While markers capable of distinguishing mature 5-HT neuron subtypes are wanting, at hand are markers that, when viewed in combinations, can resolve 5-HT progenitor cells into discrete subsets. From these subsets may arise physiologically relevant groupings of mature 5-HT neurons; this is because developmental programs that define the fate and function of neurons are often set in motion by the action of factors differentially expressed among their antecedent progenitor cells. 5-HT progenitor cells reside in the embryonic hindbrain in bilateral territories flanking the floor plate and spanning much of the anteroposterior (AP) extent of the hindbrain. This progenitor territory can be subdivided along the AP axis into molecularly distinct subsets based on the broader partitioning of the hindbrain into segments (rhombomeres) with distinguishing gene expression profiles (reviewed in ⁵). Thus, aspects of 5-HT neuron subtype identity may be determined through the action of rhombomere(r)-specific genetic programs on resident 5-HT progenitor and precursor cell subsets. We have set out to deconstruct the 5-HT neural system based on rhombomere-defined 5-HT sublineages. Our investigations have begun with studies of 5-HT progenitor cells situating in r1, r2 or r3.

Our approach extends the recently developed paradigm of intersectional and subtractive genetic fate mapping^6,7 (and reviewed in ⁸) through the generation of (1) a novel, broadly applicable dual recombinase-responsive indicator allele, RC::Fela (Fig. 1a,b) that provides enhanced single-cell resolution by comparison with our previously generated intersectional alleles^6,7; and (2) a highly efficient Flpe recombinase driver line, ePet::Flpe (Fig. 1d,e; Supplementary Fig. 1a–q), capable of mediating recombination in 5-HT precursors defined by expression of the ETS-domain transcription factor Pet-1 – thus, in most if not all 5-HT neurons (Supplementary Fig. 1a–q). Placing RC::Fela and ePet::Flpe in combination with a cre driver line active in a specific rhombomere allows for determining which mature 5-HT neurons arise from that specific rhombomere (with its associated unique molecular code); moreover, it presents a means to gain genetic access for further manipulation of just that 5-HT neuron subset in isolation – such cannot be achieved using single recombinase-based strategies.

To study the subset of 5-HT neurons arising from r1, we partnered RC::Fela and ePet::Flpe with an En1-cre knock-in allele (En^Cki)⁹ – while the latter allele is active in midbrain progenitor cells in addition to the entirety of r1 (Fig. 1f, data not shown, and ⁹) this is not confounding given that 5-HT neurons do not arise from midbrain^4,10. In triple transgenic animals, we show that the intersectional marker, nuclear-β-galactosidase (nβgal), is efficiently activated and can be used to trace r1-derived 5-HT precursor cells; in other words, those 5-HT cells having a history of both En1 and Pet-1 expression (Fig. 1g, blue cells; Fig. 1h–l, red cells). In these same triply transgenic embryos, 5-HT precursors derived from progeny situated caudal to r1 (caudal to En1) expressed enhanced green fluorescent protein (eGFP) as a lineage tracer instead of nβgal, as these cells underwent only Flpe-mediated recombination of RC::Fela (Fig. 1h,i; the “subtractive” green population). No cells were co-labeled with both nβgal and eGFP (Fig. 1h,i), demonstrating that nβgal was not produced at detectable levels from the Flpe/non-Cre-recombined (eGFP⁺) RC::Fela allele; thus, intersectional and subtractive populations can be distinguished unequivocally (Fig. 1h,i).

Upon tracking r1 (En1-cre)-derived 5-HT precursors into mature postnatal stages, we found that they comprise entirely 5-HT nuclei B7, B6, and even B4, despite its relatively caudal position (Fig. 2h and data not shown – red cells). r1 (En1-cre)-derived 5-HT neurons were also found populating ventral pontine 5-HT nuclei (B9, B8 and B5)(Fig. 2g,i,j – red cells), along with contributions from progenitor cells situated caudal to r1(En1 territory). These findings extend the work of Zervas et al¹⁰ and Simon et al¹¹: (1) they reveal an unpredicted developmental and genetic heterogeneity among B9, B8, and B5; (2) they substantiate that the rostral 5-HT system should not only include B9-B5 but also B4 (a point variably represented in the literature); and (3) they support the notion that the requirement for En1 to generate B7 and B6 neurons (and likely all r1 5-HT-descendants) is autonomous to the parental 5-HT progenitor cells and not secondary to alterations in the 5-HT progenitor cell environment within r1.

