Long-COVID cognitive impairments and reproductive hormone deficits in men may stem from GnRH neuronal death

Abstract

Background: We have recently demonstrated a causal link between loss of gonadotropin-releasing hormone (GnRH), the master molecule regulating reproduction, and cognitive deficits during pathological aging, including Down syndrome and Alzheimer's disease. Olfactory and cognitive alterations, which persist in some COVID-19 patients, and long-term hypotestosteronaemia in SARS-CoV-2-infected men are also reminiscent of the consequences of deficient GnRH, suggesting that GnRH system neuroinvasion could underlie certain post-COVID symptoms and thus lead to accelerated or exacerbated cognitive decline.

Methods: We explored the hormonal profile of COVID-19 patients and targets of SARS-CoV-2 infection in post-mortem patient brains and human fetal tissue.

Findings: We found that persistent hypotestosteronaemia in some men could indeed be of hypothalamic origin, favouring post-COVID cognitive or neurological symptoms, and that changes in testosterone levels and body weight over time were inversely correlated. Infection of olfactory sensory neurons and multifunctional hypothalamic glia called tanycytes highlighted at least two viable neuroinvasion routes. Furthermore, GnRH neurons themselves were dying in all patient brains studied, dramatically reducing GnRH expression. Human fetal olfactory and vomeronasal epithelia, from which GnRH neurons arise, and fetal GnRH neurons also appeared susceptible to infection.

Interpretation: Putative GnRH neuron and tanycyte dysfunction following SARS-CoV-2 neuroinvasion could be responsible for serious reproductive, metabolic, and mental health consequences in long-COVID and lead to an increased risk of neurodevelopmental and neurodegenerative pathologies over time in all age groups.

Funding: European Research Council (ERC) grant agreements No 810331, No 725149, No 804236, the European Union Horizon 2020 research and innovation program No 847941, the Fondation pour la Recherche Médicale (FRM) and the Agence Nationale de la Recherche en Santé (ANRS) No ECTZ200878 Long Covid 2021 ANRS0167 SIGNAL, Agence Nationale de la recherche (ANR) grant agreements No ANR-19-CE16-0021-02, No ANR-11-LABEX-0009, No. ANR-10-LABEX-0046, No. ANR-16-IDEX-0004, Inserm Cross-Cutting Scientific Program HuDeCA, the CHU Lille Bonus H, the UK Medical Research Council (MRC) and National Institute of Health and care Research (NIHR).

Keywords: COVID-19; Cognition; GnRH; Hypothalamus; Infertility; Neurodevelopment.

Fig. 1

Serum testosterone, and gonadotropin (LH & FSH) concentrations in a male patients hospitalized in the intensive care unit (ICU) of the Lille Medical University Hospital (CHU Lille). (a–d) COVID-19 infected patients in the ICU were divided into 4 groups depending on total testosterone and LH levels, where Group 1 = testosterone <2.88 ng/ml with LH ≤ 2 IU/L, Group 2 = testosterone <2.88 ng/ml with LH > 2 but <12 IU/L, Group 3 = testosterone <2.88 ng/ml with LH > 12 IU/L, Group 4 = testosterone >2.88 ng/ml. A two-tailed unpaired t-test was used to estimate the significance of the difference between groups (total n = 60 patients) during their first week in the ICU. (a) Testosterone (two-tailed unpaired t-test, Group 4 (n = 3) vs Group 1 (n = 13), t (14) = 6.195, ∗∗∗∗p < 0.0001; Group 4 (n = 3) vs Group 2 (n = 38), t (39) = 6.124, ∗∗∗∗p < 0.0001; Group 4 (n = 3) vs Group 3 (n = 6), t (7) = 4.392, ∗∗p = 0.0032 (b) Luteinizing hormone (LH) (two-tailed unpaired t-test Group 4 (n = 3) vs Group 1 (n = 13), t (14) = 6.616, ∗∗∗∗p < 0.0001; Group 4 (n = 3) vs Group 2 (n = 38), t (39) = 2.333, ∗p = 0.0249; Group 4 (n = 3) vs Group 3 (n = 6), t (7) = 1.460, p = 0.1877 (c) Follicle-stimulating hormone (FSH) (Group 4 (n = 3) vs Group 1 (n = 13), t (14) = 3.137, ∗∗p = 0.0073; Group 4 (n = 3) vs Group 2 (n = 38), t (39) = 2.977, ∗∗p = 0.0050; Group 4 (n = 3) vs Group 3 (n = 6), t (7) = 0.8167, p = 0.4410) (d) Percentage of mortality among patients with normal (n = 37) and abnormal HPG status (n = 23). Pearson's Chi-squared test with Yates' continuity correction, Χ2 (1) = 10.38, p = 0.0013 (e–g) Non-COVID-19 male ICU patients (n = 50) were divided into 4 groups depending on total testosterone and LH levels, where Group 1 = testosterone <2.88 ng/ml with LH ≤ 2 IU/L, Group 2 = testosterone <2.88 ng/ml with LH > 2 but <12 IU/L, Group 3 = testosterone <2.88 ng/ml with LH > 12 IU/L, Group 4 = testosterone >2.88 ng/ml. A two-tailed unpaired t-test was used to estimate the significance of the difference between groups (total n = 50 patients). (e) Testosterone (two-tailed unpaired t-test, Group 4 (n = 3) vs Group 1 (n = 26), t (27) = 14.45, ∗∗∗∗p < 0.0001; Group 4 (n = 3) vs Group 2 (n = 16), t (17) = 5.399, ∗∗∗∗p < 0.0001; Group 4 (n = 3) vs Group 3 (n = 5), t (6) = 7.346, ∗∗∗p = 0.0003 (f) Luteinizing hormone (LH) (two-tailed unpaired t-test, Group 4 (n = 3) vs Group 1 (n = 26), t (27) = 8.926, ∗∗∗∗p < 0.0001; Group 4 (n = 3) vs Group 2 (n = 16), t (17) = 3.495, ∗∗p = 0.0028; Group 4 (n = 3) vs Group 3 (n = 5), t (6) = 1.654, p = 0.1493 (g) Follicle-stimulating hormone (FSH) Group 4 (n = 3) vs Group 1 (n = 26), t (27) = 6.139, ∗∗∗∗p < 0.0001; Group 4 (n = 3) vs Group 2 (n = 16), t (17) = 1.635, p = 0.1204; Group 4 (n = 3) vs Group 3 (n = 5), t (6) = 1.134, p = 0.2999.

