The Vagal Autonomic Pathway of COVID-19 at the Crossroad of Alzheimer’s Disease and Aging: A Review of Knowledge

The pathophysiology of COVID-19

COVID-19 disease progression comprises three distinct but overlapping phases: an early infection phase, a pulmonary phase, and a severe hyper-infl-ammation phase [13, 18].

In the early infection phase, the virus infiltr-ates the upper airway tract, lung parenchyma, eyes, and gut. From a recently published exten-sive survey of literature [13], the first symptoms appear in average 5.2 days after initial cont-amination and encompass fever, cough, dysp-nea, fatigue, nausea, anosmia, and ageusia. This initial phase is associated with innate immu-nity response, which is marked by an increased plasma concentration of C-Reactive Protein (CRP) and activation of monocytes and mac-rophages, along with a rise of plasmatic amy-loid-A protein [13]. Activated monocytes and macrophages have been indeed detected in bro-nchial and alveolar epithelia by immunohis-tochemistry after ex-vivo infection of human biopsies with SARS-CoV-2 [19].

This late phase involves a “cytokine storm” (also called macrophage activation syndrome), as observed in various infectious and non-infectious diseases, i.e., an excessive inflammatory response caused by a dysregulated host immune response [25]. The COVID-19 cytokine storm is marked in the systemic circulation by rises of CRP and of activated monocytes-macrophages and affects multiple organs [26]. The mechanism underlying cytokine storm is the failure of inflammatory response regulation back to homeostasis, which normally results from the “inflammatory reflex” defined by Tracey as “a neural loop in which afferent inflammatory signals activate an opposing motor response of the vagus nerve that suppresses cytokine production in order to limit or prevent damage” [27].

Beyond the lung, SARS-Cov-2 displays tropism for the vagus nerve and its centers

The initial event of COVID-19 is an infection of human cells by SARS-Cov-2 via stereospecific binding of its envelope S protein (SPIKE) to Angiote-nsin-Converting Enzyme-2 (ACE2) at the cell membrane and subsequent internalization of the complex [28], which occurs initially in the upper airway and the lung. ACE2 expression has indeed been dem-onstrated in all cell types of the human airway epithelium by tissue- and single-cell RNA-seque-ncing from bronchoscopy biopsies of 267 subjects [29]. The viral infection has been observed by applying SARS-CoV-2 particles from nasopharyngeal aspirate of a human COVID-19 patient, on ex-vivo cultures of respiratory tract mucosae and in-vitro cultures of alveolar epithelial cells from healthy human donors [19]. In human biopsies, SARS-CoV-2 has been detected by immunohistochemistry in all differentiated cell types of the airway mucosa and of the alveolar epithelium, aging being related with increased ACE2 protein expression and decreased apoptosis effectors [30]. Its replication period is 2–4 days in both airway- and alveolar tissues [19]. The lung-originating infection, propagation, and viral multiplication rate have been confirmed and detailed in a transgenic mouse overexpressing the human ACE2 gene [31]. Importantly, in the human olfactory epithelium, ACE2 was recently revealed by immunohistofluorescence to be localized exclusively in the non-neuronal sustentacular cells, and not in the centrally projecting olfactory sensory neurons [32]. Consistently, single-cell transcriptomal analysis of the murine nasal mucosa lately revealed that SARS-CoV-2 entry genes are expressed exclusively in non-neuronal cells of the olfactory epithelium [33]. These observations render unlikely the nose-to-brain route of SARS-Cov-2 neuroinvasion that has been repetitively postulated since the beginning of COVID-19 pandemy.

ACE2 is ubiquitously expressed in numerous other tissues, encompassing about 150 different cell types [34]. Quantitative comparisons with transcriptomic assays and or comparative immunohistochemistry have all established that the highest ACE2 expressions are found in intestine, followed by testis, gallbladder, kidney, heart muscle, all far above lung and brain levels [34–36]. Remarkably, SARS-CoV-2 triggers ACE2 downregulation from the infected cells’ plasma membranes and thus lowers tissue capacity to degrade the bioactive vasoconstrictor hormone angiotensin-II into its vasorelaxant and anti-inflammatory metabolite angiotensin-(1–7), which in turn yields increased concentration of bioactive angiotensin-II locally in SARS-CoV-2-infected tissues [37].

