Ligation of Dectin-1 by fungal glucans elicits a Th17 response that is necessary for clearing many fungal pathogens. Laminarin is a (1→3, 1→6)-β-glucan that is widely reported to be a Dectin-1 antagonist, however, there are reports that laminarin is also a Dectin-1 agonist. To address this controversy, we assessed the physical properties, structure, purity, Dectin-1 binding, and biological activity of five different laminarin preparations from three different commercial sources. The proton nuclear magnetic resonance analysis indicated that all of the preparations contained laminarin although their molecular mass varied considerably (4400–34,400 Da). Two of the laminarins contained substantial quantities of very low m.w. compounds, some of which were not laminarin. These low m.w. moieties could be significantly reduced by extensive dialysis. All of the laminarin preparations were bound by recombinant human Dectin-1 and mouse Dectin-1, but the affinity varied considerably, and binding affinity did not correlate with Dectin-1 agonism, antagonism, or potency. In both human and mouse cells, two laminarins were Dectin-1 antagonists and two were Dectin-1 agonists. The remaining laminarin was a Dectin-1 antagonist, but when the low m.w. moieties were removed, it became an agonist. We were able to identify a laminarin that is a Dectin-1 agonist and a laminarin that is Dectin-1 antagonist, both of which are relatively pure preparations. These laminarins may be useful in elucidating the structure and activity relationships of glucan/Dectin-1 interactions. Our data demonstrate that laminarin can be either a Dectin-1 antagonist or agonist, depending on the physicochemical properties, purity, and structure of the laminarin preparation employed.
Laminarin is a storage glucan that was first isolated from the Laminariaceae family of brown algae seaweed by Schmiedeberg (1). Typically, laminarins are low m.w. (1→3)-β-d-glucans with varying degrees of (1→6)-β sidechain branching (2, 3). Laminarin has been widely studied and has been increasingly used in recent years because of its interaction with the glucan-specific pattern recognition receptor, Dectin-1 (4). A diverse array of bioactivities has been reported for laminarin, including inhibition of apoptosis (5), antitumor activity (6), anticoagulant properties (7), and antioxidant activity (8). However, the primary use of laminarin in recent years has been as a ligand for pattern recognition receptors in the innate immune system (9). By way of example, Brown et al. (4, 10) reported that the C type lectin receptor Dectin-1 is the principal receptor responsible for binding fungal β-glucans and eliciting antifungal innate immune responses.
Early work with laminarin and Dectin-1 suggested that laminarin blocked the agonist activity of zymosan, a crude yeast cell-wall extract (4). This suggested that laminarin was a Dectin-1 antagonist. Subsequent studies have provided considerable support for the concept that laminarin acts as a Dectin-1 antagonist (11–14). Indeed, laminarin has become a very useful reagent for studying antifungal host defenses due to its ability to block Dectin-1–mediated signaling. However, some studies have provided contradictory evidence by demonstrating that laminarin can act as a Dectin-1 agonist by activating cells in the innate immune system (15). It is not clear why some reports have indicated that laminarins are agonists, whereas others have reported that laminarin is a Dectin-1 antagonist. These conflicting and contradictory results are further confounded by the commercial availability of multiple laminarin preparations that may vary in purity, m.w., and physicochemical properties.
As a result of these conflicting reports, we investigated the relationship between the physical properties, structure, purity, and biological activity of various commercially available laminarins. The laminarins evaluated in this study varied widely with respect to their physicochemical properties, purity, and bioactivities. As an example, many groups have reported that laminarin is a small m.w. compound (3). We found that the m.w. of laminarins varied considerably depending on the preparation. Certain preparations also contained up to 60% (based on the area under the curve) of low m.w. compounds, some of which were not laminarin. Although dialysis removed these low m.w. contaminants, it did not affect the presence of macromolecular aggregates also found in two of the preparations. Receptor-binding assays indicated that all of the laminarins were bound by mouse (m) and human (h) Dectin-1, but the binding affinities varied considerably, and binding did not correlate well with any specific physical property. Biological evaluation of the laminarins demonstrated that two were agonists, two were antagonists, and a third was an antagonist, but became an agonist upon removal of the low m.w. compounds.
