The genome of novel O. splanchnicus 57 strain (isolated from a healthy fecal donor) was obtained by whole genome sequencing (WGS) to accurately identify the bacterium on species level with Kraken (Wood and Salzberg, 2014). The assembled genome size was 4.2 Mb with a G + C content of 43.3% (Table 1). Of the 3544 genes, 1687 (47.6%) were hypothetical proteins without a putative function assigned. O. splanchnicus 57 genome assembly matched the O. splanchnicus type strain 1651/6^T genome (NC_015160.1) with 99.84% identity and 85% query coverage. The assembled genome data of O. splanchnicus 57 was screened for genes that encode enzymes for Kdo2-lipid A modification using BLAST (Basic Local Alignment Search Tool, BLASTn algorithm). The genes encoding LpxA-LpxL acyltransferases for the Kdo2-lipid A moiety synthesis were found, but the genome lacked the gene lpxM needed for the hexa-acylation of the lipid A, which indicates that O. splanchnicus 57 is likely to produce less toxic, penta-acylated form of LPS.

We studied the adherence of O. splanchnicus 57 to (8 days post-plating) Caco-2 and HT-29 cell lines as well as to porcine mucus. The relative adherence percentage of the bacterium to the enterocyte cell lines and mucus was below 1%, which is considered to be unspecific, background level binding (Vesterlund et al., 2006; Figure 1A). Thus, O. splanchnicus 57 can be considered as non-adherent. In addition, we studied the ability of the isolate to enhance Caco-2 monolayer integrity using transepithelial electrical resistance (TER) measurements (Figure 1B). O. splanchnicus 57 did not induce changes in the epithelial integrity as the change in TER values remained at the same level as the medium control indicating that the isolate did not reinforce or compromise the epithelial monolayer.

Next, we examined the immunomodulatory effect of O. splanchnicus 57 cells and the culture supernatant, i.e., spent medium, on the TNF-α and IL-10 production by healthy human peripheral blood mononuclear cells (PBMCs). Heat-killed E. coli K12 was used as a proinflammatory control stimulus in both experiments. There was no significant difference in the TNF-α/IL-10 ratio released from PBMCs following a 24-h stimulation with O. splanchnicus 57 cells or E. coli K12 (Data not shown). However, the spent growth medium of O. splanchnicus 57 induced significantly higher levels of IL-10 in relation to TNF-α compared to E. coli K12 suggesting different amounts of pro- and anti-inflammatory bacterial effector molecules present in the spent medium as compared to intact bacterial cells (Figure 2A; Supplementary Figure 1). TNF-α and IL-10 production by PBMCs is presented as ratios collectively from all donors in Figure 2A and as total concentrations from each donor separately in Supplementary Figure 1. PBMCs from each individual donor produced significantly higher IL-10 and lower TNF-α levels after a stimulation with O. splanchnicus 57 spent medium compared to E. coli (Supplementary Figure 1). The O. splanchnicus 57 spent medium stimulated IL-10 release in human PBMCs ranging between 100–330 pg/ml depending on the donor (Supplementary Figure 1). The GAM growth medium for O. splanchnicus 57 itself (i.e., unspent medium control) did not affect the cytokine production in PBMCs. Stimulation of human PBMCs with LPS isolated from O. splanchnicus 57 and E. coli K12 showed no significant difference in the ratio of proinflammatory cytokine TNF-α to anti-inflammatory cytokine IL-10 measured after 24 h of stimulation (Figure 2B). The ratio of TNF-α/IL-10 produced by PBMCs was lower but not statistically significant after induction with O. splanchnicus 57 or B. fragilis type strain LPS as compared to E. coli K12 LPS. While B. fragilis type strain LPS stimulated statistically higher IL-10 production in PBMCs from all donors, O. splanchnicus 57 LPS stimulated higher IL-10 production in PBMCs of one out of four donors as compared to E. coli LPS (Supplementary Figure 2).

