Increased Neutrophil H2O2 Production and Enhanced Pulmonary Clearance of Klebsiella pneumoniae in G6PD A- Mice

The X-linked A− variant (rs1050828, Val68Met) in G6PDX accounts for glucose-6-phosphate (G6PD) deficiency in approximately 11% of African American males. This common, hypomorphic variant may impact pulmonary host defense and phagocyte function during pneumonia by altering levels of reactive oxygen species produced by host leukocytes. We used CRISPR-Cas9 technology to generate novel mouse strain with “humanized” G6PD A- variant containing non-synonymous Val68Met single nucleotide polymorphism. Male hemizygous or littermate wild-type (WT) controls were inoculated intratracheally with K. pneumoniae (KP2 serotype, ATCC 43816 strain,103 CFU inoculum). We examined leukocyte recruitment, organ bacterial burden, bone marrow neutrophil and macrophage (BMDM) phagocytic capacity, and hydrogen peroxide (H2O2) production. Unexpectedly, G6PD-deficient mice showed decreased lung bacterial burden (p=0.05) compared to controls 24-h post-infection. Extrapulmonary dissemination and bacteremia were significantly reduced in G6PD-deficient mice 48-h post-infection. Bronchoalveolar lavage fluid (BALF) IL-10 levels were elevated in G6PD-deficient mice (p=0.03) compared to controls at 24-h but were lower at 48-h (p=0.03). G6PD A- BMDMs show mildly decreased in vitro phagocytosis of pHrodo-labeled KP2 (p=0.03). Baseline, but not stimulated, H2O2 production by G6PD A- neutrophils was greater compared to WT neutrophils. G6PD A- variant demonstrate higher basal neutrophil H2O2 production and are protected against acute Klebsiella intrapulmonary infection.


Background
Glucose-6-phosphate dehydrogenase (G6PD) de ciency is among the most common red blood cell (RBC) enzymopathies worldwide 1 .It affects over 500 million individuals 1 .While it is predominantly clinically silent, complete loss of enzyme activity is lethal in human and mouse models 1 .G6PD de ciency results in decreased production of NADPH via the pentose phosphate pathway, leading to impaired antioxidant activity by key NADPH-dependent pathways, impairing glutathione reductase 1 , and increasing oxidized glutathione and hydrogen peroxide (H 2 O 2 ) levels.Over 200 causative mutations in the G6PDX gene are associated with known clinical variants that are classi ed into ve categories based on residual enzyme activity 2,3 .The A-, or African, variant is a class III variant, indicating it causes "mild" disease 4 .It is the most common G6PD variant and affects up to 11% of African American men 5 .
Despite being one of the most widely-studied enzyme de ciencies, key questions remain regarding the functional relevance of G6PD de ciency due to seemingly con icting ndings.A meta-analysis previously reported that G6PD de ciency confers protection against malaria potentially due to enhanced early phagocytosis of infected erythrocytes 6,7 .This protection is thought to have exerted selective pressure in malaria-endemic regions, thus accounting for its global prevalence.This is in contrast to recent investigations suggesting that G6PD de ciency may impair key immune functions, including in ammasome activation 8 , reduced NETosis 9 , and impaired macrophage and neutrophil oxidative burst 10 .Trauma patients with G6PD de ciency have longer duration of antibiotics and hospital stays, and circulating monocytes from these patients demonstrate higher levels of oxidative stress 11 .Furthermore, severe G6PD de ciency de ned as < 5% enzyme activity, can phenocopy chronic granulomatous disease (CGD), which is due to an inherited mutation in one of ve NADPH oxidase genes leading to severely decreased ROS production, and presents with a predisposition to recurrent catalasepositive bacterial and fungal infections 9 .In G6PD de cient patients, increased susceptibility to infections have been reported and include, but are not limited to, pneumonia without microbiologic diagnosis 12 , and speci c pathogens Serratia marcescens, Chromobacterium violaceum, Aspergillus spp.causing disseminated aspergillosis 13 , and bronchial infections in G6PD A-trauma patients 11 .
The role of G6PD activity in the immune response to infection remains controversial, with a variety of observtions noted in different experimental models, with a variety of methods used to inhibit enzyme activity 14,10 , which may impact generalization of results across studies.For example, one report observed impaired in ammasome activation in human G6PD-de cient peripheral blood monocytes (PBMCs) and in siRNA-mediated G6PD knockdown in LPS-primed THP1 cells 8 .Another demonstrated no difference in survival between G6PD-de cient mice and littermate controls after cecal ligation and puncture-induced sepsis 14 , while still another used a small-molecule inhibitor of G6PD to show that G6PD inhibition suppresses neutrophil oxidative burst 10 .Individuals with severe G6PD de ciency may be predisposed to recurrent bacterial lung infections, suggesting that residual G6PD activity may play a role in determining host leukoctye function.Moreover, the role of reactive oxygen species (ROS) has not been well-understood in the context of G6PD Aleukocytes given that G6PD catalyzes the generation of NADPH from NADP + and curtails hydroxyl free radical conversion from H 2 O 2 .While there are multiple cellular sources of NADPH in nucleated cells, evidence suggests they may be insu cient to maintain an adequate pool of NADPH 15 .Therefore, inhibition of G6PD may alter a cell's ability to handle its oxidative load; consistent with this idea, one study suggests that total ROS levels are increased in G6PD de cient monocytes from trauma patients 11 .Furthermore, hypochlorous acid (HOCl) production by neutrophil myeloperoxidase (MPO), a vigorous antimicrobial killing compound 16,17 , uses H 2 O 2 as its substrate.It is therefore possible that leukocytes with de cient G6PD activity may be able to harness the excess H 2 O 2 for enhanced antimicrobial activity.
To our knowledge, no study has examined the direct effect of the hypomorphic G6PD A-variant in a mouse model of bacterial acute lung infection.Closer examination of this variant may have important clinical implications since approximately 11% of African American males are hemizygous for this Xlinked variant.We hypothesized that G6PD de ciency due to the A-variant alters pulmonary host defense, neutrophil and macrophage function, by altering ROS levels produced by host leukocytes.To address this question we used a novel humanized murine model of G6PD A-to examine the host response to K. pneumoniae acute lung infection, a model where early neutrophil antimicrobial function is essential for host defense 18 .Our study demonstrates that decreased G6PD activity is protective against K. pneumoniae infection due to altered neutrophilic ROS production.