Materials and Methods
Five different laminarins were obtained from three different commercial sources. The vendor, lot number, m.w., and identifiers for each laminarin are shown in Table I. Prior to use in bioassays the laminarins were depyrogenated to remove contaminating endotoxin using Affi-Prep Polymyxin Support (Bio-Rad, Hercules, CA), filter sterilized (0.45 μm), and dissolved in culture media. The laminarins were screened for endotoxin using the InvivoGen (San Deigo, CA) HEK-Blue hTLR4 reporter cell line. The HEK-Blue null cell line was used as a control. All of the Sigma laminarins and the Carbosynth laminarin contained endotoxin prior to depyrogenation. Interestingly, the InvivoGen laminarin was endotoxin negative. After depyrogenation all of the laminarins were endotoxin negative.
RAW264.7 (RAW) and THP-1 cells were obtained from American Type Culture Collection (Manassas, VA) and maintained in complete DMEM or RPMI 1640 media, respectively, which was supplemented with 10% FBS and penicillin/streptomycin/glutamine (Thermo Fisher Scientific). PBMCs obtained via leukapheresis were purchased from AllCells (Alameda, CA) and cryopreserved in 50% RPMI 1640 with 40% FBS/10% DMSO, after extensive washing in 1× PBS. Purified Candida albicans β-glucan was prepared as described by our laboratory (16).
The proton nuclear magnetic resonance analysis of laminarin
Proton spectra were collected on a Bruker Avance 400 nuclear magnetic resonance (NMR) spectrometer using a 5 mm probe. Approximately 2 mg of glucan was dissolved in 550 μl deuterium oxide (CAS 7789-20-0; Sigma-Aldrich, St. Louis, MO). Proton NMR (1H-NMR) spectra were obtained at 60°C in 5 mm NMR tubes. Chemical shift referencing was accomplished using trimethylsilylpropanoic acid set to 0.0 ppm. NMR spectra were processed using TOPSPIN 2.1 running on the Avance 400 NMR.
m.w. determination and polymer distribution of laminarins
The m.w., polydispersity, polymer distribution, and Mark-Houwink (α) values were derived from gel permeation chromatography (GPC) with a Viscotek/Malvern GPC system consisting of a GPCMax autoinjector fitted to a TDA 305 detector (Malvern, Westborough, MA). The TDA 305 contains a refractive index detector, a low-angle laser light scattering detector, a right-angle laser light scattering detector, an intrinsic viscosity detector, and an ultraviolet detector (λ = 254 nm). Three Waters Ultrahydrogel columns, i.e., 1200, 500, and 120, were fitted in series (Waters, Milford, MA). The columns and detectors were maintained at 40°C within the TDA 305. The system was calibrated using dextran and pullulan standards in mobile phase (Malvern). Laminarin samples were dissolved (1–3 mg/ml) in mobile phase [50 mM sodium nitrite (pH 7.6)]. The samples were incubated for ∼15 min at ambient temperature, followed by sterile filtration (0.2 μm), and injected into the GPC (200 μl) over an injected mass range of 100–600 μg. Sample recovery was routinely ≥95%. Each sample was analyzed in duplicate or triplicate over at least three concentrations. The data were analyzed using Viscotek OmniSec software v. 22.214.171.1244. Differential index of refraction was calculated using the OmniSec software (v. 126.96.36.1994). Differential index of refraction for the laminarin samples was determined to be 0.155. Initially data were analyzed using a single peak assignment to obtain an average m.w. for the entire polymer distribution. Subsequently, each peak was quantified, and the data expressed as area under the refractive index curve, adjusted for calculated concentration. The percentage that each peak contributed to the total polymer distribution was normalized to 100% total polymer distribution. Replicate analysis of calibration standards indicated reproducibility of ±3%, which is well within the limits of the technique.
