In this work, we showed that specific inactivation of Meis2 TF in Isl1-expressing cells in mice severely impaired inhibitory baroreflex function independently of any developmental cardiac malformations or contractility defects of the heart and cardio-vascular system. In addition, the Meis2 expression in subclasses of vagal neurons that we and others reported and that are predicted to have proprioceptive and mechanosensitive properties46, together with the recent demonstration that Piezo2-positive vagal neurons are essential for the cardiac baroreflex42, strongly suggest that Meis2 inactivation in those neurons could be responsible for the blunted inhibitory cardiac reflex we report. In this scenario, Meis2-expressing mechanosensitive neurons, including those from the JNG and possibly the DRG whose function is to sense stretch induced by artery and/or heart deformations fail to properly encode the information necessary to trigger the normal inhibitory baroreflex feedback. Thus, our model reinforces current hypotheses on alterations of primary sensory neurons function in ASD disorder47, and underline the importance of conditionally targeted mouse models to disentangle intermingled and complex phenotypes found in human mutants.
The baroreflex is a classical and complex mechanism that coordinates adaptive cardio-vascular tone through both autonomic and sensory components48. Elevated blood pressure promptly triggers a compensatory decrease in cardiovascular output to maintain body and brain blood pressure within homeostatic ranges49. There is no real consensus about the sensory neuron subtypes involved. They are commonly called baroreceptors, display mechanosensitive properties and project to precise locations on arteries where they sense arterial wall distortion. This arterial baroreceptor reflex system plays a dominant role in preventing short-term wide fluctuations of arterial blood pressure, as recurrently demonstrated in experiment where arterial baroreceptor denervation leads to an increase of the beat to beat variability of blood pressure and related heart rate50.
The baroreflex is associated with some pathological conditions48,51, but only recently, imbalance of cardiac autonomic regulation in patients with intellectual disabilities and ASD is emerging7–19. However, the origins of dysautonomia in ASD is still unclear and somehow controversial with highly variable profiles depending on the studies. Many studies report that ASD patients present a higher heart rate, and that exposure to external stimuli leads to a blunted heart rate response compared to healthy subjects8. HR is increased in ASD patients compared to control due to a lower parasympathetic activity17,18, but other reports revealed on the contrary an increased parasympathetic activity15. Moreover, intermittent neuro-cardiovascular autonomic dysfunction affecting heart rate and blood pressure was also over-represented in ASD52,53.
Interpretation of results in human patients has proven complicated due to the genetic variability causing the different syndromes and the combinatory effect of multiple affected organs other than the nervous system. In most investigations related to cardiac autonomic regulation and HRV analysis in ASD, patients are rigorously matched in age and gender but cohorts usually do not take into account the genetic basis of the diagnosed ASD. Indeed, the large number of neurodevelopmental genes supporting ASD symptoms, but also the variability of the symptoms accompanying different mutations within the same gene could account for discrepancies between studies. A good example linking gene dosage effect to the severity of phenotypic manifestation is Rett syndrome. Rett syndrome is associated with MECP2 gene mutations, but the type of mutation, i.e. loss-of-function, gene duplication or triplication, and the degree of mosaicism for these mutations within cell types lead to highly heterogeneous phenotypic manifestations and clinical presentation ranging from microcephaly to normal brain size, shortened lifespan or not54. Nevertheless, studies have shown modified autonomic function both in children and adult ASD patients overall characterized by a lower autonomic nervous activity than healthy subjects.
The genetic links between ASD and congenital heart malformation in humans also prevent unmasking deleterious effects on cardiac autonomic regulation in ASD full knockout mouse models55. Recent advances in the understanding of the biology of the MEIS family of TFs and their well-known partners PBX members emphasized their essential contribution to cardiac morphogenesis and physiology. In humans, non-synonymous variants for PBX1, PBX2, PBX3, MEIS1 and MEIS3 have been identified in patients with congenital cardiac defects56, and humans carrying MEIS2 mutations present cardiac septal defects27,28. Similar phenotypes are also described in full-knockout models for those genes33,34. In mouse, genetic ablation of Pbx1-3 at specific developmental stages lead to heart malformations34. Pbx1 deficiencies results in persistent truncus arteriosus, whereas Pbx2 and 3 inactivation leads to Pbx1 haploinsufficiency with overriding aorta, ventricular septal defect, and bicuspid aortic valves34. Meis1 and Meis2 mutant mice also exhibit cardio-vascular and septal defects31,33,34,57,58.