Fig. 2

Serum testosterone, and gonadotropin (LH & FSH) concentrations in male patients followed at Imperial College London 3 months after contracting COVID-19. COVID-19 infected male patients (n = 47) were divided into 4 groups depending on total testosterone and LH levels, where Group 1 = testosterone <2.88 ng/ml with LH ≤ 2 IU/L, Group 2 = testosterone <2.88 ng/ml with LH > 2 but <12 IU/L, Group 3 = testosterone <2.88 ng/ml with LH > 12 IU/L, Group 4 = testosterone >2.88 ng/ml. A two-tailed unpaired t-test was used to estimate the significance of the difference between groups (total n = 47 patients). (a) Testosterone (two-tailed unpaired t-test, Group 4 (n = 36) vs Group 1 (n = 4), t (38) = 2.384, ∗p = 0.0222; Group 4 (n = 36) vs Group 2 (n = 7), t (41) = 3.262, ∗∗p = 0.0022. (b) Luteinizing hormone (LH) (two-tailed unpaired t-test, Group 4 (n = 36) vs Group 1 (n = 4), t (38) = 2.093, ∗p = 0.0431; Group 4 (n = 36) vs Group 2 (n = 7), t (41) = 0.2810, p = 0.7801. (c) Follicle-stimulating hormone (FSH) (two-tailed unpaired t-test, Group 4 (n = 36) vs Group 1 (n = 4), t (38) = 1.535, p = 0.1330; Group 4 (n = 36) vs Group 2 (n = 7), t (41) = 0.6067, p = 0.5474. (d) Testosterone levels of patients between visits. The colour of the dot represents the HPG axis status (Group1 in red, Group 2 in orange, and Group4 in blue). The size of the dot is representative of the testosterone levels (ng/ml). Except Pt 14 and Pt 22 who switched to Group1/Group 2, all other patients were in Group 4 in the follow-up visit (median days since initial presentation: 464; median days since first post-COVID visit: 237 days). The 9 patients who displayed lower testosterone levels during the second visit than the first are outlined. (e) Testosterone levels at the second visit and the BMI of patients. Two men, both of whom had a BMI above 30, displayed extremely low total testosterone levels without compensatory high LH levels. The colour of the dot represents the HPG axis status (abnormal in red (n = 2) and normal in blue (n = 21)).

Fig. 3

Expression of viral transcripts and proteins, ACE2 and TMPRSS2 in the hypothalamus of COVID-19 patients and non-infected controls. (a) Quantitative PCR analysis of N-protein mRNA in the hypothalamus of COVID-19 patients using three distinct sets of probes (N1, N2 and N3). n = 5 control patients and 4 SARS-CoV-2-infected patients. Note that one COVID-19 patient (yellow point) had viremia at the time of death. Two-tailed unpaired t test, for N1 t (7) = 3.097, p = 0.0174, for N2 t (7) = 3.143, p = 0.0163 and for N3 t (7) = 3,587, p = 0.0089). (b) RNAscope labelling for S-protein mRNA (pink) in the hypothalamus of a control (non-infected) patient and a COVID-19 patient. U6 RNA was used as a positive control. Blue: DAPI. Scale bar: 20 μm. (c) Immunolabelling for SARS-CoV-2 N-protein (white), viral dsRNA (green) and vimentin (magenta) to identify tanycytes, indicating that viral markers are absent in the hypothalamus of controls (left) but heavily expressed in a 63-year-old COVID-19 patient (right). 3 V: third ventricle. Scale bar: 30 μm. (d) N-protein (white), dsRNA (green) and vimentin (magenta) immunolabelling showing abundant N-protein colocalization with vimentin in numerous tanycytic fibres (white arrows) and its presence in non-tanycytic cell bodies (empty arrows) near the median eminence (ME). dsRNA is not present in the vimentin-rich tanycytic cell body layer lining the ventricular wall. Blue: DAPI. Scale bar: 30 μm. (e,f) Extremely strong labelling for S-protein (white) seen in the end-feet (white arrowhead) of tanycytes (vimentin; magenta in e; green in f), which also express ACE2 (green in e) and TMPRSS2 (magenta in f), at the pial surface of the ME, where tanycytic processes (white arrows) contact fenestrated capillaries. Blue: DAPI. Scale bar: 20 μm.