Binding interactions of laminarins with recombinant mouse and recombinant human Dectin-1
Samples were analyzed for binding to recombinant human (rh) or recombinant mouse (rm) Dectin-1 (catalog number 1859-DC-050 and 1756-DC-050 respectively, both are C-terminal, extracellular truncated versions containing a 10 His N terminal tag; R&D Systems, Minneapolis, MN) with an Octet Ks bio-layer interferometry instrument manufactured by ForteBio (Fremont, CA). All samples were run at 30°C, 1000 rpm, using 10× kinetics buffer [PBS (pH 7.4), 0.1% BSA, 0.02% PS2] and using Ni-NTA coated SA Biosensors. Saturation curves were constructed using 10× kinetics buffer as a baseline over a ligand concentration range of 7.716–493.8 μM.
THP-1 or RAW cells were seeded at 5 × 105 or 3 × 105 cells per well, respectively, in 96-well plates, and treated with increasing concentrations of the indicated compound for 18–20 h. Prior to use, stock solutions of compounds were sonicated at 45°C for 5 min for homogenization. Treated cells were pelleted by centrifugation at 200 × g for 5 min, followed by harvesting of supernatants. TNF-α levels were determined using a mouse- or human-specific TNF-α DuoSet ELISA kit (R&D Systems) per the manufacturer’s guidelines. ODs were determined by reading 450 nm wavelength on a Molecular Devices plate reader with SoftMax pro software v. 5.4.6.005. Cytokine concentration of each sample was calculated through fitting the standard curve to a four-parameter logistic equation using XLFit software (IDBS, Guildford, U.K.).
Dectin-1 knockdown cells
RAW or THP-1 cells were infected with MISSION short hairpin RNA (shRNA)–expressing lentiviral vectors targeted to the Dectin-1 coding regions according to the manufacturer’s instructions (Sigma-Aldrich). Four different shRNA particle subsets targeting different regions of the transcript were used for each cell type. The constructs used are outlined in Supplemental Table I.
For evaluation of knockdown efficiency, cells were infected with the indicated lentiviral vectors and selected for 3 d in either 2 μg/ml (THP-1) or 6 μg/ml (RAW) puromycin. Total RNA was isolated and quantitative RT-PCR amplification was performed using a Bio-Rad MyIQ Detection System. Two picomoles of ABI TaqMan primer for murine Dectin-1 (Mm01183349_m1) or GAPDH (Mm99999915_g1) (for RAW cells) or hDectin-1 (Hs01902549_s1) (for THP-1 cells) or 18 sRNA (Hs99999901_s1) were added to a TaqMan RNA-to-Ct One Step Kit reaction per the manufacturer’s instructions (Applied Biosystems, Foster City, CA). Reactions were subjected to the following protocol: 48°C for 15 min and 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min. Expression levels were calculated as percent reduction in δ CT values compared with shRNA nontarget control.
Cell populations showing the greatest knockdown percentage via real-time PCR analysis were selected for single-cell isolation via limiting dilution. After monoclonal populations were established, total RNA was isolated and assessed as above using quantitative RT-PCR. Expression levels were compared with shRNA nontarget control and single-cell clones with the greatest knockdown were subjected to additional analysis via Western blotting with an anti–Dectin-1 Ab (PA5-34382; Invitrogen). Cell lines showing optimal knockdown via PCR and Western blot analysis (clone 6 for RAW and clone 3 for THP-1) were subjected to functional studies.
Multiplex cytokine analysis
Cryopreserved PBMCs were thawed into complete RPMI 1640 media and plated at a density of 7.5 × 105 cells per well in a 96-well plate. Laminarins were serially diluted in water from the stock of 10 mg/ml and added to the wells containing cells to give the indicated final compound concentration. Cells were incubated with compound for 18–20 h followed by harvesting of supernatants. Supernatants were frozen at −80°C until analyzed. Cytokine concentrations were measured per the manufacturer’s instructions using a six-plex multiplex kit against hTNF-α, hIL-6, hIL-1β, IL-23, IFN-ɣ, and IL-12p70 (R&D Systems), collected using a Luminex 200 instrument (Luminex) and analyzed with xPonent version 3.1 software.
