Amphotericin B stimulates γδ T and NK cells,and enhances protection from Salmonella infection

Preparation of PBMCs

Whole blood was drawn from newborn Holstein bull calves (<12 h old), older Holstein bull calves (1–3 mo old) and adult (>2 yr old) Holstein cows into sodium heparin tubes (Becton Dickinson, Franklin Lakes, NJ, USA). Histopaque 1077 (Sigma-Aldrich, St. Louis, MO, USA) was used to separate the leukocytes from whole blood, as per the manufacturer's instructions, as previously described. ³² Cell preparations were maintained in X-VIVO 15 medium (Cambrex; East Rutherford, NJ, USA) at 1 × 10⁶ cells/ml at 37℃, 10% CO₂ in the presence of known or test agonists, or medium only as controls for 24 h, unless otherwise indicated. Samples from a minimum of three calves were obtained and analyzed per experiment. Cells from human subjects were derived from peripheral blood and similarly purified using Histopaque 1077. Mouse spleens were collected from TCR-α^–/– mice, under aseptic conditions, and spleens were homogenized. This strain has increased numbers of γδ T cells, facilitating their analyses. ³³ Red blood cells were water lysed before culturing cells in X-VIVO media. All experiments were performed in accordance with and approved by the institutional animal care and use committee of Montana State University, and the use of human subjects was approved by the institutional review board of Montana State University.

Preparation of AmB formulations

Commercially available sources of AmB were screened for their ability to stimulate γδ T cells. A water-soluble deoxycholate source for use in cell culture (Cat# A2942; Sigma-Aldrich) performed better than other commercial sources and was thus chosen for these studies.

Endotoxin screening

To analyze the concentration of endotoxin that may have been present in the preparations of AmB, a kinetic turbidimetric Limulus amebocyte lysate (LAL) (Associates of Cape Cod; East Falmouth, MA, USA) assay was performed as per the manufacturer’s instructions. AmB was diluted to concentrations ranging from 56.25 to 7.03 µg/ml in LAL reagent water and analyzed for endotoxin using LAL reagent re-suspended in Glucashield buffer. A ThermoMax Microplate Reader (Molecular Devices, Sunnyvale, CA, USA) was utilized to measure LAL agglutination, and each of the samples was analyzed in triplicate. All samples were found to be negative for detectable endotoxin.

FACS-based analyses

A BD FACSCalibur equipped with a high-throughput screening (HTS) loader (Becton Dickinson) or a BT FACSCanto (Becton Dickinson) was used to analyze cells by flow cytometry. Bovine cells were stained with mAbs against IL-2Rα (LCTB2A) (VMRD) and/or γδTCR (clone GD3.8). The mAbs were directly labeled (FITC, PE, PE–Cy5, PE–Cy7, APC or APC–Cy7) or indirectly labeled with goat-antimouse PE, FITC or APC (Jackson ImmunoResearch, West Grove, PA, USA). Human cells were stained for γδ TCR (11F2; BD Biosciences) and anti-human CD69 (BioLegend, San Diego, CA, USA). Mouse γδ T cells were stained with Ab specific for γδ TCR (clone GL3; BD Biosciences). In order to determine the level of background staining in the FACS analyses, second-stage reagents were used alone.

IL-2Rα expression on gated γδ T cells was determined by multicolor FACS after cells were cultured in the presence or absence of varying concentrations of AmB for 48 h. In order to determine cell proliferation after incubation with AmB, PBMCs or mouse splenocytes were labeled with 2.5 μM carboxyfluorescein diacetate succinimidyl ester (CFSE) for 5 min in Hank’s Balanced Salt Solution (HBSS; Mediatech, Manassas, VA, USA), washed and then cultured at 37℃, 10% CO₂, in X-VIVO medium for 5 d in the presence or absence of AmB. AmB-stimulated bovine cells proliferated in the absence of additional IL-2, while IL-15 or IL-2 (each at 1 ng/ml) was added to human or mouse CFSE cultures, respectively. Quantification of cell division was done with multicolor FACS analysis to determine the percentage of γδ T cells that had divided at least once, based on a 50% or greater reduction in the intensity of CFSE. Con A (1 µg/ml; Sigma-Aldrich) was used as a positive control. Appropriate negative controls were also employed, including buffers and assay medium only.

IL-18 activation assays

To test for priming effects by AmB, bovine PBMCs were isolated and incubated in X-VIVO 15 medium at 37℃ and 10% CO₂ in the presence of AmB (125 ng/ml) or medium only for approximately 24 h. Cells were then washed with X-VIVO and re-suspended in X-VIVO 15 medium in the presence or absence of 1 ng/ml (ELISA) or 10 ng/ml (flow cytometry) recombinant human (rhu) IL-18 (R&D Systems, Minneapolis, MN, USA). To measure secreted IFN-γ, cells were subsequently incubated approximately 20 h and the supernatant fluids were collected for IFN-γ quantification by a bovine IFN-γ ELISA kit (MABTECH, Cincinnati, OH, USA).

