VACCIMEL is a mixture of four 70 Gy-irradiated allogeneic cutaneous melanoma cell lines, which are injected intradermally in patients, not later than 3 h after thawing. To investigate the fate of the thawed cells over time, we observed their morphology up to 168 h in tissue culture. The results are shown in Figure 1. Immediately after thawing, the cells appeared intact, with dense cytoplasm. Most cells were refringent, suggesting an intact cell membrane, whereas some had lost refringence, indicating altered permeability. Vacuoles were observed beneath the cell membrane. This pattern was altered after 24 h since most remaining cells were non-refringent, vacuoles appeared in the cytoplasm, and some cell membranes were ruffled, indicating a process of apoptosis; only 13.7% of the cells remained intact. Many cells appeared loaded with vesicles, some of which were described as Radiation Induced Lipid Bodies, and their lipid composition has been described (15). After 48 h, the vacuoles increased in the cytoplasm of the remaining cells, and some cell fusions started to be noticed; the number of non-refringent cells increased. This process intensified after 72 h and 96 h, and fused cells with bizarre forms were more abundant. After 168 h, only a few fused cellular masses were observed. Therefore, even as soon as 24 h after thawing, a process of cellular apoptosis and necrosis appeared, which increased with time. The total cell number and integrity decreased significantly during the first 48 h, with only 39.4% of the starting cell number persisting, of which 8.1% retained integrity as analyzed by generalized linear mixed models (GLMMs) with gamma errors (Table 1A).

The persistence of some relevant markers was determined by immunohistochemistry: i) Ki67 to measure residual DNA synthesis and nuclear integrity; ii) HLA-I to determine the fate of this allogeneic marker; iii) PMEL/gp100 and MelanA/MART-1 as MDA; and iv) MAGEA1 antigen, representative of cancer testis antigens (Table 1B, Figure 2). The expression of these markers did not change significantly up to 48 h from thawing as analyzed by generalized linear models (GLMs), with binomial errors. Examples of the immunohistochemical appearance of these markers are shown in Figure 2. It was noteworthy that 48 h after thawing, Ki67 staining was not restricted to the nuclei, since it was also observed in the cytoplasm and in the vicinity of the cell membrane, suggesting a loss of nuclear membrane integrity (Figure 2). Therefore, a passage of DNA into the cytoplasm and its eventual release into the extracellular milieu is suggested. Thus, it was found that although the number of irradiated cells diminished over time, their antigen expression remained relatively constant, suggesting that the cellular antigenic offer to the immune system persisted for several days after inoculation. Also, these results imply that the antigenic material that was released into the extracellular milieu increased at least during the first 48 h after inoculation.

During vaccination, each VACCIMEL dose (1.6x10⁷ irradiated allogeneic melanoma cells) is injected i.d. (intradermal) along with 120 µg protein BCG (~2–10 x10⁵ colony forming units) and 400 µg rhGM-CSF (divided in four 100 µg/day i.d. injections at the vaccination site). Once VACCIMEL is injected into the dermis in conjunction with BCG and GM-CSF, monocytes are among the first innate immune cells arriving at the injection site (16). We investigated the in vitro uptake of PKH67–labelled VACCIMEL by monocytes obtained from healthy donors (HD), either in the presence or absence of BCG at two different MOI, and 100 ng/ml GM-CSF during co-culture. It is difficult to calculate the actual amount of BCG and GM-CSF to which monocytes are exposed in vivo after the first and successive vaccinations, since it would be a balance between the amount of injected BCG and BCG clearance, which has been demonstrated to be also quite efficiently carried out by neutrophils (17) and the number of recruited monocytes at the inoculation site. For the in vitro assays we have chosen a 15µg/ml protein BCG dose (MOI 0.04 BCG: monocyte), taking as a reference a similar BCG concentration that has been previously shown to be active in trained immunity in vitro assays (18); we also tested a 10 x dose (MOI 0.4 BCG: monocyte).

