In our previous studies as well in this study, we demonstrate the superior ability of osteoclasts (OCs) to condition NK cells for greater expansion and heightened function (Figure S1) (31). Here, we compared cell expansion, IFN-γ secretion, and NK cell-mediated cytotoxicity of untreated, IL-2 treated NK, IL-2+anti-CD16mAbs treated NK, and IL-2+anti-CD16mAbs+sAJ2 treated NK cells with IL-2+anti-CD16mAbs+sAJ2+OCs treated NK cells, and found higher cell expansion and increased function in the presence of OCs (Figure 1). Probiotic bacteria, sAJ2 is a combination of 7-8 different strains, and is prepared as described previously (32). Due to their unusually high expansion rate and potent function, we coined IL-2+anti-CD16mAbs+sAJ2+OCs treated NK cells as supercharged NK (sNK) cells to differentiate them from all the other NK cell subsets that we had tested in our laboratory throughout the last 30 years (28). To further understand the differences between IL-2 treated primary NK (NK+IL-2) cells and sNK cells, we performed single-cell RNA sequencing. In the analysis of NK+IL-2 and sNK cells, we also integrated untreated NK cells derived from donor PBMC to help characterize the NK cell subsets. By studying untreated NK, NK+IL-2, and sNK cells, we were able to identify 4 transcriptionally unique NK cell clusters (Figure 2A). All 4 clusters have a consistent expression of NK cell genetic markers (IL2RB, CD7, NKG7). Among the 4 clusters, Cluster 1 has the highest expression of IL7R and NCAM1, resembling the transcriptional signature of previously characterized CD56^bright NK cells (33). Cluster 3 has a comparably higher expression of genes related to cytotoxicity (GZMB, PRF1) and FCGR3A, which are identified as the genetic signature of cytotoxic CD56^dim NK cells. The gene expression pattern of cluster 2 follows a subset of transitional NK cells between CD56^bright and CD56^dim NK cells. When the single-cell clusters are split by the three conditions, cluster 4 is shown to be exclusive in untreated NK cells (Figure 2B). We have also analyzed the expression of the main effector molecules in NK cell cytotoxic granules: perforin-1, granzyme B and cathepsin C (cathepsin C is responsible for the activation of granzyme B) (32). A significantly higher expression of granzyme B and cathepsin C in the presence of slightly decreased perforin was seen in sNK cells in comparison to primary IL-2 treated NK cells (Figures 2C, S2A). Through the cell-cycle score analysis on the single-cell RNA sequencing result, a considerably higher amount of sNK cells is assigned to the G2M phase, indicating a more active proliferation program in the sNK cells compared to untreated and NK+IL-2 cells (Figure 2D). Also, by performing SCENIC analysis on the sequencing data, a distinct regulon network is utilized in the sNK cells compared to either untreated or NK+IL-2 cells. Among the predicted regulon activities, sNK cells have upregulated regulon activities associated with NK cell survival (STAT2, IRF9) and effector functions (IRF1, JUN, STAT1, HIF1A) (Figure 2E) (27, 34–37). When assessed NK cell-mediated cytotoxicity of IL-2 treated primary NK and sNK cells against oral squamous carcinoma stem cells (OSCSCs), Mia PaCa-2 (MP2), and K562 cell lines, significantly higher cytotoxicity was mediated by sNK cells in comparison to NK+IL-2 cells (Figures 2F–H). We also observed higher secretion of IFN-γ and TNF-α by sNK cells in comparison to primary NK cells treated with IL-2 (Figures 2I–K, S2B). These results exhibited higher anti-cancer activity of sNK cells in comparison to primary activated NK cells.

Our previous findings have demonstrated that differentiated tumors were more sensitive to chemotherapeutic drugs in comparison to CSCs/poorly differentiated tumors (12). Here, we determined the extent of cell death of oral cancer stem-like cells (OSCSCs), NK-diff-OSCSCs, CSCs/poorly differentiated pancreatic cancer MP2, and differentiated pancreatic cancer (PL12) and NK-diff-MP2 with or without the treatments with chemotherapeutic drugs cisplatin (CDDP) and paclitaxel (Figure 3). We observed higher cell death induced by CDDP (Figures 3A, C, S3) and paclitaxel (Figures 3B, D) against differentiated tumors in comparison to their CSCs/poorly differentiated counterparts. Thus, differentiation of tumors by the NK cells is an important step not only in curtailing the tumor growth, but more importantly in the response to chemotherapy drugs.