To determine the cellular source for IFN-γ production by bovine cells by flow cytometry, intracellular staining was performed. Following cell stimulation as described above, brefeldin A (eBioscience, San Diego, CA, USA) was added to the culture for 4–6 h. Cells were then stained with a directly labeled mAb against γδ TCR (GD3.8) or Alexa Fluor 488-conjugated CD335 (AKS1, specific for bovine NK cells; AbD Serotec, Raleigh, NC, USA). Cells were then fixed with 2% paraformaldehyde for at least 10 min, washed once with PBS + 2% horse serum and then washed once with 0.2% saponin (Sigma-Aldrich) in PBS + 2% horse serum. Intracellular bovine IFN-γ was detected using a PE-conjugated mouse IgG1 mAb against IFN-γ (MCA1783E; ABD Serotec). For isotype controls, cells were stained with a PE-conjugated mouse IgG1 Ab (BioLegend). IFN-γ Abs and isotype controls were re-suspended in 0.2% saponin. Cells were stained for 25 min at room temperature. After staining, cells were washed, then analyzed using a FACSCalibur with HTS.

Infection model

Calves were housed in ABSL2 specific pathogen-free conditions from 1–6 wk of age to ensure equivalent health status prior to ST infection studies. Calves were tested pre-infection for common pathogens, and all were found negative for ST infection. Calves did not naturally acquire any infections in our facilities. At 6 wk, calves were orally infected with 6 × 10⁶, 1 × 10⁷ or 6 × 10⁷ CFU of a natural calf isolate of ST (Type O, Group B). This isolate has been utilized in prior studies, ³¹ and causes a predictable and severe infection in calves. Prior to infection, 1 μl of ST stock was added to 4 ml of LB media and shaken for 18 h at 37℃. Bacteria were then subcultured 1:40 then shaken again for 2 h at 37℃. CFU counts were verified by plating serial dilutions of bacterial suspensions on LB agar plates. Plasma and fecal samples were collected at infection and at 24 h intervals after infection. Calves were monitored twice daily, and hourly when disease symptoms became apparent. Enterocolitis was defined as fever (>40.6℃) and diarrhea between 24 and 48 h after infection. Fecal samples were collected daily and utilized for bacterial DNA extraction, and also submitted to the Montana State Veterinary Diagnostic Service. Those from experimentally infected calves tested positive for the input strain of ST and negative for other infectious agents. All animal protocols were reviewed and approved by the MSU Institutional Animal Care and Use Committee.

Bacterial DNA extraction and q-PCR

Fecal samples were collected daily from infected calves, and small aliquots were weighed and stored at –80℃ until DNA was extracted. Bacterial DNA was extracted using the QIAamp DNA Stool Mini Kit (QIAGEN, Frederick, MD, USA) according to manufacturer's instructions and stored at –80℃ until further processing. This approach effectively removes PCR inhibitors found in fecal samples. Real-time qPCR was run using SYBR Green Mastermix (Bio-Rad, Berkely, CA, USA) and read on the MYiQ PCR Detection System (Bio-Rad), both according to the manufacturer’s instructions. All primers were ordered from IDT (Coralville, IA, USA), and standard curves were generated for each primer pair. Values for Salmonella species were detected using primers specific to invA, a gene involved in invasion of the intestinal epithelium. ³⁴ These values were normalized to total bacteria levels using primers that are specific to a conserved region among many bacterial species. These values were then normalized to the mass of the fecal aliquot.

Statistical analyses

Data presented were analyzed in GraphPad (GraphPad, La Jolla, CA, USA) and statistical significance was assessed using the unpaired Student’s t-test(two-tailed) or two-way ANOVA as indicated. Data were considered significant when P-values were <0.05.