Equilibrium concentration-binding relationships were determined from receptor-binding experiments. Immobilized rmDectin-1 or rhDectin-1 were exposed to increasing concentrations of laminarin for 300 s followed by a dissociation period of 600 s in buffer without laminarin. Equilibrium binding was measured at 10 s into the dissociation period and corrected using an identical biosensor with immobilize Dectin-1 but not exposed to laminarin. Concentration-binding data were fit using GraphPad Prism 6.07 (La Jolla, CA) to progressively complex zero, one-site and two-site specific binding models ± a linear nonspecific binding component. The sequential F test was used to determine the simplest binding model that best fit the data, and whether global fitting of curve parameters was appropriate. When individual curve fits were appropriate, geometric means of the Kd values and their 95% confidence intervals were used for statistical analysis.
Biological activity was analyzed by determination of cytokine concentration in cellular supernatants through fitting OD450 values from ELISA experiments to standard curves to determine the picograms per milliliter level of cytokine. The ensuing dose-response curves were then log(x) transformed and normalized in GraphPad Prism by setting the largest and smallest values in each data set at 0 and 100% respectively, with the replicate values of 0 and 100% defined by the mean of the replicates. To determine significance of differences in potencies, transformed normalized dose response values were fitted to a nonlinear regression with variable slope to determine logEC50 and EC50. LogEC50 values, SE of logEC50 values, and df were plotted and subjected to one-way ANOVA to determine significance between groups using Tukey correction and multiple comparisons.
The 1H-NMR analysis revealed that all preparations contained laminarin, but some also contained nonlaminarin impurities
1H-NMR spectra of the laminarins (Fig. 1A) are consistent with (1→3)-β- glucan with no evidence of (1→6)-substituted glucosyl sidechains. When examining only the carbohydrate spectral regions, laminarins 02, 03, and 05 are of comparable purity to the reference laminarin 01. Laminarin 04 shows additional resonances suggestive of multiple solution conformations resulting from the higher m.w. In the spectra for laminarins 02 and 03, there is an AB quartet resonance (arrow) between 2.8 and 3.2 ppm resulting from an unknown impurity that possibly contains an asymmetrically substituted methylene group. The AB quartet resonance impurity is not laminarin and it is not a derivative of laminarin. We speculate that the AB quartet is one of the chemical(s) used to isolate the laminarin, which was not completely removed following the isolation process. We did not detect any other impurities aside from the AB quartet in the laminarin preparations.
There are substantial differences in the polymer distributions and m.w. of the laminarins
We evaluated all of the laminarins by high performance, multidetector GPC as previously described (17). The m.w. of the respective laminarins are given in Table I. As can be seen, four of the five laminarins have an average molecular mass within the range of 4400–4800 Da (Table I). In contrast, laminarin 04 has an molecular mass of 34,400 Da, which is an order of magnitude larger than the other laminarins evaluated in this study. We also found that there were substantial differences in the polymer distribution of the laminarins. The 01 (reference) laminarin was a very uniform preparation with a single peak (molecular mass = 4840 Da) (Fig. 1B). In striking contrast, laminarins 02 and 03 showed polydisperse elution profiles with multiple peaks (Fig. 1B). There was a clearly identifiable laminarin peak in each of these preparations and the m.w. for the laminarin peaks was consistent with the reference laminarin (Table I). However, there were multiple small molecular mass (<2000 Da) peaks downfield (≥30 min retention volume) of the laminarin peak that accounted for 40–60% of the total polymer distribution as determined by AUC. Thus, only 40–60% of the polymer distribution could be attributed to laminarin in these two preparations. The remaining amount contained unknown compounds, some of which were not carbohydrates, (see Fig. 1A). Laminarin 05 showed a single uniform peak with a molecular mass of 4750 Da. Laminarin 04 also showed a single uniform peak, but the molecular mass for this laminarin was 34,400 Da.
Dialysis effectively removed low m.w. impurities from laminarins 02 and 03, but did not impact the macromolecular aggregate found in these laminarins
Based on the 1H-NMR and GPC data for laminarins 02 and 03, we sought to determine if the low m.w. impurities could be removed by dialysis. Dialysis in ultrapure water against a 1000 Da molecular mass cutoff membrane removed most, but not all, of the low molecular mass compounds (Fig. 2). As can be seen, 1H-NMR (left) and chromatographic analysis (right) after dialysis show that the unknown compound(s) are removed and the low m.w. compounds that were eluted after the laminarin peak are dramatically reduced or completely eliminated (Fig. 2). We did not detect any other impurities in the laminarin preparations.
