Tetraspanins are involved in Burkholderia pseudomallei-induced cell-to-cell fusion of phagocytic and non-phagocytic cells

Tetraspanins are four-span transmembrane proteins of host cells that facilitate infections by many pathogens. Burkholderia pseudomallei is an intracellular bacterium and the causative agent of melioidosis, a severe disease in tropical regions. This study investigated the role of tetraspanins in B. pseudomallei infection. We used flow cytometry to determine tetraspanins CD9, CD63, and CD81 expression on A549 and J774A.1 cells. Their roles in B. pseudomallei infection were investigated in vitro using monoclonal antibodies (MAbs) and recombinant large extracellular loop (EC2) proteins to pretreat cells before infection. Knockout of CD9 and CD81 in cells was performed using CRISPR Cas9 to confirm the role of tetraspanins. Pretreatment of A549 cells with MAb against CD9 and CD9-EC2 significantly enhanced B. pseudomallei internalization, but MAb against CD81 and CD81-EC2 inhibited MNGC formation. Reduction of MNGC formation was consistently observed in J774.A1 cells pretreated with MAbs specific to CD9 and CD81 and with CD9-EC2 and CD81-EC2. Data from knockout experiments confirmed that CD9 enhanced bacterial internalization and that CD81 inhibited MNGC formation. Our data indicate that tetraspanins are host cellular factors that mediated internalization and membrane fusion during B. pseudomallei infection. Tetraspanins may be the potential therapeutic targets for melioidosis.


Expression of tetraspanins on uninfected A549 and J774A.1 cells. The role of tetraspanins in
B. pseudomallei infection and MNGC formation were investigated in two cell lines. A549 represented a nonphagocytic human epithelial cell, and J774A.1 represented a phagocytic mouse macrophage cell. To determine if tetraspanins are expressed on the cell surface of the cells, we used flow cytometry and antibodies that recognise CD9, CD63, and CD81. We observed in both A549 and J774A.1 that the MFI level of cells treated with antibodies against CD9 and CD81 were significantly higher than those treated with isotype controls ( Supplementary  Fig. S1). In contrast, the MFI level of cells treated with the antibody against CD63 was not different from isotype controls. These results indicated that CD9 and CD81 expressed on A549 and J774A.1 at a relatively high level but that CD63 was poorly expressed on the cell surface.
Expression of tetraspanins on B. pseudomallei infected A549 and J774A.1 cells. We next investigated the level of these tetraspanins expression on A549 and J774A.1 cells during B. pseudomallei infection. We used flow cytometry and specific antibodies to determine CD9, CD63, and CD81 expressions on cells at 1, 4, and 12 h after infection at MOI 100 or 30. The result showed that expressions of CD9, CD63, and CD81 on both cells were not significantly different between B. pseudomallei infected cells and non-infected cells at all-time points ( Supplementary Fig. S2). These results demonstrate that B. pseudomallei infections did not alter CD9, CD63, and CD81 expression on these cells.

Effect of anti-tetraspanin MAbs and recombinant EC2 proteins on B. pseudomallei adhesion and internalization.
To determine the role of tetraspanins in B. pseudomallei adhesion and internalization in A549 and J774A.1 cells, we used MAbs specific to tetraspanins as well as recombinant EC2 proteins to treat cells for 1 h before infection. The controls were cells incubated with PBS or IgG isotype-matched MAbs or GST. The result showed no difference in bacterial adherence between cells pretreated with these reagents and isotype controls or PBS in both cell types ( Supplementary Fig. S3). Since J774A.1 cells are capable of phagocytosis, we considered that the measurement of associated bacteria at this point might be both bacterial adhesion and internalization. We, therefore, investigated the role of tetraspanin in adhesion in J774A.1 after blocking phagocytosis with 2 µg/ml cytochalasin D 17 . The result also showed no significant difference in adherence between J774A.1 cells pretreated with anti-tetraspanin MAbs and isotype controls or PBS, which was similar to the result of the experiment without cytochalasin D. The data suggest that tetraspanins are not involved in bacterial adhesion in J774A.1 cells (Supplementary Fig. S4). For internalisation, only pre-treatment with anti-CD9 MAb and CD9-EC2 significantly increased the percentage of bacterial internalisation into A549 cells (P = 0.003 for anti-CD9 MAb; P = 0.002 for CD9-EC2) (Fig. 1A,B). In contrast, none of these MAbs or EC2 proteins inhibited B. pseudomallei internalisation in J774.A1 cells. The data suggest different roles of CD9 for bacterial uptake between A549 and J774.1 cells.