Animals
We used CRISPR-Cas9 technology 19,20 to create a humanized murine model of G6PD de ciency containing the G6PD A-variant (rs1050828, Val68Met).Valine 68 was substituted to Methionine using an oligouncleotide template for homology-directed repair (HDR) containing the homologous human DNA sequence.Founder mice and N1 offspring were con rmed through sequence analysis and a stable colony was obtained through strategic back-crossing to wild-type mice.We previously reported that erythrocyte G6PD activity from hemizygous male mice was approximately 5% of WT-G6PD activity 21 .
For in vivo pneumonia studies, a sample size of 8 animals per group was determined a priori to achieve a power of 0.83 for a 2-tailed test using an α of 0.05.A skilled technician who was blinded to the biological hypothesis performed the inoculations, monitoring, and harvest.Only male mice were used in this study as G6PD de ciency is X-linked.Male mice aged 12 -14 weeks were used for experiments with WT littermates used as controls.Brie y, mice were inoculated via direct intratracheal instillation with K. pneumoniae at 10 3 CFU inoculum.Animals were sacri ced and tissues were harvested after 24-and 48hours in independent experiments.

Euthanasia and Ethical Approval
Animals were euthanized at time of experimental procedures using overdose iso urane followed by exsanguination via inferior vena cava.The mice were monitored carefully and euthanized if they met prede ned criteria for euthanasia.All procedures were performed with approval of the Institutional Animal Care and Use Committee at the University of Pittsburgh (protocol numbers 18052522 and 18063096).All procedures are in accordance with relevant guidelines and regulations including ARRIVE guidelines.Klebsiella pneumoniae strains K. pneumoniae strain: The strain used in this study was K. pneumoniae strain 43816 serotype 2 (American Type Culture Collection, Manassas, VA).K. pneumoniae inoculation K. pneumoniae strains were grown overnight in tryptic soy broth (TSB) for 18 hours shaking at 250 rpm at 37°C.A 1:100 dilution of bacteria in TSB was then incubated at 37°C for 1.5 hours.The bacterial inoculum was prepared in PBS to OD 600nm = 0.2.Mice were instilled with 10 3 CFU inoculum by the intratracheal route.We have previously described a detailed method of intratracheal administration of bacteria by direct visualization 22,23 .

