The polymorphism L412F in TLR3 inhibits autophagy and is a marker of severe COVID-19 in males

ABSTRACT

The polymorphism L412F in TLR3 has been associated with several infectious diseases. However, the mechanism underlying this association is still unexplored. Here, we show that the L412F polymorphism in TLR3 is a marker of severity in COVID-19. This association increases in the sub-cohort of males. Impaired macroautophagy/autophagy and reduced TNF/TNFα production was demonstrated in HEK293 cells transfected with TLR3^L412F-encoding plasmid and stimulated with specific agonist poly(I:C). A statistically significant reduced survival at 28 days was shown in L412F COVID-19 patients treated with the autophagy-inhibitor hydroxychloroquine (p = 0.038). An increased frequency of autoimmune disorders such as co-morbidity was found in L412F COVID-19 males with specific class II HLA haplotypes prone to autoantigen presentation. Our analyses indicate that L412F polymorphism makes males at risk of severe COVID-19 and provides a rationale for reinterpreting clinical trials considering autophagy pathways.

Abbreviations: AP: autophagosome; AUC: area under the curve; BafA1: bafilomycin A1; COVID-19: coronavirus disease-2019; HCQ: hydroxychloroquine; RAP: rapamycin; ROC: receiver operating characteristic; SARS-CoV-2: severe acute respiratory syndrome coronavirus 2; TLR: toll like receptor; TNF/TNF-α: tumor necrosis factor

KEYWORDS:

• Autophagy
• COVID-19
• HLA
• L412F
• TLR3

Introduction

In December 2019, a new virus was isolated in Wuhan, China, which was called Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2). SARS-CoV-2 is an enveloped positive-sense RNA virus that caused a new pandemic, which WHO named COVID-19 (coronavirus disease-2019).

To date, many characteristics of SARS-CoV-2 are still unclear and, although its ability to be transmitted from one person to another has been ascertained, uncertainties remain about the exact modes of transmission and pathogenicity. In addition, a high variability of symptoms in infected patients and between different populations has been reported; one of the possible explanations of such variability is the genetic background of the host that may affect immune responses to the virus. Among host genetic factors that might impact on symptoms severity there are genes involved in virus entry and mediators of innate immunity [1,2]. TLRs (toll like receptors) are a class of proteins that play a key role in host innate immunity, causing the production of pro-inflammatory cytokines (TNF, IL1, and IL6) and type I and II Interferons, that are responsible for innate antiviral responses. Among TLR genes, TLR3 encodes an interferon‐inducing dsRNA sensor, whose activation is involved in protection against different RNA viruses [2–4]. Upon viral infection, TLR3 signaling leads to the activation of two factors, NFKB and IRF3 (interferon regulatory factor 3), which play an essential role in the immune response. This results in the production of various cytokines, including TNF (tumor necrosis factor), activating immune responses. However, increased inflammatory responses can make the patient more susceptible to pneumonia and autoimmune diseases. Accordingly, a protective effect against fatal pneumonia has been reported in the absence of TLR3 [5–7]. Among TLR3 variants, the functional L412F polymorphism (rs3775291; c.1234 C > T) is known to decrease TLR3 expression on the cell surface [8]. This polymorphism also leads to poor recognition of SARS-CoV-2 dsRNA, during replication, compared to its wild-type (WT) counterpart [9] and has been recently associated with SARS-CoV-2 susceptibility and mortality [10].

There is evidence that TLR3, as with other TLRs, acts through autophagy in determining susceptibility to infections [11]. The autophagic pathway is essential during infection and for molecular processes such as cell maintenance and homeostasis [12,13]. Indeed, autophagy is one of the major cell defense mechanisms against pathogens [14]. A role for autophagy is reported in different studies on other coronaviruses such as the mouse hepatitis virus/MHV and the transmissible gastroenteritis virus/TGEV [15,16]. A role in SARS-CoV-2 infection has also been described [17–19]. In particular, SARS-CoV-2 can inhibit autophagy resulting in accumulation of autophagosomes and inhibition of viral clearance that, together with immune dysfunction and the activation of numerous inflammatory cytokines, leads to a more severe form of COVID-19 [20–22].

To shed light on the mechanisms underlying the diverse susceptibility to COVID-19, we performed a nested-control study within our GEN-COVID cohort, confirming the role of L412F polymorphism in the TLR3 gene in susceptibility to SARS-CoV-2 and further defining the potential mechanisms by which this effect is exerted.

Results and discussion

Comparing the extreme phenotypes of SARS-CoV-2 infection, severe COVID-19 patients (cases) versus SARS-CoV-2 PCR-positive oligo-asymptomatic subjects (controls), and using LASSO Logistic regression on common bi-allelic polymorphisms from whole-exome sequencing, we identified the L412F polymorphism (rs3775291; c.1234 C > T) in TLR3 as a severity marker (Figure 1A). The grid search curve of the cross-validation score (Figure 1B) shows a maximum of the regularization parameter in 10. With this calibration setting, the 10-fold cross-validation provides good performances in terms of accuracy (73%), precision (74%), sensitivity (73%), and specificity (73%) as shown in Figure 1C. The confusion matrix is reported in Figure 1D, whereas the receiver operating characteristic (ROC) curve (Figure 1E) provides an area under the curve (AUC) score of 80%.

Figure 1. The histogram of the LASSO logistic regression weights represents the importance of each feature for the classification task, (A) The positive weights reflect a susceptible behavior of the features to the target COVID-19 disease, whereas the negative weights a protective action. (B) Cross-validation ROC-AUC score for the grid of LASSO regularization parameters; the error bar is given by the standard deviation of the score within the 10 folds; the optimal regularization parameter is chosen by selecting the one with highest cross-validation score (red point). (C) Boxplot of accuracy, precision, sensitivity, specificity, and ROC-AUC score for the 10-fold of the cross-validation. The box extends from the Q1 to Q3 quartile, with a line at the median (Q2) and a triangle for the average. (D) Confusion matrix for the aggregation of the logistic regression predictions in the 10 folds of the cross-validation. (E) ROC curve for the 10 folds of the cross-validation. (F) Distribution of carriers of the polymorphism L412F in homozygous or heterozygous states stratified by clinical category.