AmB induces cell division and activation of γδ T cells

HTS assays identified both AmB and a related antifungal drug, Nystatin (data not shown), as potent inducers of IL-2Rα expression on bovine γδ T cells. Non-γδ T cells, including αβ T cells had minimal responses to these drugs. AmB had higher specific activity and was utilized in all further studies concerning the effects of polyenes on γδ T cells. We first measured cell activation induced by AmB by evaluating the percentage of γδ T cells expressing IL-2Rα and the change in mean fluorescence of this receptor. AmB induced increased expression of IL-2Rα on γδ T cells, similar to the response induced by another T-cell agonist, ConA (Figure 1). Figure 1 demonstrates that AmB affected γδ T cells at very low (ng/ml) concentrations. The cells responded strongly after 48 h of incubation with the antifungal drug, but 24 h was sufficient to induce the increased IL-2Rα expression (data not shown). The effect of AmB on IL-2 receptor expression was specific to the γδ T cells within the PBMC population (Figure 1A), but not all γδ T cells were activated even after exposure to the highest (non-cytotoxic) concentrations. This suggests that the drug may only activate specific subsets of γδ T cells. Furthermore, it was observed that removal of monocytes from the PBMC preparation caused a decrease in the AmB effect on γδ T cells (data not shown), indicating a contributory role for monocytes in the response of γδ T cells to AmB. A similar assay was performed using human PBMCs, measuring the increased expression of CD69 as an indicator of activation for human γδ T cells. Similar to bovine cells, human γδ T cells increased expression of CD69 after stimulation with AmB (Figure 1D, E). These data indicate that AmB stimulates γδ T cells from both cattle and humans at low doses. OPEN IN VIEWER

In addition to enhanced IL-2Rα expression, AmB induced proliferation of γδ T cells (Figure 2A, B), detected in a 5-d CFSE assay, and calculated based upon a decrease in CFSE staining at each cell division. The AmB-induced proliferation of γδ T cells was dose-dependent, as lower doses caused few cell divisions above medium-treated cells, while doses of AmB above 112.5 ng/ml were slightly toxic to the cells and caused less proliferation (Figure 2B). These results were consistent with the dose curve related to AmB-induced IL-2Rα expression on γδ T cells, as doses higher than 112.5 ng/ml showed decreased activation, likely owing to cytotoxicity at higher concentrations. Proliferation was also specific to cells staining positive for the γδ TCR, and had little effect on γδ TCR negative cells in the lymphocyte gate (data not shown). A similar proliferation assay was performed using human PBMCs and mouse splenocytes. Figure 2C demonstrates that subsets of human and mouse γδ T cells and γδ TCR-negative cells also proliferated in response to AmB in combination with IL-15 (human) or IL-2 (mouse). These data indicate that the γδ T-cell stimulatory effect is conserved amongst these three species. OPEN IN VIEWER

AmB enhances IFN-γ production by bovine lymphocytes

As γδ T cells are a major source of IFN-γ, we assessed the capacity of AmB to stimulate the production of this cytokine in combination with IL-18. IL-18 is produced by stimulated macrophages and, together with IL-12, stimulates the production of IFN-γ. First, we assessed expression of the activation marker IL-2Rα in response to AmB in combination with IL-18. We determined that when combined with IL-18, AmB greatly augmented γδ T-cell activation compared with IL-18 or AmB alone (Figure 3A, B). In addition to IL-2Rα, we determined that the combination of AmB and IL-18 consistently enhanced IFN-γ production by PBMCs (from multiple animals) compared with IL-18 alone (Figure 3C). These results indicate that AmB in low doses acts as an agonist that primes cells for activation by other stimulants in vitro. As IL-18 is expressed in vivo during infection, these data suggest that AmB could enhance the production of this important cytokine during infection. We also sought to determine the cellular source of IFN-γ in AmB- and IL-18-stimulated PBMCs using intracellular flow cytometry. When cells were stained for intracellular IFN-γ and γδ TCR, it was clear that the IFN-γ was emanating from a non-γδ T-cell source (data not shown). This was followed up by staining with an Ab specific for CD335, which is a marker for bovine NK cells. ³⁵ The plots in Figure 3D suggest that the primary source of IFN-γ in AmB- and IL-18 stimulated PBMCs were NK cells. Thus, AmB primes γδ T cells for IL-2-induced proliferation and NK cells for enhanced production of IFN-γ in the presence of IL-18. OPEN IN VIEWER

Next we assessed the effects of age differences on responses to AmB in calves. The in vitro production of IFN-γ was assessed in PBMCs derived from newborn calves (<12 h old), 6-12-wk-old calves and adult (>2 yr old) cattle. When 6–12-wk-old calves were compared with Holstein cows, PBMCs derived from calves produced significantly more IFN-γ in response to AmB alone and AmB combined with IL-18 than those derived from mature cow/heifers (Figure 4A). In addition, when newborn calves were compared to 6–12-wk-old calves, newborn calf PBMCs produced significantly more IFN-γ upon stimulation than PBMCs from the older calves (Figure 4B). Together, these data suggest that the ability for AmB to enhance IFN-γ production by NK cells may decrease with increasing age of the animal. Therefore, AmB may be optimally applied in younger populations. OPEN IN VIEWER

