Clinical and Immunological Characteristics of Human Infections With H5N6 Avian Influenza Virus

Abstract

Infection with various subtypes of avian influenza virus (AIV) in humans typically results in severe respiratory disease. H5N1 infection was first documented in humans in 1997. Since then, at least 860 infections with 454 deaths (case fatality rate [CFR], 52.8%) were attributed to H5N1 virus worldwide, as of 2 March 2018 [1]. H7N9 infection emerged in China in 2013 [2] and caused 1567 infections with 615 deaths (CFR, 39.2%) to date [1]. Concomitant to the H7N9 outbreak was the emergence of H10N8 (3 infections with 2 deaths; CFR, 67%) and H6N1 infections (1 infection, survived) [1]. H7N9 virus was found to be more transmissible to humans compared to H5N1 virus, because H7N9 virus can bind to both avian- and human-origin sialic acid receptors [3–5], a feature not always observed with H5N1 virus. The first human infection with severe disease caused by the novel H5N6 AIV subtype was reported in China in April 2014 [1, 6, 7]. Only China has reported human H5N6 infections thus far, with a total of 19 cases and 13 deaths (CFR, 68.4%) [1, 8, 9]. However, recent studies showed that some H5N6 virus strains also possessed the ability to bind both avian- and human-origin sialic acid receptors and were shown to be more transmissible than H5N1 virus in a ferret model [10, 11], indicating that this pathogen may be of high public health risk.

The acute phase of AIV infection is usually accompanied with elevated and abnormal production of multiple cytokines and chemokines, such as interferon gamma induced protein 10 (IP-10), macrophage inflammatory protein 1β (MIP-1β), interleukin (IL) 6, and IL-8, suggestive of a dysfunctional immune response. Higher concentrations of multiple cytokines/chemokines are known to be indicative of AIV disease severity and unfavorable survival outcomes [12–14]. Clinical data from 5 H5N6 infections in humans are available [6, 15–20]. However, H5N6 virus-specific immune responses and clinical characteristics of H5N6 infections relative to other high-profile AIV subtypes have not yet been systematically evaluated. In this study, we analyzed the clinical characteristics among patients infected with H5N6 (n = 19), H5N1 (n = 53), and H7N9 (n = 160) viruses. Of these, data from 5 H5N6 cases (with severe disease), 2 H5N1 cases, and 37 H7N9 cases were collected by our research group and previously unpublished. We described and compared the clinical data and concentrations of serum cytokines/chemokines from H5N6 patients with other related influenza A viruses and characterized specific immune responses from the surviving (n = 1) and nonsurviving (n = 2) H5N6 patients.

MATERIALS AND METHODS

Patient Information

Subjects presented in this study were hospitalized patients with laboratory-confirmed H5N6 (n = 5), H5N1 (n = 2), H7N9 (n = 37), and 2009 pandemic H1N1 (pH1N1) (n = 7) infections or bacterial pneumonia (n = 4). Healthy controls (n = 10) were also included. All 5 H5N6 patients had severe disease. Three H5N6 cases (patients 6, 8, and 10) were hospitalized in the Shenzhen municipality of Guangdong Province, and 2 H5N6 cases (patients 4 and 5) were identified by the Yunnan Center for Disease Control and Prevention. Samples from patients 6 (nonsurvivor 1), 8 (survivor), and 10 (nonsurvivor 2) were analyzed for immune responses after infection and compared based on survival. The H7N9, pH1N1, and bacterial pneumonia patients and the healthy controls were obtained at the same laboratory.

Data Collection and Analysis of Clinical Findings

Clinical information, including complete blood counts and blood biochemistry from our H5N6 patients, were collected at the earliest time-point after hospitalization. These results were combined with 14 additional H5N6 cases reported or previously published [6, 15–20]. Individual parameters were compared with data collected from H5N1 (n = 53, 2 previously unpublished) and H7N9 (n = 160, 37 previously unpublished) patients during 2013–2015 [21–23].