The L412F polymorphism has an overall allele frequency of about 20%, ranging from 30% in European to 0.88% in African (mainly sub Saharan) populations [8]. It is intriguing that a relatively COVID-19-free population such as sub Saharan has a very low frequency (0.88%) of this polymorphism and that Asian (26.97%) and European (30.01%) have a much higher frequency. The variant protein with phenylalanine is under-represented on the cell surface, it is not efficiently secreted into the culture medium when expressed as the soluble ectodomain, and it has reduced capability to activate the expression of TLR3-dependent reporter constructs [8]. In order to confirm the role of the polymorphism, we compared individuals showing severe COVID-19 (cases) and those with no sign of the disease (controls). We subdivided patients into two categories, those having the polymorphism in heterozygous or homozygous state and those homozygous for the WT allele. We found that the prevalence of L412F polymorphism is significantly higher in cases compared to controls (p-value 2.8 × 10⁻²) (Table 1). The global allele frequency of L412F in our cohort (cases and controls) is 29.38%, comparable to the allele frequency of 29.79% reported in the European (non-Finnish) population in the gnomAD database (https://gnomad.broadinstitute.org/). The identified frequencies were in Hardy-Weinberg equilibrium.

Table 1. L412F and COVID-19 outcome (both sexes).

	Cases	Controls	Marginal Row
			Totals
L412F	186 (55.0%)	139 (46.3%)	325 (50.9%)
Wild-Type	152 (45.0%)	161 (53.7%)	313 (49.05%)
Marginal Column Totals	338 (52.97%)	300 (47.02%)	638 (Grand Total)

p-value (cases vs controls) = 2.8x10⁻²

Sex-related differences of TLRs activation following stimulation by viral nucleic acid may be involved in the sex-related variability in response to viral infections [23]. Several rare TLR3 loss of function mutations are known to be linked both to influenza and SARS-CoV-2 virus as well as hyperfunctioning mutations [24,25]. In agreement with these data, when we stratified by gender, the statistically significant difference increased in the sub-cohort of males giving an Odds Ratio of 1.94 (95% confidence interval, 1.23 to 3.06; p = 3.8x10⁻³), whereas it was lost in the sub-cohort of females (p-value 5.8 × 10⁻¹) (Tables 1, 2, 3).

Table 2. L412F and COVID-19 outcome (males only).

	Cases	Controls	Marginal Row
			Totals
L412F	131 (55.3%)	45 (38.8%)	176 (49.8%)
Wild-Type	106 (44.7%)	71 (61.2%)	177 (50.1%)
Marginal Column Totals	237 (67.13%)	116 (32.86%)	353 (Grand Total)

p-value (cases vs controls) = 3.6x10⁻³

Table 3. L412F and COVID-19 outcome (females only).

	Cases	Controls	Marginal Row
			Totals
L412F	55 (54.5%)	94 (51.1%)	149 (5.,3%)
Wild-Type	46 (45.5%)	90 (48.9%)	136 (47.7%)
Marginal Column Totals	101 (35.43%)	184 (64,56%)	285 (Grand Total)

p-value (cases vs controls) = 5.8x10⁻¹

We then investigated the prevalence of patients carrying L412F in heterozygous or homozygous states in all 4 categories of COVID-19 clinical severity, considering only male subjects regardless of age (n = 665). We found that the prevalence of carriers directly increased with the severity of COVID-19, from a clinical condition not-requiring hospitalization to intratracheal intubation (Figure 1F)

The L412F substitution in TLR3 falls in the ectodomain, in the 14 leucine-rich repeats/LRR domain, a motif of 22 amino acids in length that folds into a horseshoe shape [26]. Proteins containing leucine-rich repeats are involved in a variety of biological processes, including signal transduction, cell adhesion, DNA repair, recombination, transcription, RNA processing, disease resistance, apoptosis, and the immune response [27]. The L412F substitution is expected to have a limited structural impact with minimal rearrangement of near hydrophobic amino acids such as tryptophan 386 (Figure 2). However, the absence of one of the leucines probably determines a different rearrangement of the motif and consequently of the near glycosylation site Asn414, having an impact on protein-protein interaction and in signal transduction process [28].

Figure 2. Superposition of wild-type and mutated TLR3 protein. (A) TRL3 human protein tridimensional structure of 2Z7X crystal structure. In green cartoon representation of TLR3 protein. (B) and (C) Zoom of the mutated region with Leu412 in red sticks and Phe412 in magenta. The hydrophobic core of Leu377, Leu389, and Trp386 is in blue sticks.

Germline knockout of TLR3 inhibits autophagy and upregulation of TLR3 promotes damage after myocardial infarction mainly because of autophagy rather than inflammatory activation [29]. Interestingly, we could notice a statistically significant (p = 3.8x10⁻²) reduced survival at 28 days in TLR3_L412F COVID-19 patients treated with hydroxychloroquine (HCQ) (Figure 3A). As this drug is a well-established inhibitor of autophagy, we reasoned that alterations of this important biological process might have a role in the increased severity of the clinical phenotypes of SARS-CoV-2 infection in patients with the TLR3^L412F polymorphism. Notably, beside being entirely ineffective at changing the clinical evolution of COVID-19, which led to retraction of the paper reporting the clinical trial [30], HCQ may have been responsible for reportedly increased fatality rates among patients treated with this drug [31,32]. Poly(I:C) stimulation of the TLR3 receptor has already been shown to stimulate autophagy [29]. Therefore, we decided to compare the efficacy of transfected WT and L412F-mutated receptors in inducing autophagy upon poly(I:C) treatment. To monitor autophagy, we used indirect immunofluorescence microscopy to score the formation of punctate intracellular vacuoles stained for LC3B. Moreover, to better appreciate autophagosomal formation, we also used bafilomycin A₁ (Baf A1), an inhibitor of lysosomal acidification, to prevent lysosomal degradation of autophagosome-associated LC3B [33]. Baf A1 allows, in fact, the detection of each autophagosome formed in the time-lapse between addition of the drug to cells and harvesting [33]. Importantly, to avoid potentially confounding effects of the endogenous TLR3 receptor in transfected cells, we used TLR3 knockout HEK cells. In these cells, when transfected with a plasmid encoding WT TLR3 (TLR3_WT), we observed a progressively increasing number of autophagosomes (APs) when stimulated with poly(I:C) for different time points in the presence of BAF A1 (compared to BAF A1 alone) (Figure 3B-C), indicating a stimulation of the synthesis of these vesicles and a positive autophagic flux. Conversely, in HEK cells transfected with TLR3_L412F, the number of AP was reduced by poly(I:C) stimulation in the presence of BAF A1 (Figure 3B-C), demonstrating a block in AP synthesis and a reduced autophagic flux. Interestingly, in the absence of BAF A1, AP numbers did not increase upon poly(I:C) stimulation, suggesting that, in HEK cells, fast degradation of AP may compensate for a small increase in the synthesis. Indeed, also rapamycin (RAP), a strong stimulus for autophagy [34], induced only a small increase of AP in the absence of BAF A1 in these cells upon transfection of both TLR3_WT and TLR3_L412F, supporting slow rates of AP synthesis in these cells upon stimulation with different stimuli. The fusion process of autophagosomes with hydrolase-containing lysosomes represents the final step in the degradation process along the autophagic route and its evaluation provides important information for flux analysis [33]. Therefore, as a further confirmation of a reduced autophagic flux in HEK cells expressing the TLR3^L412F mutant as compared to TLR3_WT-transfected cells, we decided to score fusion of autophagosomes to lysosomes in these cells upon poly(I:C) stimulation, by measuring colocalization rate of LC3B and LAMP-1, through analysis with the Volocity software of immunocytochemistry experiments (Fig. S1).