of tetraspanins during B. pseudomallei-induced MNGC formation in A549 and J774A.1 at 12 h post-infection. In A549 cells, the number of MNGC formation and MNGC size of cells treated with MAbs against CD9 and CD63 and their EC2 proteins were not different from those treated with isotype or GST controls (Figs. 2, 3A,B). Interestingly, treatment with MAb specific to CD81 and CD81-EC2 significantly inhibited the number of B. pseudomallei-induced MNGC formation by 68.3% and 53.8% (P < 0.001 for both comparisons), respectively (Figs. 2, 3A,B) and reduced in MNGC size by 60.1% (P < 0.001) and 41.1% (P = 0.003), respectively compared with cells treated with isotype-matched controls or GST control (Figs. 2, 3C,D).
To investigate whether the inhibition of B. pseudomallei-induced MNGC formation by CD9-EC2 and CD81-EC2 may increase at higher EC2 concentrations, pretreatments of J774A.1 were performed with CD9-EC2 and CD81-EC2 at 10, 20, 50, and 100 µg/ml. We did not observe different MNGC formation between cells treated with PBS or GST control and those pretreated with ten µg/ml CD9-EC2, but there was a decrease in MNGC formation when cells were treated with 20, 50, and 100 µg/ml of CD9-EC2 concentrations ( Supplementary Fig. S5A). We further tested the effect of combined CD9-EC2 and CD81-EC2 pretreatment in J774A.1 cells, but there was no increase in inhibitory effect from the results of pretreatment with CD9-EC2 alone or with CD81-EC2 alone ( Supplementary Fig. S5B).
We next confirmed the role tetraspanins in B. pseudomallei-induced MNGC formation by treating the cells with anti-tetraspanin MAbs at 5 h post-infection, the time in which after bacterial internalization and before MNGC formation took place. Nevertheless, we observed no different results from those of adding anti-MAbs before infection (Supplementary Fig. S6). Furthermore, Giemsa staining of MNGC revealed that B. pseudomallei   www.nature.com/scientificreports/ uted in all areas of the cells ( Fig. 4 and Supplementary Fig. S8). J774A.1 and A549 cells were infected with B. pseudomallei K96243. Our investigations revealed the presence of bacteria within the cytoplasm of the infected J774A.1 and A549 cells (Fig. 4). CD81 was observed in all areas of infected A549 cells (Fig. 4A), and CD9 was distributed in infected J774A1 cells (Fig. 4D). Cells treated with isotype-matched controls were negative for CD81 and CD9 (Fig. 4B,E, respectively). The results also demonstrated that tetraspanins CD9 and CD81 were located with B. pseudomallei at membrane protrusion ends.

Generation of tetraspanins gene knockout in A549 cells.
To confirm the role of tetraspanins, CD9, and CD81 in B. pseudomallei internalization and MNGC formation in A549 and J774A.1 cells, we generated CD9 and CD81 gene knock out cell lines using the CRISPR/Cas9 system. sgRNA sequences targeting both human and mouse CD9 and CD81 were designed (Fig. 5A) and cloned into the PX459V2.0. A549 or J774A cells were transfected with individual plasmids expressing Cas9 and the corresponding sgRNA. As assessed by flow cytometry, gRNA1 and gRNA2 were able to disrupt CD9 and CD81 expression in A549 cells with efficiencies of 18% and 35%, respectively (Fig. 5B). Transfection of J774A.1 with the plasmids resulted in massive cell death, and the number of living cells was not enough for further analysis. Only CD9 and CD81 knockout A549 cells were further sorted by FACS. The sorted cells were isolated into a single cell by limiting dilution. Two single clones, A549-CD9KO1 and A549-CD81KO2, were expanded and selected for further analysis. Flow cytometry analysis confirmed the absence of CD9 and CD81 expression on the cell surface of those cells when compared with the unedited control (Fig. 5C). To characterize the nature of the mutation in these cells, PCR amplicon spanning the Cas9-sgRNA cleavage site was amplified from the two individual clones and analyzed the products on an agarose gel. Only one PCR fragment was amplified in the A549-CD9KO1 clone, while two PCR fragments were detected in the A549-CD81KO2 clone. The products from both A549-CD9KO1 and A549-CD81KO2 clones were smaller than the unedited control DNA, indicating that gene deletions were introduced into the genomic sequence of those cells ( Fig. 5D and Supplementary Fig. S9). To confirm these findings, individual PCR fragments were purified from the gel and subjected to direct sequencing. Sequence from A549-CD9KO1 displayed a 259-bp deletion in the 5′ untranslated region of CD9 in both alleles (Fig. 5D). This would inhibit CD9 expression via translation control. Two deleted alleles were identified from A549-CD81-KO2 (Fig. 5D). One exhibited a 15-bp deletion in the 5′ untranslated region of CD81 gene, which would affect its expression. Another allele revealed a 96-bp deletion that generated truncated protein that lacked the intracellular N terminal domain and EC1. The lack of EC1 would destabilize EC2 folding and further disrupt EC2 function 24 . The lack of the protein expressions on the cell surface was confirmed by FACs analysis, Western blotting ( Supplementary Fig. S10), and sequencing analysis, which indicated successful CD9 and CD81 gene knock out in A549-CD9KO1 and A549-CD81KO2 cell lines using the CRISPR/Cas9 system.