Upon tracking r2-derived 5-HT precursor cells into postnatal brain, by partnering Rse2::cre⁷ with RC::Fela and ePet::Flpe, we found that they contribute to nuclei B9, B8 and B5 (Fig. 2,l,n,o – red cells), intermingling with non-r2-derived 5-HT neurons (Fig. 2,l,n,o – green cells). Consistent with the r1-derived fate maps, r2 5-HT progenitor cells do not contribute to nuclei B7, B6, and B4 (Fig. 2m and data not shown) – instead these nuclei derive entirely from r1 (Fig. 2h and data not shown). Similar results were found upon incorporation of Egr2::cre (r3/r5)¹² as the cre driver: contribution to nuclei B9, B8 and B5 (Fig. 2q,s,t – red cells), intermingling with r1- and r2-derived 5-HT neurons (Fig. 2q,s,t - green cells), and no contribution to B7, B6, and B4 (Fig. 2r and data not shown). Because the Egr2::cre driver marks as the intersectional (nβgal+) population those 5-HT neurons arising from either r3 or r5, we cannot rule out a contribution from r5 to the B9, B8, and B5 nuclei. Arguing against such an r5 contribution, however, is that the intervening rhombomere, r4, is thought not to produce 5-HT neurons (reviewed in ⁴), creating a gap separating development of the B9-B4 nuclei (the “rostral” 5-HT neural system) from the B3-B1 nuclei (the “caudal” 5-HT neural system). Thus, the simplest interpretation of our r3/r5 5-HT fate map is that the nβgal+ cells (red cells) seen in B9, B8 and B5 (Fig. 2q,s,t) arise from r3 while the nβgal+ cells seen in the nucleus B3 (Fig. 2u) come from r5; the latter cells are dispersed among eGFP+ cells likely deriving from r6-r7. Collectively, the above findings (summarized in Fig. 2a-f) reveal that the distribution of 5-HT neurons derived from distinct rhombomere-defined gene expression domains differs from the anatomically defined groupings of mature 5-HT neurons¹³.

The transcription factor Nkx2.2, expressed throughout the 5-HT primordium, has been reported necessary for the generation of most 5-HT neuronal precursors except a rostral cohort (reviewed in ⁴), suggesting that some rostral 5-HT progenitors, despite expressing Nkx2.2, do not require it for 5-HT neuron production. Towards localizing this Nkx2.2-independent subset of 5-HT progenitor cells, we examined the distribution of 5-HT neurons in the postnatal Nkx2.2^−/− brain stem using our 5-HT fate map as a basis for interpretation. We found that in Nkx2.2^−/− mutants, 5-HT neurons comprising B6 and B7 are present in qualitatively appropriate number and density (Fig. 3d, compared to 3a). Additionally, we found that neurons comprising the B4 nucleus as well as a subset of neurons within B8, B5, and B9 are also present (Fig. 3e, compared to 3b, and data not shown). This distribution is highly similar to that of the r1-derived 5-HT fate map, suggesting that it is largely 5-HT progenitors situated caudal to r1 that require Nkx2.2 activity. Unexpectedly, we also found a cohort of 5-HT neurons in the B1-B3 areas in mutant tissue (Fig. 3f, compared to 3c, and data not shown). Given our fate map showing that these most caudal nuclei do not receive contributions from r1 progenitor cells, it is possible that there exists a subset of caudal progenitors, in addition to those within r1, that are capable of producing 5-HT progeny in the absence of Nkx2.2 activity. Alternatively, in the absence of Nkx2.2, r1-derived 5-HT progenitors may expand their distribution to include aspects of the B1-B3 complex, permitting a skeletal distribution of 5-HT neurons in all nuclei. To distinguish between these possibilities, we mapped in Nkx2.2^−/− mutants the fate of r1 (En1-cre)-derived (nβgal+) and non-r1-derived (eGFP+) 5-HT neurons (Fig. 3j-l). In the vicinity of B1-B3 in mutant tissue, we detected eGFP+ but not nβgal+ cells (Fig. 3l), suggesting that these 5-HT neurons do not arise from r1 but rather there exists a subset of caudal cells that, like r1 5-HT progenitor cells, can produce 5-HT neurons without Nkx2.2 activity. The precise location and molecular signature of these newly identified progenitor cells remains unknown and of great interest.

By developing and applying intersectional and subtractive gene activation technologies, we have shown that 5-HT neuron subtypes with distinct molecular histories and requirements can be isolated in situ. From our findings, we define a novel molecular framework for the 5-HT system. Because this model is based on genetic lineage, we predict it to have functional and pathophysiological relevance. Further, the tools presented here provide a powerful template from which select 5-HT lineages can be manipulated in vivo in virtual isolation – such capabilities open new avenues for considering 5-HT neuron functions and the etiological differences among 5-HT neuron disorders.