Figure 3. Analysis of autophagy in TLR3_L412F-expressing cells. (A) 28-day survival study of TLR3- L412F carriers vs not-carriers in the group treated with hydroxychloroquine. N = 156, with 73 carriers of TLR3- L412F. Three carriers and 1 not-carrier died in the first 28 days of treatment. (B) Analysis of autophagy in HEK-KO cells expressing wild type or L412F mutant proteins. HEK-KO cells were transfected for 24 h with plasmids encoding TLR3_WT and TLR3_L412F. Cells were next incubated in full medium (FM) or FM + 400 nM bafilomycin A₁ (BAF A1) for 3 h and stimulated for increasing times with 50 μg/ml poly(I:C), as indicated. Cells were next fixed, permeabilized with 100 μg/ml digitonin and stained with anti-LC3B antibodies and revealed with Alexa Fluor 488-conjugated secondary antibodies. Nuclei were stained with DAPI. Where indicated, RAP (500 nM, for 2 h) was used as positive control for induction of autophagy. (C) Same as in B, but the amount of autophagosomes (scored as LC3B-positive dots) per cell was quantified by Volocity software. Measures were obtained by analyzing at least 400 cells/sample from 3 different experiments (n = 3). (D) Analysis of TNF mRNA expression in HEK-KO cells expressing wild type or L412F mutant proteins. HEK-KO cells were transfected for 24 h with plasmids encoding empty vector (CTR), TLR3_WT and TLR3_L412F and next stimulated for increasing times with 50 μg/ml poly(I:C), where indicated. TNF levels were evaluated by Real Time PCR. The gene expression levels were evaluated by the fold change versus TLR2 WT sample using the equation 2^−DDCt. Data are presented as the mean ± SEM. Data significance was analyzed using One-way ANOVA test with Holm-Sidak’s correction. Asterisks were attributed for the following significance values: P > 0.05 (ns), P < 0.05 (*) and P < 0.01 (**). (E) Normal human fibroblasts (NDHF) from subjects expressing the TLR3_WT receptor were stimulated with 50 μg/ml poly(I:C) or RAP (1 μ M) for 4 h, in full medium alone or containing 400 nM bafilomycin A₁ for 3 h. Cells were next fixed, permeabilized with 100 μg/ml digitonin and stained with anti-LC3B antibodies and revealed with Alexa Fluor 488-conjugated secondary antibodies. Nuclei were stained with DAPI. (F) Same as in E, but the number of autophagosomes (scored as LC3B-positive dots) per cell was evaluated for each sample by Volocity software. (G and H) same as in E-F, but fibroblasts are homozygous for the TLR3_L412F receptor. Statistical analysis was performed using Student’s t test. Means § SEM for each value are shown in the graphs. ns = not significant; ** = p < 0.01; *** = p < 0.001. < 0.001.

During SARS-CoV-2 infection, adipocytes produce pro-inflammatory cytokines like TNF, IL6 and IL1B/IL-1β which recruit immune cells to the site of infection [35]. Autophagy is stimulated and regulated by these pro-inflammatory cytokines [35]. TNF is a potent immunomodulator and proinflammatory cytokine that has been implicated in the pathogenesis of autoimmune and infectious diseases. It is produced by activated monocytes and macrophages, as well as by many other cell types, including lymphocytes, as a transmembrane protein. Through cell modulation, TNF can activate both cell death and survival mechanisms. TNF induces autophagy through a feedback mechanism, causing further recruitment and activation of lymphocytes and contributing to the excess inflammation typical of SARS-CoV-2 infection [36]. As chloroquine, a powerful inhibitor of autophagy, inhibits production of different cytokines, among which TNF [37], we next decided to test if the inhibitory effect of the L412F mutation on autophagy was also able to mimic the effect of the pharmacological inhibitor of this process in HEK cells. Indeed, while poly(I:C) readily stimulated TNF expression in HEK cells transfected with the TLR3_WT receptor, this effect was completely abolished in TLR3_L412F-transfected cells (Figure 3D).

In order to validate data obtained on transfected HEK cells, we next isolated and cultured skin fibroblasts from healthy donors with different genotypes relative to the TLR3 locus: wild-type (WT/WT) and L412F (L412F/L412F) homozygous. In these primary fibroblasts, immunofluorescence analysis revealed that the number of LC3B-positive vesicles increased upon poly(I:C) stimulation both in the absence and in the presence of BAF A1 (Figure 3E-F), showing an overall positive autophagic flux in WT cells, while the flux resulted significantly reduced in L412F/L412F fibroblasts (Figure 3G-H). As a control, RAP stimulation of both WT/WT and L412F/L412F showed a positive autophagic flux (Figure 3(E,F,G,H) confirming that the mutation specifically affected TLR3-dependent autophagy and not the general autophagic process. Also in these cells, we confirmed a reduced autophagic flux in TLR3_L412F-expressing fibroblasts as compared to TLR3_WT-expressing cells, upon stimulation with poly(I:C), by measurement of the LC3B-LAMP1 colocalization rate (Fig. S2).

Overall, our results therefore suggest that the outcome of clinical trials with HCQ should be reinterpreted in the light of TLR3^L412F polymorphism status. Negative effects of the drug in L412F bearing subjects may have masked a possible positive outcome in L412F-free subjects. Importantly, they also support a positive role of autophagy in the anti-viral response of the organism to SARS-CoV-2, as suggested by a recent report by Hayn and colleagues [38], demonstrating that at least 3 viral proteins are able to specifically block autophagic turnover.