Surprisingly, our Doppler-echocardiography investigations did not reveal any heart morphological or contractile defects. This might be due to a later Meis2 inactivation in cardiac neural crest compared to the AP2α-IRES-Cre strain used by others33, at a time when Meis2 is no longer required. Given that both genes are involved in heart morphogenesis, this might also result from redundant Meis1 and Meis2 activities within the timeframe of our genetic ablation. Nonetheless, we previously showed that Meis1 ablation-induced septal defect depends on the CRE strain used for neural crest gene ablation. Meis1 inactivation in early neural crest resulted in septal defects, but Meis1 inactivation in late neural crest did not produce contractile and morphological defect31.
We also showed that at baseline conditions, Meis2 mutant mice do not present symptomatic heart rhythm disturbance such as major sinus pause or arrest or atrial/ventricular ectopic beats. Instead, a large variability in sinus rhythm confirmed by high HRV, without brady- or tachycardia, was observed. A profound sinus node dysfunction in Meis2 mutant is thus unlikely. We further demonstrated that the large beat-to-beat variability in Meis2 mutant mice results from a dysregulation of the sensory-autonomic control of cardiac rhythm. We identified a lower sympatho-vagal activity at baseline reflected by the decrease LF/HF ratio that could also explain the low mean arterial pressure observed in mutant mice. When using drugs that rapidly and robustly modify blood arterial pressure, we unmasked a sensory-autonomic dysregulation characterized by a blunted cardio-inhibitory reflex. Surprisingly, only the cardio-inhibitory baroreflex was affected, but the sympathetic activation following a fall in blood pressure was maintained although both vasoconstriction and vasodilation could be pharmacologically elicited.
According to the vagal dominance in the beat-to-beat baroreflex adjustment of HR and blood pressure, and the increased variability during baroreceptor denervation reported in several animal models50, we suggest a defective baroreceptor related-vagal pathway induced by Meis2 inactivation. Moreover, because Meis2 is not expressed in sympathetic neurons31, along with the observation that basal mean heart rate is unaffected and the cardio stimulatory reflex seems to be unaltered, we can exclude that sympathetic nerves were affected by Meis2 deletion.
Instead, we conclude that Meis2 inactivation interferes with the sensory component of the vagal-mediated baroreflex. First, because of the possible Meis2 recombination sites following CRE activity when using the Isl1CRE strain. The LIM-homeodomain TF Isl1 is expressed by several neural and non-neural tissues both during embryonic development and postnatal life amongst which the peripheral and central nervous systems, the pancreas, the heart and the pituitary gland59–63. However, interaction of tissues other than the nervous system with the baroreflex is unlikely. Beside sensory and autonomic peripheral neurons, specific CNS neuronal populations express Isl1 including spinal motor neurons, retinal ganglion cells, hypothalamic, central amygdala and striatal neurons30,64−70. We therefore cannot fully exclude that Meis2 recombination also occurs in some of these central neuronal populations that somehow participate in the autonomic imbalance we report in Isl1+/CRE::Meis2LoxP/LoxP mice.
