Commensal bacteria, resident on the human skin in the order of trillions, has two different aspects such as maintaining host skin immunity and generating inflammation on the skin [1
]. As a beneficial effect, they establish immune balance and defense for our skin. One of the commensal bacteria strains, Staphylococcus epidermis
T cell accumulation on the host epidermis, so that it establishes a natural defense against other pathogens such as C. albicans
]. However, several commensal bacterial strains induce severe inflammation on the skin [3
]. The underlying mechanism of the generation of skin inflammation is still unclear in several parts; however, the inflammation may progress and become a serious situation without any proper treatment. Therefore, commensal bacteria-originated skin inflammation has been recognized as a critical factor that should be avoided not only in clinical situations, but also in daily life.
is a well-known pathogen of serious skin inflammation, and the bacteria frequently colonizes on the inflamed skin. Atopic dermatitis (AD) is a well-known skin inflammation characterized by abnormal S. aureus
]. In the inflamed site on the skin, several subsets of T cell accumulation are usually observed, together with other immune cells recruitment [5
]. For instance, IFN-ɤ+
, or IL-17A+
T cells (Th1, Th2, or Th17) and IL-17A+
ɤδT cells, are famous effector T cells (Teff) that increase in atopic dermatitis (AD) skin [6
]. These cells are orchestrated to establish the total immune response for the generation of skin inflammation. On healthy skin, commensal bacteria does not induce an inflammatory response at all without critical basic disease-like immunodeficiency, as our immune system allows for their residence as a symbiotic effect [8
]. When the skin has been injured by factors such as mechanical and chemical damage, especially invading the epidermis layer deeply, it will be a trigger to induce a strong inflammatory response against the commensal bacteria, because abundant immune cells acquire the opportunity to interact with the bacteria [9
]. In fact, it is well known that epithelial barrier disruption promotes AD or AD-like symptoms bearing a large amount of S. aureus
Even though several host side critical factors related to initiating/promoting commensal bacteria-originated skin inflammation has been understood, the key factor which has a dominant role in activating the host immune system on the bacterial side is still controversial. S. aureus
is composed by a rigid outer cell wall containing several components such as peptidoglycan (PGN), lipoteichoic acid (LTA), wall teichoic acid (WTA), and cell wall/membrane proteins including lipoproteins [11
]. These are located on the outer layer of the bacteria, so that these components potentially possess the chance to frequently interact with the host immune cell. From this concept, several studies have targeted these cell wall components, and have revealed the mechanism behind the induction of an inflammatory response in the host immune system by PGN and LTA through the Toll-like receptor 2 (TLR2), and activating the inflammatory cytokine production in immune cells [16
]. However, these antigenic functions have not simply been adapted to the pathology of skin inflammation; therefore, any convincing findings have been reported for the pathogenesis of these cell wall components in skin inflammation. This evidence provides an insight into other cell wall components that strongly activate the host immune system at not only the single cell level, but also on the whole immunological network in the tissue.
Lipoprotein is a cell wall protein in Gram-positive bacteria including S. aureus
, and has various functions for bacterial survival on the host skin [15
]. In one of the S. aureus
strains, USA300, around 70 lipoproteins have been identified with their predicted microbiological functions [15
]. The predictable function in the majority of these lipoproteins have been determined from its structure; however, the exact functions and roles of these components are still unknown. Even though the S. aureus
lipoprotein has many unclear parts, the immunological role of the immune cell has been elucidated mainly by using in vitro studies. This is the same as other potential antigens derived from S. aureus
, lipoprotein through Toll-like receptor 2 (TLR2), which subsequently activates the Nuclear Factor-kappa
B) pathway to induce inflammatory cytokine production [18
]. In addition to the evidence, one study used a systemic S. aureus
infection animal model and showed that lipoprotein recognition by the host immune system was indispensable for the establishment of the total inflammatory response [20
]. From these points, bacterial lipoprotein is expected to be a potentially strong inflammation inducer on the skin.