Bronchoalveolar Lavage
A detailed method of bronchoalveolar lavage uid (BALF) collection has been previously described 24,25 .Brie y, at the speci ed times, mice were anesthetized with iso uorane.The thoracic cavity was opened by midline incision.The trachea was exposed and cannulated with a 20-gauge catheter, which was secured with a 2-0 silk suture.The left main stem bronchus was identi ed and divided at the hilum, and the entire left lung was removed.BAL was performed on the right lung with PBS containing 0.6 mM EDTA instilled in one aliquot of 0.6 mL, followed by three aliquots of 0.5 mL.

Determination of bacterial burden
The left lung, spleen, and right medial lobe of the liver were removed, and placed in 1 ml of sterile water.
Tissues were immediately homogenized on ice with a tissue homogenizer.Blood was drawn in citratephosphate-dextrose treated 23-gauge syringes from the inferior vena cava of each anesthetized mouse.Serial 1:10 dilutions of isolated whole blood and tissue homogenates in sterile PBS were plated on tryptic soy agar.Plates were incubated for 18 h at 37˚C after which colonies were counted.
Isolation and culture of bone marrow-derived macrophages (BMDMs) Isolation and differentiation of bone marrow-derived myeloid cells to non-activated macrophages has been previously described 26 .Brie y, bone marrow cells from WT and G6PD A-mice were allowed to adhere and cultured in differentiation media (DMEM, 20% neonatal calf serum, 30% L929 conditioned medium, 1% penicillin-streptomycin) at a density of 1 x 10 6 cells/ml in 100 x 20mm sterile petri dishes for 7 days.Medium was replaced at day 5.After 7 days, differentiated BMDMs were cultured in 24-well plates at a density of 2.5 x 10 5 cells/ well in a nal volume of 500 µL 24-hours prior to subsequent use in in vitro pHrodo phagocytosis assay.

Isolation of bone marrow neutrophils
Extraction of bone marrow cells was performed as described previously 27 , and neutrophils were isolated from cell suspensions via negative selection (Miltenyi Neutrophil Isolation Kit, #130-097-658) according to the manufacturer's instructions.Isolated cells were diluted in serum-free OptiMEM to a nal concentration of 4x10 4 cells per 30 µL media.The nal cell suspension was >95% polymorphonuclear cells as determined by cytospin preparations.These cells were immediately used in the modi ed Amplex Red assay as described below.Unused cell suspensions were centrifuged at 300 x g for 10 minutes, lysed, and pellets stored at -80°C.pHrodo™ in vitro Phagocytosis Assay K. pneumoniae strain 43816 serotype 2 was grown to OD 600nm = 0.2 then heat-killed at 60°C for one hour.
Heat-killed bacteria were centrifuged at 12,000 rpm for 2 min, resuspended in 100 mM sodium bicarbonate, centrifuged at 12,000 rpm for 2 min, then resuspended in 200 µL sodium bicarbonate.pHrodo™ iFL Green STP Ester dye (ThermoFisher #P36012) at 10 mg/mL was diluted in the bacterial suspension to 1 mM, then incubated for one hour in the dark.Unconjugated dye was removed through successive washing steps with PBS and 100% methanol.BMDMs at a density of 2.5x10 5 cells/well were infected with heat-killed, labeled-bacteria at MOI=10 for 90 min.Plates were centrifuged at 1500 rpm for 5 min at 4°C to synchronize phagocytosis.As a negative control, cytochalasin D [10 µM] was applied to selected wells one hour prior to infection to inhibit phagocytic uptake.Adherent cells were isolated by vigorous washing with ice cold PBS.pHrodo™-positive macrophages were examined by ow cytometry analysis.The phagocytosis of pHrodo labeled KP2 in BMDMs was detected as green uorescence positive events by BD FACScalibur.Four wells of WT and four wells of G6PD A-BMDMs were obtained from one mouse each as technical duplicates.

In vivo Phagocytosis Assays
For measuring in vivo phagocytosis, pHrodo labeled K. pneumoniae was instilled intratracheally into mice at a concentration of ~5 x 10 7 CFU in 0.1 mL total volume.After one hour, mice were euthanized and BAL was performed using 1-mL of 0.6mM EDTA in PBS followed by three sequential lavages of 0.9-mL.BAL cells were labeled with uorochrome-conjugated antibodies against Ly6G PE and F4/80 APC.The ow cytometry data were acquired on BD FACSCalibur.The positive events were back-calculated as absolute BAL cell counts based upon total cell counts.

