The influence of activatory FcγRs on virus nAbs remains relatively understudied, particularly due to the assumption that viruses are directly neutralized via Fab-dependent mechanisms, overlooking the potential role of activatory FcγR during viral clearance. Given the decisive role of activatory FcγRs in mediating IgG-dependent effector mechanisms, we aimed to evaluate their impact on virus nAbs during an acute viral infection. Initially, we confirmed the neutralizing efficacy of two types of antibodies: broad nAbs obtained from sera of immunized mice, and a previously described monoclonal LCMV nAb, Wen3 ( Supplementary Figure S1A ) (29). Our assessment affirmed robust neutralizing activity for both antibody types. First, to rule out variations in antibody distribution or degradation between WT and Fcer1g knockout mice (Fcer1g^-/- , commonly known as FcγR-KO) lacking all activatory FcγRs, we isolated serum from actively immunized mice and evaluated its neutralizing activity. Remarkably, we observed similar neutralizing capacities in the serum from both passively immunized WT and Fcer1g-KO mice ( Figures 1A, B ). By excluding variations in antibody concentration and distribution, we can more confidently attribute any observed differences in virus control to the absence of activatory FcγR rather than fluctuations in antibody levels.

Subsequently, by utilizing WT and FcγR-KO mice, our results unveiled a significant dependence on activatory FcγRs for early virus titre control in the spleen but not in the liver by both broad nAbs (obtained from LCMV immune mice) ( Figure 1C ), and Wen3 ( Figures 1D, E ). Since the liver titre was undetectable in these experiments, we refrained from measuring it in future experiments. Interestingly, we observed a more pronounced reduction in virus titres in the spleen of WT mice compared to FcγR-KO mice. This suggests a dual mechanism of action for virus nAbs, with the direct neutralization of the virus accounting for the decrease observed in KO mice, and an additional Fc-mediated effect contributing to the further reduction seen in WT mice. These findings underscore the dependency on FcγRs specifically in the spleen, emphasizing the crucial role of FcγR in mediating the efficacy of virus nAbs. In line with these findings, levels of interferon-γ (IFNγ) and interferon-α (IFNα) were notably reduced in WT mice treated with Wen3 compared to FcγR-KO mice ( Supplementary Figures S1B ).

To further explore the dose-dependent effect of Wen3 on virus titre reduction, we administered three different doses (100 µg, 350 µg, and 500 µg) of Wen3 to both WT and FcγR-KO mice, strikingly, at the lowest concentration, the FcγR-KO mice did not exhibit any reduction in virus titres, suggesting that at lower concentrations, nAbs primarily exert their function through FcγRs ( Figure 1F ). Conversely, at medium doses, both direct neutralization and Fc-mediated mechanisms appear to contribute to virus titre reduction.

Next, we aimed to investigate whether nAbs depend on activatory FcγRs when administered before infection. Our findings revealed significant differences in virus control between WT and KO groups, even in the presence of pre-treatment with memory serum or Wen3 ( Supplementary Figures S1C, D ). Collectively, these results underscore the inadequacy of nAbs alone in conferring effective virus control and emphasize the critical contribution of activatory FcγRs-mediated mechanisms.

To further pinpoint the effect of activatory FcγRs on hematopoietic cells, we generated bone marrow chimeras by irradiating FcγR-KO mice followed by reconstitution with either WT or FcγR-KO bone marrow cells. Notably, chimeric mice receiving WT bone marrow exhibited a significant decrease in virus levels following Wen3 treatment compared to those receiving KO bone marrow ( Figure 1G ), providing additional evidence for the pivotal role of activatory FcγRs on hematopoietic cells in controlling virus infection through virus neutralizing IgG antibodies.

Next, we aimed to capture the overall changes in FcγRs expression levels following viral challenge. Hence, the FcγR expression repertoire on effector cells following infection might provide insights into responsible effector cell populations. To explore this, we assessed the expression of all FcγRs on different immune cells in naïve state as well as 24 hours after infection in both blood and spleen ( Figures 2A, B ). The gating strategy utilized to gate different monocyte subsets is detailed in Supplementary Figures S2A, B . Interestingly, we observed a significant increase in the expression of FcγRI and FcγRIV on patrolling monocytes in both spleen and blood ( Figure 2A, B ), as well as on splenic macrophages, 24 hours post-infection compared to their naïve counterparts. This finding suggests a plausible involvement of these receptors in facilitating antibody-dependent effector functions crucial for viral clearance.

