GB2571925A - Iron-based vaccine adjuvants - Google Patents

Abstract

A composition comprising iron (ferric or ferrous ions) and at least one vaccine adjuvant. The vaccine adjuvant may be LPS (lipopolysaccharide), monophosphoryl lipid A, a toll-like receptor agonists, Poly IC, a pattern recognition receptor (PRR) ligand, a NOD-like (nucleotide oligomerization domain) receptor ligand, a lectin receptor, alum, MF95, saponin, Quil-A or QS21. The iron may be iron chloride, iron oxides, iron phosphate or iron hydroxide. The composition may be suitable for use in human or veterinary medicine. The composition can be used in combination with an antigen as part of a vaccine or immunomodulatory product. Also claimed is a composition comprising monophosphoryl lipid A and iron (II) chloride (FeCl2). The composition may be used in the preparation of a vaccine or immunomodulatory product. The ambient endotoxin levels in particular matter (PM) from various environments (traffic, green and industrial) was monitored and the ability of host cell immune recognition of LPS was assessed for each sample. Analysis of metals in the environmental samples was carried out. Higher levels of iron was present in traffic samples, where there was highest levels of TLR4 stimulation by LPS.

Description

IRON-BASED VACCINE ADJUVANTS

FIELD OF THE INVENTION

The present invention in general relates to combinations of iron (ferric or ferrous ions e.g. FeCI₂) and vaccine adjuvants (e.g. Monophosphoryl lipid A) for use in vaccines or immunomodulatory products.

BACKGROUND TO THE INVENTION

Numerous studies with varying populations, regions, and methodologies have associated an increase in respiratory and cardiovascular morbidity and mortality with a dose-dependent exposure to ambient particulate matter (PM). Despite the strong link, PM encompasses a highly diverse range of components varying in chemical composition (from natural to anthropogenic sources), phase (liquid/solid particles), and particle size, which makes it difficult to pinpoint the components responsible for these adverse effects. Furthermore, although regulatory standards and limits for PM size ranges, PMw (<10 pm) and PM2.5 (<2.5 pm), and specific particle constituents exist, there is still no threshold below which adverse health effects are no longer observed (WHO, 2016). This drives the urgent need to improve our understanding of the toxicological mechanisms - including synergist effects - of specific PM compounds to provide better air quality management.

There are many suggested mechanisms and outcomes of PM-induced toxicity, but the exacerbation of inflammation and oxidative stress by PM on the respiratory system plays a major role, especially in susceptible individuals. Contributing towards this, bacterial endotoxins are suggested as apt candidates due to their ubiquitous and chemically-durable nature, their naturally high pro-inflammatory capacity, and their ability to prime immune responses resulting in a heightened immune response to other air pollutants. Subsequently, ambient endotoxins have been studied in several cities worldwide, typically collected using filter-based samplers and quantified with the LAL assay. Using these methods, endotoxins have been spatiotemporally monitored in various PM size ranges, fluctuation of endotoxin content in PM was examined, the effects of meteorological factors or endotoxin sources were investigated, and correlations of endotoxins to inflammatory biomarkers have been explored. Although higher levels of urban endotoxins have been related to higher levels of immune biomarkers, such as IL-6 and TNFa, direct proportionality is not often observed. What is currently lacking, is the link between quantified endotoxin - as a constituent of urban PM - and the host immune recognition through toll-like receptor 4 (TLR4). This relationship may vary depending on the different assays and their specificity of endotoxin recognition, interfering or synergistic PM components or different LPS potencies from Gram-negative bacterial populations.

-2Pattern recognition receptors (PRRs) such as TLR4 can be found on the surface of barrier epithelial cells and many immune cells (such as macrophages) which represent the first line of innate immune defence against microorganisms entering the lungs. TLR4, together with MD-2, is responsible for the recognition of endotoxin. Stimulation of the TLR4/MD-2 complex sets off a cascade of reactions strongly activating the release of pro-inflammatory mediators such as interleukins IL-8, IL-1 β and IL-6, and tumor necrosis factor alpha (TNFa). Recently it was also found that the transition metals nickel and cobalt could specifically activate the human TLR4/MD2 receptor complex (initially considered to be limited to endotoxin) through conserved histidine residues in the TLR4 ectodomain of humans and primates. Furthermore, nickel was shown to act synergistically with endotoxin in the production of hTNFa.

This study aimed to bridge the gap between endotoxin quantification and the inflammatory response by investigating the biological recognition of endotoxin via the human TLR4/MD-2 complex. Endotoxin concentrations were quantified from a diverse representation of outdoor urban PM samples to determine whether their associations with different sources - and possibly other PM components, such as transition metals - are relevant for their biological recognition by the TLR4/MD2 complex and subsequent inflammatory response.

We have now surprisingly found that the effect of nickel and cobalt - previously reported to activate the hTLR4/MD2 complex - was found to be negligible in comparison to that of iron. In fact, the addition of Fe as a factor significantly improved the regression model between the two endotoxin quantification assays, explaining 77% of the variation of the TLR4 stimulation and excluding the significant effect of land use class. Moreover, the effect of iron proved to be more than a correlation, since dosing LPS with Fe⁺² led to an increase up to 64% in TLR4 stimulation, while Fe⁺² without LPS was unable to stimulate a response.