Secondly, the large literature linking sensory neurons to baroreflex and our recent finding that Meis2 is necessary to normal functioning of peripheral mechanosensitive neurons place Meis2-expressing peripheral sensory neurons in best position to support the lack of cardio-inhibitory reflexes in these mice. Beside the autonomic system that includes parasympathetic and sympathetic efferents and control heart and blood vessels contractility, peripheral sensory innervation of the heart is of dual origin71,72. Anatomically, sensory fibers originate from vagal neurons located in the jugular-nodose complex and run through the vagus and the inferior cardia nerve72. Afferent sensory fibers sense local target organs activities such as tissue tension and send the information to higher brain structures in order to elaborate an adapted response. Although most afferent and efferent information to the heart navigates through the vagus nerve46,49, there is evidence that DRG sensory neurons are also involved, in particular for cardiovascular reflexes72–74. Retrotracing experiments in cats, dogs and rats injected at different locations in the heart, coronary artery or in the inferior cardiac nerve labelled neurons in the DRG indicating that heart and arteries also receive afferent sensory fibers from the DRG75–80. In addition, molecular characterization of these neurons revealed that they express a range of markers compatible with the identity of several subclasses of DRG sensory neurons79,81,82.
Vagal neurons have been studied for very long time, but the knowledge and understanding of the precise identities and physiological functions of the different subpopulations of vagal sensory neurons remain fragmented mainly because of the lack of molecular knowledge and tools to specifically target them. Nerve sectioning experiments combined with the mixed nature of the vagus nerve also impedes full interpretation on their precise function. Cranial ganglia contributing to the vagal nerves are multiple and arise from the neural crest derived jugular ganglia and the placode-derived nodose and petrose complex that eventually merge during embryogenesis83. The molecular characteristics of these primary sensory neurons have only been very recently elucidated and showed that nodose and jugular neurons are molecularly fundamentally different with jugular neurons sharing many features with somatosensory DRG neurons46. In this scRNAseq study46, Meis2 was detected in 2 of the 18 nodose neuron clusters, and in 4 of the 6 jugular neurons clusters with relatively high expression in clusters displaying a molecular profile similar to myelinated DRG neurons involved in gentle touch. Functional classification of nodose clusters predicted Meis2 expressing populations to have DRG proprioceptive-like features. Interestingly, in this database (https://ernforsgroup.shinyapps.io/vagalsensoryneurons/), the mechano-sensitive Piezo2 channel recently shown to be involved in baroreflex42 was coexpressed in all Meis2-expressing clusters (Supplementary Fig. 2). Thus, mechanosensitive neurons of the DRG and of the Jugular-Nodose complex are molecularly highly similar.
In our mouse model, we could not evidence any neuronal loss of vagal neurons of the JNG, suggesting that Meis2 inactivation does not affect neuronal survival or identity as demonstrated by the normal expression of Ntrk2, Ntrk3 and Ret. These results are in line with our previous studies on Meis1 or Meis2 inactivation in different types of peripheral neurons31,36. When Meis1 is specifically inactivated in sympathetic neurons, distal innervation of target organs, including the heart, is compromised, but early sympathetic specification is unaffected31. More strikingly, using the very same mouse strain as in the present study, we found that mechanosensitive neurons of the DRG that normally express Meis2 failed to fully differentiate and to elaborate complex distal peripheral sensory terminals mediating touch sensation in the skin36. In both mouse models, these peripheral innervation defects result in physiological consequences: mice lacking Meis1 in sympathetic neurons display severe chronotropic incompetence due to sympathetic dysfunction and mice lacking Meis2 in DRG mechanosensitive neurons have impaired touch sensations. Using an another conditional Meis2 strain, Machon et al. inactivated Meis2 in the neural crest, including the neural crest-derived cranial sensory ganglia encompassing trigeminal (V), facial (VII) and vestibulocochlear (acoustic) nerves (VIII)33. Although the authors did not thoroughly detail their findings, most neural crest-derived cranial ganglia were reported to be present, but nerves exiting the ganglia seemed less numerous and less ramified than in WT embryos as seen by whole mount neurofilament staining. However, in this study, the physiological consequences have not been investigated.
To conclude, although we could not unambiguously demonstrate that Meis2 expressing vagal and/or DRG neurons are directly responsible for the blunted autonomic response and the lack of baroreflex in Isl1+/CRE::Meis2LoxP/LoxP mice, our study clearly showed that our genetically modified animal model is a very appropriate tool to study autonomic dysregulation independently of cardiac remodeling.