IgG2a antibodies, such as Wen3, are known for their potent activation of effector functions through specific FcγR binding, mainly FcγRIV (30, 31). To explore this, we administered FITC-labeled Wen3 antibody to mice infected with LCMV. Strikingly, our analysis revealed predominant binding of Wen3 to two distinct populations: patrolling and inflammatory monocytes ( Figure 2C ). Building on these findings, our investigation aimed to ascertain whether Wen3 binds to monocyte subsets via FcγRs. To investigate this, we administered FITC-labeled Wen3 to both FcγR WT and KO mice. Subsequent analysis, conducted 25 minutes post-injection, demonstrated significant binding of Wen3 to both patrolling and inflammatory monocyte populations in both blood and spleen ( Figure 2D ). Notably, the binding of Wen3 to patrolling monocytes was almost exclusively dependent on FcγR, whilst binding to inflammatory monocytes did only partially depend on activatory FcγRs ( Figure 2D ). Moreover, minimal binding to neutrophils was observed indicating a selective interaction with monocyte subsets.

To determine the role of the variable region of Wen3 in influencing the binding to monocyte subsets, we injected FITC-labeled IgG2a anti-CD45.1 antibody into LCMV infected CD45.2 mice. Remarkably, we observed a similar binding pattern ( Figure 2E ), confirming that the interaction is mediated via the Fc region. These results strongly suggest that activatory FcγRs on monocyte subsets might play a role in controlling virus infection through nAbs.

Next, we aimed to determine which effector cells, as well as activatory receptors, are responsible for Fc-mediated effector functions. Our earlier observations indicate that the LCMV nAb Wen3 binds to both patrolling and inflammatory monocytes via Fc receptors. To determine whether inflammatory monocytes play a role in mediating the function of Wen3, we selectively depleted inflammatory monocytes using a well-established antibody targeting CCR2 (MC-21), known for effectively removing inflammatory monocytes while leaving patrolling monocytes unaffected (32, 33). Successful depletion of inflammatory monocyte was confirmed in blood ( Figures 3A, B ); while patrolling monocytes population remained intact ( Figure 3B ). Interestingly, depletion of inflammatory monocytes did not hinder Wen3-mediated virus control ( Figure 3C ), reinforcing the dispensability of inflammatory monocytes in this context.

Given the crucial role of macrophages in phagocytosis of opsonized virus particles as well as their ability to express all classes of FcγRs (34), we sought to investigate the contribution of macrophages in nAb-mediated virus control during LCMV infection. To specifically evaluate the effect of macrophages, we employed a previously described method capitalizing on the rapid regeneration of monocytes in the blood (35). Macrophages were depleted using a single dose of clodronate ( Figure 3D ), ensuring the depletion of tissue-resident macrophages throughout the infection period while facilitating earlier monocyte appearance ( Figure 3E ). Indeed, macrophage depletion led to increased LCMV propagation ( Figures 3F ), consistent with the known protective role of type 1 interferon production by macrophages (36). Nevertheless, and strikingly, even in the absence of tissue-resident macrophages and the heightened virus propagation, in presence of monocytes subsets, passive administration of Wen3 provided comparable protection to control-treated mice ( Figures 3F, G ). This underscores the dispensability of inflammatory monocytes as well as the macrophages in antibody-mediated virus control. Moreover, it highlights the multifaceted nature of the immune response, prompting further investigation into the specific effector mechanisms responsible for antibody-mediated viral clearance.

Next, we aimed to explore the involvement of other innate and adaptive immune cells in Fc-mediated control of LCMV through Wen3. Neutrophils were initially considered due to their previously described role in controlling viruses via FcγRs (37). Additionally, they express FcγRIV, which is crucial for IgG2a activity (11), making them ideal candidates for effector function. Interestingly, depletion of neutrophils did not affect the effectiveness of Wen3 ( Supplementary Figure S3A ), suggesting that neutrophils are dispensable for FcγR-mediated virus control. Moreover, we assessed the roles of NK cells and T-cells, two key components of the immune system known for their cytotoxic functions and ADCC (38). These cell types were included not only due to their well-established effector functions but also to explore the broader immunological landscape potentially engaged by monoclonal antibody therapy. NK cells, through FcγRIII, can directly mediate ADCC (39), while T cells may be indirectly activated via cytokines produced by other immune cells following mAb engagement, contributing to immune orchestration and viral clearance (40). Utilizing NK cell-depleted and T-cell knockout mouse models, we evaluated their contributions to antibody-mediated virus control. Remarkably, depletion of NK cells or genetic deficiency of T-cells ( Supplementary Figures S3B, C ) did not impact the effectiveness of Wen3-mediated virus control, suggesting that both cell types are dispensable for this process. This comprehensive analysis suggests that neutrophils, NK cells, and T-cells are dispensable for Wen3-mediated virus control.