Lymphocyte Purification and Intracellular Cytokine Staining

Peripheral blood mononuclear cells (PBMCs) were isolated from the whole blood of donors by density gradient centrifugation using Ficoll-Hypaque (TBD Science) and washed twice in RPMI 1640 medium containing 10% fetal bovine serum (Life Technologies). The PBMCs were cryopreserved in liquid nitrogen before analysis. PBMCs were recovered and kept at room temperature for 2 hours, and then cultured with H5N6 virus in the Biosafety Level 3 laboratory for 2 hours. The cells were incubated for an additional 4 hours with GolgiStop/monesin (BD Biosciences) at 37°C in 5% carbon dioxide. Cells cultured with medium alone or phytohemagglutinin were used as the negative and positive controls, respectively. To determine the phenotypes and functional profiles of H5N6 virus-specific T cells, stimulated PBMCs were harvested and stained with anti–CD3-FITC, anti–CD4-APC-CY7, anti–CD8-BV510, anti–CCR7-AF700, and anti–CD45RA-PE-CF594 surface markers (BD Biosciences). The cells were then fixed and incubated in permeabilizing buffer (BD Biosciences), and then stained with anti–IFN-γ–PE-Cy7 and anti–IL-2–PE (BD Biosciences). All fluorescent lymphocytes were gated on a flow cytometer LSR Fortessa (BD Biosciences) and analyzed with FlowJo software.

Cytokine and Chemokine Measurements

The sera of patients with laboratory-confirmed infection of H5N6 (n = 4), H5N1 (n = 2), H7N9 (n = 6), and pH1N1 (n = 7) viruses were collected at the earliest possible time-point after hospitalization. The sera of healthy subjects (n = 10) and hospitalized patients with bacterial pneumonia (n = 4) were included as the negative and positive control groups, respectively. The concentrations/activities of 48 cytokines and chemokines were measured using the Bio-Plex Pro Human Cytokine Array 27-Plex Group I and 21-Plex Group II Kits (Bio-Rad) on a Luminex 200 (Luminex Multiplexing Instrument, Merck Millipore) following manufacturer instructions.

Hemagglutination Inhibition and Neutralizing Antibody Assays

Patient sera were first treated by receptor-destroying enzyme (RDE; Denka Seiken UK). Sera were mixed with RDE at a ratio of 1:4, incubated at 37°C overnight, heated at 56°C for 30 minutes, and then incubated with a final concentration of 20% chicken red blood cells to inactivate any remaining RDE. The treated sera were used to determine hemagglutinin inhibition (HI) and neutralizing antibody (nAb) concentrations using the A/Shenzhen/Th002/2015(H5N6) virus, following the method described in a previous study [24].

Real-time Reverse-transcription Quantitative Polymerase Chain Reaction

The viral load was measured on sputum from the respiratory tract and/or oral swabs, collected from patients at various time-points after hospitalization. Viral RNA was extracted and real-time reverse-transcription polymerase chain reaction (RT-qPCR) was performed as previously described [25].

Quantification of Hypoxia and Lung Injury

The partial pressure of oxygen (PaO₂) in arterial blood taken from the patients at various time-points after hospitalization is measured by the ABL90 blood gas analyzer (Radiometer). The fraction of inspired oxygen (FiO₂) is calculated by the following formula: FiO₂ = (21 + oxygen flow [in units of L/minute] × 4) / 100. The PaO₂/FiO₂ ratio (in units of mm Hg) is calculated by dividing the PaO₂ value with the FiO₂ value. A PaO₂/FiO₂ ratio ≤100 mm Hg is considered one of the criteria for severe acute respiratory distress syndrome (ARDS).

Statistical Analysis

The unpaired, 2-tailed t test was used to determine whether the differences in the cytokine and chemokine levels between H5N6 and another chosen group was statistically significant. The Pearson correlation coefficient analysis was used to analyze linear correlation. Both tests were performed with GraphPad Prism. P values of .01–.05, .001–.01, and .0001–.001 were considered statistically significant, very significant, and extremely significant, respectively.

RESULTS

Timeline and Features of H5N6 Infection in Humans

The timeline for all reported 19 human infections with H5N6 virus in China to date is shown in Figure 1, with patients numbered chronologically in order of disease report date. Almost all H5N6 patients were found to have had prior exposure to poultry or live poultry markets. Except for case 1 (female, 5 years old) and case 19 (female, 3 years old) who were outpatient cases with mild respiratory symptoms [1, 17], the other 17 cases all had severe disease [6, 15, 16, 18–20]. Generally, the hospitalized H5N6 patients (patients 2–18) initially showed flu-like symptoms including fever, sore throat, headache, chills, cough, and myalgia, quickly developing into shortness of breath, cough with blood-tinged sputum, and vomiting in some cases at 1–7 days after symptom onset. ARDS, organ failure and death between 6–31 days after symptom onset were often the outcomes in moribund patients. Chest radiographs and computed tomographic scans of patients 4, 5, 6, 8, and 10 are shown (Supplementary Figure 1), in which severe lung lesions could be observed in all patients after 5 days after symptom onset.