Without any dependency for each character of the antigens derived from commensal bacteria including S. aureus
, the substances must be recognized by the skin dendritic cells (DCs) in one of the initial steps to activate the host immune response [21
]. Due to having a strong recognition and phagocytic ability against an ectopical substance, skin DCs capture various substances through a receptor-mediated manner [22
]. The incorporated substance is processed as an antigen in general, so that it is presented to naive T cells through antigen presenting molecules such as the major histocompatibility complex (MHC) for priming, which is an important mechanism for antigen-specific Teff generation [23
]. There are indispensable steps for the establishment of inflammatory responses based on Teff accumulation in infected/inflamed sites on the skin; hence it must be adopted for understanding the pathogenesis for S. aureus
-originated skin inflammation.
In this study, we revealed that S. aureus lipoprotein (SA-LP) induced strong inflammation on the skin with the activation of the DC-Teff axis. We confirmed that SA-LP-activated DCs in a TLR2 dominant manner. The SA-LP activated skin DCs, which migrate into the skin-draining lymph nodes (dLNs), and SA-LP was processed as an antigen for priming naive T cells. In addition to these fundamental findings, a SA-LP intradermal (ID) injection into the ear exhibited dramatic thickness in the murine model. The inflamed skin possessed abundant IFN-γ-producing CD4+ and CD8+T cells (Th1 and Tc1) in the model. The underling mechanism of Teff accumulation was regulated by the dermal DC (DDC), which migrated from the skin into dLN as it specifically reacted to SA-LP. Furthermore, SA-LP-stimulated DC transfer into the mouse ear showed a similarly inflamed condition on the skin as the SA-LP ID-injected mice, and the inflamed skin possessed Th1 and Tc1 accumulation. This is the first report to address the evidence of S. aureus-derived lipoprotein inducing serious skin inflammation. These findings will form a valuable direction towards an understanding of the novel mechanisms in S. aureus-related skin inflammation.
The S. aureus
cell wall is composed of several substances, and has been traditionally investigated as an antigen to induce the host immune response, and inflammation in each tissue and organ [36
]. PGN and LTA are the most famous cell wall components in Gram-positive bacteria; therefore, many studies have focused on these substances and revealed the mechanisms behind induction of the inflammatory response [37
]. According to these studies, we initially investigated the antigenicity of PGN and LTA in ID-injected mice. We found that PGN and LTA induced inflammation on the injected skin; however, there was much less severity of thickness when compared with whole bacteria (live or HK) injections (Figure 3
A and Figure S3
). In addition to the findings, PGN and LTA failed to induce the accumulation of abundant Teff on the treated skin, while granulocytes and macrophages were increased in a dominant manner (unpublished data). On the other hand, live-SA application showed severe epidermal and dermal hyperplasia caused by the accumulation of abundant Teff in the inflamed site (Figure 3
A). This evidence provided the possibility of the existence of other dominant inflammation inducers in the S. aureus
cell wall. Lipoprotein located on the S. aureus
cell wall was revealed as a strong antigen from in vitro study by using BMDCs, and it generated serious inflammation on the treated skin (Figure 1
and Figure 3
A). The inflammatory microenvironment in the SA-LP ID injected skin was similar to the live-SA treated case (Figure 3
A), which means that SA-LP has a potentially important role in establishing systemic immune response in S. aureus
-originated skin inflammation. Throughout the study, our results indicated the important role of DCs in SA-LP-derived skin inflammation. In the SA-LP ID-injected mice, a high population of mDC-loaded SA-LP as an antigen was observed in the skin-dLN (Figure 2
C), and it possessed the responsibility of generating Teff accumulated on the inflamed site (Figure 3
B–D and Figure S4A
). Ex vivo antigen presentation also showed the antigenicity of SA-LP in DCs for the generation of Th1 and Tc1 (Figure 5
C). Hence, we concluded that SA-LP is a strong factor in establishing skin inflammation, rather than other cell wall components, in S. aureus
Although we confirmed that SA-LP induces severe skin inflammation, we also considered its antigenicity for other substances derived from S. aureus
, as SA-LP failed to induce an increase in Th17 or γδT, which is generally observed in S. aureus
-related skin inflammation [7
]. These T cell subsets were slightly increased in the SA-LP ID injected mice, but there was not a dramatic change when compared with the vehicle control (Figure S4C,E
). Notably, we found that LTA promoted γδT and NKT cell accumulation on the ID injected skin (unpublished data). The exact mechanism for the increase of these T cell subsets are still unknown; however, it is obvious that several factors are orchestrated to establish a whole T cell-mediated response in S. aureus
-related skin inflammation. Furthermore, several cell wall components and toxins in S. aureus
also induced skin inflammation with accumulation and activation of various cells [39
]. These responses also seem to be important factors for generating total inflammatory responses against for S. aureus
. We have suspected that it is the reason for why SA-LP ID injection showed less severity for skin inflammation compared with cases with live S. aureus
When we detected a strong induction of Th1 and Tc1 on the skin, and dLN in the SA-LP ID-injected mice (Figure 4
A and Figure S4C,D
), we suspected the existence of a relationship with memory T cells (TCM
) as previous reports have indicated that commensal bacteria possesses the ability to establish a memory T cell pool in LN and peripheral organs, including the skin [2
]. In fact, both Th1 and Tc1 were detected in the vehicle control mice, though these absolute numbers were less than that of the SA-LP treated mice (Figure 4
A and Figure S4C,D,F
). The skin-accumulated T cells did not express CD69 and CD103, common TRM
markers, in the SA-LP ID injected mice. Furthermore, the T cells detected in dLN did not contain the CD62L+
) population (unpublished data). Taken together, we concluded that the accumulated Teff on the SA-LP ID-injected skin mainly originated from the priming effect for naive T cells in the skin-dLN through the SA-LP-activated DCs.