Viable VACCIMEL cells were labeled with PKH67 and monocytes were purified as described under Materials and Methods. After irradiation, freezing, and thawing, their uptake by monocytes was evaluated by flow cytometry and transmission electron microscopy. For every co-culture experiment, the single-cell populations were analyzed from the monocytes gate to exclude cell clumps (doublets), as observed in the flow cytometry strategy (Supplementary Figure 1). This analysis revealed that after coincubation, progressive amounts of PKH67-labelled VACCIMEL material were incorporated by CD14⁺ cells (Figure 3A). When the assay was performed using purified monocytes from different HD (n=3) (Figure 3B), significant increases of the percentages of double labelled CD14⁺PKH67⁺ cells were found from 6 to 24, and 48 h in the presence and in the absence of BCG (MOI 0.4) plus GM-CSF (p<0.05, GLMMs with Gaussian errors). At 6 h, PKH67-labelled VACCIMEL phagocytosis was significantly higher in the absence of BCG (MOI 0.4) plus GM-CSF (p= 0.0022) than in its presence, but this difference was lost at longer times. After 48 h of co-culture, phagocytosis reached a mean of 74.6% in the absence of BCG and 72.1% in the presence of BCG (MOI 0.4) plus GM-CSF, respectively. At each time point, a control culture performed at 4°C significantly inhibited PKH67-labelled VACCIMEL phagocytosis by monocytes at all the times assayed (p<0.0001). Similar results were obtained when BCG was assayed at MOI 0.04 plus GM-CSF (not shown). Thus, monocytes were able to phagocytose VACCIMEL equally, either in the presence or absence of BCG plus GM-CSF.

To investigate how long the phagocytosed material took to be digested, we asked if PMEL/gp100, a highly immunogenic MDA expressed in ~40% of VACCIMEL cells, could still be detected by immunohistochemistry in monocytes which would mean that the immunogenic peptide had not yet been digested. After 48 h co-culture, monocytes were purified by CD14⁺ affinity columns to eliminate remaining VACCIMEL cells. Purified monocytes were fixed, paraffin-embedded, and processed for immunohistochemistry for CD14 and PMEL/gp100. As observed in Figure 4, the PMEL/gp100 epitope could be detected inside purified monocytes, cultured both in the presence and absence of BCG plus GM-CSF. This result implies that antigenic material from VACCIMEL incorporated into monocytes is incompletely digested after 48 h, a finding that is supported by cross-presentation analysis (see below). It is interesting to observe that CD14, which in baseline monocytes is restricted to the cell membrane, after 48 h phagocytosis may be clearly observed in the cytoplasm, accompanying VACCIMEL phagocytosis (Figure 4).

To our knowledge, for the first time, the combined phagocytosis of tumor cells and BCG by monocytes was analyzed by transmission electron microscopy. VACCIMEL cells were easily recognized by their smooth cell membranes, pale and sometimes disintegrated cytoplasm. In some cells, melanosomes were detected. Scarce mitochondria were observed. Monocytes may be recognized by their horseshoe-shaped nuclei with heterochromatin attached to the nuclear membrane. In Figures 5A, after 6 h of co-incubation, a monocyte presenting phagolysosomes containing one to several BCG is observed. In Figures 5B, after 24 h co-culture, a tumor cell and a monocyte may be observed in contact; the coexistence within phagolysosomes of BCG and amorphous material was observed. In Figure 5C, D two types of contacts may be observed between monocytes and tumor cells: pseudopodia (Figure 5C), which capture particulate material, and membrane bulges in monocytes making contact with tumor cells, allowing passage of particulate material between both cells (Figure 5D). In Figure 5E a mature monocyte after 96 h in culture is observed, with many lysosomes in the cytoplasm. In Figures 5F, a monocyte pseudopod is observed making contact with a tumor cell and through which small particles about 100 nm in diameter may be seen flowing between both cells. BCG uptake by monocytes was quantified from the transmission electron microscopy images and showed that the percentage of monocytes that ingested BCG bacilli increased significantly with the incubation time from 6 to 48 h reaching around 60% (Figure 5G). At 48 h, a significantly higher proportion of monocytes that incorporated >5 bacilli per cell was found after phagocytosis of BCG at MOI = 0.4 as compared to MOI = 0.04 (p<0.0001) as analyzed by GLMMs assuming Gaussian errors.