Figure 1.

A timeline of human infections with H5N6 virus in China. Patients are ordered in chronological order based on disease report date with the geographical location in parentheses. Various milestones in the disease course are indicated with different-colored circles. The numbers of H5N6 cases presented in this study are marked in red. Abbreviations: ARDS, acute respiratory distress syndrome; ICU, intensive care unit.

Open in new tabDownload slide

Clinical Comparison of H5N6, H7N9, and H5N1 Infections in Humans

To compare the severity of H5N6 disease to that of past major AIV outbreaks of human infections, clinical information was next compiled and summarized for hospitalized patients infected with H7N9 (n = 160, collected during 2013–2015), H5N1 (n = 53, collected during 2003–2015), and H5N6 (n = 17, collected during 2014–2017) viruses in China (Supplementary Table 1). The median age for H7N9, H5N1, and H5N6 cases presented in this study was 59, 27, and 40 years old, respectively. Due to the advanced age of some H7N9 patients, there was also a higher instance of coexisting medical conditions, such as hypertension (47%) and diabetes (17%), compared to that of H5N1 and H5N6 patients. The median intervals from onset to admission (in days) were similar, and all the hospitalized H5N6 cases (patients 2–18) were admitted into hospital within 7 days after symptom onset except patient 17, who was admitted 10 days after symptom onset. H7N9 patients were predominantly male (84%), while the male-to-female ratio of H5N1 and H5N6 patients was relatively equal. The smoking history was similar across the groups (12.5%–23%), and the majority of patients in all groups reported prior exposure to poultry (68%–94.1%).

A complete blood count with differential was assessed for each patient either on the date of hospital admission or at the earliest time-point thereafter (Table 1). Interestingly, the incidences of lymphopenia, as well as elevated alanine aminotransferase, serum creatinine, and lactate dehydrogenase, were higher in H5N6 patients, whereas the incidence of elevated aspartate aminotransferase was lower (Table 1).

Expression Profiles of Cytokines/Chemokines Among AIV and pH1N1 Infections in Humans

Because concentrations of various cytokines/chemokines after AIV infection can indicate disease severity, these parameters were measured and compared among hospitalized patients infected with H5N6 (n = 4), H5N1 (n = 2), and H7N9 (n = 6), as well as pH1N1 (n = 7) viruses and bacterial pneumonia (n = 4). All samples were collected at the earliest time possible after admittance. Healthy volunteers (n = 10) were sampled as a negative control. The results show that patients with H5N6 infections were higher on average in almost every parameter measured, including IL-2Rα, IL-3, IL-6, IL-10, IL-12p40, IL-18, interferon (IFN) α2, IP-10, angiotensin II, monokine induced gamma interferon (MIG), stem cell factor (SCF), stromal cell-derived factor 1α (SDF-1α), tumor necrosis factor related apoptosis-inducing ligand (TRAIL), hepatocyte growth factor (HGF), leukemia inhibitory factor (LIF), monocyte chemotactic protein 3 (MCP-3), and β nerve growth factor (β-NGF), with the exception of IL-8 (H5N6 comparable with pH1N1), MCP-1 (H5N6 comparable with pneumonia), stem cell growth factor (SCGF) (H5N6 comparable with H5N1), and macrophage colony stimulating factor (M-CSF) (H5N6 comparable with H5N1) (Figure 2). Importantly, of the 48 cytokines/chemokines tested, H5N6 patients were found to be significantly higher than healthy controls, H7N9 and pH1N1 patients in 21, 11, and 5 parameters, respectively (Figure 2), indicating that infection with different AIV subtypes could result in distinct cytokine/chemokine production profiles. H5N6 appears to result in more severe hypercytokinemia among infected patients compared with non-H5 AIV subtypes.

Figure 2.

