The impact of the level and distribution of methyl-esters of pectins on TLR2-1 dependent anti-inflammatory responses

Pectins have anti-inflammatory effects via Toll-like receptor (TLR) inhibition in a degree of methyl-esterification-(DM)-dependent manner. However, pectins also vary in distribution of methyl-esters over the galacturonic-acid (GalA) backbone (Degree of Blockiness - DB) and impact of this on anti-inflammatory capacity is unknown. Pectins mainly inhibit TLR2-1 but magnitude depends on both DM and DB. Low DM pectins (DM18/19) with both low (DB86) and high DB (DB94) strongly inhibit TLR2-1. However, pectins with intermediate DM (DM43/DM49) and high DB (DB60), but not with low DB (DB33), inhibit TLR2-1 as strongly as low DM. High DM pectins (DM84/88) with DB71 and DB91 do not inhibit TLR2-1 strongly. Pectin-binding to TLR2 was confirmed by capture-ELISA. In human macrophages, low DM and intermediate DM pectins with high DB inhibited TLR2-1 induced IL-6 secretion. Both high number and blockwise distribution of non-esterified GalA in pectins are responsible for the anti-inflammatory effects via inhibition of TLR2-1.


Introduction
A lower intake of dietary fibers in Western society compared to more traditional diets is associated with a higher chance of developing diseases with a dysregulated immunity such as type 2 diabetes, obesity, inflammatory bowel disease, and autoimmune disorders [1][2][3][4][5] . In contrast, a high dietary fiber intake in traditional societies coincided with a lower frequency of those diseases [1,6] . The mechanisms by which dietary fiber intake prevents immunity-related disease is not fully understood. Several studies have shown that dietary fibers can influence immunity by supporting intestinal microbiota and enhancing production of metabolic fermentation products such as short-chain fatty acids (SCFA), aryl hydrocarbon receptor (Ahr)-ligands or other microbial-derived molecules [7,8] . Moreover, dietary fibers are also known to directly stimulate the immune system [9,10] by binding to Toll-like receptors (TLRs) [11][12][13] . TLRs are a family of pattern recognition receptors (PRRs) which play an important role in intestinal immune regulation [14] . PRRs serve as sensors for innate immunity and may after activation stimulate transcription factors NF-κB and AP-1, which induce upregulation of pro-and anti-inflammatory genes, depending on the activated receptor interactions [15] . This may activate not only innate immune responses but also activate adaptive immune responses [16][17][18] . In the intestine, TLRs are expressed on most immune and gut epithelial cells [19,20] . Each TLR recognizes specific pathogen-associated molecular patterns (PAMPs), damage-associated molecular patterns (DAMPs), or food associated molecules [15] .
Pectin is one of the dietary fiber molecules with TLR binding capacity and has been shown to have anti-inflammatory effects depending on its chemical structure [12,13,[21][22][23][24] . Native plant pectin consists of homogalacturonan, rhamnogalacturonan I (RG-I), and II (RG-II). RG-I segments consist of a backbone of repeating disaccharide backbone structures of alternating GalA and rhamnose residues. The rhamnose residues can be branched with neutral side chains. RG-II segments contain a backbone of GalA residues, with short side chains which contain 12 different sugar residues [25,26] .
These pectins consist mainly (≥ 60%) of linear 1,4-D-galacturonic acid (GalA) (homogalacturonan) segments and minor amounts of branched rhamnogalacturonan segments [27] . The homogalacturonan backbone can be methyl-esterified (Figure 1), and the amount of esters on the backbone is referred to as the degree of methylesterification (DM) [28] . Dependent on the DM, pectins have different functional properties. Sahasrabudhe et al., showed that TLR2-1 was inhibited in a DM-dependent manner by lemon pectins in which a gradual decreasing DM increased TLR2-1 inhibiting properties of pectins. In addition, TLR2 ectodomains bound stronger to pectins with a lower DM pectins than to pectins with a higher DM [13] . However, pectins not only differ in DM but also in distribution of methyl-esters over the backbone. The degree of blockiness (DB) is a structural parameter for the distribution of nonesterified GalA residues in pectins ( Figure 1). When comparing pectins with similar DM, high DB (HB) pectins have a more blockwise distribution of non-esterified GalA residues compared to low DB (LB) pectins. This in contrast to LB pectins that have a more random distribution of non-esterified GalA residues ( Figure 1) [29] . When comparing pectins with a different DM (DM40 and DM80), but a similar DB, the total number of non-esterified GalA residues that are blockwise distributed is larger on pectins with DM40 than on pectins with DM80. This is because the DM40 pectin contains a larger number of non-esterified GalA residues than DM80 pectin ( Figure 1) [29] . How DB contributes to TLR signaling is not known. and DM80. Homogalacturonan pectins consist of a galacturonic acid (GalA) backbone structure in which GalA residues can be methyl-esterified (degree of methyl-esterification; DM). Low degree of blockiness (LB) pectins contain a more random distribution of non-esterified GalA residues, whereas high degree of blockiness (HB) pectins contain a more blockwise distribution of non-esterified GalA residues [29] .