Comparison of the effect of anti-tetraspanin MAbs on MNGC formation induced by B. thailandensis and B. pseudomallei. The inhibition of B. pseudomallei-induced MNGC formation by CD9
and CD81 GST-EC2 proteins is consistent with their effect on this process in B. thailandensis infected mouse macrophages 16 . However, our previous report demonstrated that in this system, anti-CD9 and anti-CD81 MAbs enhanced MNGC formation, while here, the same antibodies inhibited B. pseudomallei-induced MNGC formation (Figs. 2 and 3). We, therefore, compared the effect of the anti-tetraspanin MAbs on MNGC formation in J774A.1 and A549 cells infected with B. thailandensis and B. pseudomallei K96243. Indeed, the results showed different effects of anti-CD9 and anti-CD81 MAbs between cells infected with the two Burkholderia species. We observed that pretreatment with anti-CD9 and anti-CD81 MAbs enhanced B. thailandensis-induced MNGC formation and MNGC size in J774A.1 cells (Figs. 8, 9A,B), consistent with our previous work 16 . The results of B. pseudomallei infection were the same as in previous experiments (Figs. 2 and 3), demonstrating that in this system, anti-CD9 and anti-CD81 MAbs significantly inhibited MNGC formation and reduced MNGC size in J774A.1 cells (Fig. 9C,D).

Discussion
Here, we demonstrated a role for host tetraspanins in B. pseudomallei internalisation and MNGC formation during B. pseudomallei infection in A549 and J774A.1 cell lines. In A549 lung epithelial cells, our results revealed that CD9 facilitated B. pseudomallei internalisation and that CD81 inhibited in MNGC formation. By blocking with a specific antibody to CD81 or recombinant CD81-EC2 protein, we found a decrease in the number and size of     www.nature.com/scientificreports/ MNGC formation. By blocking with specific antibodies or EC2 proteins, we showed a significant decrease in MNGC formation of infected cells. The tetraspanins CD9, CD63, and CD81 are known to be widely expressed in mammalian cells, but whereas CD9 and CD81 are predominantly expressed on the cell surface, CD63 is mainly distributed in intracellular vesicular compartments 25 . Our data are consistent with those reports confirming that CD9 and CD81 but not CD63 were expressed on the cell surface of A549 and J774A.1 cells. www.nature.com/scientificreports/ Tetraspanins are known to be involved in infections caused by other intracellular pathogens, particularly viruses 4,26,27 . There is also growing evidence for roles for tetraspanins in bacterial infection; for example, CD81 has been demonstrated to be involved in Listeria monocygenes invasion into human epithelial cells 28 . TEMs appear to act as "adhesion platforms" for a wide range of bacteria, and that agents that disrupt TEM can prevent bacterial attachment to host cells [29][30][31] . B. pseudomallei employs several virulence factors for adhesion and invasion to host cells such as PilA and adhesion proteins; BoaA and BoaB 32 . For invasion, BopE, type III secretion system (T3SS) effectors facilitate B. pseudomallei entry into host cells 33 . However, our study did not find that CD9, CD63, and CD81 mediate the adherence of B. pseudomallei to epithelial cells and macrophages. The interplay between host cells and B. pseudomallei is complex and not very well understood but presumably in macrophage uptake is mainly by phagocytosis.
