oriented with the spinal cord at the top. A and B, The bundles of DRG containing neuronal cell bodies and their associated axonal fiber tracts are interspersed between the vertebral bodies (vb). Examples of fiber tracts emanating from DRG neurons projecting laterally from the midline are indicated and designated with lines. A, NCAM immunostaining. A low-magnification photomicrograph identifies NCAM immunopositive neurons in the spinal cord, DRG, and fiber tracts. Not all fiber tracts are evident because some are out of the plane of the section. B, NeuN immunostaining, a marker of mature neurons, identifies DRG neurons in different stages of development that are not spatially homogeneous. NeuN distinguishes more mature neurons in the developing DRG bundle from all neurons (DAPI nuclear staining). DRG fiber tracts located within individual DRG bundles are indicated and lack NeuN immunostaining. C-C″, BRN3A and TrkB immunostaining. DRG identity is confirmed by expression of sensory neuron-specific transcription factor BRN3A (C) and the TrkB neurotrophic receptor (C′). BRN3A immunolabels cell bodies and TrkB immunolabels cell bodies and fiber tracts (merged image, C″). Together with NeuN, these markers further demonstrate the nonhomogenous spatial organization of neurons and fibers within DRG bundle. Vb, Vertebral bodies. Panels (C-C″) are taken at the same magnification, shown by the bar in panel C.

As developing neurons mature, they begin to express the neuron-specific nuclear antigen NeuN (45). Staining with DAPI (to identify nuclei) and neuronal nuclei (NeuN) shows that the neurons within some DRG are in various stages of development in this 10-wk fetus and that there is a heterogeneous gradient of increasing neuronal maturation, as assessed by increased NeuN immunostaining (Fig. 1B). Additionally, the morphology of fiber tracts can be delineated within some DRG in regions that are devoid of DAPI and NeuN signals (Fig. 1B). NeuN is restricted to neuronal cells of the spinal cord and DRG and not found elsewhere in the sample. BRN3A is a transcriptional regulator found in the cell bodies of all sensory neurons, including those of the DRG (46, 47). DRG are also characterized by expression of the neurotrophin receptors TrkB and TrkC (48) and TrkA (49, 50). Immunohistochemistry identified BRN3A (Fig. 1C) (also known as transcription factor POU4F1), TrkB (Fig. 1C′) and TrkC (not shown) in the structures we identified as DRG (by staining for NCAM) confirming their molecular identity as DRG (Fig. 1C), consistent with their morphological appearance.

Expression of P450scc and P450c17 in dorsal root ganglia

Our prior studies showed that the steroidogenic enzymes P450scc and P450c17 are expressed in the DRG of fetal mice (13, 51). P450scc, encoded by the CYP11A1 gene, catalyzes the cleavage of the cholesterol side chain to yield pregnenolone, which is the first and rate-limiting step in steroidogenesis; it is expression of P450scc that renders a cell steroidogenic (1). Immunohistochemical staining with antisera specific for human P450scc identified this steroidogenic enzyme in the DRG at 10 wk of human gestation (Fig. 2B). The rabbit nonimmune control sera did not identify any specific structures, and background fluorescence is detectable in vertebral bodies (Fig. 2A). The level of P450scc immunostaining (Fig. 2B) is consistent throughout most of the DRG examined, similar to the broadly expressed BRN3A (Fig. 1C). Neural expression of P450scc is limited to the DRG; we never detected P450scc in fiber tracts from the ganglia. However, P450scc is also expressed at lower levels in unidentified nonneuronal cells (NeuN negative) in the lateral region occupied by the location of fiber tracts. We did not detect definitive staining for P450scc in the spinal cord (not shown), although we (51) and others (52, 53) have detected it in the mouse and rat spinal cord.

Fig. 2.