Building on our earlier observation of the selective binding of Wen3 to patrolling monocytes, as well as the increased FcγRIV expression on these cells upon infection ( Figures 2A-D ), we aimed to delve deeper into their potential role in antibody-mediated immune responses. To directly assess the significance of patrolling monocytes in IgG-mediated virus control, we employed a targeted depletion approach using a low dose of clodronate, as illustrated in Figure 4A . Administration of this clodronate dose effectively resulted in the complete elimination of patrolling monocytes throughout the experiment period ( Figures 4B, C ). Interestingly, depletion of patrolling monocytes resulted in loss of the Fc-dependent virus control of nAbs. ( Figures 4D, E ). To further investigate whether this effect is based on specific features of the Wen3, or it is generalized to other nAbs, we repeated the monocyte depletion procedure as described earlier and treated mice with neutralizing serum. Intriguingly, we observed a reduction in the activity of the neutralizing serum in the absence of patrolling monocytes ( Figure 4F ), thus further supporting the generalization of our findings.

To address concerns regarding potential toxicity induced by clodronate, we opted to deplete monocytes using CD115, a receptor known for its high expression on monocyte subsets (41). Administration of CD115 resulted in more than a 60% reduction in the numbers of patrolling monocytes while leaving inflammatory monocytes unaffected ( Figure 4G ). Consistent with our earlier observations, depletion of patrolling monocytes led to loss of Fc-mediated virus control ( Figure 4H ). Moving forward, we utilized a mouse model (CD11c-cre/iDTR) in which diphtheria toxin receptor (DTR) expression is induced under the CD11c/Itgax promoter (42), allowing for the depletion of dendritic cells as well as patrolling monocytes by diphtheria toxin administration. Consistent with the previously observed role of patrolling monocytes, CD11c-cre/iDTR mice exhibited a loss of the Fc-dependent virus control of nAbs ( Figure 4I ).

Having identified patrolling monocytes as essential for Fc-mediated virus control by nAbs, our next objective was to investigate which activatory FcγR is involved in this process. IgG2a antibodies, such as Wen3, were reported to have a very weak or no affinity for FcγRIII (11). To further investigate which FcγR involved in Wen3-mediated virus control we utilized FcγRIII-KO mice, and as expected, FcγRIII-KO mice did not influence the efficacy of Wen3 ( Supplementary Figure S3D ), demonstrating a comparable reduction in virus titres to WT mice. Having confirmed the dispensability of FcγRIII, we turned our attention to FcγRIV, given its high affinity to IgG2a antibodies like Wen3. Remarkably, blocking of FcγRIV led to loss of Fc-mediated virus control ( Figure 4J ), which aligns well with the observed increase in expression of FcγRIV on patrolling monocytes during infection ( Figure 4K ). These findings collectively confirm the pivotal role of patrolling monocytes in regulating the Fc-dependent virus control of nAbs.

To further dissect the mechanism underlying patrolling monocyte-mediated viral control, we investigated whether direct elimination of infected cells by patrolling monocytes contributes to this phenomenon. We employed an in vitro co-culture system consisting of patrolling monocytes and LCMV-infected target cells to assess the cytolytic activity of patrolling monocytes against virus-infected cells. We utilized bone marrow-derived dendritic cells (BMDCs) from FcγR-KO mice as target cells and infected them in vitro for 48 hours to allow for viral propagation and antigen presentation. After 48 hours, most of the infected BMDCs were found to express the epitope for Wen3 ( Supplementary Figure S4A ). As effector cells, we utilized patrolling monocytes, inflammatory monocytes, and neutrophils. Interestingly and in line with our earlier observations, among all tested effector cells, both blood-derived and spleen-derived patrolling monocytes exhibited the highest killing capacity against virus-infected cells ( Figures 5A, B ). To further investigate whether the killing capacity of patrolling monocytes is specific to virus-infected cells or represents a broader activity, we utilized naïve B-cells as target cells. Interestingly, patrolling monocytes exhibited a significant ability to mediate killing of B-cells using CD20 antibody in vitro compared to inflammatory monocytes ( Figure 5C ). Indeed, our in vivo data further supported this observation, revealing a reduction in the number of LCMV-infected cells expressing high levels of PD-L1 following Wen3 treatment ( Figure 5D ). This decrease is likely attributed to their elimination through ADCC mechanisms.