We used humanized-BLT (hu-BLT) mice model to demonstrate the efficacy of combinational treatment with sNK cells and chemotherapy against human oral CSCs/poorly differentiated tumors. Hu-BLT mice were generated by surgically implanting pieces of human fetal liver and thymus tissues under the renal capsule of NSG mice, followed by tail vein IV injection of same-donor CD34⁺ hematopoietic cells to support full reconstitution of the human bone marrow (38–40). In this study, hu-BLT mice after human immune cell reconstitution were surgically with human OSCSCs in oral cavity followed by IV injections of sNK cells and CDDP sequentially as depicted in Figure (Figure 4A). Upon sacrifice, the tumors were harvested and weighed. sNK cells alone or sNK cells combined with CDDP treated mice had smaller tumors in comparison to untreated mice with tumor (Figure 4B). Next, we dissociated tumors and recovered single cells to determine tumor cell counts. Tumors from tumor implanted mice treated with sNK cells alone or sNK cells combined with CDDP treated mice had significantly lower numbers of tumor cells as compared to tumor alone implanted mice (Figure 4C). When the same numbers of dissociated tumor cells from hu-BLT mice were cultured, significantly lower tumor cell expansion was seen in tumors from sNK cells alone or sNK cells combined with CDDP treated mice as compared to untreated tumors from tumor-bearing mice until day 14-19, after which the tumor growth rate gradually increased and at day 30 the tumor cultures were terminated. Even though the levels of tumor growth approached the levels seen with tumor alone implanted mice, we could still see a higher decrease in tumor growth with sNK+CDDP group as compared to sNK group, and both groups had on average less growth when compared to those from tumor alone implanted mice (Figure 4D). When the dissociated tumors of hu-BLT mice were used as targets for primary IL-2 activated NK cells in Cr-release assay, tumors from sNK cells alone or sNK cells combined with CDDP treated mice were killed much less when compared to untreated tumor-bearing mice (Figure 4E). In addition, there was statistically significant differences in the resistance of tumor cells to NK cell-mediated cytotoxicity dissociated from sNK+CDDP group as compared to sNK group, and both these groups had highly significant decreases in cytotoxicity when compared to those obtained from tumor alone implanted hu-BLT mice (Figure 4E). Our previous studies have demonstrated that CSCs/poorly-differentiated tumors grow at higher rate and are excellent targets of NK cell-mediated cytotoxicity, whereas differentiated tumors grow slow and are resistant to NK cell-mediated cytotoxicity (13, 41–43). Results shown in Figures 4D, E indicates that tumors from sNK cells alone or sNK cells combined with CDDP treated mice exhibited characteristics of differentiated tumors. In addition, when all the floating immune cells were removed by changing the media from tumor cultures throughout the days of 7-27 the differentiated nature of tumor cells grown from sNK cells or sNK in combination with CDDP gradually reverted to their CSC/poorly differentiated tumors, and their growth rate gradually increased and approached to those grown from tumor alone implanted BLT mice (Figures 4D). We have previously shown that reversion of NK differentiated tumors occurs after two weeks in culture without immune cells, and it correlates with the decreased MHC class I expression on tumor cells (44).

We then assessed IFN-γ secretion and NK cell-mediated cytotoxicity of splenocytes, peripheral blood and bone marrow derived cells of tumor-bearing hu-BLT mice with or without treatments. We observed suppression of both secretion of IFN-γ (Figures 4F, S4, S5A) and NK cell-mediated cytotoxicity (Figures 4G, S5B) in tumor-bearing untreated mice in comparison to those obtained from healthy mice without tumor implantation. Increase or restoration of IFN-γ secretion and NK cell-mediated cytotoxicity by bone marrow, splenocytes and PBMCs were seen in sNK cell injected tumor-bearing mice, and both these functions were further increased with the combination of sNK cells and CDDP treatment of tumor implanted mice when compared to those with tumor alone implanted mice (Figures 4F, G, S4, S5). CDDP alone injected mice either did not increase or increased slightly the secretion of IFN-γ in cells dissociated from spleen, peripheral blood, and BM (Figure S4).