Distinct expression patterns for steroidogenic enzymes in DRG bundles and fiber tracts at 10 wk gestation (gw10). A, Background fluorescence associated with the secondary antibody is low in the absence of primary rabbit polyclonal antibodies against P450scc or P450c17. There is a low background of autofluorescence associated with some structures such as the vertebral bodies (vb). B, P450scc immunostaining is enriched in DRG bundles (cb, cell bodies) and absent in the laterally extending fiber tracts. Low levels of immunostaining (slightly above background) near the extending fiber tracts are associated with unidentified nonneuronal cell types because they are not positive with NeuN or NCAM. C, P450c17 immunostaining using an antibody against bacterially expressed human P450c17 (42). P450c17 is expressed in DRG neurons and fibers. This DRG shows nonuniform morphology and P450c17 immunostaining; several other examples of DRG bundles display various amounts of P450c17 immunostaining (shown in later figures). In contrast to P450scc, the laterally projecting fiber tracts express P450c17. The destination of these projections is unknown. D, TrkB immunostaining. TrkB expression was analyzed in a section adjacent to the section shown in C and shows a distribution similar to that seen for P450c17 in DRG bundles and laterally extending fiber tracts. E, Higher-magnification photomicrograph of P450c17 immunostaining shows spinal cord (sc) fiber tract and DRG immunopositive staining.

We detected P450c17 in the DRG (Fig. 2C) in a pattern similar to that seen in mouse embryonic dorsal root ganglia (13). P450c17 was found in both the soma and fiber tracts emanating from the cell bodies within the DRG in the 10-wk sample. Compared with BRN3A and to P450scc, expression of P450c17 is more restricted among the various DRG we analyzed: P450c17 is robustly expressed in most DRG but is barely expressed in some regions of other DRG in this 10-wk sample (Fig. 2, C and E). Although BRN3A and P450scc appear diffuse within the soma, the appearance of P450c17 appears sharper with distinct boundaries. Because of the uncertain orientation of our specimen, it is unclear whether this difference in P450c17 expression is related to the position of the DRG along the anterior/posterior axis or a reflection of the dorsal/ventral organization of an individual DRG.

The distribution of P450c17 expression resembles that of TrkB (Fig. 2D) because both are expressed in DRG cell bodies and fiber tracts. Because we previously showed colocalization of TrkB and P450c17 in some mouse DRG neurons (13), it is likely that both proteins colocalize in some human DRG neurons as well because their immunostaining patterns appear to overlap. P450c17 is also detected in smaller projections that branch off from the main fiber tract bundles emanating from the DRG. Although these scattered bundles of P450c17 positive axons originate from the DRG and appear to have directional organization, we do not know where their axons terminate. It is unlikely that the cells (identified by DAPI staining) depicted in the field and surrounded by P450c17 positive axons are neurons because they do not express NCAM or NeuN (Fig. 1A; additional data not shown). It is possible, but unlikely, that these nonneuronal cells could receive terminal endings from the neighboring axons that contain P450c17.

During development, DRG are closely associated with the ventral region of the spinal cord and receive direct innervation from fiber tracts of the spinal cord, primarily motor neurons; we found that P450c17 is also expressed in some of these fiber tract projections (Fig. 2E). Although NCAM is expressed broadly in the spinal cord and DRG, expression of P450c17 is much more scattered than NCAM in the spinal cord in the 10-wk sample.

Expression of P450c17 in the developing nervous system at 9 and 11 wk

The 11-wk (Fig. 3A), 9-wk (Fig. 3, B–D) samples were largely intact with several discernible morphological landmarks. The rostral/caudal orientation was preserved during the cryosectioning procedure. In contrast to the 10-wk specimen, expression of P450c17 in the 11-wk fetal nervous system appeared more homogenous along the anterior/posterior axis (Fig. 3A). Whereas immunohistochemistry is only semiquantitative, the expression of P450c17 in the adrenal (Fig. 3, A and B) and the spinal cord appeared to be greater and was more distinct and more organized in the 11-wk DRG than in the 9- and 10-wk samples and similar to the more mature pattern seen in the midgestation mouse and rat spinal cords (13, 53–55). By contrast, expression of P450c17 in the 9-wk sample was more diffuse and different from the 10- and 11-wk samples. The 9-wk sample lacked distinct DRG fiber tracts except for the DRG innervating the leg (Fig. 3, C and D), a feature that is shared with the 11-wk sample (Fig. 3A). However, P450c17 abundance at 9 wk appeared to be low and less distinct. There were small amounts of P450c17 in fiber tracts between the spinal cord and some DRG, but this was less obvious than in the 10- and 11-wk samples (not shown). Some DRG contained areas of P450c17 expression characterized by very short neural fibers and the absence of long neural fibers, which are seen at 10 and 11 wk.