In this study, we revealed that DDCs have the responsibility of generating Teff through specific recognition of SA-LP. It has already been reported that each DDC has a special character for Teff generation [43
]. As a general priming mechanism for CD4+
naive T cells, these DDCs have strong activity. On the other hand, a specific DDC subset such as CD207+
DDC was revealed as being important for cross-presentation, which is an indispensable mechanism for the generation of CD8+
]. Our finding is strongly supported by these reports, as the SA-LP ID injected mice showed a significant increase in mDDCs in the skin-dLN, and the DCs strongly induced the generation of Th1 and Tc1 with high specificity (Figure 4
A and Figure S4C,D,F
). Furthermore, the evidence of antigen presentation through both MHC class II for naive CD4+
T cells and MHC class I for naive CD8+
T cells were also confirmed by MHC blocking for in vitro antigen presentation using SA-LP as an antigen (Figure 4
C). However, we have not yet revealed the exact mechanism of the cross-presentation for SA-LP by CD207+
DDC. To reveal this mechanism, further investigation is required.
Throughout this study, we have revealed that the DDC and Teff axis is a major immunological response to establish skin inflammation originating from the S. aureus lipoprotein. Our findings imply that lipoprotein produced from Gram-positive bacteria, including S. aureus, is one of the strong potential inflammation inducers on the skin. This study is the first comprehensive report for the relationship of bacterial lipoprotein and skin inflammation. Furthermore, this study used an in vivo approach to reveal the mechanism of total immune response against bacterial lipoprotein, so it has a high impact for the novel understanding of S. aureus pathogenesis on the skin. Our data suggested that S. aureus lipoprotein may be a therapeutic target, so that it has a chance to be a novel approach for the treatment of serious acute/chronicle skin inflammation originating with S. aureus. The immunological character of S. aureus lipoprotein still has several unclear parts; however, our findings will be a great help for the strategy of prevention and cure in S. aureus-related serious skin inflammation.
4. Materials and Methods
Mice. C57BL/6J specific pathogen free (SPF) mice were purchased from The Jackson Laboratory (Bar Harbor, ME, USA). Gender matched mice between 8–12 weeks of age were used for each experiment.