The lack of gene lpxM in the O. splanchnicus 57 genome indicated that the bacterium potentially harbors a penta-acylated lipid A, making its LPS less proinflammatory as compared to hexa-acylated LPS of E. coli. LPS extract from O. splanchnicus 57 was used to stimulate human THP-1 dual reporter cells to study the induction of NF-κB signaling (Figure 2C). After a 24-h stimulation, the level of NF-κB activity induced by O. splanchnicus 57 LPS was equivalent to B. fragilis type strain that was used as control (Figure 2C). The lipid A of B. fragilis type strain LPS has a penta-acylated, less endotoxic form compared to E. coli LPS (Weintraub et al., 1989), as also shown by the level of NF-κB signaling in human THP-1 cells induced by different bacterial LPSs in our experiments. Taken together, the results suggest that O. splanchnicus 57 LPS is less pro-inflammatory as compared to E. coli LPS and that the spent medium of O. splanchnicus 57 contains substances that counteract the pro-inflammatory effect of LPS.

Considering the potentially bioactive effector molecules secreted by O. splanchnicus 57, we studied the presence of OMVs in the bacterial culture. We used a previously published protocol (Elhenawy et al., 2014) with modifications to isolate OMVs. Transmission electron microscopy (TEM) was used to confirm the presence of OMVs in the preparations. Also, O. splanchnicus cells were also subjected to TEM. In the electron micrographs, O. splanchnicus 57 bacterial cells appeared as short rods as previously described in the literature (Göker et al., 2011). OMVs were clearly visible in the electron micrographs (Figure 3). The OMVs of O. splanchnicus 57 were observed to be smaller than 100 nm in size.

Next, we assessed the effect of O. splanchnicus 57 cells, spent medium and OMVs on the induction of proinflammatory cytokine IL-8 in enterocytes. HT-29 cell line was used to assess the ability of the strain to attenuate LPS-induced inflammatory responses i.e., IL-8 release from enterocytes as Caco-2 cell line has defects in the TLR4 signaling, which causes unresponsiveness to LPS stimulation (Van De Walle et al., 2010). HT-29 cells were incubated with different concentrations of the bacterium, spent medium and vesicles (Figures 4A–C). E. coli K12 and its OMVs were used as proinflammatory controls and B. fragilis as a control that does not elicit a reaction in HT-29 cells. McCoy 5A medium, the growth medium for HT-29 cells, was included as the background control. The 1:10 dilution of O. splanchnicus 57 and B. fragilis cells (corresponding to 10⁷ cells per ml) prompted IL-8 responses that were above background but were four times lower as compared to the responses induced by E. coli K12 (Figure 4A). The 1:100 and 1:1000 dilutions of O. splanchnicus 57 and B. fragilis cells induced IL-8 release statistically at the similar level as the background showing a dose-dependent decrease in response, whereas E. coli K12 triggered a strong response with all the concentrations showing a saturated induction already with the most diluted sample. Thus, the difference in the proinflammatory capacity of E. coli and the two other commensals was clear. The same result could be observed when using spent medium, as both O. splanchnicus 57 and B. fragilis spent media diluted to 1/2 or 1/4 of the adjusted concentration elicited only a minor IL-8 response, which did not differ significantly from the background and GAM medium control (Figure 4B). In contrast, the 1/2 dilution of E. coli K12 spent medium was excessively toxic to the HT-29 cells (not shown due to partial cell death resulting in unreliable quantification), and 1/4 spent medium dilution triggered a major induction of IL-8 in HT-29 cells (582 ± 78 pg/ml) significantly above the background level. Also, OMVs isolated from O. splanchnicus 57 and E. coli K12 induced significantly distinct IL-8 responses with all the amounts used (10⁷ – 10¹⁰ vesicles) as E. coli K12 OMVs elicited a four-fold stronger response than O. splanchnicus 57 OMVs (Figure 4C).

Subsequently, we also examined the anti-inflammatory capacity of O. splanchnicus 57 by assessing whether the bacterium could attenuate E. coli LPS –induced IL-8 production in HT-29 cells (Figure 4D). In three biological experiments out of five, incubation with O. splanchnicus 57 with HT-29 monolayer prior to LPS-induced inflammation significantly decreased the IL-8 release. In the two other experiments, the IL-8 levels after O. splanchnicus 57 treatment were also lower than LPS control, but the difference did not reach statistical significance. The culture supernatant of O. splanchnicus 57 drastically attenuated the LPS-induced IL-8 response to the background level with both, 1/2 and 1/4, dilutions (Figure 4E). Medium control, i.e., GAM broth, did not have an effect as the induced IL-8 levels were the same or higher than the LPS control. Interestingly, also O. splanchnicus 57 OMVs dose-dependently decreased the IL-8 production in LPS-induced HT-29 cells suggesting an anti-inflammatory capacity (Figure 4F). The decrease was significant only with the two highest concentrations, 10¹⁰ and 10⁹ OMVs per well, which yielded a clear, 50% or more decrease as compared to the LPS control.