Fig. 3.

Spatial and temporal distribution of P450c17 during human development. P450c17 is detected by immunohistochemistry at 11 wk gestation (gw11) (A) and 9 wk gestation (gw9) (B–D). A, Low magnification of gw11 human embryo. P450c17 fibers are detected in all DRG along the rostral-caudal axis. The prominent fibers innervate the leg (leg DRG, line). High levels of P450c17 are expressed in the fetal adrenal (oval structure, lower left); the adrenal is also immunopositive for P450c21 (not shown). P450c17 expression is highly organized in the spinal cord (boxed inset). B–D, P450c17 expression at 9 wk gestation (gw9) characterizes early DRG development. B, High expression in human fetal adrenal. The expression of P450c21 had the same distribution (not shown). C and D, Limb DRG fibers develop earlier than other DRG along the rostral-caudal axis. C, Limb DRG fibers express P450c17 at low levels. D, P450c17 expression is limited to DRG cell body bundles. Magnification bars, 200 μm (A and B); 175 μm (C); 150 μm (D).

Specificity of the P450c17 immunodetection

We identified P450c17 using the same high-affinity rabbit polyclonal antibody raised against bacterially expressed human P450c17 (42) that was used to map P450c17 expression in the developing mouse nervous system (13), suggesting that this antibody detects similar antigens in both human and mouse nervous systems. Because it is hypothetically possible that the antiserum is detecting some antigen other than P450c17, we sought to demonstrate that our antiserum truly detected only P450c17.

First, the expression of P450c17 was confirmed using commercially available goat antisera raised against different regions of human P450c17 (Fig. 4). These three goat antisera identified indistinguishable patterns of expression in the cell bodies of the DRG and in their associated fibers; this pattern of expression was identical to that seen with our in-house rabbit antihuman antiserum (42). Results from goat antibodies N17:SC46084 (Fig. 4A) and C12:SC46080 (Fig. 4B) confirm P450c17 expression in the cell bodies and fibers, fibers going from the DRG and fibers going to the spinal cord, and the spinal cord. Results with antibody N18:SC46085 were equivalent (not shown). P450c17 was also found in fiber tracts that branched from the main fiber bundles from the DRG (Fig. 4C).

Fig. 4.

The spatial distribution of P450c17 is confirmed with multiple antisera. The localization of P450c17 identified with our rabbit antihuman P450c17 antiserum (Figs. 2 and 3) is confirmed with two goat antisera directed against the distinct regions of human P450c17 (A and B). Both antisera also label neurons in the spinal cord (sc; A) as well as interneuron fiber tract projections between the spinal cord and DRG (B). C, High-resolution P450c17-positive fibers within the fiber tract. The neurons associated with these fiber tracts are from the DRG bundles located out of the field of view and not from the adjacent nonneuronal cells (DAPI). Magnification bars, 100 μm (A and B); 10 μm (C).

Second, we removed the IgG directed against P450c17 from both our rabbit antihuman antiserum and the commercial goat antisera by immunodepleting each antiserum with pure, bacterially expressed, full-length human P450c17 and clearing the IgG/P450c17 complexes from the antisera with protein A (Fig. 5A). In comparison with the control, nonimmunodepleted antiserum (Fig. 5B), immunocytochemistry using the immunodepleted antisera generated no detectable fluorescent signals (Fig. 5, C and D), confirming that the immunostaining was specific for P450c17.

Fig. 5.