TLR3 variant L412F has been associated with a wide range of autoimmune diseases including Addison disease and hypothyroidism [39]. TLR3 rare variants resulting in partial loss of function and occurring together with the common variant L412F, or with another rare variant, have been identified in Addison disease [40]. Persistent viral infections in a background of defective innate immunity lead to overexpression of HLA allotypes prone to present autoantigen. Defects of autophagy have been observed in many infectious and autoimmune diseases. Alteration of autophagic processes causes the onset of autoimmunity due to increased survival and reduced apoptosis of self-reactive lymphocytes [41–43]. HLA has been shown to be implicated in disease severity and clinical outcome of patients with COVID-19 [44]. Accordingly, an increased frequency of autoimmune disorders as co-morbidity was found in our cohort in L412F COVID-19 patients with specific HLA class II haplotypes prone to autoantigen presentation. In particular, we analyzed the DR3-DQ2 haplotype which predisposes to different types of autoimmune diseases [45,46]. The frequency of autoimmune disorders is indeed significantly increased in male patients with HLA DR3/DQ2 haplotype and L412F, especially diabetes (25%) (Table 4 and Table 5). These results suggest that the combination of L412F in TLR3 and a specific class II HLA haplotype puts male patients at risk of post-COVID autoimmune exacerbation emphasizing the need for appropriate follow-up.

Table 4. Association between DR3-DQ2 + L412F haplotype and autoimmune disorders in male patients.

	With autoimmune disease	W/O autoimmune disease	Marginal Row
			Totals
DR3-DQ2 + L412F	12 (3.59%)	12 (3.59%)	24 (7.18%)
Other	64 (19.16%)	246 (73.65%)	310 (92.81%)
Marginal Column Totals	76 (22.75%)	258 (77.24%)	334 (Grand Total)

p-value (cases vs controls) = 0.000951

Table 5. Male patients with L412F and HLA DR3/DQ2 haplotype.

PatientID	Age	Ethnicity (white = 1, hispanic = 4)	Clinical Category	GT TLR3 (L412F)	Clinical known comorbidities
AR-COV-25	69	1	4	0/1	Hypertension, Kidney failure, COLD
BS-COV-102	59	1	3	0/1	Diabetes Mellitus, diabetic neuropathy with transmetarsal amputation, bilateral fachiectomy, left hernioplasty, diabetic retinopathy
BS-COV-58	47	1	3	0/1	Hypertension
BS-COV-65	72	1	2	0/1	Rheumatoid Arthritis
BS-COV-70	65	1	3	0/1	Autoimmune hepatitis, autoimmune polyghiandol syndrome, previous autoimmune thyroid
CR-COV-17	61	1	2	0/1	Inflammatory disease, Hypertension
CR-COV-4	66	1	3	0/1	None
LS-COV-10	37	1	1	0/1	DLB-CL EBV+, Immunodeficiency, Bipolar Disorder
LS-COV-19	58	4	2	0/1	Diabetes Mellitus
LS-COV-8	48	1	4	0/1	Hypertension
MORE-COV-16	65	1	3	0/1	Hypertension, Asthma, Dyslipidemia
MORE-COV-7	72	1	4	0/1	Hypertension
PG-COV-15	87	1	2	0/1	Congestive heart failure, Dyslipidemia, Stroke
PG-COV-23	68	1	1	0/1	N/A
PV-COV-02	65	1	2	0/1	Rheumatoid Arthritis
PV-COV-23	72	1	3	0/1	Hypertension, Diabetes Mellitus
PV-COV-24	79	1	2	0/1	Atrial Fibrillation, Diabetes Mellitus, Kidney failure
PV-COV-39	33	4	4	0/1	Diabetes Mellitus, Obesity
PV-COV-73	54	1	3	0/1	None
PV-COV-97	66	1	4	0/1	Congestive Heart Failure, Hypertension, Diabetes Mellitus
RUF-COV-1	52	1	0	0/1	None
SPC-COV-14	56	1	3	0/1	Asthma, HCV
TV-COV-81	44	1	1	0/1	None
TV-COV-96	51	1	2	0/1	Hypertension

No association was found between AIRE loss of function variants and COVID-19 outcome, as outlined by the absence of the gene in Figure 1.

In conclusion, we have identified the second protein-encoding polymorphism that modulates COVID-19 outcome. These results indicate that L412F polymorphism in the TLR3 gene makes males, in whom after puberty testosterone lowers TLR3 expression, at risk of severe COVID-19 in a context of a polygenic model. Moreover, based on impairment of autophagy, these data provide a rationale for reinterpreting clinical trials with HCQ stratifying patients by L412F. Finally, the combination of L412F in TLR3 and specific HLA class II haplotypes may put male patients at risk of post-acute sequelae of SARS-CoV-2 infection pointing to the need for an appropriate follow-up. Our experiments suggest an important role of autophagy downstream of the TLR3 receptor, possibly affecting TNF production and susceptibility to infections, including SARS-CoV-2, pinpointing to IFNs treatment (especially IFN γ) avoiding hydroxiclorochine.

Materials and methods

Patients

We performed a nested case-control study (NCC). We used a cohort of 1319 subjects (cases and controls) from the Italian GEN-COVID Multicenter study, infected with SARS-CoV-2 diagnosed by RT-PCR on nasopharyngeal swab [47]. Cases were defined as patients needing endotracheal intubation or CPAP/biPAP ventilation. Controls were oligo-asymptomatic subjects not requiring hospitalization.

Ethics approval

The GEN-COVID study was approved by the University Hospital of Siena Ethical Review Board (Protocol n. 16,929, dated 16 March 2020).

LASSO logistic regression

We adopted the $\lambda \sum _{k=1}^p\left|{\beta }_k\right|$Least Absolute Shrinkage and Selection Operator (LASSO) logistic regression model for the classification of severe COVID-19 patients (cases) versus SARS-CoV-2 PCR-positive oligo-asymptomatic subjects (controls), able to enforce both the sparsity and the interpretability of the results. By denoting with ${\beta }_k$ the coefficients of the logistic regression and by lambda (λ) the strength of the regularization, the LASSO regularization [48] term of the loss, has the effect of shrinking the estimated coefficients to 0. In this way, the weights of the logistic regression algorithm can be interpreted as the feature importances of the subset of the most relevant features for the task [49]. The input features are the common bi-allelic polymorphisms from whole-exome sequencing as well as gender, and the age, the latter as a continuous variable normalized between 0 and 1. Common bi-allelic polymorphisms are defined as combinations of two polymorphisms, each with minor allele frequency above 1%, with frequency above 5% in the cohort.