Comparison of serum cytokine/chemokine concentrations between healthy volunteers as well as patients infected with H5N6, H5N1, H7N9, or pH1N1 viruses or bacterial pneumonia. Samples from patients infected with H5N6 (n = 4), H5N1 (n = 2), H7N9 (n = 6), or pH1N1 (n = 7) were collected at the earliest possible time-point after hospitalization for assays measuring the concentrations of 21 cytokines and chemokines. Healthy subjects (n = 10) and patients with bacterial pneumonia (n = 4) were involved as controls. Values were graphed on a logarithmic scale and presented in units of picograms per milliliter. The mean ± standard error is shown for each group, and the P values (as determined by t test) above the various groups represent the comparison of means between the H5N6 patients and the group in question. P values of .01–.05, .001–.01, and .0001–.001 were considered statistically significant (*), very significant (**), and extremely significant (***), respectively. Abbreviations: Ctrl, control; HGF, hepatocyte growth factor; IFN, interferon; IL, interleukin; IP-10, interferon gamma induced protein 10; LIF, leukemia inhibitory factor; M-CSF, macrophage colony stimulating factor; MCP, monocyte chemotactic protein; MIG, monokine induced gamma interferon; NGF, nerve growth factor; ns, not significant; Pneu, pneumonia; SCF, stem cell factor; SCGF, stem cell growth factor; SDF, stromal cell-derived factor; TRAIL, tumor necrosis factor related apoptosis-inducing ligand.

Open in new tabDownload slide

Dynamic Changes of Cytokine/Chemokine Levels in the Survivor Versus Nonsurvivors of H5N6 Infection

Concentrations of various cytokines/chemokines after H5N6 infections were assayed at the indicated times from the sera of patients 6 (nonsurvivor 1), patient 8 (survivor), and patient 10 (nonsurvivor 2). Aside from IL-10, elevated concentrations of both pro- and anti-inflammatory cytokines were observed in nonsurvivors including IL-2Rα, IL-3, IL-6, IL-8, IL-12p40, IL-18, IFN-α2, IP-10, angiotensin II, MCP-1, MCP-3, MIG, SCF, SCGF, SDF-1α, TRAIL, HGF, LIF, M-CSF, and β-NGF (Figure 3). Elevated concentrations of IL-6, IL-8, and MCP-1 were also found in serous fluids from the pleural cavity of nonsurvivor 1 (Supplementary Figure 2), suggesting hydrothorax and lung failure. In contrast, for the survivor there was either a reduction in certain cytokines/chemokines, such as IL-12p40, IP-10, and MIG, or very little change in the concentrations during acute H5N6 disease and recovery (Figure 3). Concentrations of IL-2Rα, IL-3, IL-12p40, IFN-α2, LIF, MCP-3, TRAIL, and β-NGF significantly correlated with the PaO₂/FiO₂ ratio in patients 6, 8, and 10 at the time-points when taken during the course of H5N6 disease (Supplementary Figure 3). Additionally, concentrations of certain cytokines and chemokines (MIG, angiotensin II, IL-8, SCGF-β, and M-CSF) negatively correlated with cycle threshold values of the survivor and nonsurvivor 2 determined by RT-qPCR (Supplementary Figure 4).

Figure 3.

Changes of cytokine/chemokine concentrations in H5N6 patients over time. Samples from 3 H5N6 patients (patient 6, nonsurvivor 1; patient 8, survivor; patient 10, nonsurvivor 2 denoted as circle, square, and triangle, respectively) were collected over time following hospitalization and assayed for the concentrations of 21 different cytokines and chemokines, which showed significant differences vs those in healthy subjects. Values are graphed on a logarithmic scale and presented in units of picograms per milliliter. Abbreviations: HGF, hepatocyte growth factor; IFN, interferon; IL, interleukin; IP-10, interferon gamma induced protein 10; LIF, leukemia inhibitory factor; M-CSF, macrophage colony stimulating factor; MCP, monocyte chemotactic protein; MIG, monokine induced gamma interferon; NGF, nerve growth factor; SCF, stem cell factor; SCGF, stem cell growth factor; SDF, stromal cell-derived factor; TRAIL, tumor necrosis factor related apoptosis-inducing ligand.

Open in new tabDownload slide

H5N6 Virus-specific Immune Responses in the Survivor Versus Nonsurvivors of H5N6 Infection

Substantial changes in white blood cell (WBC) and lymphocyte (LYM) counts were not observed in nonsurvivor 1 due to an early death after disease onset. Absolute counts of WBCs and LYMs were considerably increased in nonsurvivor 2. However, the percentage of LYM (LYM%) did not rise, suggesting that other components of WBC, such as neutrophils, macrophages, and eosinophils, were largely responsible for the observed increase of WBC count as a result of a dysregulated inflammatory response (Supplementary Figure 5). In contrast, an increase in WBC count was not observed in the survivor, whereas a modest and substantial increase was observed with LYM and LYM%, respectively (Supplementary Figure 5), suggesting that a controlled immune response may have contributed to survival.