Immunodepletion of P450c17 antibody confirms antisera specificity for P450c17. A, Bacterially expressed and purified human P450c17 protein [called pCWH17mod(G3H6); arrowhead] was used to immunodeplete both rabbit and goat antisera. The purity of two different P450c17 preparations, E1 and E2, is apparent on the Coomassie-stained gel. B, Control, nondepleted rabbit P450c17 antibody identifies DRG bundle and fiber tracts. This section shows interneuron fiber tract projections between the spinal cord and DRG. C and D, P450c17 immunostaining is absent from P450c17-immunodepleted rabbit and goat antisera. Magnification bars, 100 μm.

P450c17 in the trigeminal ganglia

The 11-wk sample included part of the head and face, which were absent from the other samples. This area contained a prominent structure that stained for P450c17 (Fig. 6). A morphologically very similar structure was seen in the head of the embryonic d 16 mouse; because those samples were wholly intact, they were previously identified as the trigeminal ganglion (13) (Fig. 6, A and B). Thus, we identified this structure in the 11-wk human fetus as also being the trigeminal ganglion (Fig. 6, C and D). Fibers from the trigeminal ganglion are both sensory and motor and innervate the face (scalp, forehead, eyes, nose, and cheeks) and mouth.

Fig. 6.

Trigeminal ganglia show immunopositive staining of P450c17. A and B, Trigeminal ganglia (tgg; line) in the embryonic d 16 mouse head provides a comparison for the human tgg. A, Medial mouse section at low magnification. This section shows P450c17 in the brain cortex. B, Section adjacent to A at high magnification, showing tgg cell bodies and fibers. C and D, Immunohistochemistry detects P450c17 in the tgg of the gw11 human head. C, Lateral section of the gw11 head at low magnification. The tgg fibers innervate the head under the eye and around the jaw. D, High magnification of the gw11 tgg cell bodies and fibers (arrow). P450c17 tgg fibers project toward the anterior and dorsal aspect. Magnification bars, 500 μm (C); 200 μm (D).

P450c17 in immature DRG

The availability of samples from approximately 9, 10, and 11 wk permits an assessment of DRG maturation (Fig. 7). In addition to varying levels of NeuN reflecting differential maturation of DRG neurons, P450c17 expression may also coincide with DRG neuronal maturation. At 10 wk, short, immature fibers containing P450c17 are prominent (Fig. 7A), and small P450c17-positive circular structures are seen at higher magnification (Fig. 7B). In addition to the short, immature fibers, the 10-wk sample contains longer, more mature fibers (Fig 7C) that resemble those seen at 11 wk (Fig. 3A). By contrast, at 9 wk, only a few fibers are P450c17 positive, and the circular structures are smaller and more numerous (Fig. 7, D–F). We propose that these circular structures represent lamellipodial patches or budding sites for axons that then become neuritic growth cones.

Fig. 7.

Immature DRG show immunopositive staining of P450c17. Immature DRG were seen at 10 wk gestation (gw10; A–C) and gw9 (D–F). A, Short fiber tracts, representative of immature neurons, are P450c17-positive and seen at gw10. B, P450c17-immunopositive staining in small circular structures. These structures may represent sites for axon formation. Small fibers are seen protruding from some of these circles and may represent later events in axonogenesis. C, P450c17 in long fiber tracts may represent later stages of DRG development at gw10. P450c17 expression in small fiber projections emanate from the main axon bundle. The neurons associated with these fibers are from the DRG bundles located out of the field of view and not from the nonneuronal cells, indicated by DAPI staining. D--F, The majority of DRG at gw9 lack P450c17-positive fibers. D, Low magnification of a DRG with minimal P450c17 expression. E, Boxed inset from D. Higher magnification reveals small circular structures revealed by P450c17 immunohistochemistry as well as P450c17-positive fibers. F, An adjacent DRG with a higher density of small circular structures. Magnification bars, 50 μm (D); 25 μm (A and C); 10 μm (E); 5 μm (B); 3 μm (F).

