The Paraventricular Hypothalamus Regulates Satiety and Prevents Obesity via Two Genetically Distinct Circuits

Abstract

SIM1-expressing paraventricular hypothalamus (PVH) neurons are key regulators of energy balance. Within the PVH^SIM1 population, melanocortin-4 receptor-expressing (PVH^MC4R) neurons are known to regulate satiety and bodyweight, yet they account for only half of PVH^SIM1 neuron-mediated regulation. Here we report that PVH prodynorphin-expressing (PVH^PDYN) neurons, which notably lack MC4Rs, function independently and additively with PVH^MC4R neurons to account for the totality of PVH^SIM1 neuron-mediated satiety. Moreover, PVH^PDYN neurons are necessary for prevention of obesity in an independent but equipotent manner to PVH^MC4R neurons. While PVH^PDYN and PVH^MC4R neurons both project to the parabrachial complex (PB), they synaptically engage distinct efferent nodes, the pre-locus coeruleus (pLC), and central lateral parabrachial nucleus (cLPBN), respectively. PB-projecting PVH^PDYN neurons, like PVH^MC4R neurons, receive input from interoceptive ARC^AgRP neurons, respond to caloric state, and are sufficient and necessary to control food intake. This expands the CNS satiety circuitry to include two non-overlapping PVH to hindbrain circuits.

Figure 1.. PVH^PDYN neurons are a MC4R-negative satiety population that accounts for the missing 50% of PVH^SIM1 neuron-mediated satiety

(A) Schematic of Mc4r-T2A-Cre::Ai9-Rosa26-LSL-TdTom::Pdyn-EGFP mouse. (B) Representative histological sections of the PVH from rostral to caudal (MC4R::TdTOM=red, PDYN-EGFP=green, double labeled=yellow) (C) Quantification MC4R::TdTOM+, PDYN-EGFP+ vs. double labeled neurons summed across the rostra-caudal extent of the PVH (n=4). (D) Schematic of AAV-DIO-hM4Di injection in the PVH of Sim1-IRES-Cre, Mc4r-T2A-Cre, Pdyn-IRES-Cre, or compound Pdyn-IRES-Cre::Mc4r-T2A-Cre mice. (E) Representative images of hM4Di-mCherry expression in PVH^SIM1, PVH^MC4R, PVH^PDYN, and PVH^PDYN::MC4R neurons in the PVH. (F) Light-cycle cumulative food intake in sated mice following saline vs. CNO/hM4Di inhibition of PVHSIM1, PVHMC4R, PVHPDYN and PVHPDYN::MC4R neurons. PVH^SIM1 n=6: Repeated Measures (RM) 2 way ANOVA Time F(3,15)=87.43 ****P<0.0001, Treatment F(1,5)=243.8 ****P<0.00001, Interaction F(3,15)=57.22. Sidak’s Multiple Comparisons: Saline vs. CNO ****P<0.00001 at 1,2,3 hours. PVH^MC4R n=7: RM 2 way ANOVA Time F(3,18)=23.47 ****P<0.0001. Treatment F(1,6)=109.4 ****P<0.0001, Interaction F(3,18)=23.01, ****P<0.0001, Sidak’s Multiple Comparisons: Saline vs. CNO **P<0.01 at hour 1, ****P<0.0001 at hour 2,3. PVH^PDYN n=5: RM 2 way ANOVA Time F (3,12)=26.03 ****P<0.0001, Treatment F(1,4)=50.07 ***P<0.001, Interaction F(3,12)=34.47 ****P<0.0001. Sidak’s Multiple Comparisons: Saline vs. CNO ****P<0.0001 at hours 1,2,3. PVH^PDYN::MC4R n=5: RM 2 way ANOVA Time F(3,12)=135. 4 ****P<0.0001, Treatment F(1,4)=245.9 *****P<0.0001, Interaction F(3,12)=88.51 ****P<0.0001, Sidak’s Multiple Comparisons: Saline vs. CNO ****P<0.0001 at hours 1,2,3. (G) Same data at the 3 hour time point quantified as change in food intake (CNO-Saline). ANOVA F(3, 18) = 25.87 ****P<0.0001, Tukey’s Multiple Comparisons between groups: PVH^SIM1 vs. PVH^{MC4R ****}P<0.0001, PVH^SIM1 vs. PVH^PDYN ****P>0.0001, PVHMC4R vs. PVHPDYN:MC4R ***P<0.001, PVHPDYN vs. **PVHPDYN::MC4R P<0.01, PVH^PDYN::MC4R vs. PVH^SIM1 N.S., PVH^MC4R vs. PVH^PDYN NS. Data are represented as mean ± SEM.

Figure 2.. PVH^PDYN neurons prevent obesity independent of and additive with PVH^MC4R neurons.