Odoribacter splanchnicus isolate 57 was isolated from the feces of a healthy, pre-screened FMT donor as described in our previous study (Hiippala et al., 2020). The use of the donor sample was approved by the Ethics Committee of Hospital District of Helsinki and Uusimaa Finland (DnroHUS124/13/03/01/11). The donor, described in a previous study (Jalanka et al., 2016), provided a written informed consent. In short, the frozen banked fecal solution (saline-10% glycerol) was thawed, serially diluted in PBS and cultivated on Gifu anaerobic medium (GAM; Nissui Pharmaceutical Co., Ltd., Japan) for 48 h under anaerobic conditions (Concept Plus anaerobic workstation, Ruskinn Technology Ltd., Bridgend, United Kingdom) at 37°C. Separate colonies were picked and re-streaked on new agar plates until they were purified. Gram-staining and microscopy were used to evaluate the purity of the isolate followed by partial 16S rRNA gene sequencing for tentative identification. Bacterial mass from the isolate was resuspended in TE buffer (10 mM Tris–HCl, 1 mM EDTA, pH 8), heated to 95°C for 15 min to break the cells and used as PCR template. PCR amplification of the partial 16S rRNA gene was carried out using 27F DegL (5′-AGR GTT YGA TYM TGG CTC AG-3′) forward and Pd (5′-GTA TTA CCG CGG CTG CTG-3′) reverse primers. Sanger sequencing of the PCR product using the forward primer was carried out in the Institute of Biotechnology core facility, University of Helsinki. The partial 16S rRNA gene sequence was compared to NCBI 16S ribosomal RNA sequences database using BLASTn to acquire a genus-level identification.

Genomic DNA was extracted from O. splanchnicus 57 cell pellet with DNeasy Blood and Tissue Kit (Qiagen, Hilden, Germany) following the manufacturer’s protocol for Gram-negative bacteria and fragmented to approximately 500 bp using a Q800R2 Sonicator (QSonica, Newtown, CT, United States). Genome libraries were prepared for paired end sequencing using the NEBNext^® Ultra^TM II Kit (New England Biolabs, Ipswich, MA, United States) and quantification was carried out using the Library Quantification Kit (KAPA Biosystems, Waltham, MA, United States). Libraries were pooled in equimolar amounts and sequenced on the MiSeq (Illumina, Inc., San Diego, CA, United States). Short read data were assembled using unmanned genome assembly pipeline (UGAP;¹) using the SPAdes genome assembler. The species of each genome was identified with Kraken (Wood and Salzberg, 2014). Read data were deposited in the NCBI SRA under BioProject PRJNA575760. Genomic comparison to O. splanchnicus type strain 1651/6^T genome (NC_015160.1) and screening of the genes that encode enzymes for Kdo2-lipid A modification we carried out using BLAST (Basic Local Alignment Search Tool, BLASTn algorithm).

Bacteroides fragilis type strain E-022248T (= DSM 2151 = ATCC 25285) from the VTT Culture Collection (VTT Technical Research Center of Finland) was grown under anaerobic conditions for 2 days at 37°C on Brucella agar with hemin and vitamin K (Sigma Aldrich, St. Louis, MO, United States) along with 5% defibrinated sheep blood (Bio Karjalohja Oy, Finland). Escherichia coli K12-derived TOP10 (Invitrogen, United States) was aerobically cultivated overnight in Luria–Bertani broth (Becton Dickinson, United States).