To address whether FcγR-mediated killing could also be mediated by non-neutralizing antibodies, we tested the KL53 clone, a non-neutralizing IgG2a antibody directed against the LCMV nucleoprotein (NP). In vitro, KL53 exhibited only minimal killing capacity when co-cultured with patrolling monocytes, substantially lower than that observed with Wen3 ( Supplementary Figure S5A ). These findings suggest that not all virus-specific antibodies are equally capable of triggering FcγR-mediated effector functions. However, given that KL53 targets NP rather than the surface-expressed glycoprotein (GP), differences in antigen accessibility must be considered when interpreting these results. Future studies will be required to more directly address the role of non-neutralizing GP-targeting antibodies in FcγR-dependent viral clearance.

These compelling results strongly suggest that the interaction between activatory FcγRs on patrolling monocytes and virus-specific antibodies, plays a pivotal role in controlling virus infection by eliminating virus-infected cells.

Next, we aimed to investigate how the levels of patrolling monocytes impact passive immunization effectiveness. We took advantage of the previously described role of muramyl dipeptide (MDP) in increasing numbers of patrolling monocytes through converting inflammatory monocytes into patrolling monocytes by triggering the NOD2 receptor (21). As expected, the treatment with MDP leads to an increase in numbers of patrolling monocytes ( Figures 6A, B ). In line with our previous observations, administering Wen3 to mice treated with MDP led to a substantial improvement in spleen virus control ( Figure 6C ). This highlights the pivotal role of patrolling monocytes in shaping passive immunization efficacy. Moreover, upon reviewing all previously conducted experiments, we found a direct correlation between Wen3-mediated virus control and patrolling monocyte counts in blood and spleen ( Figure 6D ). Overall, our findings emphasize the critical influence of patrolling monocyte frequencies on passive immunization outcomes. They shed light on the intricate relationship between patrolling monocytes and antibody-driven immune responses, highlighting their potential as key factors in antiviral defense strategies.

Although we demonstrated the direct involvement of patrolling monocytes in killing virus-infected cells via ADCC mechanisms, we did not specifically investigate the possibility of patrolling monocytes phagocytizing opsonized viruses. Considering the well-documented high phagocytic capability of patrolling monocytes, it is highly plausible that they may also engage in the phagocytosis of opsonized viruses, further contributing to the elimination of viral infections. Moreover, patrolling monocytes are known for their non-classical effector functions, including, trogocytosis, and interactions with tissue-resident immune cells that may shape the broader antiviral response. Although they are generally thought to lack robust expression of canonical cytotoxic molecules such as perforin and granzyme B, this does not exclude their participation in non-classical ADCC-like activity. Further investigations are warranted to explore this aspect comprehensively.

This study utilized mice sourced from various strains, including C57BL/6 (wild type, WT) obtained from Jackson Laboratory, Fcer1g-KO (FcγR-KO or Fcer1g^-/- ) mice, siblings served as controls, CD11c-cre and iDTR mice were provided by Prof. Dr. Ari Waisman, Johannes-Gutenberg-University Mainz, to WH, FcγRIII-KO were acquired from Prof. Dr. Falk Nimmerjahn, Friedrich-Alexander-University Erlangen, and Tcrb-KO (Tcrb^-/-) mice were acquired from Jackson Laboratory. Mice, maintained on the C57BL/6 genetic background (when needed back-crossed at least 10 times) and bred as homozygotes were utilized. Male and female mice aged between 8 to 16 weeks were selected for experimentation. To ensure unbiased allocation, age and sex-matched animals were randomly assigned to different treatment groups. All mice were maintained in single ventilated cages under specific-pathogen-free conditions. Animal experimentation was conducted under the authorization of the Nordrhein Westfalen Landesamt für Natur, Umwelt und Verbraucherschutz (Recklinghausen, Germany), and adhered strictly to the German law for animal protection and institutional guidelines at the Ontario Cancer Institute of the University Health Network and at McGill University.