Cell-mediated immunity was analyzed following H5N6 virus infection (Figure 4A–C). An H5N6 virus-specific response, as determined by measuring the presence of IFN-γ–secreting cells, was not detected in the nonsurvivors during the course of disease. In contrast, the survivor displayed a substantial increase in the number of IFN-γ–secreting cells, rising from 0.031% at day 7 to 4.15% by day 18, the date of discharge from hospital (Figure 4A). In-depth analysis of the IFN-γ⁺ T-cell response in the survivor showed a bias toward CD8⁺ T-cell–mediated immunity (Figure 4B) and increasing numbers of virus-specific effector T cells (CD45RA⁺CCR7^–) between 10 and 18 days after symptom onset (Figure 4B). Furthermore, increasing numbers of virus-specific IL-2 secreting cells were detected only in the survivor (Figure 4C), and in-depth analysis of the IL-2⁺ T-cell response showed a bias toward CD4⁺ T-cell–mediated immunity, with the effector memory T-cell population (CD45RA^–CCR7^–) increasing between days 10 and 18 (Figure 4C).

Figure 4.

H5N6 virus-specific cellular and humoral immunity in H5N6 patients after the onset of illness. Peripheral blood mononuclear cells and sera from nonsurvivor 1 (patient 6), survivor (patient 8), and nonsurvivor 2 (patient 10) were extracted at different days after symptom onset at the acute phase and used for this study. A, Detection of interferon gamma (IFN-γ)–secreting cells during the course of H5N6 disease in nonsurvivor 1 (patient 6), survivor (patient 8), and nonsurvivor 2 (patient 10). B, CD4⁺ and CD8⁺ dependency of IFN-γ–secreting cells and characterization of memory phenotypic T-cell subsets in the survivor (patient 8). C, Detection of interleukin 2 (IL-2)–secreting cells during H5N6 disease in the survivor (patient 8) and characterization of central memory and effector memory T-cell subsets based on IL-2–secreting cells. D, Hemagglutinin inhibition (HI) or neutralizing antibody (nAb) response to H5N6 infection in nonsurvivor 1 (patient 6, red), survivor (patient 8, blue), and nonsurvivor 2 (patient 10, green). Antibody titers are presented in units of reciprocal dilutions. Reciprocal dilutions >20 are considered positive for HI or nAb and are denoted by a dashed line. Abbreviations: HI, hemagglutinin inhibition; IFN-γ, interferon gamma; IL-2, interleukin 2; nAbs, neutralizing antibodies; SSC-A, side scatter area.

Open in new tabDownload slide

H5N6 virus-specific humoral immunity was also characterized after the onset of the disease (Figure 4D). In nonsurvivor 1, HI and nAb assays at 3 days after symptom onset showed that the concentrations were 10 and 40 reciprocal dilutions, respectively. In nonsurvivor 2, the patient displayed high HI and nAb titers and had increasing levels of nAb (40 to 640) from 7 to 13 days after symptom onset (Figure 4D). In the survivor, the HI and nAb titers were 40 and 160, respectively, at 7 days after symptom onset, and remained detectable until the date of discharge (Figure 4D).

DISCUSSION

H5N6 AIV has been endemic among poultry in China and Southeast Asia since 2013 [26] and gradually replaced H5N1 AIV as a dominant subtype in poultry across southern China [9], similar to H7N9 AIV [27]. Additionally, H5N6 virus has been circulating in wild birds and already caused several outbreaks in Asia and Europe [26]. A recently published study showed that avian isolates of H5N6 virus are more transmissible than H5N1 virus in ferrets [11]. Some H5N6 virus strains were found to possess the ability to bind human-origin receptors [10, 11], representing a notable step toward potential virus transmission in humans [28]. These findings suggest that H5N6 virus is an epidemic, or even pandemic threat, even though human H5N6 infections have only been reported in China thus far.