Our previous studies have shown that CSCs/poorly-differentiated tumors are excellent targets of direct NK cell-mediated cytotoxicity, whereas their differentiated counterparts are significantly more resistant (13, 41–43). We have also shown previously that differentiated tumors have higher surface expression of MICA/MICB and are susceptible to ADCC mediated by the primary NK cells in the presence of anti-MICA/MICB antibody, even though NK cells are not able to kill these tumors directly (20). Here, we evaluated NK cell-mediated cytotoxicity against untreated stem-like (OSCSCs and MP2) and untreated or anti-PD-L1 antibody treated differentiated tumors (NK-Diff-OSCSCs, OSCCs, NK-Diff-MP2, and PL12) (Figure 5A–F). We demonstrate that NK cells mediated direct cytotoxicity of OSCSCs (Figure 5A) and MP2 (Figure 5D) tumors, whereas susceptibility to NK cell-mediated cytotoxicity was substantially and significantly lower against NK-diff-OSCSCs (Figures 5B, S6A), OSCCs (Figure 5C, S6A), NK-diff-MP2 (Figures 5E, S6B), and PL12 (Figure 5F, S6B) tumors when compared to OSCSCs and MP2 cells. NK cell-mediated cytotoxicity was increased against anti-PDL1-treated NK-diff-OSCSCs (Figure 5B), OSCCs (Figure 5C), NK-diff-MP2 (Figure 5E), and PL12 (Figure 5F) tumors. In accordance, higher surface expression of PD-L1 was seen on NK-diff-OSCSCs and OSCCs (Figure 5G) and NK-diff-MP2 and PL-12 (Figure 5H) tumors in comparison to their stem-like counterparts. Differentiated tumors also expressed higher surface expression of MHC-class I (Figures S7B, C). We also assessed surface expression of PD-L1 on NK cells (Figure S7A). The results indicated that both IL-2 and IL-2+anti-CD16mAb treatment elevated the expression of PD-L1 on NK cells (Figure S7A). Taken together, the data indicated that differentiated tumors express higher PD-L1 on their surface, and treatment of these cells with anti-PD-L1 antibody mediate ADCC in the presence of NK cells. However, poorly differentiated tumors are devoid of this surface antigen and therefore NK cells may become inactivated in the presence of anti-PD-L1 antibody treatment since activated NK cells exhibit PD-L1 on the surface.

We have previously demonstrated that NK cells secrete higher levels of IFN-γ when co-cultured with CSCs/poorly differentiated tumors in comparison to differentiated tumors (42). In our current study, we co-cultured stem-like/poorly differentiated (MP2 and OSCSCs) and their differentiated counterparts (NK-Diff-MP2 and NK-Diff-OSCSCs) with NK cells with or without anti-PD1 treatment (Figures 6, S7). Lower secretion of IFN-γ was found when NK cells were co-cultured with NK-diff-OSCSCs (Figures 6A, S8) and NK-diff-MP2 (Figure 6B) in comparison to the cultures with their stem-like counterparts. Anti-PD1 treated NK cells without tumors showed slightly increased levels of IFN-γ secretion, however, the highest effect of anti-PD1 treatment was seen when NK cells were co-cultured with OSCSCs (Figures 6A, S8) or MP2 (Figure 6B) cells in comparison to those co-cultured with their NK-differentiated counterparts.