Measurement of Extracellular H 2 O 2 Production
The measurement of extracellular H 2 O 2 production was performed using BMDMs and bone marrow neutrophils using a modi ed Amplex red assay.Brie y, H 2 O 2 production was quanti ed in whole cells by Amplex® Red as previously described [28][29][30] .Either bone marrow derived macrophages or bone marrow derived neutrophils were prepared as described above, resuspended in OptiMEM and then plated (4 × 10 4 cells/well) into 96-well plate in assay buffer (25 mM Hepes, pH 7.4, containing 0.12 M NaCl, 3 mM KCl, 1 mM MgCl 2 ) supplemented with 0.1 mM Amplex® Red, and 0.32 U/ml of horse radish peroxidase (HRP).
The reaction was started by addition of 5µM phorbol 12-myristate 13-acetate (PMA).Fluorescence measurements were made using a Biotek Synergy 4 Hybrid Multi-Mode Microplate Reader with a 530/25 excitation and a 590/35 emission lter.The reaction was monitored at 37 o C for 1 h.A standard curve of known H 2 O2 concentrations was included on each plate.NADPH oxidase (NOX) activity was obtained by calculating the rate of H 2 O 2 production as RFU/min/40,000 cells after subtracting the equivalent value given by cells treated with 300 U/ml of catalase.Data are expressed as fold change of WT vehicle control.

Statistical Analysis
A Student t test was used to compare differences between two groups.A 2-tailed Mann-Whitney U test was used for data that was not normally distributed.Differences were considered signi cant for p-values < 0.05, and signi cance was Bonferroni-adjusted for multiple comparisons where appropriate.All statistical analysis was performed using GraphPad Prism 9.5.0.

G6PD A-mice show lower lung bacterial burden and decreased extrapulmonary dissemination following Klebsiella pneumoniae intrapulmonary infection
We tested the hypothesis that G6PD A-alters the clearance and dissemination of K. pneumoniae from the lungs following acute intrapulmonary infection.At 24 h post-infection, G6PD A-mice showed signi cantly lower lung bacterial burden compared to WT littermates and a trend toward lower lung burden after 48 h (Fig. 1A).At 48 h post-infection, G6PD A-mice showed signi cantly less bacteremia (Fig. 1B) and reduced organ dissemination to spleen, liver, and kidney (Fig. 1C-E).
Given the enhanced host defense observed in the G6PD A-mice, we examined the in ammatory cell recruitment pro le of BALF at 24 and 48 h.Compared to WT mice, G6PD A-mice showed similar total BALF cell counts, neutrophil, and mononuclear cell numbers/mL (Fig. 2A, B, D).Lung tissue myeloperoxidase levels, a marker of total neutrophil content in lung tissue homogenates, were similar between G6PD A-and WT mice (Fig. 2C).Total BALF protein, a marker of lung microvascular permeability, was not different between G6PD A-and WT mice (Fig. 2E).These results suggest that the G6PD A-variant is associated with enhanced early host immune response and reduced extra-pulmonary bacterial dissemination.

G6PD A-mice show similar in ammatory cytokine pro le in the lung as WT mice during infection
Previous studies demonstrate that G6PD de cient monocytes from human subjects exhibit increased ex vivo LPS-induced IL-10 levels, and greater in vivo IL-1β, IL-6, and IL-10 levels in the serum and peritoneal lavage uid from LPS-treated G6PD-de cient mice 14,31 .Consistent with these prior human studies, G6PD A-mice had signi cantly higher levels of IL-10 in the BALF compared to WT mice (Fig. 2F) after 24 h.However, after 48 h, BALF IL-10 levels in G6PD A-mice were lower than those measured from WT mice.No differences were observed in the master cytokines IL-1β, TNF-α, or GM-CSF (Fig. 2F-J).Further, measurements of lung tissue malondialdehyde, a major lipid oxidation product, using the thiobarbituric acid reactive substance (TBARS) assay found similar levels in G6PD A-and WT mice (Fig. 2K).