Fig. 4

SARS-CoV-2 infects GnRH neurons and leads to their death in COVID-19 patients. (a) Immunolabelling for GnRH (red) and NRP1 (white) in hypothalamic GnRH neurons in a control patient. Arrowheads show a double-labelled GnRH neuron, while white arrows show a GnRH-immunoreactive process expressing NRP1. Scale bar: 50 μm. (b) Immunolabelling for GnRH (red), ACE2 (white), and S-protein (green) in hypothalamic GnRH neurons in a COVID-19 patient. Arrowheads show a triple-labelled GnRH neuron, white arrows show a GnRH-immunoreactive process that does not express ACE2, and empty arrows show an ACE2-immunoreactive neuron-like process that does not express GnRH. Blue: DAPI. Scale bar: 50 μm. (c) Immunolabelling for GnRH (red) and cleaved caspase 3 (green) in the infundibular nucleus (Inf)–median eminence (ME) area of the hypothalamus of a COVID-19 patient. Blue: DAPI. Scale bars: 500 μm (inset 30 μm). (d) Quantification of proportion of GnRH neurons showing normal or abnormal morphology in control and COVID-19 patients (two-sided Fisher's exact test, p < 0.0001, n = 66 GnRH neurons, n = 4 patients). (e) Proportion of cleaved caspase 3-expressing or healthy GnRH neurons with normal or abnormal morphology in COVID-19 patients (two-sided Fisher's exact test, p = 0.0387, n = 66 GnRH neurons, n = 4 patients). (f) Quantitative RT-PCR for GnRH and NRP1 in the infundibular nucleus-median eminence of COVID-19 and control patients showing the almost complete disappearance of GnRH expression in patients (two-tailed unpaired t-test t (7) = 2.197, p = 0.0320 for control vs COVID-19 brains. n = 5 for control patient and n = 4 for COVID-19 brains).

Fig. 5

Human fetal GnRH neurons are susceptible to SARS-CoV-2 viral infection. (a) Schematic representation of a horizontal section through the nose and brain of a gestational week (GW) 14 human fetus, showing region immunolabelled in (b–d). (b–d) TMPRSS2 (b,c, red) and ACE2 (d, red) immunolabelling in the olfactory epithelium (OE), vomeronasal organ (VNO) and olfactory nerve (ON) of a GW 14 fetus. Blue: DAPI. Scale bars: 1 mm in b and 100 μm in c-d. (e) In a GW 11 human fetus, many GnRH neurons (white) migrating out of the VNO also express NRP1 (green) and/or ACE2 (red), host cell proteins that mediate SARS-CoV-2 infection (white arrows), while NRP1 and ACE2 are also expressed by some olfactory and vomeronasal nerve axons that form the scaffold for GnRH neurons. Blue: DAPI. Scale bar: 40 μm (f) Differentiating FNC-B4 cells in culture showing the presence of NRP1 (magenta) and ACE2 protein expression (red) in cells that have begun to express GnRH (green, top view). Non-GnRH cells (bottom view) also express NRP1, in keeping with the more widespread expression of this guidance molecule in the fetal nose and brain. Blue: DAPI. Scale bar: 10 μm. (g) RT-PCR analysis demonstrating the expression of mRNAs for GnRH, NRP1, and ACE2 by FNC-B4 cells. The housekeeping 18 S RNA was used as a control. n = 3 wells. (h) Fluorescence-activated cell sorting (FACS) of FNC-B4 cells (red) infected with pseudotyped lentiviral particles carrying a full-length SARS-CoV-2 spike protein and a ZsGreen reporter gene (green), showing infection of some cells by the pseudovirus. (i) ZsGreen expression, indicating pseudotyped viral particle entry, is almost undetectable in uninfected negative control cells treated only with vehicle (two-tailed unpaired t-test t (4) = 3.566 p = 0.0235, n = 3 wells). (j) Immunolabelling of the SARS-CoV-2 S-protein in FNC-B4 cells. Scale bar: 20 μm.

Figure S1

COVID-19 patient survival and disease severity.

Figure S2

SARS-CoV2 infection of ACE2-producing olfactory neurons projecting to the olfactory bulb.

Figure S3

Immunolabeling for ACE2 and TMPRSS2 in the embryonic human nose and olfactory bulb.

Figure S4

Evolution of hypothalamic function over time in COVID patients.