Hu-BLT mice were surgically implanted with human MP2 tumors followed by injections of sNK cells and anti-PD1 sequentially as depicted in Figure 7A. Upon sacrifice, the tumors were harvested and weighed, and the results were compared to tumor bearing mice in the absence of treatments. sNK cells alone or sNK cells combined with anti-PD1 antibody injected mice had smaller tumors in comparison to untreated tumor implanted mice (Figures 7B, S9A). Anti-PD1 antibody alone treated mice had smaller tumors compared to untreated tumor-bearing mice but the size was larger in comparison to either sNK treated or sNK + anti-PD1 antibody treated group (Figure 7B). Tumors were then dissociated and counted and the numbers were adjusted to 1.5X10⁵ per well. Tumor growth were slightly with anti-PD1 alone treatment, and was much less with sNK cells alone or sNK cells combined with anti-PD1 treatments when compared to untreated tumor implanted mice (Figures 7C). When the same numbers of dissociated tumor cells from hu-BLT mice were cultured, significantly lower tumor cell expansion was seen in tumors from sNK cells alone or sNK cells combined with anti-PD-1 antibody treated mice as compared to those from untreated tumor-bearing mice until day 14-18, after which the tumor growth rate gradually increased and approached to those obtained from tumor alone implanted mice, and at day 27 the tumor cultures were terminated. Even though the levels of tumor growth gradually approached the levels seen with tumor alone implanted mice at the days 18-24, we could still see a lower tumor growth with sNK+anti-PD-1 antibody treated group as compared to sNK treated group, and both groups had on average lower growth when compared to those from tumor alone implanted mice (Figures 7D, S9B). Tumors dissociated from anti-PD-1 antibody group exhibited a lower rate of tumor growth when compared to the tumor alone implanted group but the levels of tumor growth were higher than those obtained from either the sNK treated or the sNK+anti-PD-1 antibody treated group (Figures 7D, S8B). In addition, when all the floating immune cells were removed by changing the media every three days from tumor cultures throughout the days of 7-27 the differentiated nature of tumor cells grown from sNK cells or sNK in combination with anti-PD-1 gradually reverted to their CSC/poorly differentiated tumors, and their growth rate gradually increased and approached to those grown from tumor alone implanted BLT mice (Figure 7D).

We observed suppression of both secretion of IFN-γ (Figures 7E, G, H, S10A, S11A) and NK cell-mediated cytotoxicity (Figures 7F, S10B, S11B) in tumor-bearing hu-BLT mice in comparison to those from healthy mice. Restoration or increased IFN-γ secretion and NK cell-mediated cytotoxicity were seen in tumor-bearing mice injected with sNK cells alone or anti-PD1 alone, and both of these functions were further increased with the combination of sNK cells and anti-PD1 antibody injection in tumor bearing mice (Figures 7E–H, S9, S10). It is important to note that the levels of IFN-γ secretion were comparable between sNK cells treated mice and anti-PD-1 treated mice; however, the NK cell mediated cytotoxicity was much higher in sNK treated tumor bearing mice than in the presence of anti-PD-1 treated tumor bearing mice (Figures 7E–H, S10, S11). The combination of both sNK and anti-PD-1 treatment significantly elevated IFN-γ secretion when compared to each treatment alone and increased cytotoxicity more than those seen in the presence of sNK treatment alone (Figures 7E–H, S10, S11). Taken together, these results indicated that sequential treatment of sNK cells with anti-PD-1 antibody is capable of increasing cytotoxic function in tumor-implanted hu-BLT mice and significantly augmented the secreted IFN-γ in immune cells from bone marrow, spleen and peripheral blood.

Oral squamous carcinoma stem cells (OSCSCs) and oral squamous cell carcinoma (OSCCs) were isolated from patients with tongue tumors at UCLA (13, 42, 55). NK cells, OSCSCs, and OSCCs were cultured in RPMI 1640 (Invitrogen by Life Technologies, CA), supplemented with 10% fetal bovine serum (FBS) (Gemini Bio-Products, CA). Recombinant IL-2 was obtained from NIH-BRB. Antibodies to CD16 were purchased from Biolegend (San Diego, CA). Antibodies used for flow cytometry – IgG2, MHC-class I, and B7H1 (PD-L1) were purchased from Biolegend (San Diego, CA). MIA PaCA-2 (MP2), PL12, and Capan human pancreatic cancer cell lines were provided by Dr. Nicholas Cacalano (UCLA, School of Medicine, CA, USA), and were cultured in DMEM supplemented with 10% FBS. Cisplatin and Paclitaxel were purchased from Ronald Reagan Pharmacy at UCLA. ELISA kits for IFN-γ were purchased from Biolegend (San Diego, CA), and multiplex analysis kit was purchased from Millipore (Billerica, MA). Propidium iodide (PI) and chromium-51 was purchased from PeproTech (Cranbury, NJ, USA) and Perken Elmer (Waltham, MA, USA), respectively. Chromium Single cell 3’ Reagent kit v3, Cat#1000075 was purchased from10X Genomics (Pleasanton, CA, USA).

