The molecular bases of chromaffin progenitor specification are still enigmatic. A classic model had postulated a common progenitor cell for sympathetic neurons and neuroendocrine chromaffin cells (the SA progenitor), and a crucial role for glucocorticoids in blocking neuronal and promoting neuroendocrine differentiation [7, 11, 12, 40]. Analysis of the glucocorticoid receptor knockout [15, 41] failed to support this hypothesis: glucocorticoid receptor-deficient mice had normal numbers of adrenal chromaffin cells, which resembled their wild-type counterparts in virtually all structural and chemical aspects.
In our search for alternative cues, we report in this study that adrenal cortical cells (interrenal cells in the chick) express BMP-4 starting at the beginning of cortical cell assembly. At early stages, BMP-4 mRNA is detected in the wall of the dorsal aorta and in adjacent tissues, particularly those extending ventrally and laterally. These regions include the developing adrenal cortical cells expressing the orphan nuclear receptor SF-1. In addition, they engulf the area lateral to the aorta where NC-derived cells are found and adrenal chromaffin cells differentiate. With ongoing development, BMP-4 expression outside the wall of the aorta becomes restricted to adrenal cortical cells, which by then intermingle with the differentiating adrenal chromaffin cells to form the chick adrenal gland. BMP-4 is continuously expressed by cortical cells at least until E15. Thus, throughout this developmental period, differentiating chromaffin cells are surrounded by cells with high BMP-4 expression. This differs from the situation encountered by the cells destined to form secondary sympathetic ganglia. At their initial differentiation site, the primary sympathetic ganglia, the wall of the dorsal aorta is the major source of BMP-4. On the migration route from primary to secondary sympathetic ganglia, BMP-4 expression is hardly detectable. In secondary sympathetic ganglia, BMP expression is detectable by RT-PCR  (K. Tsarovina and H Rohrer, unpublished), yet barely detectable by in situ hybridisation (present study; UE unpublished observations). Taken together, BMP-4 expression levels differ dramatically between sites of adrenal gland and secondary sympathetic ganglion formation, provoking the question of whether continuously elevated BMP-4 levels constitute a molecular cue required for chromaffin cell differentiation and counteracting neuronal differentiation.
BMPs have been firmly established in their role in the development of autonomic sympathetic and parasympathetic neurons (see [6, 43, 44] for reviews). BMPs synthesized by cells in the wall of the dorsal aorta trigger the initial development of NC cells towards noradrenergic sympathetic neurons [3–5]. BMPs induce expression of the transcription factors MASH1, Phox2a/b, HAND2, and Gata2/3 [3, 45, 46]. Overexpression of BMP-4 [3, 22] and BMP depletion by noggin  demonstrates that BMPs are sufficient and necessary to induce the expression of the enzymes of noradrenalin biosynthesis, TH and dopamine β-hydroxylase (DBH), and the neuronal markers neurofilament-L, SCG10, neurexin I, and synaptotagmin I in NC-derived precursors. Thus, the available data suggest that BMPs are the decisive stimulus triggering a network of transcription factors necessary for the differentiation of NC cells into noradrenergic sympathetic neurons.
The observation that withdrawal of BMP-4 after a short period in NC cell cultures promotes the formation of ganglion-like aggregates of cells extending neurites and expressing TH, neurexin 1, and synaptotagmin I  supports the idea that short exposure to BMP suffices to induce noradrenergic and neuronal differentiation. To test whether prolonged BMP availability suppresses neuronal properties in vivo, the effect of virus-mediated overexpression of BMP-4 at sites of sympathetic ganglion formation was analyzed. The expression of the neuronal marker neurofilament-M mRNA in sympathetic neurons, which is barely detectable in chick chromaffin cells throughout embryonic development ( and this study), appears unaltered in ganglia at sites of BMP-4 overexpression (this study). The grafted BMP-4-secreting cells were clearly biologically effective, as judged by the massive local overproduction of cartilage, alterations in spinal cord patterning and structure of the dorsal aorta. There are several possible explanations for the failure of BMP-4 to ectopically increase numbers of chromaffin cells, beyond the possibility that high levels of BMP-4 may not specifically induce a chromaffin phenotype. Thus, BMP-4 may not have the capacity to convert committed sympathetic neuronal into chromaffin progenitors, at least not in the given cellular and molecular context. Or chromaffin cells might have been generated, but did not survive.