The fundamental hyper-parameter of the logistic regression algorithm is the strength of the LASSO term, which is tuned with a grid search method on the average area under the ROC curve for the 10-fold cross-validation. The regularization hyperparameter varies in the range [10⁻³, 10²] with 50 equally spaced values in the logarithmic scale. The optimal regularization parameter is chosen by selecting the parameter with the highest cross-validation score. During the fitting procedure, the class slight unbalancing is tackled by penalizing the misclassification of the minority class with a multiplicative factor inversely proportional to the class frequencies. The data pre-processing was coded in Python, whereas for the logistic regression model the scikit-learn module with the liblinear coordinate descent optimization algorithm was used. Performances of the model were evaluated using the cross-validation confusion matrix as well as by computing precision, sensitivity, specificity, and the ROC curve.

Cell culture and transfection

HEK-Dual™ Null (NF/IL8) cells (Invivogen, hkd-nullni) cells were cultured in Dulbecco modified Eagle medium (DMEM; Euroclone, ECB7501L) supplemented with 10% fetal bovine serum (FBS; Euroclone, ECS0180L), 2 mM L-glutamine (Carlo Erba, ABP379-100) and 100 units/ml penicillin-streptomycin (Life Technologies, 15,140,148) at 37°C in an atmosphere of 5% CO₂:air. Transfections were performed with 1 μg DNA plasmid using lipofectamine LTX (Life Technologies, 15,338,500). The cells were seeded to be 70% to 80% confluent, then DNA was diluted in DMEM with 10 mM HEPES, pH 7.2. Lipofectamine LTX was next added to the complex (5 μl) to allow creation of complexes (30 min at RT). Ultimately, DNA-lipid complexes were added to cells. Bafilomycin A₁ was from Santa Cruz Biotechnology (sc-201,550). Human primary fibroblasts were obtained from the Genetic Biobank of Siena (http://biobanknetwork.telethon.it/; cell line number: Rett 2250, 2980/18, 1031/15, Rett 1200). Fibroblasts were cultured in Dulbecco Modified Eagle medium supplemented with 10% FBS, 2% L-glutamine and 1% penicillin-streptomycin, according to standard protocols, and routinely passed 1:2 with trypsin-EDTA (0.05%) solution (Irvine Scientific, 9342).

Immunofluorescence (IF)

Cells were washed with phosphate-buffered saline (PBS; Oxoid, BR0014G), then fixed with 4% paraformaldehyde in PBS for 20 min, washed with PBS and permeabilized with digitonin solution (Life Technologies, BN2006) for 20 min. Then, the cells were washed three times in PBS. Permeabilized cells were incubated with anti-LC3B (MBL, M152–3) and/or anti-LAMP1 (Cell Signaling Technology, 9091) primary antibodies for 1 h, washed three times with PBS, and then incubated with anti-mouse Alexa Fluor 488-conjugated (Life Technologies, A21202) and/or Alexa Fluor 647 (Life Technologies, A21245) secondary antibodies; subsequently cells were washed three times with PBS. Nuclei were stained with a solution of 6 μM of 4ʹ,6-diamidino-2-phenylindole (DAPI; Sigma Aldrich, D9542) in PBS for 10 min. Coverslips were mounted in a fluorescence mounting medium (Dako, S3023). Samples were visualized on a TSC SP5 confocal microscope (Leica Microsystems, 5,100,000,750) installed on an inverted LEICA DMI 6000CS (Leica Microsystems, 10,741,320) microscope and equipped with an oil immersion PlanApo 63 × 1.4 NA objective. Images were acquired using the LAS AF acquisition software (Leica Microsystems, 10,210). Poly(I:C) was from InvivoGen (31,852–29-6).

Dot count and statistical analysis for autophagy

For the LC3B-positive dot counts, we performed intensitometric analysis of fluorescence using the Quantitation Module of Volocity software (PerkinElmer Life Science). LC3B-LAMP1 colocalization rate was measured by the Quantification tool of LAS AF software (Leica Microsystems). Dot counts and colocalization rate were subjected to statistical analysis. Measures were obtained by analyzing at least 400 cells/sample (dot counts) or 250 cells/sample (colocalization rate) from 3 different experiments. Significance (P value) was assessed by Student’s t test, using GraphPad Prism6 software. Asterisks were attributed for the following significance values: P > 0.05 (ns), P < 0.05 (*), P < 0.01 (**), P < 0.001 (***).

Real time qPCR analysis of TNF expression

Total RNA was isolated using the RNAeasy Mini Kit (Qiagen, NC9677589) according to the manufacturer’s instructions. cDNA synthesis was performed using the Maxima First Strand cDNA Synthesis Kit (Life Technologies, EP0751). Neo-synthetized cDNA was used to perform Real Time PCR using the PowerUp Sybr Green (Life Technologies, A25779). The following primers were used: TNF Fw CTATCTGGGAGGGGTCTTCC; TNF Rw GGTTGAGGGTGTCTGAAGGA; HPRT1 Fw GTCTTGCTCGAGATGTGATG and HPRT1 Rw GTAATCCAGCAGGTCAGCAA. Target transcripts were analyzed with the QuantStudio 7 System (Applied Biosystems, CA, USA). The comparative threshold cycle (Ct) method was used for quantification analysis. The Ct values of each gene were normalized to the Ct value of HPRT1. The gene expression levels were evaluated by the fold change using the equation 2-ΔΔCt.

HLA sequencing

HLA-class I and II genes were targeted for DNA sequencing using a biotinylated DNA probe-based capture method [50], with modifications as follows. Genomic DNA (500 ng from each sample) was fragmented enzymatically using the NEBNext Ultra ii FS module (New England Biolabs, E7810S). Individual samples were labeled uniquely using 3 μl of 15 μM custom dual-index adapters (Integrated DNA Technologies, Coralville, IA, USA) and the NEB ligation module. Post ligation cleanup was based on the Kapa Hyper Prep protocol (Kapa Biosystems, Wilmington, MA) and followed by dual size selection. Paired ends of 250 bp each were sequenced using a NovaSeq instrument and SP Reagent Kit (Illumina Inc, San Diego, CA, USA). HLA alleles were determined from the sequence data using the consensus from three algorithms: NGSengine 2.10.0 (GenDX, Utrecht, The Netherlands), HLA Twin (Omixon Biocomputing Ltd. Budapest, Hungary) and HLA*LA [51].