Written informed consents, approved by UCLA Institutional Review Board (IRB), were obtained from healthy individuals, and all procedures were approved by the UCLA-IRB. Peripheral blood was separated using ficoll-hypaque centrifugation, after which the white layer, containing peripheral blood mononuclear cells (PBMCs) was harvested. NK cells and monocytes were negatively selected from PBMCs using the EasySep^® Human NK cell enrichment and EasySep^® Human monocytes enrichments kits, respectively, purchased from Stem Cell Technologies (Vancouver, BC, Canada). Purified NK cells and monocytes were stained with anti-CD16 and anti-CD14, respectively, to measure purity using flow cytometric analysis. Samples showing greater than 95% purity were used for the study.

Purified NK cells were activated with rh-IL-2 (1000 U/ml) and anti-CD16 mAb (3 µg/ml) for 18-20 hours before the supernatant was harvested, and was used in differentiation of OSCSCs, and MP2 cells. The supernatant volume was determined based on IFN-γ required, and was accessed by ELISA specific to IFN-γ. Differentiation of OSCSCs and MP2 cells were conducted with an average total amount of 2000 pg and 5000 pg, respectively, over the course of 5 days. On day 0, 1 × 10⁶ tumor cells were cultured, on day 1 unattached tumor cells were removed and attached tumor cells were treated with NK cell supernatants on days 1, 2, 3 and 4. On day 5, tumor cells were rinsed with 1 X PBS, detached and used for experiments.

To generate osteoclasts (OCs), monocytes were cultured in alpha-MEM media supplemented with M-CSF (25 ng/mL) and RANKL (25 ng/mL) for 21 days, media was replenished every three days. Human purified NK cells were activated with rh-IL-2 (1000 U/ml) and anti-CD16 mAb (3 µg/ml) for 18-20 hours before they were co-cultured with OCs and sAJ2 (OCs : NK:sAJ2; 1:2:4) in RPMI 1640 medium containing 10% FBS. Probiotic bacteria, AJ2 is a combination of seven to eight different strains of gram-positive probiotic bacteria (Streptococcus thermophiles, Bifidobacterium longum, Bifidobacterium breve, Bifidobacterium infantis, Lactobacillus acidophilus, Lactobacillus plantarum, Lactobacillus casei, and Lactobacillus bulgaricus) selected for their superior ability to induce optimal secretion of both pro-inflammatory and anti-inflammatory cytokines in NK cells (32). The medium was refreshed every three days with RPMI containing rh-IL-2 (1500 U/ml).

Cells were washed twice with an ice-cold PBS and lysed in NP-40 lysis buffer supplemented with protease inhibitor cocktail (APExBIO). Lysates were centrifuged at 16000g at 4°C for 20 minutes to obtain post-nuclear cell fraction. Protein concentration was determined with Pierce BCA protein assay kit (ThermoFisher Scientific). Non-reducing SDS-PAGE was performed, and separated proteins were transferred to nitrocellulose membrane. Membranes were blocked for 1 hour in 5% non-fat dry milk in PBS. Membranes were incubated in primary antibodies overnight at 4°C and HRP conjugated secondary antibodies for 1h at room temperature. Bands were visualized with Clarity Max Western ECL substrate (BioRad). Images were acquired with ChemiDoc ML imaging System (Biorad). The following primary antibodies were used: mouse anti-granzyme B (sc-8022, Santa Cruz Biotechnology), mouse anti-cathepsin C (sc-74590, Santa Cruz Biotechnology), mouse anti-perforin-1 (sc-136994, Santa Cruz Biotechnology). We used anti-mouse HRP conjugated secondary antibodies (405306 BioLegend). Stain free technology (BioRad) was used for loading control.

Single-cell RNA sequencing was performed using a 10X Chromium machine. Single-cell cDNA libraries were prepared using the 10X Chromium Single cell 3’ Reagent kit v3 and sequenced via Illumina Novaseq 6000 (Illumina) to a depth of around 30 thousand reads per cell. Raw data from each sample were demultiplexed and aligned to a custom reference genome (GRCh38), and UMI counts were quantified using 10X Genomics CellRanger software (v3.0.0) with default parameters. Single-cell clustering and cell-cycle scoring are performed using the Seurat package (v3.0). Single-cell regulatory network inference and clustering (SCENIC) is done by using the SCENIC R package (1.2.0) with the hg38 database (https://resources.aertslab.org/cistarget/).