Even so, BMP-4 distinctly promoted differentiation of chromaffin properties in the sympathetic ganglia, as judged by the increase in the number and size of chromaffin granules. In sympathetic ganglia of chick embryos at E6-9, two cell populations with differently sized granules are present [47, 48]. A growing population of cells shows scarce dense core vesicles with approximate diameters of 100 nm while a transient population of cells, which is predominantly located in upper lumbar sympathetic ganglia  and was not found in the upper thoracic ganglia analysed in the present study, shows larger granules of up to 300 nm in diameter. It is not clear which of the cell populations is affected by BMP-4 overexpression, and it is currently not clear whether neurofilament-M expression differs between the two populations. One possibility is an effect of BMP-4 on granule size in sympathetic neuron precursors while neurofilament-M expression levels remain refractory. Alternatively, cells with larger granules that may show low neuronal marker expression similar to chromaffin cell precursors may respond with an increase in granule size to BMP treatment. The growing evidence of an early divergence between sympathetic neuron and adrenal chromaffin cell precursors may support the latter scenario.
Several lines of evidence argue for both similarities and differences in the signalling networks involved in the generation of sympathetic neurons and chromaffin cells (see  for a review). An initial study analyzing the SA cell lineage in mice lacking MASH1 reported that the neuronal progeny of the SA lineage was largely eliminated, whereas adrenal chromaffin cells were hardly affected . Our re-analysis of the MASH1 knockout revealed that MASH1 was required for orchestrating the normal differentiation program of a majority, but not all, chromaffin cells . Similar to the MASH1 knockout, our analysis of Phox2B-/- mice provided evidence that chromaffin cells are apparently distinctly different from sympathetic neurons in their requirement for Phox2B . Recently, it has been shown that the zinc-finger factor Insm1 (IA1) is expressed early during SA development and that a null mutation of Insm1 affects the development of chromaffin cells and sympathetic neurons differentially . Thus, these analyses of MASH1, Phox2B and Insm1 deficient mice suggest that the SA progenies populating sympathetic ganglia and adrenal glands are not identical with regard to their requirements for these transcription factors. The distinct requirements may reflect distinct origins and distinct characters of SA progenitors, and/or differences in the cellular/molecular environments of sympathetic ganglia and the adrenal gland, respectively. Similar to mice, in the chick embryo differences between sympathetic neuron and adrenal chromaffin precursors have been noted from the earliest stages of differentiation . The presence of neuron-like cells with high neurofilament-L expression levels in adrenal tissue throughout development  and the presence of chromaffin-like cells with low neurofilament-M expression in sympathetic ganglia (this study) suggest an early diversification of lineages that cannot be overcome by differences in the environment.
What, then, is the function of BMP-4 in the adrenal gland? Our experiments applying noggin for 3 or 5 days to adrenal explants isolated at S23 strongly suggest that adrenal BMP-4 augments numbers of TH-positive cells, and does so by inducing TH in TH-negative SA progenitors rather than by stimulating proliferation monitored by BrdU incorporation. The notion that adrenal BMP-4 induces TH in SA progenitors that are still TH-negative at the time of their migration into the adrenal anlagen is also supported by our observation that a substantial number of SA progenitors, which are Sox10-, MASH1- and/or Phox2B-positive, but TH-negative, can be found in the vicinity and inside the chick (S23) and E12.5 mouse adrenal anlage [38, 45]. Thus, adrenal BMP-4 would serve a similar role as BMP-4 secreted from the dorsal aorta, that is, to induce NC cells to become SA progenitor cells. It is also conceivable that adrenal cortical BMP-4 may act in an autocrine fashion on BMPR-bearing cortical cells in functions that remain to be elucidated.