In the murine brain, extensive mapping of the ACE2 protein by immunohistochemistry showed that ACE2 is expressed exclusively by neurons, and not by glial cells, with the highest densities of ACE2 protein occurring in the three interconnected components of the dorsal vagal complex of the brainstem (Fig. 1): 1) the viscerosensory nucleus tractus solitarius (NTS), 2) the dorsal motor nucleus of the vagus (providing efferent innervation of all viscera, including the heart), and 3) the neuro-hemal structure: area postrema [38–40]. These ACE2-enriched structures of the posterior brain represent the superior autonomic nuclei [15, 41], i.e., the centers of the major homeostatic reflexes in mammals (including the baroreflex, the chemoreflex, and the inflammatory reflex, as detailed below). Importantly, the thresholds of these homeostatic reflexes are modulable via cellular plasticity mechanisms such as synaptic efficacies, neurotrophin signaling, axonal sprouting, and neurogenesis [42]. The vagal afferents from the lung arise from peripheral sensory endings that were observed by electron microscopy in direct juxtaposition with the alveolar epithelium in the adult rat, on the basal pole of and in-between the epithelial cells [43, 44]. These vagal sensory fibers belong to glutamatergic neurons having their perikarya in the peripheral nodose ganglion and projecting directly into the brainstem nucleus tractus solitarius [15]. Since SARS-CoV-2 displays neurotropic property, it could be uptaken by the vagal nerve endings of the lung alveolae up to the NTS [28], and probably spread away trans-synaptically in the nervous system [45]. Consistently, experimental infection of transgenic ACE2-overexpressing mice with another SARS-coronavirus, either by nasal inspiration or by transcranial injection, led to the specific expression of the viral antigen in neurons of the three components of the brainstem dorsal vagal complex by four days after infection [46], i.e., not much longer than the time lag for the initial replication in the lung [19, 31]. This delay for SARS-CoV to migrate from the lung to the brainstem via the vagus nerve to invade the NTS, seems likely to fit with the time-course of atypical hypoxemia and delayed acute respiratory distress syndrome in severe COVID-19 patients, as already suggested by Li et al. [47].

In line with this statement, the neurotropism of SARS-CoV-2 per se has been recently demonstrated in vitro, using cultures of neural cells from human induced pluripotent stem cells [48]. In these neural cells in vitro, moreover, SARS-CoV-2 infection was shown to significantly increase apoptotic death incidence by three times over controls at 1–3 days after infection [48]. The latter observations altogether suggest that the neuroinvasion of the brainstem dorsal vagal complex by SARS-CoV-2 in the course of COVID-19 can rapidly trigger neurodegeneration in this integrative center of the autonomic nervous system; this issue should now be assessed directly by histopathology on postmortem brainstem samples from patients deceased with severe COVID-19.

Besides, SARS-CoV-2 can also gain access to the dorsal vagal complex via blood circulation at the level of the area postrema (Fig. 1), due to the lack of blood-brain-barrier in this neurohemal structure [15, 47].

To sum up, the dorsal vagal complex of the brainstem can be a target of SARS-CoV-2 because of its specifically high enrichment in ACE2 and could be reached readily by the virus through two distinct lung-to-brain routes: the vagus nerve and the blood circulation (Fig. 2).

OPEN IN VIEWER

Fig. 2

The vagal routes of SARS-CoV-2 neuroinvasion. In COVID-19, SARS-CoV-2 can gain access to the ACE2-rich dorsal vagal complex either via blood at the level of the neuro-hemal structure area postrema, or via intraneuronal transport through the vagal sensory innervation of the lung alveaolar epithelium. SARS-CoV-2 binding to ACE2 in the dorsal vagal complex yields decreased tissue concentration of ACE2, which in turn increases the ratio between local concentrations of angiotensin-II above its counter-regulatory angiotensin-(1–7) metabolite. This imbalance in the superior autonomic centers worsens the aging-related decline in baroreflex sensitivity and cardiovascular function. Concomitantly, SARS-CoV-2 invasion of the brainstem counteracts the efferent arm of the inflammatory reflex, leading to multi-organ failure through a cytokine storm. Non-medicamentous stimulation of the auricular vagus nerve is likely to restore vagus nerve functions compromised by SARS-CoV2 infection. Green plain circles represent neuronal perikarya; the yellow arrow indicates the therapeutic auricular vagus nerve stimulation (see text).