The human colonic epithelial cell lines Caco-2 (ACC 169) and HT-29 (ACC 299) were obtained from the German Collection of Microorganisms and Cell Cultures (DSMZ). Cell lines were grown at 37°C in an incubator under an oxic atmosphere with 5% CO₂ and passaged after reaching 80% confluence (approximately every 3–4 days) using TryplExpress (Lonza, United States) to detach the cells. Passages 6–28 were used in the experiments. HT-29 cells were cultivated in McCoy 5A (Lonza, Belgium) medium containing 10% heat-inactivated (30 min at 56°C) fetal bovine serum (FBS; Gibco) and 100 U ml^–1 PEST. Caco-2 cells were grown in RPMI 1640 medium (Sigma-Aldrich, United States) supplemented with 20% FBS, non-essential amino acids (1%, NEAA; Lonza, Belgium), 15 mM HEPES (Lonza, Belgium), 100 U ml-1 penicillin and streptomycin (PEST; Lonza, Belgium) and 2 mM L-glutamine (Lonza, Belgium). Human THP-1 dual reporter monocytic cell line (thpd-nfis, Invivogen, United States) were cultured at 1 × 10⁵ cells/ml in 96-well plates in RPMI 1640 (Gibco, United States) supplemented with 10% fetal bovine serum (Gibco) and 100 U ml^–1 penicillin and streptomycin.

The adherence of O. splanchnicus isolate 57 to Caco-2 and HT-29 cell lines (8 days post-plating) and mucus was studied as previously described (Kainulainen et al., 2013; Reunanen et al., 2015). The bacterium was grown in GAM medium supplemented with 10 μl ml^–1 of [6′-³H]thymidine (17,6 Ci mmol^–1, Perkin Elmer, United States) to metabolically radiolabel the cells. Six technical replicates (parallel wells) were used in each experiment. To assess the adherence to intestinal mucus, porcine mucus (Sigma-Aldrich, 50 μg well-1 in PBS) was allowed to absorb to Maxisorp microtiter plate wells overnight at 4°C. 10,000 Caco-2 or HT-29 cells per well were seeded onto 96-well microplate. [³H]Thymidine-labeled O. splanchnicus isolate 57 cells were washed with an appropriate medium (McCoy 5A for HT-29 cells, RPMI 1640 for Caco-2 cells and PBS for the mucus assay) without the supplements and adjusted to OD_600nm 0.25 corresponding to approximately 10⁸ cells/ml. After 1 h of incubation on the epithelial cell monolayer or mucus at 37°C in a CO₂ incubator, the bacterial suspensions were removed, and wells were washed three times to remove the non-adherent bacteria. Adhered bacteria were lysed with 1% SDS-0.1 M NaOH solution and radioactivity was measured with a liquid scintillator (Wallac Winspectral 1414, Perkin-Elmer, Waltham, MA, United States). The percentage of bound bacteria was calculated relative to the radioactivity of the bacterial suspension initially added to the wells.

Caco-2 cell line is suitable for the monolayer integrity studies measuring TER due to the cells’ enterocytic differentiation and expression of intercellular junctional complexes (Hidalgo et al., 1989; Klingberg et al., 2005; Hsieh et al., 2015). The effect of O. splanchnicus 57 isolate on the TER of Caco-2 monolayer was studied as previously described (Kainulainen et al., 2015; Reunanen et al., 2015). Briefly, 50,000 Caco-2 cells were seeded on PET inserts with a pore size of 3 μm (Sarstedt, Nümbrecht, Germany) and grown for 8 days. The medium in the well was changed every 3 days. E. coli TOP10 was used as a negative control in the experiments due to its detrimental effect on monolayer integrity (Reunanen et al., 2015). O. splanchnicus 57 was washed with RPMI medium including the supplements, adjusted to OD_600nm 0.25 and 100 μl of bacterial suspension was added onto cell monolayer. The TER was measured after 0 and 24 h using an EVOM epithelial voltmeter with an electrode (World Precision Instruments, United Kingdom). The blank resistance (measurement at time point 0) was subtracted from the measurements made after 24 h of incubation, and the unit area resistance (Ω cm²) was calculated by multiplying the tissue resistance values by the surface area of the filter membrane. The change in TER during the 24 h was calculated comparing the samples to the medium control.

Odoribacter splanchnicus 57, B. fragilis and E. coli K12 were grown in an appropriate culture broth. Cells were harvested by centrifugation (4,000 rpm, 10 min) and resuspended in single-phase Bligh and Dyer mixture consisting of chloroform/methanol/water (1:2:0.8 v/v) to remove phospholipids (Vatanen et al., 2016). After 20 min of incubation at room temperature, the suspension was centrifuged, and the cell pellet was washed twice with Bligh and Dyer mixture and the remaining LPS preparation was suspended in PBS. The yielded LPS concentration was quantified using the chromogenic LAL assay (Thermofisher, United States), by following the manufacturer’s instructions.