Natural product carbohydrate polymers can self-aggregate to form very high m.w. macromolecular complexes (18). Accordingly, we found that laminarins 02 and 03 contained high m.w. macromolecular complexes (Fig. 3, top). Not surprisingly, dialysis did not eliminate these high m.w. complexes (Fig. 3). In contrast, laminarins 04 and 05 did not contain any high m.w. macromolecular complexes (Fig. 3). In addition, the reference laminarin 01 did not show high m.w. complexes (data not shown).
Recognition and interaction of laminarins with rh and rmDectin-1
The interaction of laminarins with m- and hDectin-1 was examined using bio-layer interferometry (BLI) (Tables II, III). Interactions of the laminarins with m- and hDectin-1 were best characterized by a one-site specific binding model; most of the interactions with hDectin-1 and one with mDectin-1 had an additional nonspecific binding component. The interaction of laminarins with rh and rmDectin-1 was dose dependent and saturable (Fig. 4). Interestingly, the maximum BLI signal did not appear to reflect differences in molecular mass between the laminarins.
For the laminarins with a molecular mass between 4000 and 4840 Da, the half-maximal binding concentrations, apparent Kd values, ranged between 0.205 and 0.998 μg/ml for hDectin-1 (Table II), and between 0.162 and 1.170 μg/ml for mDectin-1 (Table III). The Kd values for the larger laminarin 04 was significantly higher, i.e., lower affinity, for rh and rmDectin-1.
Bioactivity of laminarins in murine and human cells
To initially assess the biological activity of the five laminarins, as well as several of their dialyzed counterpoints, the cytokine outputs from mouse and human macrophage cell lines were assessed. The cytokine TNF-α was used as a biomarker for activity because it has been well established to be induced in response to activation of the C type lectin receptor family (19). Five different laminarins demonstrated agonist activity in the RAW cells and reached similar peak levels at the highest concentration evaluated (1000 μg/ml), whereas three of the laminarins induced no or very low levels of TNF-α (Fig. 5). When data were normalized to account for any variation in maximal output and curve shapes, and fitted using a variable slope nonlinear regression model, among the five highly active laminarin preparations distinctly variable relative potencies (EC50) were noted (Fig. 5B).
The most potent compound was 03 dialyzed, followed by 04 (regardless of dialyzed state), 03 undialyzed, and 02 dialyzed (Fig. 5). Although dialysis seemed to have no impact on the potency of laminarin 04, interestingly, dialysis of the 03 preparation enhanced its potency roughly 30-fold (Fig. 5, solid versus dashed blue lines), and increased the stimulatory activity of laminarin 02 from largely inactive to active (Fig. 5A, solid versus dashed red line). The remaining laminarins, undialyzed 02, reference 01, and 05, were generally inactive at cytokine induction, although at the highest concentrations were able to induce minimal amounts of cytokine.
Because laminarins 01, 02, and 05 demonstrated very little agonistic activity, it was hypothesized that they were antagonists. To evaluate antagonistic properties, laminarin pretreated cells were subsequently stimulated with a potent Dectin-1 agonist, purified C. albicans β-glucan. Addition of the Dectin-1 agonist to the agonist laminarins did not greatly change their dose-response curves or activities. The differential potencies remained and all were able to induce similar peak cytokine levels (Fig. 5C, 5D). The three inactive laminarins, 01, 02, and 05, decreased the β-glucan–induced TNF-α cytokine up to an average of 24, 24, and 48%, respectively (Fig. 5C). This decrease was more evident at midlevel concentrations for laminarins 01 and 02, whereas it was more dose responsive for 05 laminarin.