Acknowledgments

This study is part of the GEN-COVID Multicenter Study, https://sites.google.com/dbm.unisi.it/gen-covid, the Italian multicenter study aimed at identifying the COVID-19 host genetic bases. Specimens were provided by the COVID-19 Biobank of Siena, which is part of the Genetic Biobank of Siena, member of BBMRI-IT, of the Telethon Network of Genetic Biobanks (project no. GTB18001), of EuroBioBank, and of RD-Connect. We thank the CINECA consortium for providing computational resources and the Network for Italian Genomes http://www.nig.cineca.it for its support. We thank private donors for the support provided to A.R. (Department of Medical Biotechnologies, University of Siena) for the COVID-19 host genetics research project (D.L n.18 of 17 March 2020). PJN was supported by a grant from fastgrants.org and US NIH R56 AI151549. We also thank the COVID-19 Host Genetics Initiative (https://www.covid19hg.org/) and MIUR project “Dipartimenti di Eccellenza 2018-2020” to the Department of Medical Biotechnologies University of Siena, Italy. We thank Dr. Margherita Leonardi for the experimental contribution in the autophagy data analysis. TDJF received the Post-Doctoral fellowship from Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) of Brazil.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed here.

Additional information

Funding

MIUR project “Dipartimenti di Eccellenza 2018-2020” to Department of Medical Biotechnologies University of Siena, Italy (Italian D.L. n.18 March 17, 2020). Private donors for COVID-19 research. “Bando Ricerca COVID-19 Toscana” project to Azienda Ospedaliero-Universitaria Senese. Charity fund 2020 from Intesa San Paolo dedicated to the project N. B/2020/0119 “Identificazione delle basi genetiche determinanti la variabilità clinica della risposta a COVID-19 nella popolazione italiana”. The Italian Ministry of University and Research for funding within the “Bando FISR 2020” in COVID-19 and the Istituto Buddista Italiano Soka Gakkai for funding the project “PAT-COVID: Host genetics and pathogenetic mechanisms of COVID-19” (ID n. 2020-2016_RIC_3).

Appendix

GEN-COVID Multicenter Study (https://sites.google.com/dbm.unisi.it/gen-covid)

Mirella Bruttini^1,2,17, Rossella Tita¹⁷, Sara Amitrano¹⁷, Anna Maria Pinto¹⁷, Maria Antonietta Mencarelli¹⁷, Caterina Lo Rizzo¹⁷, Valentina Perticaroli^1,2,17, Massimiliano Fabbiani¹⁹, Barbara Rossetti¹⁹, Giacomo Zanelli^2,19, Elena Bargagli²⁰, Laura Bergantini²⁰, Miriana D’Alessandro²⁰, Paolo Cameli²⁰, David Bennett²⁰, Federico Anedda²¹, Simona Marcantonio²¹, Sabino Scolletta²¹, Federico Franchi²¹, Maria Antonietta Mazzei²², Susanna Guerrini²², Edoardo Conticini²³, Luca Cantarini²³, Bruno Frediani²³, Danilo Tacconi²⁴, Chiara Spertilli²⁴, Marco Feri²⁵, Alice Donati²⁵, Raffaele Scala²⁶, Luca Guidelli²⁶, Genni Spargi²⁷, Marta Corridi²⁷, Cesira Nencioni²⁸, Leonardo Croci²⁸, Gian Piero Caldarelli²⁹, Maurizio Spagnesi³⁰, Paolo Piacentini³⁰, Maria Bandini³⁰, Elena Desanctis³⁰, Silvia Cappelli³⁰, Anna Canaccini³¹, Agnese Verzuri³¹, Valentina Anemoli³¹, Agostino Ognibene³², Alessandro Pancrazi³², Maria Lorubbio³², Massimo Vaghi³³, Antonella D’Arminio Monforte³⁴, Esther Merlini³⁴, Federica Gaia Miraglia³⁴, Raffaele Bruno^35,36, Marco Vecchia³⁵, Serena Ludovisi³⁷, Massimo Girardis³⁸, Sophie Venturelli³⁸, Marco Sita³⁸, Andrea Cossarizza³⁹, Andrea Antinori⁴⁰, Alessandra Vergori⁴⁰, Arianna Emiliozzi⁴⁰, Stefano Rusconi^41,42, Matteo Siano⁴², Arianna Gabrieli⁴², Agostino Riva^41,42, Daniela Francisci⁴³⁴⁴, Elisabetta Schiaroli⁴³⁴⁴, Francesco Paciosi⁴⁴, Pier Giorgio Scotton⁴⁵, Francesca Andretta⁴⁵, Sandro Panese⁴⁵, Renzo Scaggiante⁴⁷, Francesca Gatti⁴⁸, Saverio Giuseppe Parisi⁴⁸, Melania degli Antoni⁴⁹, Isabella Zanella^51,41, Matteo Della Monica⁵², Carmelo Piscopo⁵², Mario Capasso^53,54,55, Roberta Russo^53,54, Immacolata Andolfo^53,54, Achille Iolascon^53,54, Giuseppe Fiorentino⁵⁵, Massimo Carella⁵⁶, Marco Castori⁵⁶, Filippo Aucella⁵⁷, Pamela Raggi⁵⁸, Carmen Marciano⁵⁸, Rita Perna⁵⁸, Matteo Bassetti^59,60, Antonio Di Biagio^59,60, Maurizio Sanguinetti^61,62, Luca Masucci^61,62, Serafina Valente⁶³, Marco Mandalà⁶⁴, Alessia Giorli⁶⁴, Lorenzo Salerni⁶⁴, Patrizia Zucchi⁶⁵, Pierpaolo Parravicini⁶⁵, Elisabetta Menatti⁶⁶ Stefano Baratti⁴⁵, Tullio Trotta⁶⁷, Ferdinando Giannattasio⁶⁷, Gabriella Coiro⁶⁷, Fabio Lena⁶⁸, Domenico A. Coviello⁶⁹, Cristina Mussini⁷⁰, Giancarlo Bosio⁷¹, Enrico Martinelli⁷¹, Sandro Mancarella⁷², Luisa Tavecchia⁷², Mary Ann Belli⁷², Lia Crotti^73,74,75,76, Gianfranco Parati^73,74, Marco Gori^77,78, Maurizio Sanarico⁷⁹, Stefano Ceri⁸⁰, Pietro Pinoli⁸⁰, Francesco Raimondi⁸¹, Filippo Biscarini⁸², Alessandra Stella⁸², Marco Rizzi⁸³, Franco Maggiolo⁸³, Diego Ripamonti⁸³, Claudia Suardi⁸⁴, Tiziana Bachetti⁸⁵, Maria Teresa La Rovere⁸⁶, Simona Sarzi-Braga⁸⁷, Maurizio Bussotti⁸⁸, Katia Capitani^2,89, Kristina Zguro², Simona Dei⁹⁰, Sabrina Ravaglia⁹¹, Rosangela Artuso⁹², Antonio Perrella⁹³, Francesco Bianchi^2,93, Giuseppe Merla^53,94, Gabriella Maria Squeo⁹⁴, Mario Tumbarello^2,19, Ilaria Rancan^2,19, Davide Romani³⁰, Manola Pisani³¹, Stefano Busani³⁸, Andrea Tommasi⁴³, Francesco Castelli⁴⁹, Eugenia Quiros-Roldan⁴⁹, Alessandra Guarnaccia⁶¹, Oreste De Vivo⁶³, Gabriella Doddato^1,2, Annarita Giliberti^1,2, Francesca Ariani^1,2,17, Gianluca Lacerenza⁹⁵, Elena Andreucci⁹², Giulia Gori⁹², Angelica Pagliazzi⁹², Erika Fiorentini⁹², Paola Bergomi⁹⁶, Emanuele Catena⁹⁶, Riccardo Colombo⁹⁶, Sauro Luchi⁹⁷, Giovanna Morelli⁹⁷, Paola Petrocelli⁹⁷, Sarah Iacopini⁹⁷, Sara Modica⁹⁷, Silvia Baroni⁹⁸, Francesco Vladimiro Segala⁹⁹, Francesco Menichetti¹⁰⁰, Marco Falcone¹⁰⁰, Giusy Tiseo¹⁰⁰, Chiara Barbieri¹⁰⁰, Tommaso Matucci¹⁰⁰, Davide Grassi¹⁰¹, Claudio Ferri¹⁰¹, Franco Marinangeli¹⁰², Francesco Brancati¹⁰³, Antonella Vincenti¹⁰⁴, Valentina Borgo¹⁰⁴, Lombardi Stefania¹⁰⁴, Mirco Lenzi¹⁰⁴, Massimo Antonio Di Pietro¹⁰⁵, Francesca Vichi¹⁰⁵, Benedetta Romanin¹⁰⁵, Letizia Attala¹⁰⁵, Cecilia Costa¹⁰⁵, Andrea Gabbuti¹⁰⁵, Menè Roberto¹⁰⁶, Umberto Zuccon¹⁰⁷, Lucia Vietri¹⁰⁷, Patrizia Casprini¹⁰⁸, Marcello Maffezzoni¹⁰⁹ and Marta Colaneri¹¹⁰