Following overnight incubation, THP-1 dual reporter cells were induced to M0 macrophage differentiation with 15 ng/ml PMA for 24 h. Upon differentiation, cells were stimulated for 24 h with 10 ng/ml LPS from different bacterial strains: O. splanchnicus 57; B. fragilis type strain; E. coli K12 (tlrl-eklps InvivoGen; United States), smooth LPS from E. coli 0111:B4 (tlrl-eblps, Invivogen; United States), semi-rought LPS from E. coli R515 and rough LPS from E. coli J5 (IAX-100-007-M001 and 014-M001, respectively, Adipogen; United States). PBS was used as a control. Following 24-h stimulation, the supernatant was collected for the measurement of NF-κB activity.

THP-1 dual cell line NF-κB activity was measured according to the manufacturer’s instructions (InvivoGen; United States). Briefly, the reporter cells have a stable integrator secreted alkaline phosphatase (SEAP) reporter gene for monitoring nuclear factor (NF)-κB activation. Upon NF-κB activation, the SEAP reporter gene is activated, leading to the secretion of alkaline phosphatase, which is then quantifiable by a colorimetric assay (QUANTI-Blue, InvivoGen; United States) using a microplate reader (FluoStar Omega, BMG Labtech; Germany) with absorbance set for 630 nm.

Buffy coat blood from healthy donors were obtained from the IRCCS Policlinico San Matteo, Pavia, Italy. PBMCs were isolated using density gradient centrifugation (Ficoll-Paque Plus, GE Healthcare) according to the manufacturer’s instructions. Briefly, blood sample was diluted with DPBS and layered on Ficoll-Paque Plus. After centrifugation (300 g, 30 min), the separated mononuclear cells were collected, resuspended in DPBS and washed 3 times with DBPS to remove the platelets. Cells were resuspended in DPBS + 0.5% BSA + 2mM EDTA and counted.

PBMCs were added to 48-well plates (5,00,000 cells per well) in DMEM medium (ThermoFisher Sci.) supplemented with 10% FBS. O. splanchnicus 57, B. fragilis and E. coli K12 LPS preparations were diluted to 1 ng/ml in DMEM + 10% FBS. O. splanchnicus 57 was grown in GAM broth under anaerobic conditions for 2 days. The bacterial culture was centrifuged (10, 000 rpm, 3 min) and the cell pellet was washed with DMEM + 10% FBS, adjusted to OD_600nm 0.25 and 50 μl of this suspension was added onto PBMCs. DMEM + 10% FBS was used as background control. For the experiments with O. splanchnicus 57 spent medium, the bacterial suspension was adjusted to OD_600nm 1.0, centrifuged (16 000 rpm, 3 min) and the supernatant was filtered with 0.2 μm filter. 40 μl of the spent medium preparation (5%) was added onto PBMCs. Heat-killed E. coli K12 MG1655 (10⁸ cells/ml) and the bacterial growth medium (GAM) were used as controls in the experiments with bacterial cells and spent medium. PBMCs were incubated with LPS, bacterial cells or spent medium for 24 h and the supernatants were collected for cytokine analysis. TNF-α and IL-10 (hIL-10 JES3-19F1 clone; hTNF-a MAb11 clone, Biolegend, United States) were measured by using ELISA assay according to the manufacturer’s instructions. Unstimulated PBMCs were used as control in all experiments. In the LPS stimulation, the results were calculated as a ratio of TNF-α/IL-10 using PBMCs isolated from the buffy coats of four different donors. In the experiment of O. splanchnicus 57 spent medium and bacterial cell stimulation, the TNF-α/IL-10 ratio was calculated as a mean and standard deviation of three or four experiments by using PMBCs isolated from buffy coats of three or four donors, respectively.

Outer membrane vesicles isolation from 400 ml of O. splanchnicus 57 and E. coli K12 broth culture was carried out as previously described (Elhenawy et al., 2014) with modifications by pelleting the bacteria (centrifugation 4000 rpm for 10 min), filtering the supernatant with 0.2 μm filter to remove any remaining bacteria and concentrating the filtered supernatant by using Amicon 100 kDa filters (Millipore). The concentrated supernatant was ultracentrifuged (Beckman Coulter) for 2 h at 32,000 rpm. The remaining OMV pellet was resuspended in PBS and ultracentrifuged (2 h, 32,000 rpm) again. The isolated OMVs were suspended in PBS for nanoparticle tracking analysis (NTA) for quantification and electron microscopy (EM).