General considerations: the central autonomic network

Cardio-respiratory regulation is tightly dependent on the baroreflex and the chemoreflex, both involving the vagus nerve and its brainstem centers [50, 51]. The heart rate is tightly linked with the respiratory ventilation cycle since it accelerates during inspiration and decelerates during expiration [52] due to the respiratory sinus arrhythmia, which is itself driven by variations in vagal chronotropic drive [53]. The core breathing rhythm and pattern is determined by a neural pacemaker network in the brainstem: the pre-Bötzinger complex, which is interconnected with cardiovascular regulatory centers of the brainstem via the vagal viscero-sensory NTS [54] (Fig. 1). The steady-state synchrony of the cardio-respiratory rhythm in healthy adults was shown to arise from brainstem centers that altogether determine the vagal efferent activity [55]. The myocardium-innervating vagal efferent fibers arise from two distinct nuclei of the brainstem: the dorsal motor nucleus of the vagus and the nucleus ambiguous [52]. Using pharmacogenetic approaches in awake rats, it has been demonstrated that the neuronal activities of the dorsal motor nucleus of the vagus are required and sufficient to allow exercise-related capacity and endurance adaptation [56].

The dorsal vagal complex is a target of the hyper-tensive hormone angiotensin-II, which is a key component of systemic cardiovascular regulation [57, 58]. Specific angiotensin-II receptors include two subtypes with opposite physiological effects: AT1-R and AT2-R, which respectively mediate increase- and decrease of arterial pressure and heart rate (reviewed in [37]). Angiotensin-II binds to both AT1 and AT2 receptors, whereas ACE2-generated angiotensin-(1–7) binds only the vasorelaxant AT2- R and Mas receptor (Fig. 2) [37]. The vagal visce-rosensory NTS, area postrema, and dorsal motor nucleus of the vagus (DMNV) are all enriched in the hypertensive AT1-R [59]; the hypotensive AT2-R occurs in NTS and DMNV [60] and the hypotensive Mas receptors in NTS only [61]. Angiotensin-II via AT1 receptors of NTS decreases baroreflex sensitivity and increases arterial blood pressure and heart rate [62], whereas experimental overexpression of AT2 receptors in NTS by pharmacogenetics improves baroreflex efficiency in the adult spontaneously hyp-ertensive rat [63].

In addition, the vagus nerve and its brainstem centers down-regulate inflammatory and immune responses via two neural connections: 1) the noradrenergic projection from the vagal sensory nucleus NTS to the neuroendocrine paraventricular nucleus of the hypothalamus, stimulating the secretion of the anti-inflammatory glucocorticoid hormone via the hypothalamic-pituitary-adrenal axis, and 2) the cho-linergic projection from the dorsal motor nucleus of the vagus to the spleen and activated immune cells, or cholinergic anti-inflammatory pathway (CAP) [64]. Indeed, in vivo electrical stimulation of the vagus nerve prevents the acute inflammatory response to bacterial toxin injection. Moreover, in vitro, the synthesis of the major pro-inflammatory cytokine Tumor Necrosis Factor-alpha (TNF-alpha) by macrophage cultures is inhibited by the neuro-transmitter acetylcholine [65]. Reciprocally, the surgical section of the vagus nerve (or vagotomy) resulted in vivo in enhanced TNF-alpha production and excessive response to endotoxin administration [65]. Furthermore, vagus nerve stimulation reduces inflammation in several experimental models indu-cing pro-inflammatory cytokines by activating the alpha7-nicotinic acetylcholine receptor of macropha-ges [66]. In mice deficient for the latter receptor, endotoxin-induced TNF-alpha production is increased, and electrical vagus stimulation fails to reduce serum TNF-alpha levels [66]. More precisely, such macrophage-inhibiting action of the vagus nerve is mediated by acetylcholine-secreting, splenic T-lymphocytes, as demonstrated in a mouse model [67].

The efferent immunosuppressor component of the vagus nerve arises from the DMNV (Fig. 1). It is recruited physiologically in the course of microbial infections or systemic inflammation, by afferent sti-mulation of vagal sensory fibers that express receptors for pro-inflammatory cytokines and project on the viscerosensory NTS (Fig. 1), in the context of the classical “inflammatory reflex” [64, 68]. In the adult mice, implantation of cuff electrodes at the surface of the cervical vagus nerve and compute-rized neural-to-noise analysis of electrical discharge pattern allowed to identify vagal subsets of sensory fibers [69]. These fibers respond to specific cytokines (especially pro-inflammatory ones) in a dose-dependent manner [69]. It should be stressed that the progress in understanding the mechanisms behind neuro-immune communication has allowed for decoding vagal activity patterns in relation to specific diseases [70].

Altogether, the dorsal vagal complex of the brainstem is an integrative center involved in both cardi-orespiratory regulation and post-inflammation return to homeostasis.