92
In the present study, it was hypothesized that the level and distribution of methylesters in pectin determine the efficacy of pectins as TLR signaling molecule. Therefore, the relationship between pectin structures and Toll-like receptors signaling was determined by comparing the impact of pectins with different DM and DB on activation or inhibition of different TLRs in reporter cells expressing TLRs. First, structural different orange and lemon pectins were studied on having similar DM-dependent effects on activation and inhibition of TLR2, 2-1, 2-6, 3, 4, 5, 7, 8, and 9. Next, it was studied which combination of the structural parameters DM or DB induced most pronounced TLR2-1 inhibition. Furthermore, the effects of pectins on TLR2 binding was also studied. In addition to the effects of pectins on TLR2 reporter cell line, the stimulating or attenuating effects of pectins on cytokine secretion by human macrophages in vitro were studied.

Cell lines
To study the influence of pectins on Toll like receptor (TLR) signaling various HEK-Blue TM reporter cell lines (Invivogen, Toulouse, France) were used [12,30] . These reporter cell lines express Soluble Embryonic Alkaline Phosphatase (SEAP). The SEAP reporter gene is placed under the control of a NF-κB and an AP-1 responsive promotor. Upon activation of the TLRs by a specific agonist, high levels of intracellular NF-κB will lead to secretion of SEAP which can be quantified by QUANTI-Blue (Invivogen, Toulouse, France). HEK-Blue cells containing a construct of human TLR2, 3, 4, 5, 7, 8, or 9 (Invivogen) were used to study the effect of pectins on single TLRs. HEK 293/hTLR2-HA (Invivogen) was used for studying the interaction of TLR2 and pectins. All HEK-

Determination of monosaccharide composition
Neutral sugar composition of the pectins was analysed after pre-hydrolysis with 72% (w/w) H2SO4 (1 h, 30 °C) followed by further hydrolysis with 1 M H2SO4 (3 h, 100 °C) [31] . Neutral sugars released were reduced with sodium borohydride to form their corresponding alditols and then acetylated to yield their volatile derivatives. These alditol acetates were separated and quantified by gas-liquid chromatography (GLC Trace 1300; Interscience Focus-GC, Thermo Fisher Scientific) as described by Englyst and Cummings [31] equipped with a flame-ionisation detector (FID) and a 15 meter DB-225 column (Agilent J&W, Santa Clara, CA, USA). Inositol was used as internal standard. The uronic acid content was determined by the automated colorimetric mhydroxydiphenyl method [32] .

High performance size exclusion chromatography (HPSEC)
Pectin before and after enzymatic digestion were analysed using an Ultimate 3000 system (Dionex, Sunnyvale, CA, USA). A set of four TSK-Gel super AW columns (Tosoh Bioscience, Tokyo, Japan) was used in series: guard column (6 mm ID × 40 mm) and the columns TSK super AW 4000, 3000 and 2500 (6 mm × 150 mm). The column temperature was set to 55 °C. Samples (10 μL, 2.5 mg/ml) were eluted with filtered 0.2 M NaNO3 at a flow rate of 0.6 ml/min. The elution was monitored by refractive index detection (Shodex RI 101; Showa Denko K.K., Tokyo, Japan). The HPSEC system was calibrated using polydisperse pectin standards having molecular weights ranging from 10 to 100 kDa as estimated by viscosimetry [34] . To display clearly the molecular weight of pectins larger than 100 kDa, 150 kDa has been calculated from the standards.
Molecular weights presented were estimated at the top of the curve, despite slight differences in elution patterns of the various pectins pointing to differences in polydispersity.

High performance anion exchange chromatography (HPAEC)
The pectin digests were analysed and subsequently quantified using an ICS5000 High Performance Anion with or without methyl esterification was calculated from the peak area.

