Functional and developmental identification of a molecular subtype of brain serotonergic neuron specialized to regulate breathing dynamics

Abstract

Serotonergic neurons modulate behavioral and physiological responses from aggression and anxiety to breathing and thermoregulation. Disorders involving serotonin (5HT) dysregulation are commensurately heterogeneous and numerous. We hypothesized that this breadth in functionality derives in part from a developmentally determined substructure of distinct subtypes of 5HT neurons each specialized to modulate specific behaviors. By manipulating developmentally defined subgroups one by one chemogenetically, we find that the Egr2-Pet1 subgroup is specialized to drive increased ventilation in response to carbon dioxide elevation and acidosis. Furthermore, this subtype exhibits intrinsic chemosensitivity and modality-specific projections-increasing firing during hypercapnic acidosis and selectively projecting to respiratory chemosensory but not motor centers, respectively. These findings show that serotonergic regulation of the respiratory chemoreflex is mediated by a specialized molecular subtype of 5HT neuron harboring unique physiological, biophysical, and hodological properties specified developmentally and demonstrate that the serotonergic system contains specialized modules contributing to its collective functional breadth.

Figure 1. Intersectional alleles RC::FrePe and RC::FPDi for labeling and activity manipulation of subtypes of 5HT neurons

(A) Gt(ROSA)26Sor knock-in alleles RC::FrePe and RC::FPDi offer fluorescent labeling and inducible Di-mediated manipulation of neuron subtypes, respectively, when partnered intersectionally with Flp- and Cre-encoding transgenes. (B) 5HT neurons fluorescently marked with eGFP or mCherry from RC::FrePe, detected via indirect immunofluorescence. (C) Co-immunodetection of serotonergic marker Tph2 and GFP expressed in 5HT neurons using RC::Fe (a Flp-only responsive derivative of RC::FrePe in which the loxP-mCherry-stop-cassette has been removed) partnered with Pet1::Flpe. (D) Endogenous eGFP or mCherry fluorescence in live 5HT neurons in a 200 μm slice suitable for patch-clamp recording (inset - action potentials recorded from fluorescently labeled 5HT neuron). (E) Indirect immunofluorescent detection of eGFP-labeled projections (Pet1::Flpe, RC::Fe tissue) forming the median-forebrain bundle (schematized in G). Scale bar is 300 μm. (F) HA-tag immunodetection of Di in somata of 5HT neurons using derivative allele RC::FDi (Flpe-responsive) partnered with Pet1::Flpe, as compared to J, control single transgenic RC::FDi (no Flpe, thus not expressing Di) sibling tissue. Scale bar is 50 μm in B–D, F and J. (H–I) Schematic coronal sections indicating location of panels B, C, D, F, and J. Abbreviations: DRN – dorsal raphé nucleus, MRN – median raphé nucleus. See also Figure S1.

Figure 2. Molecular and developmental subtypes of 5HT neurons accessed by intersectional genetics

(A) The embryonic hindbrain is segmented along the anterior-posterior (AP) axis into rhombomeres (r1–r7), each expressing distinct genes; En1 expressed in r1 (light blue), HoxA2 in r2 (orange), and Egr2 in r3 and r5 (dark and light pink, respectively). Serotonergic progenitors are situated ventral in the hindbrain ventricular zone and span the AP axis of the developing hindbrain, giving rise to post-mitotic 5HT precursor cells that express Pet1 (Hendricks et al., 1999) (dark blue) in all but r4. (B) Recombinase driver lines use gene enhancer/promoter elements to express Flpe and cre in specific domains, for example En1-cre (yellow) in r1 and Pet1::Flpe (cyan) in 5HT precursors. (C) Pairing recombinase drivers En1-cre and Pet1::Flpe with RC::FrePe (see Figure 1A) switches on eGFP in r1-derived 5HT precursors and mCherry in all other 5HT precursors. (D–G) The entire Pet1::Flpe expressing 5HT neuron population is labeled with eGFP upon pairing with RC::Fe. (D) Sagittal brainstem schematic shows approximate rostrocaudal levels of E, F, and G. (H–V) Adult fate-mapped 5HT neurons labeled with eGFP or mCherry in triple transgenics. (J, O, T) DRN and MRN showing the intersectional subtype of 5HT neuron labeled with eGFP+ (green), and remaining 5HT neurons with mCherry (red); detection by indirect immunofluorescence. Higher magnification of boxed regions in I, N, S. (K, P, U) Raphé magnus (RMg) and rostral raphé pallidus (RPa). (L, Q, V) Raphé obscurus (Rob), caudal RPa and lateral paragigantocellularis (LPGC). Scale bars is 300 μm in E, F, G and applies to all images in a column. Scale bar is 100 μm in I and inset of U.