Mice were intravenously injected with LCMV-WE at doses of 2x10⁵ PFU, unless otherwise stated. Wen3 was administered intravenously 15–18 hours following infection unless otherwise indicated. The LCMV strain WE was originally obtained from Prof. Rolf Zinkernagel (Institute of Experimental Immunology, ETH, Zurich, Switzerland). The viral propagation process involved culturing on BHK-21 cells, which were purchased from ATCC (CRL-8544), at a multiplicity of infection (MOI) of 0.01. Following, the concentration of LCMV was quantified using the methodology described below (Fink et al., 2006).

Mice were intravenously infected with LCMV at the specified dosages. Virus titres were assessed using a plaque-forming assay. Evaluation of LCMV titres involved smashing of organs in Dulbecco’s modified Eagle medium (DMEM) supplemented with 2% fetal calf serum (FCS). Serial 1:3 dilutions of the samples were prepared, with alternate dilutions transferred onto preseeded confluent MC57 fibrosarcoma cells. Following a 2h incubation at 37°C, an overlay was applied, and the viral preparation was re-incubated at 37°C. Plaques were counted 48h later via LCMV NP staining. Cells were fixed with a 4% formaldehyde solution, permeabilized using a 1% Triton-X solution and subsequently blocked with 10% FCS in phosphate-buffered saline. Blocked cells were stained with anti-LCMV NP (VL4) antibody (made in house), followed by incubation with an ECL-conjugated anti-rabbit-IgG secondary antibody. Plaques were visualized utilizing a color reaction employing a solution composed of 0.2 M Na₂HPO₄, 0.1 M citric acid, 30% H₂O₂ and o-phenylenediamine dihydrochloride, with all chemicals obtained from Sigma-Aldrich.

LCMV immune serum (Neutralizing serum) was collected from C57BL/6 mice 90 or 120 days after infection with 2*10⁵ PFU LCMV-WE, mice were re-boosted with 2*10⁵ PFU LCMV-WE 15 days before serum collection. Serum samples were pooled from 20–30 mice and subsequently tested for LCMV titres and virus neutralizing activity using a focus-forming assay. A hybridoma cell line producing Wen3 or KL53 was originally obtained from Prof. Rolf Zinkernagel (Institute of Experimental Immunology, ETH, Zurich, Switzerland). Antibodies (Wen3 and KL53) were purified from hybridoma supernatants by affinity chromatography using Protein G Sepharose.

For measuring neutralizing activity of Wen3 or serum from immunized mice, serum was prediluted (1:40) followed by complement system inactivation at 56°C for 30 min. For analysis of IgG kinetics, diluted samples were treated with 2-mercaptoethanol (0.1 M) for removal of IgM. Wen3 was utilized at an initial concentration of 1 mg/ml. Sera/Wen3 was titrated 1:2 over 12 steps and was incubated with 60 PFU of LCMV. After a 90-min incubation at 37°C, MC57 mouse fibroblast cells were added to each well and evenly distributed on a plate shaker. After cell adhesion (typically 2 to 3 hours later), 1% methylcellulose was added, and the cells were further incubated for 48 hours. Upon confluence, the cells were fixed with 4% formaldehyde in PBS and permeabilized with 1% Triton X-100. Foci were visualized by staining for the NP using the VL-4 antibody. Titres were determined based on half-maximal inhibition values. Antibody titres are presented as twofold dilution steps (−log₂) times the predilution (that is, × 40).

For bone marrow chimera experiments, female Fcer1g-KO mice aged 10–12 weeks underwent irradiation with a dose of 9.5 Gy. The following day, bone marrow was aseptically isolated from donor mice and intravenously administered to the irradiated recipients. On day 10 post-irradiation, mice received treatment with clodronate liposomes to deplete tissue resident macrophages. Following a 50-day reconstitution period, the mice were considered suitable to be used in experimental procedures.