Determination of degree of methyl-esterification
Pectin samples (5mg) were saponified using 1 ml of 0.1M NaOH for 24 h (1 h at 4 °C, followed by 23 h incubation at RT). To the pectin blank, 1 ml of ultra-pure water was added. The head-space vials were immediately sealed with a Teflon lined rubber septum. To determine the degree of methyl-esterification (DM) a gas chromatography method was used as previously described [36] .

Determination of degree of blockiness
The degree of blockiness (DB) is calculated as the number of GalA residues present as non-methyl-esterified mono-, di-and triGalA released by endo-polygalacturonase related to the total amount of non-methyl-esterified GalA residues present and expressed as a percentage [29,37,38] . The amount of mono-, di-and triGalA after the PG/PL digestion of pectins was determined by HPAEC-PAD. For the quantification GalA, GalA2 and GalA3 were used. Since the alkaline elution conditions removes all methyl esters from the oligo-uronides, no distinction could be made between methylesterified and non-methyl-esterified GalA3. The amount of GalA3 1 (1 methyl ester) as measured by HILIC-ESI-IT-MS was used to calculate the amount of nonesterified GalA3. DB was calculated using the following formula:

Reporter cell assays
To study whether pectins can activate TLRs or inhibit TLRs, activation or inhibition assays were performed with pectins using HEK-Blue TM cells expressing human TLRs (Invivogen). HEK-Blue TM hTLR cells were seeded in 96 wells plates at the indicated concentrations ( Table 2) in 180 µl/well and were incubated overnight. The next day, the DMEM medium was replaced by DMEM medium containing pectins in the concentration 0.5 mg/ml, 1 mg/ml or 2 mg/ml. Experiments to compare lemon and orange pectins were tested at 1 mg/ml only. Activation of the TLRs was studied by treating the cells with the pectins for 24 hours. Inhibition of the TLRs was studied by pre-treating the cells with pectins for 1 hour followed by addition of 20 µl of the TLR specific agonist (Table 2). Culture medium was used as negative control and the TLR specific agonist was used as positive control for 24 hours (  Protein immunoprecipitation and ELISA for binding of TLR2 to pectin hTLR2-HA protein was isolated from HEK 293/hTLR2-HA (Invivogen) as described before [13] . HA-tagged proteins were immunoprecipitated using Pierce® anti-HA To confirm that specific pectins bind to TLR2, a capture ELISA was performed as described before [13] . Isolated TLR2-HA was applied in the concentrations 0.1 µg, 1 µg and 10 µg/well. For each pectin, rat-anti pectin antibody LM20 (1:100; Plantprobes, Leeds, UK) was used as positive control for pectin binding, to confirm even pectin immobilization. Each experiment was performed at least five times.

TLR2-1 inhibitory effect of pectins on IL-6 and IL-10 production
In addition to the TLR2-1 inhibition assay on reporter cell lines, TLR2-1-dependent inhibition of immune responses by pectins was also tested on THP-1 cells differentiated to macrophages [39] . THP-1 cell differentiation was induced by stimulation of THP-1 cells (1x10 6 cells/ml) with 100 ng/ml Phorbol 12-myristate 13-acetate (PMA, Sigma) in a 12 wells plate (in 0.5 mL medium) for 48 hours at 37°C and 5% CO2. The adherent cells were washed with PBS (Westburg, Grubbenvorst, the Netherlands) to remove PMA. Next, they were treated with pectins at 100 µg/ml dissolved in culture media.
This concentration of pectins has previously been shown to be effective in activating and inhibiting macrophage responses [13] . Non-treated THP-1 cells were used as negative control. After 1 h of pre-treatment with the pectins, 10 ng/ml of Pam3CSK4 was added. THP-1 cells treated with Pam3CSK4 or pectin only were used as control.
After 24 h incubation, media supernatant was collected. IL-6 and IL-10 were quantified in the supernatant by ELISA according to manufacturer's protocol (eBioscience, San Diego, USA).

Chemical composition of pectins
Pectins obtained from lemon and orange were characterized for the degree (percent) of methyl-esterification (DM), molecular weight, and sugar composition. The degree of blockiness was calculated after enzymatic fingerprinting of the pectins and subsequent analysis of the released oligosaccharides by HPAEC and HILIC-MS. The characteristics of the pectins are given in Table 3.