Dysautonomia in COVID-19

The first acute consequence of SARS-CoV-2 neuroinvasion into the dorsal vagal complex consists of lowering the tissue concentration of its receptor ACE2 and therefore yields a local increase in angiotensin-II concentration (see above). This hypertensive hormone acts primarily in the dorsal vagal complex to raise blood pressure and heart pumping, which are subsequently potentiated in COVID-19. In parallel, tissue concentration of the ACE2-generated inhibitory feedback metabolite of angiotensin-II: angiotensin-(1–7) decreases [37]. This angiotensinergic imbalance in the superior autonomic center has been established to cause or worsen a decline in baroreflex sensitivity and switch cardiovascular regulation to hypertension [58]. Reciprocally, experimental overexpression of ACE2 by gene transfer into the NTS of an adult spontaneously hypertensive rat was shown to significantly improve the baroreceptor heart rate reflex above controls [71].

The delayed cardiorespiratory symptoms in se-vere COVID-19 cases include heart palpitations, arrhythmias, myocardial injury, hypertension, chest tightness, and ventilatory dysfunction [72, 73, 72, 73]. Furthermore, in COVID-19 patients, the regulation of respiratory ventilation displays aberrant responses to afferent hypoxia stimuli from the peripheral che-moreceptors (the carotid bodies) [20]. This symptom indicates a drop in the NTS relay of chemoreflex, which may be due to SARS-CoV-2-triggered neurodegeneration (see above section).

A neurogenic model of direct SARS-Cov-2 impact on the brainstem autonomic centers has been recently modelized using a computational approach [74]. Also, most of the clinical characteristics of severe COVID-19, including the “cytokine storm” [26, 75] during the third phase of SARS-CoV2 infection, can be explained by dysregulation of the CAP pathway [76, 77].

Dysautonomia in Alzheimer’s disease

Interestingly, the recently reported COVID-19-related autonomic dysfunction has been characterized in AD for three decades [7] and proposed to develop in the preclinical phase of the disease [78]. The most common symptoms of AD-associated dysautonomia are cardiac failure or infarction [79], blood pressure dysregulation [80], decreased baro-reflex function [81], bronchopneumonia [82], par-asympathetic drive hypofunction [83], altered pain perception and reactivity [84], and orthostatic hypo-tension [7]. In AD patients, lower cognitive performance (especially regarding memory) was related to significantly higher cardiac sympathetic and lower parasympathetic function, independently of age, sex, academic years, diabetes, hypertension, and cho-linesterase inhibitor use [85].

These symptoms are attributed to neurodegeneration, including neurofibrillary pathology, that have been mapped in the brainstem centers: in the dorsal motor nucleus of the vagus (which fits with the parasympathetic hypofunction) and to various extents in the NTS, nucleus ambiguous, pre-Bötzinger complex [49, 86–88]. This brainstem neuropathology appears at the Braak’s stage VI of AD, i.e., at mid-course of a 20–25 year-long pathological process [88].

Dysautonomia in aging without dementia

Aging is classically associated with cardiovascular decline and morbidity, which involve decreased parasympathetic activity, blunted baroreflex sensitivity [89], and decline of vagally-modulated heart rate variability [90]. In addition, aging is associated with a systemic rise of inflammation, known as “inflammaging”, and is “macrophage-centered” [8, 91]. Since macrophages are both a major source of systemic inflammation and the main target of the cholinergic anti-inflammatory pathway, inflammaging is likely to result from the hypofunction of the CAP. Strengthening the CAP has indeed been proposed as a new therapeutic perspective in aging-related disorders [92].

In summary, the dysautonomia is associated with COVID-19, AD, and dementia-free aging, which all share some similarities (Table 1). It is particularly noteworthy that the baroreflex in AD patients is chronically reduced, which can underlie the much higher incidence of severe COVID-19 in this sub-population [6]. The failure of the CAP is shared by COVID-19, AD, and dementia-free aging, although along a very different time course (chronic in AD and aging versus acute in COVID-19), which likely corroborates aging and or AD as major risk factors for COVID-19-related mortality. The survey of underlying mechanisms and pathways helps now to propose original, non-serological tools for COVID-19 therapy and or diagnostic.

Vagus nerve stimulation (VNS) as a polyvalent tool in human clinics

VNS in human patients relies on the permanent implant of a cuff electrode on the left vagus nerve (to avoid cardiac side-effects) at the neck level, which is connected by a subcutaneous wire to a pac-emaker-like programmable electronic device. This intervention was approved three decades ago by the Food and Drug Administration (FDA) for treatment-resistant patients with epilepsy and major depressive disorder. The appropriate treatment consists typically in daily 30-second stimulation at 1mA output current with a 20 Hz frequency of 500 microsecond pulses during several months [97]. The parameters of the vagus nerve stimulator are adjusted during in-office visits by treating physicians to optimize therapeutics by-side effects. VNS also proved beneficial in various cardiovascular and inflammatory diseases [98].