We investigated if VACCIMEL uptake by monocytes altered their immunophenotype based on the expression of CD14 and CD16, and if such phenotype was modified by the addition of BCG and GM-CSF. In Figure 6, flow cytometry plots of one representative experiment are shown; the analysis of the classical monocytes (CD14⁺⁺CD16⁻), intermediate monocytes (CD14⁺⁺CD16⁺), and non-classical monocytes (CD14⁺CD16⁺⁺) from four independent HD at different times and in the presence of BCG at two different MOI (0.4 and 0.04) and GM.CSF are presented in Figure 7. At baseline, and by 6 h of being placed in culture, monocytes displayed the classical phenotype, either with or without VACCIMEL. After 24 h in culture, most monocytes either alone or plus VACCIMEL shifted to the intermediate phenotype (around 70%). This transition was significantly blocked in the presence of BCG (MOI 0.4) and GM-CSF and less inhibited at a BCG MOI of 0.04 and GM-CSF. After 48 h in culture, a slight increase in the intermediate phenotype was observed at MOI 0.4 and an almost complete transition to intermediate phenotype was observed at a BCG MOI 0.04.

Another experiment was carried out with monocytes obtained from a single HD incubated with BCG at two different MOI (0.4 and 0.04) plus GM-CSF for 0, 24 and 48 h. Similar results as those obtained when monocytes incubated with VACCIMEL, BCG and GM-CSF were obtained (Supplementary Figure 2).

These experiments suggest that BCG at a 0.4 MOI strongly blocks the transition of classical to intermediate monocyte phenotypes almost completely after 24 h co-culture and that this effect is less pronounced after 48 h. At a lesser BCG MOI (0.04) the same effect, albeit less pronounced, is observed.

Antigen presenting cells (APC) may cross-present antigens derived from exogenous antigens to CD8⁺ T cells, loading the cognate peptides in HLA class I molecules. In order to examine whether monocytes share such characteristic, and to which degree, we measured first several markers related to antigen presentation and co-stimulation, such as CD80, CD86, HLA-I, HLA-A*0201, HLA-DR, and CD11c. As observed in Figure 8, even in the absence of VACCIMEL or BCG, cultured monocytes at 0 h already expressed the above-quoted markers, although they generally increased after co-culture. Both the percentage of positive monocytes and the mean fluorescence intensity (MFI) were analyzed. Comparisons between the different timepoints in each experimental condition and between the different experimental conditions at each timepoint, are shown in Supplementary Table 1, for each marker (GLMs assuming Gaussian errors; Turkey´s multiple comparisons post test). The most remarkable findings were that: i) For CD86, 100% of monocytes expressed the marker, and it significantly increased with culture time; ii) CD80 had a low expression; a significant increase was observed between 0 and 96 h culture, except for monocytes incubated with VACCIMEL plus BCG (MOI 0.4) plus GM-CSF; iii) HLA-I: every cell expressed the marker and at 96 h, incubation with VACCIMEL plus BCG plus GM-CSF significantly diminished MFI at MOI 0.04 (p=0.031) and MOI 0.4 (p=0.025), respectively; iv) HLA-A*0201 similar results that for HLA-I were observed; v) HLA-DR: only the increase in monocytes expression after 96 h of incubation was statistically significant (p = 0.027); vi) CD11c: MFI increased significantly with culture time (0–96 h) in monocytes alone (p=0.0002), monocytes plus VACCIMEL (p<0.0001) and monocytes plus VACCIMEL and BCG at low MOI and GM-CSF (p<0.0001) but did not reach statistical significance at the high MOI.

To quantify cross-presentation, a readout of IFN-γ release by the CD8⁺ T cell clone G154, HLA-A*0201-restricted, PMEL/gp100-specific was used. VACCIMEL is a mixture of four allogeneic, irradiated cell lines, two of them without MDA expression. The irradiated MEL-XY3 cell line (MEL-XY3i), one of the cell lines that compose VACCIMEL (VAC) which strongly expresses PMEL/gp100, was used as a positive control. It was investigated if, after 48 h co-culture with VAC or XY3i, monocytes that phagocytosed tumor cells, correctly processed and presented the PMEL/gp100 cognate peptide via HLA-A*0201 to G154 T cells. To avoid possible antigen presentation by remaining tumor cells in the co-culture, monocytes were re-purified after phagocytosis (CD14⁺ purity ranged between 92-97%). Afterwards, purified monocytes were exposed to the G154 clone for 24 h stimulation, and IFN-γ released to the supernatant was quantified by ELISA. In Figure 9A (left bars) it may be observed that monocytes loaded with XY3i during 48 h, but not with VAC, were able to cross-present PMEL/gp100 peptide to G154 T cells. Assuming that 48 h were not sufficient to process the phagocytosed antigenic material, purified monocytes were cultured for an additional 48 h after phagocytosis and exposed for 24 h to the G154 clone (Figures 9A, right bars). IFN-γ release was significantly increased ~20-fold for MEL-XY3i (p<0.001), and ~5-fold for VAC, probably reflecting that monocytes could achieve at longer times proper antigen processing to degrade PMEL/gp100 to the corresponding peptides, load them onto HLA-A*0201 molecules, and stimulate G154 cells to produce IFN-γ. At both times, VACCIMEL cross-presented PMEL/gp100 directly to G154 cells less efficiently than MEL-XY3i, perhaps due to a smaller relative amount of PMEL/gp100 antigen. Positive control (viable XY3) and negative control (unstimulated G154 clone) were satisfactory.