Figure 3. Egr2-Pet1 5HT neurons specialize in regulating breathing in response to tissue CO₂ elevation and acidosis in mice

(A) Paradigm for plethysmographic assessment at 34 C of respiratory responses to 5% inspired CO₂ during quiet wakefulness in the daytime before and during CNO application (10mg/kg i.p. injection); boxes (a, b, c, d) represent stretches analyzed from continuous recordings. (B) Pet1::Flpe; RC::FDi mice (left panel) show reduced minute ventilation (V̇_E) responses to 5% CO₂ during as compared to before CNO administration (significant interaction between 5% CO₂ and CNO, p < 0.001), contrasting littermate controls (right panel) which have indistinguishable V̇_E responses to 5% CO₂ under either CNO conditions. V̇_E responses to 5% CO₂ in triple transgenics En1-cre, Pet1::Flpe, RC::FPDi (C, left panel) and HoxA2::cre, Pet1::Flpe, RC::FPDi (D, left panel) and littermate controls (right panels) are indistinguishable in the presence or absence of CNO. Triple transgenic Egr2-cre, Pet1::Flpe, RC::FPDi mice (E, right panel) showed reduced V̇_E responses to 5% CO₂ upon CNO administration as compared to pre-CNO baselines (significant interaction between 5% CO₂ and CNO, p < 0.001); littermate controls (right panel) showed no change post-CNO. Data shown as mean ± s.e.m. and statistically analyzed using 2-way RM ANOVA followed by Tukey post-hoc analysis. Brainstem schematics show the 5HT neuron subtype (green outline) expressing Di (gray) and remaining mCherry-expressing 5HT neurons (red).

Figure 4. Cellular chemosensitivity to hypercapnic acidosis maps to the Egr2-Pet1 subtype of 5HT neuron

(A) Egr2-Pet1 (green) and r6/7-Pet1 (red) subtypes of 5HT neurons intersperse within the RMg, as revealed by eGFP and mCherry detection in brainstem sections from Egr2-cre, Pet1::Flpe, RC::FrePe mice. (B) Live Egr2-cre, Pet1::Flpe, RC::FrePe brainstem slice showing eGFP and mCherry fluorescence suitable for neuron subtype identification prior to patch-clamp recording. Right arrowhead, eGFP-labeled Egr2-Pet1 neuron, shown at higher magnification in C, and its firing rate in relation to pH (C′), indicating chemosensitivity. Left arrowhead, mCherry-labeled r6/7-Pet1 neuron, shown again in D, and its firing rate in relation to pH (D′), indicating lack of chemosensitivity under these conditions. Scale bars are 100 μm. (E–F) Percent chemosensitive neurons of each subtype recorded from Egr2-cre, Pet1::Flpe, RC::FrePe slices ages P14–P18 (E) or P23–P26 (F). Number of chemosensitive neurons over total neurons recorded of that subtype shown in parentheses over each bar. ***P<0.0001, **P<0.002 (Fisher’s Exact Test). Insets show anatomical breakdown of neurons from the r6/7-Pet1 subtype; note, all Egr2-Pet1 neurons reside within the RMg. (G–H) Chemosensitivity index of neurons recorded at P14–P18 (Egr2-Pet1 n=29, r6/7-Pet1 n=31, ***P<0.0001) and P23–P26 (Egr2-Pet1 n=8, r6/7-Pet1 n=17, **P=0.0002). Data points for neurons that fit the criteria of chemosensitive outlined in black. (I–J) Spontaneous firing rates (no current injection) of neurons at P14–P18 (Egr2-Pet1 n=29, r6/7-Pet1 n=29, ***P<0.0001), and P23–P26 (Egr2-Pet1 n=8, r6/7-Pet1 n=16, *P = 0.0107). (K–L) Average change in firing rate measured over pH shifts, **P=0.0005 at P14–18 and **P=0.0009 at P23–25. Two-tailed unpaired t-test was used for statistical analysis in (G–L); error bars show mean ± s.e.m. (M–N) Average change in firing rate versus chemosensitivity index. Dotted line demarcates chemosensitivity index of 120. P<0.0001 at P14–18 and P23–25 (Pearson correlation).