¹⁹Department of Medical Sciences, Infectious and Tropical Diseases Unit, Azienda Ospedaliera Universitaria Senese, Siena, Italy

²⁰Respiratory Diseases Unit, Department of Medicine, Surgery and Neurosciences, Siena University Hospital, Siena, Italy

²¹Department of Emergency and Urgency, Medicine, Surgery and Neurosciences, Unit of Intensive Care Medicine, Siena University Hospital, Italy

²²Department of Medical, Surgical and Neurosciences and Radiological Sciences, Unit of Diagnostic Imaging, University of Siena

²³Rheumatology Unit, Department of Medicine, Surgery and Neurosciences, University of Siena, Policlinico Le Scotte, Italy

²⁴Department of Specialized and Internal Medicine, Infectious Diseases Unit, San Donato Hospital Arezzo, Italy

²⁵Department of Emergency, Anesthesia Unit, San Donato Hospital, Arezzo, Italy

²⁶Department of Specialized and Internal Medicine, Pneumology Unit and UTIP, San Donato Hospital, Arezzo, Italy

²⁷Department of Emergency, Anesthesia Unit, Misericordia Hospital, Grosseto, Italy

²⁸Department of Specialized and Internal Medicine, Infectious Diseases Unit, Misericordia Hospital, Grosseto, Italy

²⁹Clinical Chemical Analysis Laboratory, Misericordia Hospital, Grosseto, Italy

³⁰Department of Preventive Medicine, Azienda USL Toscana Sud Est, Italy

³¹Territorial Scientific Technician Department, Azienda USL Toscana Sud Est, Italy

³²Laboratory Medicine Department, San Donato Hospital, Arezzo, Italy

³³Chirurgia Vascolare, Ospedale Maggiore di Crema, Italy

³⁴Department of Health Sciences, Clinic of Infectious Diseases, ASST Santi Paolo e Carlo, University of Milan, Italy

³⁵Division of Infectious Diseases and Immunology, Fondazione IRCCS Policlinico San Matteo, Pavia, Italy

³⁶Department of Internal Medicine and Therapeutics, University of Pavia, Italy

³⁷Fondazione IRCCS Ca’ Granda Ospedale Maggiore Policlinico, Milano, Italy

³⁸Department of Anesthesia and Intensive Care, University of Modena and Reggio Emilia, Modena, Italy

³⁹Department of Medical and Surgical Sciences for Children and Adults, University of Modena and Reggio Emilia, Modena, Italy

⁴⁰HIV/AIDS Department, National Institute for Infectious Diseases, IRCCS, Lazzaro Spallanzani, Rome, Italy

⁴¹III Infectious Diseases Unit, ASST-FBF-Sacco, Milan, Italy

⁴²Department of Biomedical and Clinical Sciences Luigi Sacco, University of Milan, Milan, Italy

⁴³Infectious Diseases Clinic, Department of Medicine , University of Perugia, Santa Maria della Misericordia Hospital, Perugia, Italy

⁴⁴

⁴⁵Department of Infectious Diseases, Treviso Hospital, Local Health Unit 2 Marca Trevigiana, Treviso, Italy

⁴⁶Clinical Infectious Diseases, Mestre Hospital, Venezia, Italy.

⁴⁷Infectious Diseases Clinic, ULSS1, Belluno, Italy

⁴⁸Department of Molecular Medicine, University of Padova, Italy

⁴⁹Department of Infectious and Tropical Diseases, University of Brescia and ASST Spedali Civili Hospital, Brescia, Italy

⁵⁰Department of Molecular and Translational Medicine, University of Brescia, Italy;

⁵¹Clinical Chemistry Laboratory, Cytogenetics and Molecular Genetics Section, Diagnostic Department, ASST Spedali Civili di Brescia, Italy

⁵²Medical Genetics and Laboratory of Medical Genetics Unit, A.O.R.N. “Antonio Cardarelli”, Naples, Italy

⁵³Department of Molecular Medicine and Medical Biotechnology, University of Naples Federico II, Naples, Italy

⁵⁴CEINGE Biotecnologie Avanzate, Naples, Italy

⁵⁵Unit of Respiratory Physiopathology, AORN dei Colli, Monaldi Hospital, Naples, Italy

⁵⁶Division of Medical Genetics, Fondazione IRCCS Casa Sollievo della Sofferenza Hospital, San Giovanni Rotondo, Italy