These results suggest that monocytes can phagocytose PMEL/gp100-containing structures, adequately process them, and cross-present the PMEL/gp100 peptide to T cells.

In the presence of BCG and GM-CSF, PMEL/gp100 cross–presentation to the T cell clone was significantly inhibited (~70-fold; p< 0.01) at the higher BCG MOI, but much less so at the 0.04 MOI, both for VACCIMEL and MEL-XY3i.

Since IL-10 is a potent immunosuppressive cytokine, and based in our results on cross-presentation assays, we evaluated IL-10 production in the monocytes/VACCIMEL co-cultures exposed to G154 T cells (Figure 9B). BCG induced a high production of IL-10 by monocytes alone, either at high and low MOI, demonstrating that BCG phagocytosis at both MOI was sufficient to trigger IL-10 production. Interestingly enough, after phagocytosis of VACCIMEL or XY3i cells, IL-10 secretion dampened several folds at the low and high BCG MOI (p<0.05), respectively.

Our results suggest that: i) the VACCIMEL phagocytic process dampens monocyte IL-10 production induced by BCG plus GM-CSF, more so at the lower BCG MOI; ii) IFN-γ production by G154 lymphocytes increases under conditions of lower IL-10 production by monocytes.

VACCIMEL consists of lethally γ-irradiated cells (apoptotic/necrotic) derived from four cutaneous melanoma cell lines established in-house from human metastatic melanoma tumors. The cell lines MEL-XY1, MEL-XY2, MEL-XY3, and MEL-XX4 were grown in a GMP core facility at the Centro de Investigaciones Oncológicas-FUCA and were cultured as previously described (6). Vaccine preparation was formerly described (7). For all the in vitro experiments, lyophilized AJV BCG: Danish 1331 strain was used, which contains 0.75 mg of BCG protein and 2–8 x 10⁶ bacilli. This strain has minimal differences with the Pasteur strain 1173, and both are widely used to vaccinate against tuberculosis (42). Lyophilized recombinant human granulocyte-macrophage colony-stimulating factor (rhGM-CSF, Laboratorio Pablo Cassará, Argentina) was used.

After irradiation and freezing, VACCIMEL cells were thawed and plated in complete melanoma medium (Dulbecco’s Modified Eagle Medium: nutrient Mixture F12 (1:1) (Gibco) supplemented with 2 mM glutamine, 20 nM sodium selenite, 100 μM ascorbic acid, 0.3 mg/ml galactose, 0.15 mg/ml sodium pyruvate and 5 μg/ml insulin, 100 IU/ml penicillin, 10 μg/ml streptomycin plus 10% fetal bovine serum (Natocor, Córdoba, Argentina) for 0, 24, 48, 72, 96 and 168 h and examined under phase contrast microscopy (Olympus CKX41 microscope). Images were digitized (MC190 HD camera, Leica) and analyzed with ImageJ software (NIH). For marker expression determination, 18 x106 irradiated cells were thawed and allowed to recover from the freezing-thawing process in melanoma culture medium for 3 h; this was considered time 0. After that, they were split into three different tissue culture plates, from which at 0, 24, and 48 h, samples were obtained for counting and cell integrity determination by Trypan blue exclusion. The rest of the cells were centrifuged, washed once with PBS, fixed in 10% formalin, and processed for immunohistochemistry.