OMVs and bacterial samples were prepared for electron microscopy (EM) by loading to carbon coated and glow discharged 200 mesh copper grids with pioloform support membrane (Puhka et al., 2017). Samples were fixed with 2.0% PFA in NaPO4 buffer, stained with 2% neutral uranyl acetate, further stained and embedded in uranyl acetate and methyl cellulose mixture (1.8/0.4%). OMVs and bacterial cells were viewed with transmission EM using Jeol JEM-1400 (Jeol Ltd., Tokyo, Japan) operating at 80 kV. Images were taken with Gatan Orius SC 1000B CCD-camera (Gatan Inc., United States) with 4008 × 2672 px image size and no binning.

The measurement of IL-8 response in 8 days post-plating HT-29 cells by bacterial cells, spent medium and OMVs was carried out as previously described (Kainulainen et al., 2013). In brief, the bacterial suspensions were washed with McCoy 5A medium supplemented with 10% FBS and adjusted to OD_600nm 0.25. Bacteria were diluted to 1:10, 1:100, and 1:1,000 and incubated on the HT-29 cells for 3 h at 37°C in a CO₂ incubator. Spent media from broth cultures of O. splanchnicus 57, B. fragilis type strain and E. coli K12 were prepared by adjusting the bacterial culture to OD_600nm 1.0, pelleting the bacteria by centrifugation and filtering the supernatant through 0.2 μm filter to remove any remaining cells. Spent medium preparation was diluted to 1:4 and 1:2 using McCoy 5A medium with supplements. Bacterial growth medium (GAM) diluted with McCoy 5A medium was used as control. OMVs were diluted to McCoy 5A with supplements and added on monolayers using 10¹⁰, 10⁹, 10⁸, and 10⁷ concentrations. Three technical replicates (parallel wells) were used in each experiment. OptEIA Human IL-8 ELISA kit (BD Biosciences, United States) was used according to the manufacturer’s instructions to measure the concentration of the chemokine in the culture media.

IL-8 assays were only performed using the HT-29 cell line, because Caco-2 cells are known to be unresponsive to LPS stimulation possibly due to defects in TLR4 signaling (Funda et al., 2001; Van De Walle et al., 2010; Hsu et al., 2011). The capacity of O. splanchnicus 57 isolate or its effector molecules to attenuate LPS-induced IL-8 production in HT-29 cells was measured as previously described (Kainulainen et al., 2015; Hiippala et al., 2020). In brief, HT-29 cells were seeded 10,000 cells per well onto 96-well microplates for the attenuation assay. Broth culture of O. splanchnicus 57 was washed once with McCoy 5A medium supplemented with FBS (10%) and adjusted to OD_600nm 0.25 using the same medium. 100 μl of bacterial suspension was added onto 8 days old HT-29 cells and incubated at 37°C for 1 h under oxic atmosphere with 5% CO₂. Spent medium and OMV dilutions were prepared as in the experiments measuring IL-8 release. McCoy 5A medium containing 10% FBS without LPS was used as the background control. In the experiments with spent medium, GAM broth with LPS was used as medium control. Next, the bacterial suspension, spent medium or OMVs were removed from the HT-29 monolayer and 200 μl McCoy 5A medium with 1 ng/ml of E. coli LPS (Sigma) was added. After a 4-h incubation with LPS, supernatants were collected and IL-8 levels were measured by using an ELISA assay (BD OptEIATM Set). The IL-8 levels were calculated using four parametric logistic curves. The attenuation capacity of the isolate, spent medium and OMVs were assessed by comparing the LPS-induced IL-8 production (%) of the sample to that of the LPS control (100%). The experiment with O. splanchnicus 57 bacterial cells was repeated five times.

All the experiments were repeated two to five times (biological replicates) depending on the assay to confirm the results. An unpaired t-test was used to determine significant differences between two groups, such as the sample and the control. Homoscedasticity testing was performed with Levene’s test to identify equal or unequal variances. All statistical analyses were carried out with GraphPad Prism 8.4.1 (GraphPad Software, United States). A p-value of <0.05 was considered statistically significant.