Reagents and antibodies. Lipoteichoic acid (LTA, S. aureus origin), peptidoglycan (PGN, S. aureus origin), Pam3CSK4, sparstolonin B (SsnB), phorbol 12-myristate 13-acetate (PMA), ionomycin, ovalbumin (OVA) and α-galactosylceramide (α-GalCer) were all purchased from Sigma Aldrich (St Louis, MO, USA). Dispase and collagenase were purchased from Thermo Fisher Scientific (Waltham, MA, USA). Percoll was purchased from GE Healthcare (Chicago, IL, US). The Cytofix/Cytoperm kit was purchased from BD Bioscience (Franklin Lakes, NJ, USA). Recombinant murine granulocyte macrophage-colony stimulating factor (rmGM-CSF) was purchased from Peprotech (Rocky Hill, NJ, USA). Anti-CD11c (N418), anti-CD11b (M1/70), anti-CD207 (4C7), anti-CD103 (2E7), anti-CCR7 (4B12), anti-CLA (HECA-452), anti-CD80 (16-010A1), anti-CD86 (GL-1), anti-CD45 (30-F11), anti-CD3 (17A2), anti-CD4 (GK1.5), anti-IL-17A (BL168), anti-CD16/CD32 (2.4G2) (93), and 5-(and -6)-carboxyfluorescein diacetate succinimidyl ester (CFSE) were all purchased from Biolegend (San Diego, CA, USA). Anti-MHC-II (M5/114.15.2), anti-CD11b (M1/70), anti-TCRβ (H57–597), anti-ɤδTCR (GL3), anti-IFN-γ (XMG1.2), anti-IL-4 (8D4–8), and anti-Toll-like receptor 2 (TLR2) were all purchased from Thermo Fisher Scientific (Waltham, MA, USA). CD1d-tetramer was purchased from Proimmune (Oxford, UK). Anti-MHC class I (HB159) and anti-MHC class II (M5/114) was purchased from Bio X Cell (West Lebanon, NH, USA). The isotype-matched control for each antibody was purchased from the same company.
S. aureus culture. The frozen S. aureus (MRSA; USA300) stock was thawed on ice, then transferred to a tryptic soy broth (TSB; BD bioscience, Franklin Lakes, NJ, USA) and cultured at 37 °C for 18 hr with shaking. The colony forming units (CFU) were calculated in each culture. Heat-killed S. aureus (HK-SA) was prepared with heating at 95 °C for 30 min. The heated S. aureus suspension was centrifuged at 10,000 rpm for 1 min to harvest the bacteria cells, then the cell pellet was resuspended in phosphate buffered saline (PBS) or 0.9% NaCl.
Lipoprotein isolation and preparation of the cell wall component from S. aureus.
Lipoprotein was isolated from S. aureus
by following a method described in previous reports with modifications [18
]. Briefly, cultured S. aureus
CFU/mL) was harvested by centrifuging at 5000× g
for 20 min. The pellet was washed twice by 20 mM Tris-HCl (pH 8.0). The pellet was resuspended in 20 mM Tris–HCl (pH 8.0), then the bacterial cell was crushed with 0.3 mm stainless beads. The treated suspension was centrifuged at 5000× g
for 20 min, then the supernatant was harvested as the protein suspension. The suspension was mixed with 100% ethanol and kept at −20 °C overnight. The sample was centrifuged at 12,000× g
for 15 min, then the precipitated pellet was washed with 80% ethanol and centrifuged again at 12,000× g
for 5 min. The precipitated pellet was dissolved with 1 M urea/50 mM Tris–HCl, 50 mM ethylenediaminetetraacetic acid (EDTA) (pH 8.0) (Crude Protein Extract; CPE). Triton X-114 was added to the protein suspension (final 1%), then the suspension was incubated at 4 °C with gentle mixing. The incubated suspension was heated at 37 °C, forming the micelle phase-containing lipoprotein. The micelle phase was extracted and lipoprotein (Clude S. aureus
-lipoprotein; SA-LP) was harvested by following a method for CPE precipitation. The SA-LP was separated to each fraction (L1 to L4) in a size dependent manner by using a molecular weight cut-off filter (Amicon ultra; Darmstadt, Germany). For preparation of the cell wall extract (CWE), the twice-washed S. aureus
pellet was resuspended in 20 mM Tris–HCl (pH 8.0). The suspension was kept at −80 °C for 30 min, then sonicated for 20 min. The suspension was centrifuged at 5000× g
for 20 min, and the pellet was harvested as CWE.
Sodium dodecyl sulfate-poly-acrylamide gel electrophoresis (SDS-PAGE) and silver stain. The extracted protein solution was diluted with 5 × SDS sample buffer (2% SDS, 62.5 mM Tris–HCl (pH 6.8), 10% glycerol, 0.01% bromophenol blue, 50 mM dithiothreitol (DTT). The proteins, separated by SDS-PAGE, were visualized with a Silver Stain Kit (Thermo Fisher Scientific, Waltham, MA, USA). Whole staining procedure was followed with the manual.