Coexpression of P450c17 and POR in DRG

The small amount of human fetal tissue available precluded studies of enzymatic activity. To determine if the P450c17 found in the human fetal DRG and spinal cord could have enzymatic activity, we determined whether POR was also expressed in the same cells and cellular locations as P450c17 (Fig. 8). Immunocytochemical analysis showed that POR was expressed at low levels in the cell bodies of DRG (Fig. 8, A–D). However, POR was not found in the fiber component from DRG in which we observed P450c17 (Fig. 8, E and F). These data suggest that P450c17 could be enzymatically active within the DRG cell bodies because it could receive electrons from POR but that it may not be enzymatically active in the fibers because it does not have the appropriate electron donor. Thus, these data suggest that P450c17 may have nonenzymatic functions in specific fibers.

Fig. 8.

Expression of P450c17 and POR in DRG bundles and fiber tracts. Dual-label immunohistochemical analysis using a goat secondary Alexa 546-conjugated antiserum for analysis of P450c17 (red) and rabbit secondary Alexa 488-conjugated antiserum for analysis of POR (green). A–C, Photomicrographs of the same sample captured through different emission channels. Panel D merges panels A and B. A, P450c17 is expressed specifically in DRG bundles and fiber tracts. B, POR is expressed at low levels by all cells. There is a higher concentration of POR-expressing cells in the fiber tract region as visualized in the DAPI panel (C). Most of these cells are nonneuronal. At low magnification, it appears that there are regions in the fiber tract area that coexpress P450c17 and POR (A–D). At high resolution, in a region away from the major cell-dense fiber tract, a few nonneuronal cells are found adjacent to the fiber tracts (E, E′, E″), F, F′, and F″). Panels (E, E′, E″), F, F′, and F″) are photomicrographs of the same samples captured through different emission channels. E, E′, and E″, P450c17-positive fibers appear in close association with POR-expressing nonneuronal cells. Panels F, F′, and F″ clearly display P450c17-positive fibers that do not contain POR. Panels E and F are taken at the same magnification, shown by the bar in panel E. Panels G, G′, G″, and G⁗ are high-magnification photomicrographs of the same samples showing colocalization of P450c17 and POR within the cell bodies of the DRG. The expression domain of POR is greater within neurons than is P450c17. Magnification bars, 200 μm (A–D); 10 μm (E–G).

Discussion

Our data demonstrate that the steroidogenic enzymes needed for the synthesis of DHEA exist in the early gestation human nervous system. These data suggest that the product of these enzymes, (e.g. DHEA or other neurosteroids) may have important local effects on the development of the human fetal nervous system. Our data further suggest that P450c17 may have additional, nonenzymatic functions in the nervous system.

Studies in the early 1980s provided evidence for the synthesis of DHEA and DHEAS in the rat brain (56, 57), and endogenous synthesis of neurosteroids was demonstrated by finding the mRNA for steroidogenic enzymes in the rat brain (58). Most studies of neurosteroids have been done with rodents brains, in which there is no single steroidogenic locus: steroidogenic enzymes are found throughout the mouse brain so that steroids appear to be widely produced mediating local action (13, 34, 51). However, neither targeted ablation of the mouse gene for P450scc (59) nor human P450scc deficiency (60) results in disordered brain development or an apparent neurological phenotype. Thus, the actions of neurosteroids appear to supplement and reinforce actions regulated by other factors, rather than serving indispensable functions (for review see Refs. 61 and 62). P450c17 is found in embryonic mouse neurons as early as embryonic d 9.5 and is found throughout the developing mouse brain (13) but in the adult rodent may be confined to the hippocampus (17, 18) and spinal cord (55).