The activity of the laminarins in a human macrophage cell line was assessed to ascertain if the species differences seen in the binding assays (Tables II, III) were recapitulated in the cell-based assay system. THP-1 cells, a human monocyte/macrophage line, were subjected to treatment with increasing concentrations of the different laminarins followed by assessment of TNF-α cytokine production. Similar to the results obtained from the murine RAW cells, five of the laminarins demonstrated capacity to elicit cytokine production from the human cells (Fig. 6), but did so with more distinct potency than that seen in mouse cells. As in the mouse cells, dialyzed 03 was the most potent agonist demonstrating a 5-fold reduction in EC50 over the next closest compounds, 04 (±dialysis). Also, as in mice, both dialyzed and undialyzed 04 laminarin induced roughly equivalently responses (Fig. 6B). In contrast to the murine studies, laminarin 04 demonstrated diminished efficacy when compared with the other agonist compounds. Similar to the murine RAW cells, laminarins 01, 02, and 05 were largely inactive, exhibiting low responses at even the highest concentration tested (Fig. 6A).
The purified β-glucan that demonstrated robust ability to induce TNF-α production from RAW cells (Fig. 5) did not induce the same robust cytokine production in the human THP-1 cell line (data not shown). Therefore, laminarin 04 was used as a Dectin-1 agonist to evaluate antagonistic activity of the other compounds because it repeatedly demonstrated moderate levels of TNF-α induction from THP-1 cells. Following cotreatment with laminarin 04, laminarins 01 and 05 were able to slightly diminish the 04-induced response in a dose-dependent manner, although it was not significant. Thus, unlike in mice, laminarins 01, 02, and 05 appear to not act as antagonists in these cells, but instead remained mostly inactive (Fig. 6B).
Biological activity in Dectin-1 knockdown cells
To confirm the laminarins were exerting their action primarily through the C type lectin receptor Dectin-1, shRNA-mediated knockdown cell lines were employed. Both murine RAW cells and human THP-1 cells were subjected to lentiviral transduction with particles containing shRNA directed against the species-specific Dectin-1 mRNA. Loss of Dectin-1 message and protein were confirmed in several monoclonal lines (Fig. 7A) and a single mouse and human line, demonstrating the best knockdown was used in functional tests.
Loss of the majority of Dectin-1 protein from the murine cells greatly diminished the production of TNF-α in response to the potent agonist laminarins 03 (with and without dialysis), 02 dialyzed, and 04 (with and without dialysis), with an average of loss of maximal cytokine production of ≥80% (Fig. 7B). Although the response was greatly suppressed in the RAW Dectin-1 knockdown cells, it was not completely abolished, likely as a result of residual Dectin-1 protein in these cells (Fig. 7A).
There was also a substantial loss of laminarin activity in the Dectin-1 knockdown human THP-1 cell line (Fig. 7B). The loss of response was greater in the human cells, with only the most potent compound, 03 dialyzed, demonstrating any measurable response in the knockdown cells. This 03 response was still a 55% decrease in the maximal, wild-type response, and a 66-fold decrease in potency. This nearly complete loss of Dectin-1 activity to the various laminarins is likely attributable to the greater loss of the Dectin-1 protein in the THP-1 cell line in comparison with the RAW murine knockdown line (Fig. 7A).
Cytokine production from primary human PBMCs treated with laminarins
These results clearly indicate that laminarin compounds demonstrate variable activity based on physicochemical properties and that the activity for certain ones may be species specific. Dectin-1 agonists are known to induce a Th17-biased immune response (20) and are critical to appropriate polarization of naive human CD4+ T cells into Th17 effector cells. The cytokines critical for appropriate polarization of human cells as such are thought to include IL-6, IL-1β, and IL-23 (20). Treatment of human PBMCs with the various laminarins generally resulted in production of this pro-Th17 phenotype of TNF-α, IL-6, IL-1β, and minimal amounts of IL-23 and IFN-ɣ (Fig. 8). The laminarin dose-responses for production of each of these cytokines was roughly equivalent to the TNF-α induced from the monocytic cell lines. In the PBMCs laminarin 03 dialyzed was the most potent followed by 04 (with and without dialysis) and 03 undialyzed followed by dialyzed 02 (Fig. 8). Interestingly, in this primary, heterogeneous cell population undialyzed 02 began to display reasonable activity to induce cytokine production at the highest concentrations tested (Fig. 8, red circle). However, both laminarin 01 and laminarin 05 demonstrated no cytokine induction correlating with the previous results from both mouse and human cell lines.