(A) Schematic of AAVDJ-CMV-DIO-eGFP-2A-TeNT (Tetanus Toxin) injection in the PVH of Mc4r-T2A-Cre, Pdyn-IRES-Cre, or compound PDYN-IRES-Cre::Mc4r-T2A-Cre mice and experimental timeline. (B) Representative histological images from of eGFP2A-TeNT expression in PVH^PDYN, PVH^MC4R, and PVH^PDYN::MC4R neurons in the PVH (C) Average daily food intake of animals of each genotype in the week prior to injection of AAV vs. after 4 weeks of synaptic silencing with AAV-DIO-TeNT. GFP n=6, MC4R n=7, PDYN n=11, PDYN::MC4R n=5. Repeated Measures (RM) 2Way ANOVA: Time Pre or post AAV injection: F (1, 20) = 41.28, ****P<0.0001, Sidak’s post hoc comparisons pre vs. post injection: GFP N.S., MC4R **P<0.01, PDYN **P<0.01 PDYN::MC4R **P<0.01. (D) Body weight of the same animals prior to injection of AAV and after injection, monitored biweekly for 4.5 weeks. Repeated Measures (RM) 2Way ANOVA: Time F (11, 297) = 149.45, ****P<0.0001. Group F(3,27)=6.816, **P<0.01. Interaction (F33,297)=14.83 ****P<0.0001. Sidak’s post hoc comparisons for each group: BW at X week post injection vs. baseline BW: GFP N.S. at all time points, MC4R ****P<0.0001 from1.5 weeks onwards, PDYN ****P<0.0001 from 1.5 weeks onward, PDYN::MC4R ****P<0.0001 from 1 week onward. Sidak’s post hoc comparisons between genotype groups at 4.5 weeks: GFP vs. MC4R, PDYN or MC4R::PDYN ****<P<0.0001, MC4R or PDYN vs. PDYN::MC4R ****P<0.0001, PDYN vs MC4R N.S. Data are represented as mean ± SEM.

Figure 3.. PVH^PDYN neurons and PVH^MC4R neurons share a similar projection profile in the PB, but synaptically engage different effector nodes: pLC vs. cLPBN

(A) Schematic of AAV-DIO-Syp-mCherry injections in the PVH of Pdyn-IRES-Cre, Mc4r-T2A-Cre or Sim1-IRES-Cre for anterograde tracing. (B-D) Axonal projection profiles of above PVH satiety populations in the parabrachial complex (PB) from rostral to caudal (Syp-mCherry=red co-labeled with immunohistochemistry for choline-acetyltransferase (ChAT) =grey and tyrosine hydroxylase (TH) = green). TH marks locus coeruleus (LC) neurons while TH and ChAT mark the transition from cLPBN to eLPBN. (E) Schematic of CRACM experiments to assay connectivity between PVH^PDYN, PVH^MC4R or PVH^SIM1 neurons and downstream neurons in the pLC vs. cLPBN compartments of the PB. (F-K) Schematic and representative monosynaptic light-evoked excitatory current traces and connectivity from (F-G) PVH^PDYN neurons to the pLC (n=4 mice) and to cLPBN (n=6 mice) (H-I) PVH^MC4R neurons to pLC (n=3 mice), cLPBN (n=3 mice) and (J-K) PVH^SIM1 neurons to pLC (n=3 mice) and cLPBN (n=3 mice) (L) Summary of the connectivity of PVH^MC4R vs. PVH^PDYN neurons to pLC or cLPBN compared to connectivity of PVH^SIM1 neurons in both pLC and cLPBN. Data are represented as mean ± SEM.

Figure 4.. PB-projecting PVH^PDYN neurons respond to caloric state and receive input from ARC^AgRP neurons