TLR2 is activated by high DM orange pectins while TLR2-1 is inhibited in a DM-independent manner by orange and lemon pectins
Pectins might influence immunity through Toll-like receptor (TLR) signaling 12   As shown in Figure 3, the used sets of orange and lemon pectins have a different TLR inhibiting capacity. For the inhibition studies, TLR2-1 was studied by using Pam3CSK4 as agonist, TLR2-6 by using FSL-1 as agonist, and to study total TLR2 inhibition the agonist HKLM was used. Both lemon pectin and orange pectin specifically inhibited TLR2-1 and had no inhibitory effects on FSL-1 and HKLM induced TLR2 activation.
Orange pectins did not inhibit TLR2-1 in a stronger way with gradual lower DM content, which is the opposite of what was observed before with lemon pectins 13 : orange pectin with a DM64 had a higher inhibiting effect on TLR2-1 than orange pectin with a DM32 (50.0 ± 0.05%, p < 0.0001 vs 40.4 ± 0.05%, p < 0.0001, respectively).

Pectin's degree of blockiness has overarching effects on DM induced effects on TLR2-1 inhibition
Here and in a previous study it has been demonstrated that TLR2-1 inhibition by lemon pectins was DM dependent with more pronounced inhibition of lower DM pectins [13] .
As orange pectins, with other structural features, did not seem to have this same DM dependent inhibitory effects on TLR2-1 with lowering of DM, it was questioned whether other structural properties of pectins may play a role in TLR2-1 inhibition. In search of such differences, the degree of blockiness (DB) of the tested orange and lemon pectins was determined (Table 3). Orange DM32 pectin has a lower DB than the lemon DM33 pectin (35% and 48%, respectively). Furthermore, orange DM64 pectin has higher DB than the lemon DM52 pectin (37% and 31%, respectively). To visualize the impact of the DB more clearly the TLR2-1 inhibiting capacity was expressed  subsequently stimulated with the TLR-specific agonist. The statistical differences between TLR ligand and pectin samples were quantified using the one-way ANOVA test (* p < 0.05, *** p < 0.001 and **** p < 0.0001) (n=9).
The TLR2-1 inhibitory capacity of six lemon pectins was compared. The pectins could be grouped into three levels of similar DM of 19%, 46% or 86%, but with a different degree of blockiness (Table 3)  mg/ml. The statistical differences between Pam3SCK4 and pectin samples were quantified using oneway ANOVA test (* p < 0.05, *** p < 0.001 and **** p < 0.0001). Statistical differences between LB and HB pectins were tested by repeated measures two-way ANOVA (# p < 0.05) (n=6).

Impact of DB and DM on binding to the TLR2 protein
To further substantiate the DB-dependent binding of pectin to TLR2 a capture ELISA was performed that measures the direct binding of pectins to TLR2. This approach allows us to determine true binding of pectin by the TLR2 receptor rather than neutralizing the agonist. All high DB pectins showed stronger binding to TLR2 than the low DB pectins (p < 0.05) ( Figure 6). This effect was concentration-dependent and most pronounced with 10 µg TLR2 protein (Supplementary Figure 3). The high DB pectin bound as strong as low DM pectins to TLR2. At a very high DM, there is no significant difference between the high and low DB pectins measured in TLR2 binding. Together, these findings suggest that pectins with blockwise distributed non-esterified GalA residues bind stronger to TLR2 than pectins with randomly distributed non-esterified GalA.  The statistical differences between the Pam3CSK4 and pectin samples test (* p < 0.05, ** p < 0.01, *** p < 0.001 and **** p < 0.0001) or between control and pectin samples test (#### p < 0.0001) were quantified using the twoway ANOVA test. (n=5).