Bacterial phagocytosis is not altered in G6PD A-bone-marrow derived macrophages in vitro, and alveolar macrophages or neutrophils in vivo
We tested whether multiple cell types, including BMDMs, alveolar macrophages and bone marrow neutrophils from G6PD A-mice exhibit increased phagocytosis of K. pnuemoniae.To visualize labeled bacteria entering newly formed phagolysosomes, BMDMs were isolated, differentiated and pooled from G6PD A-and WT littermate mice and exposed to pHrodo-labeled K. pnuemoniae (MOI = 10).We observed slightly decreased phagocytosis in the G6PD A-macrophages compared to WT BMDMs (Fig. 3A).In the lung, there were no genotype differences in pHrodo signal intensity in airspace macrophages (Fig. 3B) or neutrophils (Fig. 3C).Overall, phagocytosis rate was 32-45%, and there were few Ly6G-positive cells, and ~ 82% F4/80-positive macropahges.We also compared the phagocytic receptor pro le expressed by bone marrow-derived macrophages from G6PD A-and WT mice.We did not observe any differences between the receptors surveyed involved in bacterial uptake (Fig. 3D).These experiments suggest G6PD A-does not impact bacterial phagocytosis and phagocytic receptor expression.
Baseline H 2 O 2 production is increased in A-compared to WT Mononuclear phagocytes and polymorphonuclear cells are critically important for host control of K. pneumoniae infection 32,33 .We performed a modi ed Amplex Red assay to compare the rate of H 2 O 2 production between WT and G6PD A-neutrophils and macrophages.We observed signi canly higher baseline H 2 O 2 production in the G6PD A-neutrophils (Fig. 4A) but not macrophages (Fig. 4B) compared to WT.In each cell type, the H 2 O 2 production was not different following PMA stimulation, although the total H 2 O 2 levels trended higher in G6PD A-neutrophils and macrophages.The G6PD A-BMDM NADPH production was ~ 20% reduced compared to WT G6PD activity (Supplementary Fig. 1).These results suggest that decreased G6PD activity in uences H 2 O 2 buffering capacity in neutrophils while macrophages are able to compensate and maintain WT levels of H 2 O 2 even in the absence of a fully functional G6PD enzyme.