⁵⁷Department of Medical Sciences, Fondazione IRCCS Casa Sollievo della Sofferenza Hospital, San Giovanni Rotondo, Italy

⁵⁸Clinical Trial Office, Fondazione IRCCS Casa Sollievo della Sofferenza Hospital, San Giovanni Rotondo, Italy

⁵⁹Department of Health Sciences, University of Genova, Genova, Italy

⁶⁰Infectious Diseases Clinic, Policlinico San Martino Hospital, IRCCS for Cancer Research Genova, Italy

⁶¹Microbiology, Fondazione Policlinico Universitario Agostino Gemelli IRCCS, Catholic University of Medicine, Rome, Italy

⁶²Department of Laboratory Sciences and Infectious Diseases, Fondazione Policlinico Universitario A. Gemelli IRCCS, Rome, Italy

⁶³Department of Cardiovascular Diseases, University of Siena, Siena, Italy

⁶⁴Otolaryngology Unit, University of Siena, Italy

⁶⁵Department of Internal Medicine, ASST Valtellina e Alto Lario, Sondrio, Italy

⁶⁶Study Coordinator Oncologia Medica e Ufficio Flussi, Sondrio, Italy

⁶⁷First Aid Department, Luigi Curto Hospital, Polla, Salerno, Italy

⁶⁸Local Health Unit-Pharmaceutical Department of Grosseto, Toscana Sud Est Local Health Unit, Grosseto, Italy

⁶⁹U.O.C. Laboratorio di Genetica Umana, IRCCS Istituto G. Gaslini, Genova, Italy.

⁷⁰Infectious Diseases Clinics, University of Modena and Reggio Emilia, Modena, Italy.

⁷¹Department of Respiratory Diseases, Azienda Ospedaliera di Cremona, Cremona, Italy

⁷²U.O.C. Medicina, ASST Nord Milano, Ospedale Bassini, Cinisello Balsamo (MI), Italy

⁷³Istituto Auxologico Italiano, IRCCS, Department of Cardiovascular, Neural and Metabolic Sciences, San Luca Hospital, Milan, Italy

⁷⁴Department of Medicine and Surgery, University of Milano-Bicocca, Milan, Italy

⁷⁵Istituto Auxologico Italiano, IRCCS, Center for Cardiac Arrhythmias of Genetic Origin, Milan, Italy

⁷⁶Istituto Auxologico Italiano, IRCCS, Laboratory of Cardiovascular Genetics, Milan, Italy

⁷⁷University of Siena, Diism-sailab, Siena, Italy

⁷⁸University Cote d’Azur, Inria, CNRS, I3S, Maasai

⁷⁹Independent Data Scientist, Milan, Italy

⁸⁰Department of Electronics, Information and Bioengineering (DEIB), Politecnico di Milano, Milano, Italy

⁸¹Scuola Normale Superiore, Pisa, Italy

⁸²CNR-Consiglio Nazionale delle Ricerche, Istituto di Biologia e Biotecnologia Agraria (IBBA), Milano, Italy

⁸³Unit of Infectious Diseases, ASST Papa Giovanni XXIII Hospital, Bergamo, Italy

⁸⁴Fondazione per la ricerca Ospedale di Bergamo, Bergamo, Italy

⁸⁵Direzione Scientifica, Istituti Clinici Scientifici Maugeri IRCCS, Pavia, Italy

⁸⁶Istituti Clinici Scientifici Maugeri IRCCS, Department of Cardiology, Institute of Montescano, Pavia, Italy

⁸⁷Istituti Clinici Scientifici Maugeri, IRCCS, Department of Cardiac Rehabilitation, Institute of Tradate (VA), Italy

⁸⁸Istituti Clinici Scientifici Maugeri IRCCS, Department of Cardiology, Institute of Milan, Milan, Italy

⁸⁹IRCCS C. Mondino Foundation,Pavia, Italy

⁹⁰Core Research Laboratory, ISPRO, Florence, Italy

⁹¹Health Management, Azienda USL Toscana Sudest, Tuscany, Italy

⁹²Medical Genetics Unit, Meyer Children’s University Hospital, Florence, Italy

⁹³Department of Medicine, Pneumology Unit, Misericordia Hospital, Grosseto, Italy

⁹⁴Laboratory of Regulatory and Functional Genomics, Fondazione IRCCS Casa Sollievo della Sofferenza, San Giovanni Rotondo (Foggia), Italy

⁹⁵Department of Pharmaceutical Medicine, Misericordia Hospital, Grosseto, Italy

⁹⁶Department of Anesthesia and Intensive Care Unit, ASST Fatebenefratelli Sacco, Luigi Sacco Hospital, Polo Universitario, University of Milan, Milan, Italy

⁹⁷Infectious Disease Unit, Hospital of Lucca, Lucca, Italy

⁹⁸Department of Diagnostic and Laboratory Medicine, Unity of Chemistry, Biochemistry and Clinical Molecular Biology, Fondazione Policlinico Universitario A. Gemelli IRCCS, Catholic University of the Sacred Heart, Rome, Italy

⁹⁹Clinic of Infectious Diseases, Catholic University of the Sacred Heart, Rome, Italy

¹⁰⁰Department of Clinical and Experimental Medicine, Infectious Diseases Unit, University of Pisa, Pisa, Italy

¹⁰¹Department of Clinical Medicine, Public Health, Life and Environment Sciences, University of L’Aquila, L’Aquila, Italy

¹⁰²Anesthesiology and Intensive Care, University of L’Aquila, L’Aquila, Italy

¹⁰³Medical Genetics Unit, Department of Life, Health and Environmental Sciences, University of L’Aquila, L’Aquila, Italy

¹⁰⁴Infectious Disease Unit, Hospital of Massa, Massa Carrara, Italy

¹⁰⁵Unit of Infectious Diseases, S.M. Annunziata Hospital, Florence, Italy

¹⁰⁶Istituto Auxologico Italiano, IRCCS, Department of Cardiovascular, Neural and Metabolic Sciences, San Luca Hospital; Department of Medicine and Surgery, University of Milano-Bicocca, Milan, Italy

¹⁰⁷Respiratory Diseases Unit, “Santa Maria degli Angeli” Hospital, Pordenone, Italy

¹⁰⁸Laboratory of Clinical Pathology and Immunoallergy, Florence-Prato, Italy

¹⁰⁹University of Pavia, Pavia, Italy

¹¹⁰Division of Infectious Diseases I, Fondazione IRCCS Policlinico San Matteo, Pavia, Italy