Discussion
Several studies have shown the protective effects of pectins on development of mucositis, pancreatitis, diet-induced obesity or autoimmune diabetes in mouse models [13,[40][41][42] . The exact mechanisms responsible for these protective effects of pectins are not fully understood. One of the mechanisms by which pectin can protect against inflammatory disease is by modulating TLR signaling [12,13,24] . This modulation of TLRs depends on structural parameters of pectins, such as the DM [12,13] . However, the impact of other structural features such as the blockwise distribution of nonesterified GalA residues was so far unknown. Here, it was shown that the DB is an essential factor in the attenuating effects of pectins on TLR2-1 signaling and that the effects of the DB are most distinct in pectins with higher DM (Figure 8).
The current study shows that pectins strongly inhibit TLR2-1, whereas other extracellular or intracellular TLRs are inhibited to a much lower extend by pectins. This seems in contrast to other studies which showed inhibition of TLR2-6, 4 or TLR9 induced immune responses by pectins [24,43] . However, the inhibitory effects of those pectins may be related to the presence of RG-I and RG-II side chains, which are almost absent in pectins from the current study. These pectins are mainly homogalacturonan pectins [24,43] . Sahasrabudhe et al., also confirmed that homogalacturonan pectins inhibit TLR2-1 specifically and not TLR2-6, TLR4, or TLR5 [13] . Sahasrabudhe et al., provided evidence that homogalacturonan pectins interact with the TLR1 binding site on TLR2, preventing dimerization of TLR2-1 [13] . The homogalacturonan pectins in that study were not able to inhibit dimerization of TLR2-6, which corroborates the current findings. Together, the current findings show that homogalacturonan pectins are very specific in inhibiting TLR2-1 immune responses, whereas RGI and RGII pectins can inhibit other TLR mediated immune responses [24,43] .
Our data illustrate that the high DB strengthens the DM-dependent TLR2-1 inhibition. This suggests that not only the high level but also the blockwise distribution of non-esterified GalA residues of pectins ( Figure 1) is important for TLR2-1 inhibition. This is confirmed by the observation that both low DM pectins but also intermediate DM pectin with a high DB, which have a more blockwise distribution of their nonesterified GalA residues, inhibited TLR2-1 strongly. This argumentation is further supported by the observation that intermediate DM pectin with a low DB, having a more random distribution of its non-esterified GalA residues, inhibited TLR2-1 less efficiently. However, the very high DM pectins, which showed very low inhibition of TLR2-1, contain a very low number of non-esterified GalA residues [29] . Together, these findings suggest that the blockwise distribution of non-esterified GalA residues in pectins induces more inhibition of TLR2-1 than pectins with a more random distribution of non-esterified GalA residues or having a very low number of nonesterified GalA residues.
The reason that a high DB in very high DM pectins is not leading to a significant inhibition, could be simply due to the fact that there is a limited number of nonesterified GalA residues present, despite that these non-esterified galA residues are blockwise distributed. The block size of DM88 pectin might be too small to induce a strong inhibition of TLR2-1, whereas the larger blocks in DM19 and DM43 pectins still are inhibitory. In general, a DB-value does not provide information whether the corresponding pectin may contain one big block or several smaller blocks of nonesterified GalA residues [38,44] . Based on the absolute number of non-esterified galA residues present, DM19 and DM43 contain certainly more blocks than DM88 pectin.
This suggests that a combination of block size and distribution [38,44] may be involved in TLR2-1 inhibitory capacity of pectins.
Next, the binding of low DB and high DB pectins to TLR2 was investigated to confirm true binding of pectins to the receptor rather than to the agonist. Binding of pectin to the receptor was confirmed and the DB-dependent patterns of binding were similar to what was observed for TLR2-1 inhibition in the reporter cell lines.
Furthermore, in this capture ELISA less binding was observed for very high DM pectins which confirms the aforementioned reasoning that blockwise distribution and blocksizes of non-esterified GalA residues are important for the capacity of pectins to bind to TLR2 and preventing TLR1 to associate. In addition, it was confirmed that a more blockwise distribution of non-esterified GalA residues bind stronger to TLR2 than random distribution non-esterified GalA residues. This binding may be established through ionic binding between the blocks of non-esterified GalA residues and TLR2.
Ionic binding has been shown to play an important role in the interaction of pectins and TLR2 [45] . More negatively charged pectins (low DM) bound stronger to TLR2 than less negatively charged pectins (high DM). The binding between negatively charged pectins became stronger to mutant TLR2 proteins with more positively charged amino acids [13] . Pectins with a blockwise distribution (high DB) of non-esterified GalA residues have larger areas with negative charge (higher charge density) compared to random distributed non-esterified GalA residues [46] . This suggests that the larger negative charge areas of the blockwise distributed non-esterified GalA residues in pectins may be of importance in the binding of pectins to TLR2.
The current study also showed that high DB pectins were more effective in suppressing TLR2-1 induced IL-6 responses than low DB pectins, which is in line with the TLR2-1 inhibition as was observed in TLR2-1 inhibition assays. This suggests that pectins not only affect TLR2-1 signaling, but also the subsequent initiation of IL-6 secretion. Inhibition of IL-6 responses may be beneficial under inflammatory conditions, as high levels of IL-6 play an important role in intestinal inflammation [47,48] . The secretion of IL-6 strongly depends on TLR2 activation [49,50] . This has been observed in mice with mucositis in which high activation of TLR2 induces inflammation characterized by high IL-6 levels [51,52] . Low DM pectins were able to reduce this inflammatory response in mucositis by inhibiting TLR2 signaling and IL-6 secretion [13] . As HB pectins were able to inhibit the TLR2-1 induced IL-6 secretion, they may also serve as a dietary component with potential anti-inflammatory effects on mucositis.