The genetic reassortment of AIVs is known to influence the viral pathogenicity and transmissibility to mammals [29, 30] and has contributed to the emergence of at least 3 pandemic influenza viruses, H2N2/1957, H3N2/1968, and H1N1/2009 [31, 32]. Interestingly, both H7N9 and H5N6 viruses present as diverse genetic characteristics by reassortment with low pathogenicity AIVs (eg, H9N2 virus) [9, 15, 27, 33–35], and the genetic diversity of H7N9 is considered as an underlying reason for the sharp increase in human infections during 2016–2017 [27]. Although H5N6 and H7N9 viruses both possess internal genes from H9N2 virus, the pathogenicity in humans is different. H5N6 virus seems to be more virulent to humans than H7N9 and H5N1 viruses based on the overall CFRs of 68.4% (H5N6), 39.2% (H7N9), and 52.8% (H5N1). Until recently, at least 4 genotypes or reassortments of H5N6 viruses have been found in the 19 reported human infections [9, 35]. The 5 H5N6 cases presented in our study were all caused by a dominant H5N6 genotype (G1.2) whose internal genes were all originated from H9N2 virus [9], similar to circulating H7N9 viruses. Thus, the disease severity among the studied H5N6 cases was consistent.

Although there is a limited number of reported H5N6 infections in humans thus far, many of the cytokine/chemokine markers tested are still significantly higher for H5N6 patients, compared to those infected with H7N9, pH1N1 viruses, or bacteria-induced pneumonia. Statistical significance is not reached between H5N6 and H5N1 infections, while the average and individual values between many markers tend to trend higher for H5N6. These results suggest that H5N6 disease is at least comparable in disease severity to that of H5N1, and the availability of samples from more H5N6 patients will determine whether differences in the markers are significant.

Survival and recovery from AIV infection is dependent on the development of robust, specific B- and T-cell immunity in the host resulting in a memory response, as has been observed previously with H7N9 and pH1N1 infections [36–39], as well as humoral immunity for H5N1 [40]. Interestingly, there have been no reports so far on CD8⁺ T-cell immunity in patients infected with H5N1 virus [41]. In our study, 3 H5N6 cases were tested for immune responses. Comparisons at a similar observation time can be made only for 2 patients (survivor vs nonsurvivor 2), as nonsurvivor 1 died too soon to be compared at the same observation times. High levels of HI and nAb titers were detected in both the survivor and nonsurvivor 2; however, a specific T-cell response was only observed in the survivor (Figure 4). It is possible that the presence of H5N6 virus-specific T-cell responses played a role in the control of disease progression and virus clearance, thus contributing to survival. The unfavorable survival outcome in nonsurvivor 2 suggests that the infection was not fully controlled with antibody responses alone. Cytokine/chemokine assays show that the surviving patient had lower concentrations of both pro- and anti-inflammatory cytokines/chemokines compared to the nonsurvivors. Taken together, the present findings suggest it is possible that the earlier acquisition of cellular immunity and lower concentrations of cytokines/chemokines contribute to survival, but analysis of more H5N6 patient samples will be needed to draw definite conclusions.

Taking all epidemiological, virological, immunological, and clinical factors into account, H5N6 virus poses a high public health risk comparable to, and possibly greater than, other currently circulating AIV subtypes. Proactive and precautionary measures, including studies monitoring for the evolution, drug resistance, and immune response of H5N6 virus in poultry, as well as the prevalence of H5N6 virus in migratory birds worldwide, should be conducted as a part of preparation efforts against this AIV subtype.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Financial support. This work was supported by the National Key Research and Development Project of China (grant number 2016YFE0205800); the National Science and Technology Major Project (grant numbers 2016ZX10004222, 2018ZX10733403 and 2018ZX10713001-010); the Ministry of Science and Technology of China (MOST) 973 Project (grant number 2015CB910501); the Sanming Project of Medicine in Shenzhen (grant number SZSM201412003); the Shenzhen Science and Technology Research and Development Project (grant numbers JCYJ20160427153238750 and JCYJ20160427151920801); and the National Natural Science Foundation of China (grant number 31870163). G. F. G. is a leading principal investigator of the National Science Foundation of China Innovative Research Group (grant number 81621091). Y. B. is supported by the National Natural Science Fund for Outstanding Young Scholars (grant number 31822055) and Youth Innovation Promotion Association of Chinese Academy of Sciences (grant number 2017122).

Potential conflicts of interest. All authors: No reported conflicts.

All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

References

Author notes