Animal research was performed under the written approval of the UCLA Animal Research Committee (ARC) in accordance with all federal, state, and local guidelines. Combined immunodeficient NOD.CB17-Prkdcscid/J and NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG lacking T, B, and NK cells) were purchased from Jackson Laboratory. Humanized-BLT (hu-BLT; human bone marrow/liver/thymus) mice were prepared on NSG background as previously described (38, 56). To establish orthotopic tumors, mice were first anesthetized with isoflurane in combination with oxygen, and 1 x 10⁶ human OSCSCs and MP2 tumor cells suspended in 10 μl HC Matrigel were then injected directly into the floor of their mouths and pancreas, respectively. One week after tumor implantation mice received 1 x 10⁶ OC-expanded supercharged NK cells via tail vein injection. One week after NK injections, mice received CDDP (50 µg/mice) or anti-PD1 (50 µg/mice) via tail vein injection. One week later, mice were euthanized, and tumors, bone marrow, spleen, and peripheral blood were harvested.

The oral and pancreatic tumors harvested from hu-BLT mice were immediately cut into 1 mm³ pieces and placed into a digestion buffer containing 1 mg/ml collagenase II (oral tumor) or collagenase IV (pancreatic tumor), 10 U/ml DNAse I, and 1% bovine serum albumin (BSA) in DMEM media, and incubated for 20 minutes at 37°C oven on a 150 rpm shaker. After digestion, the sample were filtered through 40 mm cell strainer and centrifuged at 1500 rpm for 10 minutes at 4°C. The pellet was re-suspended in DMEM media and cells were counted. To obtain single-cell suspensions from BM, femurs were cut at both ends and flushed by using RPMI 1640 media; afterwards, BM cells were filtered through a 40 µm cell strainer. To obtain single-cell suspensions from spleen, the spleens were minced, and the samples were filtered through a 40 µm cell strainer and centrifuged at 1500 rpm for 5 minutes at 4°C. The pellets were re-suspended in ACK buffer for 2-5 mins to remove the red blood cells followed by re-suspension in RPMI media and centrifugation at 1500 rpm for 5 minutes at 4°C. PBMCs were isolated from peripheral blood using Ficoll-Hypaque centrifugation of heparinized blood specimens. The buffy coats containing PBMCs were harvested, washed, and re-suspended in RPMI 1640 medium. Cells obtained from each tissue sample were treated with IL-2 (1000 U/ml) and cultured in RPMI 1640 medium containing 10% FBS for 7 days.

Staining was performed by labeling the cells with antibodies as described previously (43, 57, 58). The percentage of dead cells was determined by propidium iodine (PI) (100 μg/ml) staining using flow cytometric analysis. Flow cytometric analysis was performed using Attune NxT flow cytometer (Thermo Fisher Scientific, Waltham, MA) and FlowJo v10.4 (BD, Oregon, USA) were used for analysis. For selected experiments Beckman Coulter Epics XL cytometer (Brea, CA) was also used, and the results were analyzed in the FlowJo vX software (Ashland, OR).

Single ELISAs were performed as previously described (43). To analyze and obtain the cytokine and chemokine concentration, a standard curve was generated by either two- or three-fold dilutions of recombinant cytokines provided by the manufacturer. For multiple cytokine array, the levels of cytokines were determined by multiplex assay, which was conducted as described in the manufacturer’s protocol for each specified kit. Analysis was performed using a Luminex multiplex instrument (MAGPIX, Millipore, Billerica, MA), and data was analyzed using the proprietary software (xPONENT 4.2, Millipore, Billerica, MA).

The ⁵¹Cr release cytotoxicity assay was performed as previously described (59). Briefly, different ratios of effectors and ⁵¹Cr–labeled target cells were incubated for four hours. After which, the supernatants were harvested from each sample, and the released radioactivity was counted using the gamma counter. The percentage specific cytotoxicity was calculated as follows:

LU 30/10⁶ was calculated by using the inverse of the number of effector cells needed to lyse 30% of target cells ×100.

Prism-9 software was used for statistical analysis. An unpaired or paired, two-tailed Student’s t-test was performed for experiments with two groups. One-way ANOVA with a Bonferroni post-test was used to compare different groups for experiments with more than two groups. Duplicate or triplicate samples were used for assessment. The following symbols represent the levels of statistical significance within each analysis: ****(p value<0.0001), ***(p value 0.0001-0.001), **(p value 0.001-0.01), *(p value 0.01-0.05).