Mouse primary cell isolation. Skin leukocytes were isolated by following a method described in a previous report with modification [47
]. Briefly, the extracted ear was washed with tissue washing buffer (RPMI 1640 supplemented with 10% fetal bovine serum (FBS), 10 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) and 100 U/mL penicillin, 100 mg/mL streptomycin) at 37 °C for 30 min with gentle shaking. The ear was separated into the ventral and dorsal sheets from the cartilage, and incubated at 4 °C overnight with dispase working solution (tissue washing buffer containing 0.25 mg/mL of dispase) to separate the epidermal and dermal sheets. These sheets were chopped with scissors, then incubated at 37 °C for 30 min in collagenase working solution (tissue washing buffer containing 1 mg/mL collagenase and 0.01% DNase). The digested ear pieces were passed through a 5 mL syringe with a 22 G needle to make single cell suspensions. Lymph node cells were prepared from skin-draining LN (dLN) by following a method described in a previous report [48
]. Briefly, isolated dLN was crushed on a dish and suspended in cell culture medium. The cell suspension was filtered through a 70 μm cell strainer, then twice washed with cell culture medium (RPMI 1640 supplemented with 10% fetal bovine serum (FBS), 100 U/mL penicillin and 100 mg/mL streptomycin). Splenocytes were obtained from the spleen by following a method described in a previous report [48
]. Briefly, isolated spleen was crushed on a 70 μm cell strainer, and the cells were suspended in cell culture medium. After being washed once with the cell culture medium, the cells were further resuspended in an erythrocyte lysis solution (155 mM NH4
Cl, 10 mM KHCO3
, 1 mM Na-EDTA, and 17 mM Tris–HCl (pH 7.3)). After being washed twice with cell culture medium, the cells were used as splenocytes. Mouse bone marrow leukocytes were obtained from the tibia and femur. After extracting the tibia and femur, bone marrow leukocytes were flushed out with a syringe containing cell culture medium. The cell suspension was filtered through a 70 μm cell strainer and washed once with cell culture medium, then the cells were treated with the erythrocytes lysis solution. After lysis, the cells were washed twice with cell culture medium, and the cells were used as bone marrow leukocytes. Pan-naïve T cells were isolated from the splenocyte by using an EasySep Mouse Pan-Naïve T Cell Isolation Kit (Stemcell Technology; Vancouver, BC, Canada). Naïve CD4+
T cells were isolated from the splenocyte by using a MagniSort mouse CD4 naïve T cell or mouse CD8 naïve T cell Enrichment kit (Thermo Fisher Scientific, Waltham, MA, USA), respectively. LN and splenic dendritic cells were isolated by using MagniSort Mouse CD11c Positive Selection kit (Thermo Fisher Scientific, Waltham, MA, USA). The whole procedure for the cell isolation kit was performed by following the manual.
Mouse BMDCs preparation. Mouse BMDCs were prepared by following a method described in a previous report [49
]. At day 0, 2.0 × 106
of bone marrow leukocytes were suspended in 10 mL of DC culture medium (cell culture medium containing 20 ng/mL of rmGM-CSF), and the cells were seeded on a 100 mm dish. At day 3, 10 mL of the fresh DC culture medium was added to the cultured cells. At days 6 and 8, half of the cultured medium was collected and centrifuged, then the cell pellets were resuspended in 10 mL of the fresh DC culture medium. The cell suspension was put back into the original plate. At day 10, cells were ready to use for each experiment.
BMDC stimulation assay. Naïve BMDCs were seeded on a 6-well plate with DC culture medium, then the cells were stimulated by the S. aureus-derived component, LTA (10 μg/mL), PGN (10 μg/mL), SA-LP (10 μg/mL or identical amounts with isolated from 106 of S. aureus), CWE (identical amount with isolated from 106 of S. aureus), CPE (10 μg/m or identical amount with isolated from 106 of S. aureus) and live-SA (106) and HK-SA (106) for 24 hr. The TLR2 agonist Pam3CSK4 (500 ng/mL) and TLR4 agonist LPS (100 ng/mL) were used for the positive control. TLR2 was blocked by the anti-mouse TLR2 monoclonal antibody (mAB) (10 μg/mL) or inhibited by Ssn B (100 μM). The stimulated cells were analyzed by flow cytometry for the detection of activation markers (CD80, DC86 and MHC class II). Cultured medium was harvested for the measurement of cytokine production by ELISA.