DHEA and DHEAS have been reported to have many functions related to neurogenesis, neural function, protection, and survival (reviewed in Ref. 63). In vivo, DHEA and DHEAS protect the hippocampus against the toxicities of glutamate, α-amino-3-hydroxy-5-methyl-4-isoxazole-propionate and kainate (64) and protect hippocampal neurons from oxidative stress (65). In rodents, DHEA and DHEAS stimulate neurite outgrowth from fetal neurons in vitro (34); DHEA stimulates neurogenesis in the dentate gyrus in vivo (66), stimulates proliferation of human neural stem cells derived from fetal cortex (67), and stimulates differentiation of neural progenitor cells derived from human embryonic stem cells into dopaminergic neuronal lineages (68). In the brain, DHEA and DHEAS appear to act through multiple mechanisms: acting as agonists of the N-methyl-d-aspartate receptor (69, 70), antagonizing GABA_A receptors (71–73), and acting as agonists for σ-receptors (74). Most recently, DHEA and DHEAS have been shown to mediate effects through binding to TrkA (nerve growth factor) receptors at nanomolar concentrations (36), thereby acting as a neurotrophic hormone. The local synthesis of DHEA in the DRG may thus directly affect TrkA-containing neurons within the DRG to promote neuronal proliferation and axonal growth.

Despite extensive studies in rodents, there has been scant evidence for the biosynthesis of steroids in the human nervous system (reviewed in Ref. 75). Studies have demonstrated the expression of P450scc, aromatase, 5α-reductase type 1, a 3α-hydroxysteroid dehydrogenase, and 17β-hydroxysteroid dehydrogenases 1, 3, and 4 in human brain (76–84). P450scc expression was reported to increase in the temporal lobe during childhood and to reach adult levels at puberty (78) and was reported to be in higher concentrations in women than in men (76, 78). Studies of human cerebral spinal fluid (85) and human postmortem brains (86–89) have focused on detection and quantitation of neuroactive steroids including DHEA and DHEAS, rather than on expression of the steroidogenic enzyme P450c17, and hence, it is not known whether the identified DHEA and DHEAS were synthesized locally. One study indicated that P450c17 was not expressed in adult human temporal lobe and limbic system (90). DHEA increases neurogenesis and neuronal survival in cultured human neural stem cells (67), but evidence for endogenous synthesis of DHEA in the human brain has been lacking. We now show that both P450c17 and P450scc are found in the developing 9- to 11-wk human fetal spinal cord and dorsal root ganglia, suggesting that DHEA could be synthesized locally.

Because we found that the expression of P450c17 is restricted to a subpopulation of DRG cell bodies, we hypothesize that P450c17 expression identifies a gradient of cellular maturation along the anterior-posterior axis. Similarly, HOX gene expression in the developing rodent neural tube begins caudally and extends rostrally (91–94).

The curious circular structures expressing P450c17 in some immature DRG neurons at 9 and 10 wk gestation suggested that this could represent the initiation of axonal formation from these neurons. These structures are reminiscent of flattened lamellipodia around the circumference of the cell body, which initiates the first stage of development that leads to establishment of polarity of neurons (95), necessary for axonal and dendritic growth. As described by Dotti et al. (95), “These flat, veil-like structures were constantly in motion, extending, undulating, and retracting, so that their shape changed from minute to minute. Sometimes the lamellipodia extended around the entire circumference of the cell; at other times they coalesced into a few discrete zones around the periphery.” These lamellipodial patches then become neuritic growth cones. The next developmental stages are the outgrowth of minor processes from these patches and the transformation of one of the minor processes into the axon. These circular structures are seen only in immature neurons and are never seen when the neurons fully mature and send out axonal projections, consistent with the established stages of axonal and dendritic development. The expression of P450c17 in this location around the circumference of the cell body suggests that it may play a role in the earliest process of axonogenesis. Because we never detected P450scc in similar structures, we believe that locally synthesized DHEA would not play a role in axonogenesis. Furthermore, because we never detected POR within similar structures, we believe that P450c17 would serve a nonenzymatic function in these structures at that time in development. Our finding of P450c17 in fibers that lack POR suggests that P450c17 may have nonenzymatic functions in those structures as well. It will be of interest to determine whether P450c17 acts as a nonenzymatic steroid binding protein in specialized circumstances.