Conclusions
In this study, we developed a humanized model of G6PD A-to explore responses to pneumonia, using a common pathogenic strain of K. pneumoniae.We found that G6PD A-mice exhibit a favorable pulmonary host defense during K. pneumoniae infection as demonstrated by reduced lung bacterial burden and extrapulmonary dissemination compared to WT littermates.Despite enhanced pulmonary host defense, the G6PD A-mice showed similar airspace neutrophil counts, in ammatory cytokine pro le, and phagocytosis as WT mice.We observed higher baseline H 2 O 2 production from G6PD A-neutrophils but not macrophages compared to WT littermates.Together, our data suggest that the enhanced immune response may be mediated in part by differences in baseline neutrophilic H 2 O 2 production.This study suggests a complex relationship between G6PD de ciency and the role of ROS production in innate immunity.The oxidative burst in neutrophils is a principal component of the innate immune defense against bacterial and fungal pathogens, and neutrophils are the main drivers of the innate response during bacterial infection 18 , especially with regard to ROS production by NADPH oxidase 34 .
Treatment with a small-molecule inhibitor of G6PD lowered in vitro oxidative phosphorylation in mouse and human neutrophils, while the NADP+/NADPH ratio was less affected in mouse macrophages 10 .Human G6PD A-circulating monocytes have higher total levels of in vitro PMA-induced ROS compared to non-G6PD-de cient monocytes, while neutrophils from the same subjects did not exhibit differences in PMA-induced ROS 11 .Increased total ROS production in G6PD A-is likely three-fold: 1) due to inadequate NADPH substrate necessary for glutathione recycling and neutralization of ROS; 2) increased MPOmediated HOCl production from H 2 O 2 , and 3) increased activity of NOX isoforms 35 .Interestingly, we did not observe genotype differences in lung tissue homogenate TBARS as a marker of oxidative stress during the course of K. pneumoniae infection (Fig. 2K).Inability to detect a difference in tissue may be due to the lower sensitivity of this method to detect cellular oxidative stress in a speci c cell type.G6PD de ciency due to the A-variant is typically clinically silent unless these patients are exposed to overwhelming oxidative stress 1 .While it is not known exactly how phagocyte H 2 O 2 production exerts its anti-microbial properties 36 , the current study raises the possibility that G6PD A-exerts its enhanced response to K. pneumoniae at least in part via differential H 2 O 2 production.We further speculate that conversion of H 2 O 2 to HOCl may account for a component of increased bacterial killing by G6PD Aleukocytes.
In the current study, we observed increased bronchoalveolar IL-10 levels in the G6PD A-mice versus WT in the initial 24 h period, followed by lower levels in G6PD A-mice at 48 h.IL-10 has been shown to enhance phagocytosis in monocytes and macrophages 37 .In bacterial infections such as S. pneumoniae, IL-10 is important for clearing extracellular bacteria despite having sustained elevation of in ammatory markers and neutrophil in ltration in the lung 38 .Still others have shown during K. pneumoniae (ATCC 43816) infection, the absence of IL-10 may lead to more robust bacterial clearance 39 , although this phenotype may be strain speci c as a second study has shown that during high doses of carbapenem-resistant K. pneumoniae, IL-10 de ciency leads to increased mortality and uncontrolled lung in ammation 40 .A previous in vitro study found that IL-10 transcript and protein levels were markedly increased in G6PD de cient macrophages compared to WT and, depending on the activation status of the macrophage with LPS, PMA, or LPS + PMA, shifted the G6PD macrophages to an anti-in ammatory cytokine pro le, with associated increases in NF-κB, Sp3, CREB, and PKCδ signaling factors 31 .Further studies are needed to determine the exact role of G6PD in the regulation of IL-10 production in vivo.
Based on the enhanced clearance of K. pneumoniae in the circulation and extra-pulmonary organs observed in our humanized mouse, this raises the possibility of parallels to the immune-protective effect of G6PD A-against malaria.The proposed protective mechanism against malaria rests on the premise of enhanced erythrophagocytosis of infected RBCs.We did not observe differences in neutrophil or macrophage phagocytosis, or in classic phagocytic surface marker pro les.However, more recent evidence from G6PD-de cient mice infected with Plasmodium berghei demonstrate a survival advantage, less severe experimental cerebral malaria, and milder acute liver injury via an attenuated proin ammatory response compared to WT controls 41 .In that study, serum IL-10 levels were not different between G6PD de cient and WT mice, although proin ammatory cytokine levels were lower in G6PD de cient mice 41 .
We utilized a humanized murine model of the G6PD A-variant and examination of phagocytosis and H 2 O 2 production across multiple cell types.Prior studies have relied upon in vitro chemical inhibition of G6PD or mutations outside the causative V68M SNP.Those methods may fail to capture the global effect of compensatory changes resulting from the G6PD A-variant as well as alternative endogenous sources of NADPH in nucleated cells.We previously reported that this humanized mouse model demonstrate G6PD activity that is ~ 5% of WT RBCs 21 .However, G6PD A-bone marrow-derived macrophage NADPH production is ~ 20% reduced compared to WT G6PD activity (Supplementary Fig. 1).It is noteworthy that the in ammatory phenotype in response to bacterial lung infection by these G6PD A-humanized mice is relatively modest, as there were no signi cant differences in lung vascular permeability or cellular in ltration.In conclusion, this study suggests that G6PD A-variant associated differences in H 2 O 2 production may drive an enhanced acute immune response to K. pneumoniae.Future studies using this humanized mouse model may be useful in de ning the impact of G6PD in regulating the ability of nonerythroid cells to handle cellular ROS and their function during states of oxidative stress.Our study directly tests the effect of the X-linked G6PD A-variant on host immune response to K. pneumoniae.Because this variant affects approximately 11% of African American men, it may re ect an important, previously unappreciated disease modi er in patients with pneumonia and sepsis.

Declarations
Author Contributions: wrote and prepared the manuscript, signi cantly contributed to the conception and design of the study, to data analysis and interpretation.HFP, ZX, LW, ECP, and ER contributed signi cantly to experimental design, and data acquisition and interpretation.MY made signi cant contributions to animal model maintenance and data acquisition.MTG and JSL were instrumental in conceptual and experimental design, as well as data analysis and interpretation.All authors have reviewed, provided edits for, and approved the manuscript for submission.All authors are responsible for accuracy and integrity of the work.Data Availability: The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.

Figures
Figures

Figure
Figure