Flow cytometry. Cell surface markers and intracellular cytokines were analyzed by a flow cytometer (FACScalibur and LSR-II; BD Biosciences, Franklin Lakes, NJ, USA) with the fluorochrome-conjugated monoclonal antibodies described in reagents and antibodies. The cells were initially incubated with FcR blocker (anti-CD16/32; 2.4G2) at 4 °C for 10 min. For surface marker staining, the cells were incubated with the antibody at 4 °C for 30 min. Intracelluler cytokine staining was performed by using a Cytofix/CytoPerm Kit (BD Biosciences, Franklin Lakes, NJ, USA) by following the manual. Briefly, the cells stained with the antibody for the surface marker were fixed and permilized. The cells were incubated with the antibody for cytokine staining at 4 °C for 30 min. The dead cells were excluded by forward scatter, side scatter, and propidium iodide gating. All data were analyzed by BD FACS Diva (BD bioscience, Franklin Lakes, NJ, USA) or FlowJo (Tree Star; Ashland, OR, USA).
Murine skin inflammation model. To establish the skin inflammation model, anesthetized mice were treated with the antigen by intradermal (ID) injection into the ear (Figure S2
). The antigen was dissolved or diluted with 0.9% saline, then applied into the dorsal side of the ear. After 48 or 120 hr of the treatment, the mice were sacrificed and the treated ear and skin-dLN was excised for use in each analysis.
In vivo antigen tracking. For antigen tracking, FITC-labeled SA-LP and OVA or non-labeled OVA (10 μg in each) were ID-injected into the mouse ear. After 48 hr of the treatment, the skin dLN was extracted from the treated mice, then the isolated LN cells were analyzed by flow cytometry.
Ex vivo antigen re-stimulation. The skin leukocytes were isolated from the SA-LP ID-injected mice ears, then the cells were labeled by CFSE. The labeled cells were cocultured with stimulated splenic DCs (SA-LP 1 μg/mL or LPS 100 ng/mL + OVA 10 μg/mL) or naïve splenic DCs at 37 °C for 24 hr. The proliferated cells were analyzed by flow cytometry.
In vitro antigen presentation. Isolated primary DCs were stimulated with SA-LP (10 μg/mL) overnight. The stimulated or naive DCs were cocultured with pan-naïve T, naïve CD4+, or CD8+T cells for 72 hr. At the last 6 hr of the coculture, the proliferated cells were re-stimulated with 100 ng/mL of PMA, 1 μg/mL of ionomycin, and protein transportation was inhibited with Golgi stop (BD Bioscience). The proliferated cells were analyzed by flow cytometry.
In vitro MHC blocking assay. MHC class I and II molecules on the DCs were blocked with antibodies by following a method described in a previous report with modification [50
]. Briefly, LN isolated DCs were pre-incubated with the blocking antibody for MHC class I or MHC class II (10 μg/mL in each) for 1 hr. An isotype antibody was also used for the control. Then, the cells were stimulated with SA-LP (10 μg/mL) overnight. The stimulated DCs were cocultured with splenic naïve CD4+
T cells for 72 hr in the presence of the blocking antibody or isotype antibody. At the last 6 hr of the coculture, the proliferated cells were re-stimulated with 100 ng/mL of PMA, 1 μg/mL of ionomycin, and protein transportation was inhibited with Golgi stop (BD Bioscience). The cells were analyzed by flow cytometry.
SA-LP-activated dendritic cell transfer. The primary DCs were isolated from WT mouse LN, then the cells were labeled with CFSE and stimulated with SA-LP (10 μg/mL) at 37 °C overnight. The treated DCs were washed with cell culture medium three times, then the cells (2.0 × 106) were transferred into WT mice ears by ID injection. After 48 and 120 hr of the treatment, the skin-dLN and ear were used for each analysis.
Cytokine measurement by Enzyme-Linked Immuno Sorbent Assay (ELISA). The cytokine (TNF-α, IL-12p40 and IL-6) produced from the stimulated cell was measured by using a Mouse ELISA kit (Thermo Fisher Scientific, Waltham, MA, USA) for each target. The whole procedure was performed by following the manual.
Statistical analyses. A Student’s t-test was used to analyze the data for significant differences. Values of * p < 0.05, ** p < 0.01, and *** p < 0.001 were regarded as significant.