Interestingly, our results demonstrated that anti-CD9 MAb and CD9-EC2 but not CD63 or CD81 reagents enhanced B. pseudomallei internalisation into A549 cells. The data suggest that anti-CD9 MAb or CD9-EC2 may stabilise or enhance the formation of entry platforms for bacteria. The interaction of CD9 with B. pseudomallei components that explains its involvement with bacterial internalisation is currently unknown. Further study is required to investigate the role of CD9 as well as other tetraspanins for B. pseudomallei internalisation and to identify their partner proteins for the organisation of "entry platforms" for B. pseudomallei infection. Our data cannot suggest that overexpression of CD9 in A549 leads to a decrease in internalisation during B. pseudomallei infection but data from Hassuna NA et al. suggest that overexpression of CD63 in HEK 293 cells is not sufficient to promote S. Typhimurium binding to non-phagocytic cells 30 .
There was no effect of CD9 or CD81 reagents on J774A.1 cells internalisation which might be expected to occur via phagocytosis but we observed the effect in A549 cells, where the mechanism presumably involves the T3SS. This suggests that the tetraspanins were not affecting phagocytosis 34 but may be affecting elements of the T3SS-mediated entry. CD81 has been shown to be necessary for internalisation of L. monocytogenes in which the mechanism involving phosphatidylinositol 4-kinase II 28 . CD81 might also affect actin polymerization, e.g., CD81 interfaces between the plasma membrane and the cytoskeleton by activating spleen tyrosine kinase, colocalizing ezrin-radixin-moesin proteins, and recruiting F-actin to facilitate cytoskeletal reorganization and cell signaling 35 . CD9 and CD81 have also been shown to be linked to the actin cytoskeleton via their interaction partner proteins EWI-2 and EWI-F 36 . Our data indicate that CD81 plays a significant role in MNGC formation in both epithelial cells and macrophages. In contrast, the data indicate that CD9 plays this role in macrophage cells only. The finding from our CRISPR/Cas9 knockout experiment in A549 cells has consistently confirmed the role of CD81 in B. pseudomallei-induced MNGC formation and CD9 in internalization. Although the CD9 and CD81 genes knockout experiment in J774A.1 cells was not successful, our data suggest that CD9 and CD81 are involved in B. pseudomallei-induced MNGC formation in this cell. The evidence for this belief comes from our consistent results that the antibodies and EC2 proteins of CD9 and CD81 can block B. pseudomallei-induced MNGC formation.
Our immunofluorescence and confocal microscopy analyses showed that CD9 and CD81 were distributed widely on the cell surface and located with B. pseudomallei during cell fusion. Other studies have also demonstrated the relevance of CD9 and CD81 on cell-to-cell fusion in other conditions. For example, anti-mouse CD9 and CD81 MAbs can inhibit sperm-oocyte fusion 37 . Gene knockout of CD9 or CD81 in mice could impair sperm-egg fusion while double knockout of CD9 and CD81 were completely interfile 38 . Anti-CD9 MAb can inhibit canine distemper virus-induced cell-to-cell fusion [12][13][14][15] . CD9-EC2 can inhibit monocyte fusion and reduce MNGC size in monocytes cultured under fusogenic conditions [39][40][41] . To our knowledge, the role of tetraspanins in MNGC formation has never been reported in B. pseudomallei. It is possible that these tetraspanins may not directly interact with bacterial molecules during the infection and MNGC formation processes, but they may be regulated by other host partner TEM proteins that may directly interact with B. pseudomallei. Other host proteins have been shown to be involved with B. pseudomallei infection including E-selectin (CD62E), a fusion regulatory protein (CD98), E-cadherin (CD324) integrin (LFA-1) complex, ICAM-1 (CD54), signal-regulatory protein α (CD172a), and integrin-associated protein (CD47). Blocking these proteins with specific antibodies could inhibit B. pseudomallei-induced MNGC formation in human leukemic monocyte lymphoma cell line 42 . We also showed the effects of antibodies to CD47, CD98, and dendritic cell-specific transmembrane protein (DC-STAMP) on B. thailandensis-induced fusion in J774.2 cells 16 . CD9 and CD81 that are involved in B. pseudomallei-induced MNGC formation may be associated with other cell surface molecules that may interact with B. pseudomallei. The antibodies and recombinant EC2 proteins target the EC2 regions of the tetraspanins. However, if the mechanism of action is affecting signalling in the cells, then this could be by the recruitment of intracellular signalling molecules via the C-terminal regions of the tetraspanins. Moreover, tetraspanin peptides are active on the host cell rather than on the bacteria themselves, as they have no toxicity to the bacteria, e.g., they have no anti-S. aureus adherence activity when added to the bacteria before adhesion to host cells 31 . Blocking of the EC2 domain of tetraspanins by antibodies may interfere with the association between tetraspanins and other partner molecules, leading to alteration of TEM organization and function 43 . This vital information warrants further investigation.