Four micrometers (µm) sections from formalin-fixed, paraffin-embedded VACCIMEL pellets were stained using the following primary mouse anti-human mAbs: Ki-67 (clone 30–9 rabbit monoclonal primary antibody, undiluted, Ventana); PMEL/gp100 (clone HMB45, 1:500, Cell Marque); HLA-I A, B, C (clone EMR8-5, 1:350, Abcam); CD14 (clone EPR3653, 1:300, Abcam); CD11c (clone EP 1347Y, 1:200, Abcam); MLANA/MART-1 (clone A103, undiluted, Ventana) and MAGEA1 (clone SP188, 1:50, Abcam). The avidin–biotin–peroxidase (ABC) system (Vectastain, Vector Labs) was afterward used, and revealed with 3, 3’-diaminobenzidine. Sections were examined by optical microscopy (Olympus BX40 microscope, DP2-BSW software, Olympus), and digitalized pictures were analyzed with ImageJ (43). The percentage expression of each marker was determined by counting 10 fields at 1000× magnification.

Peripheral blood samples (buffy-coats) were obtained from healthy donors (HD) of the Hemotherapy Service of the Instituto Alexander Fleming according to its guidelines and in accordance with the recommendations of the Ethics Committee of the Instituto Alexander Fleming. Peripheral blood mononuclear cells (PBMC) were purified using a Ficoll density gradient (GE Healthcare, UK), and monocytes were purified using an anti-CD14⁺ cell kit (Miltenyi, Germany) following the manufacturer’s specifications. Purified monocytes were cultured in 60 mm Petri dishes without cell adherence treatment, unless specified.

For these experiments, VACCIMEL melanoma cell lines were stained before γ-irradiation with PKH67 (Sigma- USA) following the manufacturer´s procedure. To evaluate the uptake/phagocytosis of VACCIMEL, 1.2 x 10⁷ purified monocytes were co-cultured with 4 x 10⁶ PKH67-VACCIMEL cells in 5 mL RPMI medium plus 10% FCS, with or without 0.75 mg BCG (MOI 0.4) or 0.075 mg (MOI 0.04) and 100 ng/mL GM-CSF at 37° C, for 0, 6, 24, 48, and 96 h. Also, monocytes alone were equally cultured as controls. At each time point, a control culture performed at 4°C was performed to inhibit the phagocytic process. Afterwards, cells from each culture were stained with FVS510 dye to select a viable cell population and with anti-CD14 and anti-CD16, and analyzed by flow cytometry as described above. To calculate the proportion of blood monocytes that have phagocytosed labeled material from VACCIMEL (Supplementary Figure 1), cells were selected as the FVS510 negative population (live cells), then the mononuclear cells were gated and the single cell fraction was selected from the FSC-H vas FSC-A plot, to exclude doublets or steaky cells. These cells were then plotted for CD14-APC and PKH67, and the percentage of PKH67⁺ cells within the CD14⁺ monocytes population was calculated. Remaining cells from the co-cultures were further processed for electron microscopy.

Samples were centrifuged at 1500 rpm, and the pellets were rinsed with wash buffer (0.1 M phosphate buffer, pH 7.4) and then fixed in 1 mL 2.5% glutaraldehyde in 0.1 M phosphate buffer, pH 7.4. The samples were fixed for a minimum of 4 h at 4 °C. Cell pellets were washed twice and post-fixed with 0.5 mL of 1% Osmium Tetroxide (OsO₄) in 0.1 M phosphate buffer, pH 7.4, at 4 °C for 1 h. After washing and dehydration, samples were embedded in Durcupan. Sections of 0.5 µm were made from the embedded blocks and stained with toluidine blue. A region of the sample was then selected, the block was trimmed with a blade, and ultrathin sections (~ 90 nm) were obtained and mounted onto 300-mesh copper grids. Ultrathin slices were contrasted with Reynolds’ solution, washed, and dried. Grids were analyzed under a Transmission Electron Microscope EM109T TEM (80 kV, Zeiss), and images were acquired with a Gatan ES1000W digital camera. Monocytes that incorporated BCG after 6, 24, and 48 h incubation were quantified from the transmission electron microscopy pictures counting at least 100 cells for each experimental condition, in two independent experiments. Cells were at 3000x magnification. The percentage of monocytes containing BCG (monocytes BCG⁺), either total (Total BCG⁺) or separated into two groups: 1–5 BCG⁺ per monocyte, and >5 BCG⁺ per monocyte were counted.