(A) Schematic for FOS assay of neuronal activity in PB-projecting PVH^PDYN neurons in different caloric states. CTB retrolabeling of PB-projecting PVH^PDYN neurons in Pdyn-IRES-Cre::Ai9-Rosa26-LSL-TdTom mice and experimental paradigm to manipulate caloric state. (B) Representative images of the PVH from rostral to caudal (PDYN::TdTOM+ neurons=red, retrolabeled with immunohistochemistry for CTB CTB=grey and FOS=green. Insets show high-magnification images corresponding to boxed region. (C) Quantification of the % of FOS+ neurons out of all PB-projecting PVH^PDYN across the rostral-caudal extent of the PVH n= 4–5 per group, RM 2 way ANOVA, PVH location and interaction= NS. Caloric state: F(1,7)=601.8, *****P<0.0001, Sidak’s Multiple Comparisons: Fasted vs. Refed at 3 locations within PVH AP-0.60 *P<0.05, AP-0.90 ***P<0.001, AP-1.20**P<0.01. (D) Schematic of CRACM experiment assaying connectivity of ARC^AgRP neurons to PVH^PDYN neurons in Agrp-IRES-Cre::PDYN-EGFP mice with circuit specific CTB-555 retrolabeling from the entire PB, the pLC only or the LPBN only. (E) Representative images of PDYN-GFP expressing neurons retrolabeled with CTB-555 across the rostral-caudal extent of the PVH (PDYN-EGFP=green, CTB-555=red, double-labeled=yellow) (F-I) Schematic and representative monosynaptic light-evoked inhibitory current traces and connectivity between ARC^AgRP neurons and downstream PVH^PDYN neurons with or without circuit specificity: (F) ARC^AgRP➔randomly selected PVH^PDYN neurons (n=3 mice) (G) ARC^AgRP➔PB-projecting PVH^PDYN neurons (n=2 mice) (H) ARC^AgRP ➔pLC-projecting PVH^PDYN neurons (n=2 mice) (I) ARC^AgRP ➔LPBN-projecting PVH^PDYN neurons (n=1 mouse) (J) Summary of above CRACM connectivity data from ARC^AgRP ➔PVH^PDYN neurons with or without circuit specificity. Data are represented as mean ± SEM.

Figure 5.. PB-projecting PVH^PDYN neurons are both sufficient and necessary for satiety, likely via downstream effector node pLC, a novel satiety-regulating site

(A) Schematic of optogenetic terminal stimulation of the PVH^PDYN➔PB circuit: AAV-DIO-ChR2-mCherry injection into the PVH of Pdyn-IRES-Cre mice and optical fiber delivery of 463nm light in the PB (B) Representative image of ChR2-mCherry expression in PVH^PDYN neurons and their axon projection field in the PB with optical fiber placement over the PB. (C) Dark-cycle cumulative food intake over 3 hours during light off vs. light on terminal stimulation trials n=10 mice, RM 2 Way ANOVA Time F(3, 27)= 70.53 ****P<0.0001, Treatment F(1, 9)=70.53 ****P<0.0001, Interaction F(3, 27)=29.92 ****P<0.0001, Sidak’s Multiple Comparisons: Saline vs. CNO **P<0.01 at hour 1, ****P<0.0001 at hours 2 and 3. (D) Schematic of optogenetic terminal inhibition of PVH^PDYN PB: AAV-DIO-ArchT-GFP injection into the PVH of Pdyn-IRES-Cre mice and optical fiber delivery of 532nm light in the PB (E) Representative image of ArchT-GFP expression in PVH^PDYN neurons and their axon projection field in the PB with optical fiber placement over the PB (F) Light-cycle cumulative food intake over 3 hours during light off vs. light on terminal inhibition trials n=6, RM 2 way ANOVA Time F(3,15)= 90.23 ****P<0.0001, Treatment F(1, 5)= 90.23 ****P<0.0001, Interaction F(3,15) =21.88 ****P<0.0001. Sidak’s multiple comparisons: Saline vs. CNO ***P<0.001 at 1,2 hours, ****P<0.0001 at hour 3. (G) Schematic of AAV-DIO-Syp-mCherry guided fasted vs. refed FOS assay in the pLC projection field of PVH^PDYN neurons (H) Representative images of FOS+ neurons (green) in the pLC axon projection field of PVH^PDYN neurons (red) co-labeled with TH (grey) in fasted vs. refed states at low (top) and high (bottom) magnification (I) Quantification of the number of FOS+ cells in the pLC of mice sacrificed in the fasted vs. refed states across the rostral-caudal extent of the PVH^PDYN pLC projection field n=5 per group. RM 2 way ANOVA Bregma level and Interaction NS, Caloric state: F(1,8)=299.9 ****P<0.0001, Sidak’s multiple comparisons: Fasted vs. Refed at each level of the pLC ****P<0.0001. (J) Schematic of chemogenetic inhibition experiment of the pLC with injection of AAV-hM4Di-mCherry in the pLC of wild type mice. (K) Representative image of GROUP 1 animals: histological cases with hM4Di-mCherry expression in the pLC (with spill over in i,d,vLPBN) (L) GROUP 1: Light-cycle cumulative food intake after saline vs. CNO treatment of animals with the above histology n=5, Time F(3,12)= 58.76 ****P<0.0001, Treatment F(1,4) =72.75 ** P<0.01, Interaction F(3,12)= 28.54 ****P<0.0001. Sidaks Multiple Comparisons: Saline vs. CNO: *P<0.05 at 1 hour, ****P<0.0001 at 2,3 hours (M) Representative image of GROUP 2 animals: histological cases with hM4Di-mCherry in i,d,vLPBN expression only, but not pLC (N) GROUP2: Light-cycle cumulative food intake after Saline vs. CNO treatment of animals with this histology n=4, RM 2 way ANOVA F(1,3) NS. Data are represented as mean ± SEM.