The effect of anti-tetraspanin MAbs on MNGC formation induced by B. thailandensis was different from those induced by B. pseudomallei. Specifically, anti-CD9 and anti-CD81 enhanced MNGC formation induced by B. thailandensis in both A549 and J774A.1 cells, whereas anti-CD9 and anti-CD81 inhibited MNGC formation induced by B. pseudomallei, suggesting that the roles of tetraspanins for cell-cell fusion are dependent on the microorganism. Although B. thailandensis is a closely-related species to B. pseudomallei and widely used as a model to study in melioidosis, our study suggests different effects of host tetraspanins transmembrane in the bacterial-induced cell fusion of these two species. The mechanism by which tetraspanins are involved in cell-cell fusion after Burkholderia infection is not well understood. Our data and data from other studies suggest that host factors and probably bacterial factors seem to be involved in the cell-to-cell spread process. The MNGC formation induced by B. pseudomallei or B. thailandensis requires the type VI secretion system (T6SS-5), which is regulated  18,44 . This mechanism could lead to cell death by activating autophagy and triggering the cGAS-STING pathway via micronuclear formation 20 . T6SS-5 plays an essential role in MNGC formation by which the C-terminal of VgrG5 processes fusogenic activity 18,45 . It is likely that the different amino acid sequences and conformation of these factors 45,46 might contribute to the interaction with associated tetraspanins. It will be of interest to determine the effect of CD9 or CD81 on MNGC formation during infection with B. pseudomallei T6SS-5 and VgrG5 mutants in future studies. It is not clear why MAb treatments were more efficient in inhibition of MNGC formation and subsequent reduction of MNGC size than recombinant EC2 protein treatments. A possible explanation is that the MAbs may be more specific to the disruption of tetraspanin function compared to the recombinant EC2 proteins.
Tetraspanins have been reported as a novel treatment for parasitic and bacterial infections. Synthetic peptides of tetraspanin, CD9 prevent Staphylococccus aureus skin and wound infections to cultured keratinocytes and human skin infection model 31 . Our finding suggests that targeting these tetraspanins using specific antibodies or recombinant EC2 proteins could inhibit B. pseudomallei-induced MNGC, thus limited bacterial spreading between different cell types and that CD9 and CD81 may be good vaccine targets. Antibodies or recombinant EC2 proteins may be useful for therapeutic purposes. However, this requires further evaluation in other cell types and testing in animal models of melioidosis.
Host and bacterial interplay are complicated, but our study provides insights into the understanding of B. pseudomallei-induced cell-to-cell fusion of human epithelial and mouse macrophage cells by which tetraspanins are associated with the process by their distribution on the cell surface. Our study in epithelial cells and macrophages revealed that CD9 and CD81 are involved in B. pseudomallei-induced MNGC formation. Antibodies specific to tetraspanins and EC2 proteins could be a potential therapeutic agent to prevent the spread of B. pseudomallei to uninfected cells.

Methods
Bacterial strain, cell lines, and growth conditions. B. pseudomallei strain K96243, a clinical isolate, and B. thailandensis strain E264, an environmental isolate from Thailand, were used in this study. The bacteria were grown on Columbia agar (Oxoid, UK) and incubated at 37 °C overnight. Human lung epithelial cell A549 (CCL-185, American Type Culture Collection, MD, USA) was cultured in RPMI 1640 (Gibco BRL, Grand Island, NY, USA) supplemented with 10% heat-inactivated fetal bovine serum (FBS) (HyClone, USA). Mouse macrophage cell line J774A.1 was cultured in DMEM (Gibco BRL). All cell cultures were performed at 37 °C in a humidified 5% CO 2 incubator. All experiments involving B. pseudomallei were performed in a biosafety level 3 laboratory.