Histological analysis of snap-frozen spleen tissues involved staining with various monoclonal antibodies targeting LCMV-NP (Clone: VL4; made in-house, 1:1), Biotin-SP anti-rat (112-065-167, Jackson Immuno Research, 1:150), Streptavidin-APC (405207, Biolegend, 1:75), anti-F4/80 (53-4801-82, Thermo Fisher Scientific, 1:100) and anti-CD169 (130-124-896, Mylteni Biotec, 1:100).

In summary, the organs were embedded in tissue-tek and promptly snap-frozen in liquid nitrogen. The frozen tissue was then sliced into 8.0 µm thick sections. Following sectioning, the slides were air-dried and fixed in acetone for 10 min. Next, the slides were placed in a staining dish and incubated with blocking buffer (PBS with 4% FBS) for 25 min at room temperature to prevent nonspecific binding. All antibody mixtures were diluted in PBS with 2% FBS and incubated for 45 min at room temperature. Between the different staining steps, the slides were washed three times in PBS. Initially, the slides were stained with the LCMV-NP antibody. Subsequently, the Biotin-SP anti-rat antibody was used to localize VL-4 binding. Lastly, the remaining antibodies were mixed at the specified ratios above and applied to the slides. Following washing, one drop of fluorescence mounting media was added to each slide, followed by a coverslip. The slides were then left to dry in darkness for at least 30 min before image acquisition. Images were captured using the Keyence BZ-9000 (Keyence, Osaka, Japan).

Administration of antibodies was conducted intraperitoneally (i.p.) one day prior to infection and one day post-infection, unless stated otherwise. Depletion antibodies utilized were CD115 (Clone: AFS98), Ly6G (Clone: 1A8), NK (Clone: PK136), all procured from BioXcell, and each was administered at a dosage of 200 µg per mouse (except CD115 100 µg). Additionally, CCR2 (Clone: MC21) was obtained from Prof. Mattias Mack and administered at a dosage of 25 µg per mouse. For the blockage of FcγRIV, anti-FcγRIV (Clone: 9E9) was administered intraperitoneally at a dosage of 200 µg per mouse. Monocyte depletion was achieved by administering 0.5 µL/g mice weight (10 µL/20 g mice) of clodronate liposomes one day prior to infection and 4 hours before Wen3 treatment, unless otherwise specified. Macrophage depletion involved the administration of 200 µL clodronate intraperitoneal 24-30h prior to infection.

Bone marrow was harvested from the femurs and tibias of Fcer1g-KO mice. Red blood cells were lysed using RBC lysis buffer (PAN Biotech, cat#P10-90100), and the remaining cells were collected following centrifugation. These bone marrow cells were then resuspended at a concentration of 1.5 × 10 (6) cells/mL in complete RPMI containing 20 ng/ml of GM-CSF and 10 ng/ml of IL4 (Peprotech) and incubated at 37°C with 5% CO2 for 6 days. On day 2 of incubation, 75 ng/ml of GM-CSF and 30 ng/ml of IL4 were added in half the amount of the original media. After 6 days, non-adherent cells were collected and washed, then infected with LCMV-WE at an MOI of 0.1. 48 hours post-infection, the cells were harvested and labeled with CFSE before being co-cultured with effector cells.

Naive neutrophils, patrolling and inflammatory monocytes were isolated from blood and spleen of Fcer1g-WT and Fcer1g-KO mice using flow cytometric sorting on a FACS Aria III (BD) after enrichment by positive selection of B cells (Cat#130-049-501) and T cells (Cat#130-094-973) using Miltenyi beads. These effector cells were then combined at specified effector to target cell ratios in 96-well plates with Wen3 antibody at a concentration of 4 µg/mL. As a target cell, infected bone marrow-derived dendritic cells from Fcer1g-KO mice was employed as described above, for some experiments naïve B cells positively selected by B220 (Cat#130-049-501) were employed as target cells with CD20 antibody in concentration of 1 µg/mL. Following incubation for 16–20 hours at 37°C, total cell numbers were determined, and flow cytometric analysis was performed to quantify the different cell types present in the culture.

Serum cytokines were analyzed with LEGENDplex™ (BioLegend, San Diego, USA) according to the instructions and recommendation of the manufacturer.

Data is expressed as mean± SD. Student’s t-test and one-way ANOVA was used to detect statistically significant differences between groups unless otherwise mentioned. The level of statistical significance was set at P<0.05.