To determine if these two laminarins were indeed antagonists in the primary human cells, multiplex cytokine analysis was performed on cells pretreated with the various laminarins followed by treatment with the C. albicans glucan, a potent Dectin-1 agonist in these cells contrary to what was seen in the human THP-1 cells (Supplemental Fig. 1). Generally, the cytokine responses seen from glucan-laminarin–cotreated PBMCs were very similar in potency trends to those seen from laminarin-only–treated cells. However, the magnitude of maximal response was increased, up to 2-fold, for TNF-α, IL-1β, IL-23, and IFN-ɣ in the glucan-cotreated cells (Fig. 9). Furthermore, in the cotreatment groups agonistic activity was demonstrated in response to laminarin 01 for all Th17-necessary cytokines when combined with β-glucan. Increased responses were also noted with laminarin 05 at the highest concentrations tested, particularly for TNF-α and IL-6 (Fig. 9).
In this study, we describe the physicochemical and immunobiological analysis of five laminarin preparations from three different commercial sources. We analyzed the structure, m.w., polymer distribution, Dectin-1 binding, and immunoactivity of each laminarin. The results clearly indicate that laminarin can act as either a Dectin-1 antagonist or agonist, depending on the laminarin preparation employed and the biological system being evaluated. From the five preparations evaluated, we identified a high m.w. laminarin that is a Dectin-1 agonist (04) and a low m.w. Dectin-1 antagonist (05), both of which are uniform and relatively pure preparations that may be very useful in elucidating the structure and activity relationships of glucan and Dectin-1 interactions. Collectively, these data emphasize the need to critically analyze commercially available carbohydrate ligands to determine their composition, purity, and bioactivity before utilizing them in assays of Dectin-1 activity and/or antifungal innate host response.
Laminarin is a natural product (1→3, 1→6)-β-d-glucan extracted from seaweed (3). As a natural product complex carbohydrate polymer, laminarin is by its nature a variable macromolecule. Accordingly, we discovered that five laminarins procured from three different commercial sources all exhibited significant differences in size and purity. Interestingly, none of the laminarins showed evidence of (1→6)-substituted glucosyl sidechains (Fig. 1). This means that these commercial sources are (1→3)-β-d-glucans and do not possess the (1→6)-β linked sidechains that have been reported previously for laminarins (3). These results could suggest that commercial methods of isolation enrich for the 1→3 linkage. Although three of the laminarins (01, 04, and 05) were relatively uniform and chemically pure preparations, two of the preparations (02 and 03) contained low m.w. compounds (Figs. 1, 2), some of which were not laminarin. These laminarins also contained macromolecular aggregates (Fig. 3). Dialysis removed the majority of the low m.w. ultraviolet-absorbing, nonlaminarin contaminants, but as expected did not remove the macromolecular aggregates.
We also examined the interaction of the various laminarins with rh and rmDectin-1. Interactions of the laminarins with m- and hDectin-1 were best characterized as saturable, and dose dependent with a one-site specific binding model and half-maximal binding occurring in the range of 0.2–1.2 μg/ml for the laminarins with molecular masses of 4000–5000 Da. The half-maximal binding concentrations for laminarin 04 (molecular mass = 34,400 D) was significantly higher, i.e., lower affinity. This was an unexpected finding because we had anticipated that larger laminarins would show higher affinity binding due to cross-linking of multiple receptors. Binding of laminarins to m- and hDectin-1 receptors occurred in the same concentration range, but the order of potency for binding was different, indicating that species differences between the receptors include the binding site for laminarins.
The potency and order of potency for laminarin binding and bioactivity through m- and hDectin-1 receptors supports the concept that laminarins use a portion of their binding energy to produce the immunological response, resulting in responses that occur at concentrations well above the measured dissociation constants in binding experiments. The difference in order of potency between binding experiments and biological responses suggests that the various laminarins differ in efficacy, affinity, the ability to produce a biological response as well as the tendency to bind to specific receptors.