AmB enhances protection against Salmonella infection in calves

Because of its profound effects on multiple responses of the innate immune system, we applied AmB in a non-fungal infectious challenge model. Calves are highly prone to infection in early life and Salmonella is a major problem for ranchers and farmers. We reasoned that, as γδ T cells and NK cells respond to low doses of AmB in vitro, a low dose delivered in vivo would not induce reported toxic side effects, yet would stimulate these important cell subsets to contribute to protection from infection. Six-wk-old calves (average mass 54–59 kg) were injected once i.v. with approximately 0.029–0.031 mg/kg AmB (1.7 mg in 10 ml of saline), or saline only. The dose of AmB is approximately 10-fold less than the recommended 0.25–0.30 mg/kg dose delivered daily to human patients for treatment of fungal infections. Injection of AmB at this low dose did not induce fever or other apparent detrimental effects in the animals (Figure 5; time 0 is 24 h post-AmB injection). After 24 h the calves were infected with ST by the oral route at a range of 6 × 10⁶–6 × 10⁷ CFU. Calves were initially monitored twice daily, and hourly after 48 h of infection. Animals were scored on a 1–4 scale for attitude, appetite, fever and fecals. Figure 5 shows that animals (n = 4 per group) infected with 1 × 10⁷ CFU began to develop fevers at about 24 h after infection, and fevers were most severe by 52 h after infection. The calves that received AmB developed significantly lower fevers than did saline injected calves (Figure 5A). In addition, the calves that received AmB injections had less severe disease based on other parameters. When the four assessments were combined and averaged, the calves that received AmB had lower scores at every time point than the saline-injected calves (Figure 5B). Thus, one injection of low-dose AmB significantly reduced ST-induced disease symptoms. OPEN IN VIEWER

In one experiment, multiple injections of AmB prior to infection were also assessed. In this case, the calves that received four daily injections prior to infection developed a more severe disease than did saline-injected calves (data not shown). One potential reason for this outcome may be excessive TLR4 agonist signaling leading to immune suppression (tolerance), rather than stimulation. ³⁶ However, our data suggest that one low-dose injection of AmB was sufficient to provide enhanced protection from ST-induced enterocolitis.

To determine if γδ T cells were activated or altered by injection or ST infection, cell populations in peripheral blood were assessed. In each experiment, we measured activation markers on lymphocytes isolated directly from blood, and purified cultured cells re-stimulated with LPS (TLR4 agonist), Pam3cys (TLR2 agonist) and ConA/IL-2 in an attempt to assess AmB-induced priming responses ex vivo. However, no consistent changes in γδ T cells or NK cells could be determined in these studies (data not shown). This is consistent with an earlier assessment of γδ T cells during ST infection in which changes were primarily detected in lymphatic cells trafficking from the mucosal tissue. ³¹ Following i.v. administration, AmB rapidly localizes to and is retained in tissues. ³⁷ Thus, its activation effect may not be reflected in the blood. Furthermore, the primary γδ T-cell subset in circulation localizes to sites of inflammation. ³⁸ It is possible that activated γδ T cells and NK cells similarly localized to the ST-infected tissues and were therefore not detected in circulation in this model. In addition, we measured IL-6 expression in serum collected daily. There were no differences in serum concentrations of IL-6 in calves injected with AmB compared with saline-only injected calves (data not shown). Thus, this low dose of AmB did not appear to cause inflammatory effects similar to those reported for human patients receiving this drug at the therapeutic dose for fungal infections.

To assess bacterial shedding, fecal samples were collected daily from ST-infected calves and assessed for relative amounts of ST. Calves were injected with one dose of AmB or saline only and infected either with either 1 × 10⁷ or 6 × 10⁷ CFU ST. All fecal samples were tested in the Montana State Diagnostic Laboratory for ST using routine culture methods before and after ST infection, and post-infection samples were positive for ST only, as expected. We also developed a quantitative assay using qPCR to detect ST DNA. Bacterial DNA was extracted from daily fecal samples. Primers specific for a conserved region of the 16 S rRNA were used to measure total bacteria for normalization, and ST-specific primers were used to quantify ST in samples. ³⁴ Samples collected at 0 and 24 h did not contain ST, as evidenced by non-specific qPCR products and low normalized CT values. Thus, only values collected at 48 and 72 h after infection were used for statistical comparisons. For both of these infection doses, prior injection with AmB clearly reduced the amount of ST shed in the feces detectible at 48 and/or 72 h after infection (Figure 6). This assay is highly quantitative, as calves infected with the higher dose of ST shed about 10-fold more ST than calves infected with the lower dose. These data suggest that calves that were injected with AmB were not only protected from disease symptoms and severity, but they also subsequently shed less bacteria into the environment than did saline-treated calves. These data indicate that AmB can ameliorate symptoms induced by bacterial infection and counter bacterial replication in calves. OPEN IN VIEWER