Figure 5. Egr2-Pet1 5HT neurons project to respiratory nuclei implicated in PCO₂/pH sensory signal transduction and integration

Representative (at least 3 mice of each genotype examined) confocal images of axonal projections (eGFP+, green, detected by indirect immunofluorescence) from either all Pet1::Flpe-expressing neurons (Pet1::Flpe, RC::Fe mice A–E, K, L) or from Egr2-Pet1 neurons (Egr2-cre, Pet1::Flpe, RC::FrePe mice F–J, M, N). Brainstem nuclei (indirect immunofluorescence detection of cell identity markers in red) implicated in the processing of sensory input related to PCO₂/pH levels: (A, F) Locus coeruleus (LC) neuron cell bodies visualized via immunodetection of tyrosine hydroxylase (TH), under conditions which highlight soma as opposed to dendrites. Note that pericoerulear regions, such as the lateral pLC (pLCl) shown here, also referred to as the parabrachial nucleus medial (PBNm), harbor extensive dendrites from LC neurons and are significant sites for synaptic input to LC neurons; (B, G) Higher magnification view of the areas boxed in A and F, respectively. Arrowheads highlight examples of puncta from eGFP-labeled processes. (C, H) Retrotrapezoid nucleus (RTN), identified via Phox2b immunodetection coupled with anatomical location; (D, I) Nucleus of the solitary tract (NTS), TH; (E, J) PreBötzinger complex (PBC), neurokinin1 receptor; (K, M) the C1 nucleus, TH; and (L, N) the spinal dorsal horn (DH). (O) Hindbrain schematic depicting 5HT raphé (gray) and Egr2-Pet1 region (green), among brainstem/spinal cord regions examined for projections (orange). Scale bars are 100 μm in all images except B, G where scale bar is 50 μm.

Figure 6. Egr2-Pet1 5HT neurons do not project to primary respiratory motor nuclei

Representative (at least 3 mice of each genotype were examined) confocal images of axonal projections (eGFP+, green, detected by indirect immunofluorescence) from either all Pet1::Flpe-expressing neurons, (Pet1::Flpe, RC::Fe) A–C, G, or from Egr2-Pet1 neurons, (Egr2-cre, Pet1::Flpe, RC::FrePe) D–F, H. Brainstem motor nuclei involved in respiratory motor output: (A,D) the hypoglossal nucleus (12N) stained for choline acetyltransferase (ChAT); (B, E) the nucleus ambiguus (NA), ChAT; and (C, F) the ventral spinal motor neurons (VMN); ChAT. (G, H) Projections to the intermediolateral cell column of the spinal cord (IML), ChAT. (I) Hindbrain schematic depicting 5HT raphé (gray) and Egr2-Pet1 region (green), among the brainstem/spinal cord nuclei examined for projections (dark blue). Scale bars are 100 μm in A and D, and 50 μm in B, C and E–H.

Figure 7. The Egr2-Pet1 subtype of 5HT neuron is specialized to modulate the respiratory CO₂ chemoreflex via selective projections to respiratory chemosensory processing centers and the ability to sense and transduce PCO₂/pH changes

Sagittal brain schematic of serotonergic raphé with the Egr2-Pet1 subtype shown in green, the r6/7-Pet1 subtype in red, and other 5HT subtypes in shades of gray. Inventory of Supplemental Information