To characterize classical, intermediate, and non-classical monocytes, APC anti-CD14 (clone M5E2) and PE Cy7 anti-CD16 (clone 3G8) mAbs were used. Also, CD14⁺ monocytes were stained with PerCp Cy5.5 anti-CD11c (clone B-Ly6), PeCy7 anti-HLA ABC (Clone G46-2.6); BV421 anti-HLA DR (Clone Tu39); APC-H7 anti-CD80 (Clone L307.4); PE anti-CD86 (Clone 2331, FUN-1) and PE anti-HLA-A*0201 (Clone BB7.2). All mAbs were from BD Biosciences. Data acquisition was performed using a FACSCanto II cytometer (BD Biosciences, USA); data analysis was performed with FlowJo software 10.0.7 (Tree Star Inc., USA).

G154, a CD8⁺ T cell clone specific for PMEL/gp100 antigen (HLA-A*0201 restricted; KTWGQYWQV) (44), was expanded as previously described (45). CD14⁺ monocytes were purified to >95% from HLA-A*0201 positive HD`s PBMC using anti-CD14 microbeads (Miltenyi Biotec, Germany). Purified monocytes (1.2 x 10⁷ cells) were incubated with VACCIMEL (4 x10⁶ cells), or VACCIMEL (4 x 10⁶ cells) plus BCG (0.4 or 0.04 MOI) plus GM-CSF (100 ng/mL) or alone (unstimulated) in 5 mL RPMI medium plus 10% FBS (culture medium) for 48 h on low adhesion plates to allow antigen uptake. To eliminate VACCIMEL remaining cells and apoptotic bodies after incubation, monocytes were re-purified with anti-CD14 beads as described above and plated (3 x 10⁵ cells) in 0.5 mL culture medium. Also, unstimulated monocytes (3 x 10⁵ cells) incubated only with BCG (0.4 or 0.04 MOI) + GM-CSF were plated. G154 CTL clone (10⁵ cells/0.1 mL) was added and incubated overnight at 37 °C to stimulate IFN-γ release (cross-presentation after 48 h). Also, the same cultures were incubated for additional 48 h (cross-presentation after 96 h) as described above, and then exposed to the G154 clone overnight at 37° C, to stimulate IFN-γ release. In parallel, monocytes were equally incubated with irradiated MEL-XY3 cells (XY3i) expressing PMEL/gp100 antigen in 100% of the cells, instead of VACCIMEL, to assess maximal monocytes cross-presentation under excess PMEL/gp100 antigen. Controls for these experiments included G154 clone incubated with viable MEL-XY3 (10⁵ cells, positive control) or VACCIMEL (10⁵ cells), or XY3i (10⁵ cells), or G154 clone alone (negative control) in the absence of monocytes.

IFN-γ and IL-10 released to the supernatants were determined in triplicate by ELISA (OptEIA IFN-γ; OptEIA IL-10, Pharmingen BD Biosciences, San Diego, CA) according to the manufacturer’s recommendations. Calibration curves were performed for each experiment, and the sample concentration was calculated by log-log regression analysis using GraphPad Prism v9.

Statistical analyses were performed using GraphPad Prism v9 and R software. Total VACCIMEL cell counts and cell integrity at different times after thawing were analyzed using generalized linear mixed models (GLMMs) assuming a gamma error distribution. VACCIMEL antigen expression at different times after thawing, as determined by immunohistochemistry, was analyzed using generalized linear models (GLMs) with a binomial error distribution.

Monocyte phagocytosis of PKH67-labeled VACCIMEL cells was analyzed by GLMMs assuming Gaussian errors, comparing experimental conditions across time points. Phagocytosis of BCG in the presence of GM-CSF was analyzed by comparing the two BCG multiplicities of infection (MOI 0.04 and 0.4) and, within each MOI, across time points. In these analyses, the proportions of monocytes that ingested BCG (BCG⁺), as well as those containing 1–5 bacilli or > 5 bacilli, were modeled using GLMs with a binomial error distribution.

Flow cytometry data for monocyte phenotyping and cross-presentation experiments were analyzed using GLMs assuming Gaussian errors, including the main effects of treatment and time and their interaction. Post hoc pairwise comparisons were conducted using estimated marginal means with Tukey or Sidak adjustments for multiple comparisons, as appropriate.

In all analyses, results were considered statistically significant when p < 0.05.