In the AD field, the above-discussed invasive VNS has already been applied during 6–12 months on cohorts of AD patients and yielded significant cognitive-enhancing effects [14, 100].

Studies in animal models demonstrated that VNS resets the baroreflex setpoint and induces inhibition of sympathetic efferents [101, 102], which was also recorded in humans [103]. Of interest, when considering the COVID-19 pandemic, VNS could impact not only on systemic inflammation but also on the coagulant-anticoagulant balance, as suggested by VNS application in endotoxemia rats [104]. In a clinically relevant porcine model of progressive sepsis, VNS significantly attenuated multiple organ dysfunctions (with partial or complete prevention of hyperlactatemia, hyperdynamic circulation, cellular myocardial depression, shift in sympathovagal balance toward sympathetic dominance, cardiac myocardial dysfunction and reduction of activated monocytes), thus reducing vasopressor and fluid resuscitation requirements [105].

Therefore, it is likely that VNS could attenuate the clinical manifestations of severe COVID-19, as recently suggested [75]. However, VNS requires an invasive surgical procedure for electrode implantation (see above), not an option in COVID-19 pandemics. Conversely, a non-invasive modality of vagus nerve stimulation has been developed, making it more accessible as a possible first-line treatment option (see next section).

Non-invasive vagus nerve stimulation as a novel therapeutic approach for COVID-19?

Transcutaneous vagus nerve stimulation (tVNS) is a non-invasive procedure consisting of electrical current application through surface electrodes at select locations, most commonly targeting either the cervical branch of the vagus nerve in the neck or the auricular branch of the vagus nerve [106]. Considering these two techniques (cervical or auricular) of tVNS, the cervical location makes selective transcutaneous stimulation of the vagus nerve fibers difficult, because of the presence of both afferent and efferent fibers [107]. By contrast, the auricular branch of the vagus nerve is an exclusively afferent, sensory nerve that innervates part of the skin of the outer ear [94–96]. Hence, trans-auricular vagus nerve stimulation (taVNS) targets only the afferent arm of the vagus nerve.

The optimal location and electrical stimulation parameters providing the greatest tVNS therapeutic effect can be efficiently assessed with heart rate variability (HRV) monitoring. HRV, a biomarker for efferent vagal activation [108], is easier to assess with taVNS than with cervical tVNS [107]. taVNS is mostly used in pharmaco-resistant epilepsy, depression, hypertension, and other diseases [93, 105–107, 93, 105–107]. A recent study has validated the use of taVNS to inhibit both peripheral and central inflammation and suggested a link between taVNS suppression of inflammation and effectiveness in reducing depression severity [109]. Closer to COVID-19 symptoms, a prospective randomized study on 100 respiratory distressed patients undergoing lobectomy via thoracotomy showed that taVNS could significantly decrease IL-6 and IL-10 levels, reduce the incidence of postoperative pneumonia and shorten duration of hospitalization time [110]. Since drugs commonly prescribed to psychiatric patients could protect them from SARS-CoV-2 infection [111], taVNS suppression of neuroinflammation and depression may turn out to be also neuroprotective in the context of COVID-19. So far, however, to the best of our knowledge, only two studies using taVNS (Gammacore®) in COVID-19 patients [112, 113] have been published reporting: 1) a clinical respiratory benefit in two ambulatory COVID-19 patients using cervical tVNS [112]; 2) an ongoing prospective, randomized controlled study on tVNS in patients with moderate to severe COVID-19 respiratory symptoms [113]. Based on this emerging awareness, an International Consortium on Neuromodulation for COVID-19 (ICNC) has been created in early 2020, in order to “provide global leadership for the rapid advancement and clinical adoption of neuromodulation technologies to treat emerging infectious diseases” [114].

Experimental animal studies confirmed the benefits of tVNS by using endotoxemic rodent models and shed light on underlying mechanisms [115, 116]. Remarkably, taVNS also reduced cognitive dysfunction in a postoperative model of aged rats [117] and in a murine model of AD [118] via a decrease of neuroinflammation. In human clinics, however, taVNS has not been tested so far in AD [14, 119].

It remains now to test taVNS impact on AD symptoms in human cohorts, as already suggested for COVID-19 [75].