Antibodies against tetraspanins and recombinant GST fusion proteins.
Three monoclonal antibodies (MAbs) that recognised human tetraspanins CD9 (clone 602.29), CD63 (clone H5C6), CD81 (clone 1D6, Bio-Rad, USA), and IgG1 isotype control (clone JC1) were used to treat A549 cells. MAbs that recognised mouse tetraspanins, including MAbs specific to CD9 (clone MF1, Bio-Rad), CD63 (clone NVG-2, BioLegend Inc., San Diego, CA, USA), CD81 (clone Eat2, Bio-Rad), and rat IgG2b, rat IgG2a, and hamster IgG1 isotype controls were used to treat J774A.1 cells. The production of recombinant glutathione S-transferase (GST) fusion proteins with CD9, CD63, and CD81 tetraspanin EC2 extracellular domains were previously described 30 . Examination of tetraspanin expression. Cell surface expression of tetraspanins on A549 and J774A.1 cells was assessed using flow cytometry. The experiment was performed as previously described 30 . The cells were cultured in the medium for 2 days. To remove cells from culture flasks, A549 cells were trypsinised with 0.25% of trypsin-EDTA (Gibco BRL), while J774A.1 cells were dislodged by cell scraper. The cells at a concentration of 1.0 × 10 5 cells were incubated with 5 µg/ml of each primary Mab against tetraspanins, or isotype control at 4 °C for 1 h. The cells were then washed with PBS followed by incubating with 5 µg/ml secondary antibodies (A549: Alexa Fluor 488 conjugated goat anti-mouse antibody (Invitrogen, Waltham, MA, USA) or FITC-conjugated anti-rat IgG; J774A.1: anti-hamster IgG (Bio-Rad, USA) at 4 °C for 30 min. The cells were centrifuged at 1,000 rpm at 4 °C for 5 min, washed with PBS, and analysed using a FACSCalibur flow cytometer (Becton Dickinson, Franklin Lakes, NJ, USA). The data was analysed using a Flowjo software version 10.1 (FlowJo, LLC). Two independent experiments were performed for each experiment.
Adhesion assay. A549 and J774A.1 cells were seeded at 1 × 10 4 cells in 100 µl per well into a 96-well plate and incubated at 37 °C for overnight. The cells were pretreated with 20 µg/ml anti-tetraspanin MAbs or recombinant EC2 proteins at 37 °C for 1 h, then washed with PBS and infected with B. pseudomallei K96243 at a multiplicity of infections per cell (MOI) at 100 for A549 and 30 for J774A.1. The plates were incubated at 37 °C with 5% CO 2 for 1 h. Non-adherent bacteria were detached by gentle washes with PBS for 5 times 47 , and the cells were lysed with 0.1% Triton X-100 (Sigma-Aldrich) for 5 min. The bacterial suspension was serially diluted with PBS, spread on Columbia agar, and incubated at 37 °C for 24 h for colony count. The number of adhered bacteria was enumerated as colony-forming units (CFU). The percentage of adhesion reflected the number of adherent bacteria at 1 h post-infection × 100/ the number of CFU added. The percentage of adhesion was calculated from three independent experiments; each was conducted in triplicate 47 .
In another experiment, J774A.1 cells were pretreated with 2 µg/ml of cytochalasin D (Sigma-Aldrich) for 2 h to block phagocytosis 17 . The adhesion assay was then performed as described above.
Internalization assay. A549  Generation of tetraspanins knock out cell lines. CD9 and CD81 genes knock out were performed with A549 and J774A.1 cells using a CRISPR/Cas9 system 50  The PCR amplicons were run on a 2% agarose gel. Each PCR fragments were purified and subjected for direct sequencing using the forward and reverse PCR.
Statistical analysis. Statistical analysis was performed using GraphPad Prism software version 6.0 (Graph-Pad Software Inc, La Jolla, CA). The data are presented as individual points and means ± standard deviation. The data were tested for the difference using one-way ANOVA for comparison among three or more groups and t tests for comparison between two groups. A statistically significant difference was considered at P value < 0.05.