Interestingly, dialysis of laminarin 03 resulted in equivalent binding affinity to rm and rhDectin-1 protein, where the undialyzed compound showed ∼6 fold tighter binding to hDectin-1 (Tables II, III), although this same effect was not seen with laminarin 02 in the presence or absence of dialysis. Discordant with the binding data, a large shift was seen in bioactivity in the human and mouse cells between the dialyzed versus undialyzed 02 and 03 compounds. Dialysis of these preparations induced large increases in potency of both, and in the case of 02, was able to induce agonistic activity of the compound whereas it remained inactive in its unpurified state. Laminarin 04, which did not contain small m.w. compounds but did show the presence of multiple solution conformations, showed no difference in bioactivity before or after dialysis. Taken together, these data suggest that the small m.w. contaminants in the laminarin preparations are acting as inhibitors of Dectin-1 signaling or activation, although not likely Dectin-1 binding because binding affinity was not indiscriminately increased in the dialyzed samples.
In contrast to the laminarin 02 and 03 preparations, the 04 and 05 laminarins were determined to be very uniform and chemically pure. However, there was a substantial and unexpected difference in the m.w. The molecular mass of 05 was 4750 Da, whereas that of 04 was 34,400 Da. This difference translated to large differences in bioactivity with 04 acting as a strong agonist in both human and mouse cells, whereas the 05 laminarin was inactive and actually antagonistic to other Dectin agonists (Figs. 5, 6, 8, 9). These data strongly suggest that it is the size of the laminarin polymers that dictates agonist versus antagonist activity. However, laminarins 02 and 03, which have an molecular mass comparable to laminarin 05 (4400 and 4800 Da, respectively), acted as agonists when the low molecular mass contaminants were removed. Therefore, the difference in purity of the 05 laminarin versus 02 and 03 products may be an important determinant of the agonist/antagonist response for lower m.w. laminarins.
Recently, Elder and colleagues have reported that the physical size of the (1→3, 1→6)-β-glucan preparation may be critical to induction of bioactivity (21). We found that the 02 and 03 laminarin preparations contained macromolecular aggregates that could not be removed by dialysis. In contrast, laminarin 05 did not contain any detectable aggregates. This suggests that the presence of these aggregates may contribute to the agonist activity of these laminarins. Thus, the molecular size (m.w.), purity, and presence or absence of macromolecular aggregates may dictate whether a given laminarin is a Dectin-1 agonist or antagonist.
In addition to the interesting findings correlating physical and molecular properties of the laminarins to their bioactivity on Dectin-1, we also observed unexpected immunological activities. Most notable was that although the antagonist compounds, 01, 02 (undialyzed), and 05 acted as such in the cell lines, in primary human PBMCs all compounds demonstrated the ability to induce at least some production of TNF-α, IL-6, IL-1β, and IL-23, especially when they were administered in combination with another Dectin-1 agonist, i.e., C. albicans cell wall (1→3, 1→6)-β-glucan (Figs. 8, 9). These results are more complex to interpret because PBMCs represent a mixed cell population. One possible explanation is that the laminarin and/or β-glucan may be acting upon other C type lectin receptors in disparate cell types. The synergy, in terms of increased maximal cytokine level, between the laminarins and C. albicans cell-wall β-glucan was also unexpected, and as yet unexplained, but is very interesting in that it suggests these two compounds could be utilized to stimulate greater immune responses in human systems.
There is an important caveat to this study. The data presented pertains only to the specific products and lot numbers evaluated in this research. We cannot speculate about the purity or characteristics of laminarins provided by other vendors or other batches of laminarin provided by these vendors. When considered as a whole, this work outlines how laminarins of different physical properties, including size, purity, and aggregation state, may be used to elucidate the structure and activity relationships of glucan/Dectin-1 interactions. These data also emphasize the need to critically analyze commercially available carbohydrate ligands to determine their composition, purity, and bioactivity before utilizing them in assays of Dectin-1 activity or antifungal host response.
This work was supported in part by National Institutes of Health Contract HHSN272201400050C to J.T.E. and National Institutes of Health Grants GM53522, GM083016, and GM119197 to D.L.W., and C06RR0306551 to East Tennessee State University.
The online version of this article contains supplemental material.
The authors have no financial conflicts of interest.