SUMMARY OF THE INVENTION

In a first aspect, the present invention provides a composition comprising iron (ferric or ferrous ions) and at least one vaccine adjuvant.

In a specific embodiment, said at least one vaccine adjuvant is selected from the list comprising: TLR4 agonists, LPS (lipopolysaccharide), Monophosphoryl lipid A, TLR2 agonists (Pam2Cys or Pam3Cys), TLR3 agonists, Poly l:C, TLR5 agonists (flagellin-Ag complexes), TLR7/8 agonists (imiquimod, resiquimod), TLR9 agonists (CpG DNA), a PRR (Pattern Recognition Receptor) ligand, a NOD-like (nucleotide oligomerization domain) receptor ligand, a lectin receptor, alum, MF95, saponin, Quil-A and QS21.

-3In another specific embodiment, said iron (ferric or ferrous ions) is selected from the list comprising: FeCI₂, FeCI₃, Fe₃O₄, Fe₂O₃, FePO₄, FeSO₄, and Fe(OH)₂.

In a very specific embodiment, the present invention provides a composition comprising FeCI₂ and Monophosphoryl lipid A.

The present invention also provides the compositions as disclosed herein for use in human or veterinary medicine. More in particular, the present invention provides the use of the compositions as disclosed herein as an adjuvant in vaccination or immunomodulation.

The present invention further provides a vaccine or immunomodulatory product comprising an antigen and a composition as defined herein.

Finally, the present invention provides the use of iron (ferric or ferrous ions) and a vaccine adjuvant in the preparation of a vaccine or immunomodulatory product.

BRIEF DESCRIPTION OF THE DRAWINGS

With specific reference now to the figures, it is stressed that the particulars shown are by way of example and for purposes of illustrative discussion of the different embodiments of the present invention only. They are presented in the cause of providing what is believed to be the most useful and readily description of the principles and conceptual aspects of the invention. In this regard no attempt is made to show structural details of the invention in more detail than is necessary for a fundamental understanding of the invention. The description taken with the drawings making apparent to those skilled in the art how the several forms of the invention may be embodied in practice.

Fig. 1 : The comparison of endotoxin quantification by the rFC assay and via TLR4 stimulation through the HEK-Blue™ hTLR4 cell line, both using the same endotoxin stock for their standard curves. Endotoxin concentrations from the separate locations within each land use class were not significantly different and further comparisons were thus made between the three land use classes (traffic, green and industrial) with each sample represented by a black dot. Statistical significance indicated by * p<0.05, ** p<0.01

Fig. 2: (A): The correlation between the rFC assay and the HEK-Blue™ hTLR4 cells for samples in the land use classes: traffic ·, urban green A, and industrial s. (B): The response ratio of the hTLR4 stimulation to the rFC assay for the considered urban land use classes.

-4Fig. 3 : Transition metal concentrations (ng m'³) measured with ICP-MS, with each dot represented by a sample from the six sampling sites within the city of Antwerp for urban green, traffic and industrial locations. Black dots represent Traffic 1, Park, and Harbour; grey dots represent Traffic 2, Campus, and Metal recycling plant respectively. Note the differences in the vertical axes. Points on dotted horizontal line were samples below the minimum detection limit.

Fig. 4 : mRNA levels of the pro-inflammatory markers IL-8, IL-1 β, and TNFa measured in human macrophage-like U937 cells after 3 h exposure with collected air samples. Values are expressed relative to the negative control, represented by an expression of one, and the geometric means of the locations are shown in dotted lines, while each sample is represented as a dot. The land use classes: traffic, green and industrial each comprise two different sampling locations. The relative expressions for these pro-inflammatory biomarkers were significantly higher than the negative control for all locations (Welch’s t-test for unequal variance). Statistical significance indicated by * p<0.05, ** p<0.01

Fig. 5: Scatter plot of endotoxin concentration as determined with the rFC assay (EU m^-3) vs the relative expression in mRNA IL-8 expression, with each dot representing a sample from either the green, industrial and traffic locations. Grey points represent Campus, Metal plant, and Traffic 2, respectively.

Fig. 6: HEK-Blue hTLR4 cells were stimulated in triplicate with LPS (5 EU ml^-1) and/or increasing iron concentrations (2 μΜ to 0.75 mM) for 20 hours. The resulting SEAP activity was converted to EU ml'¹ with a standard curve. The “no LPS” control tested the response of the iron solutions in the absence of LPS. For statistical analysis, the Fe doses were compared to the negative control (* p<0.05, ** p<0.01).

Fig. 7 : HEK-Blue hTLR4 cells were stimulated in triplicate with LPS (5 EU ml-1) and/or increasing magnesium chloride concentrations (0.1 μΜ to 1 mM)for20 hrs.

Fig. 8: Viability of HEK-Blue hTLR4 cell line measured with the CellTiter-Glo 2.0 Assay after cells were stimulated in triplicated for 20 hrs with LPS (5EU ml-1) and/or increasing iron concentrations.

-5EXAMPLES

METHODS:

1.1 Sample collection

Samples were collected from six monitoring sites within Antwerp (Belgium), a city of approximately half a million inhabitants and accommodating the second largest harbour in Europe. Antwerp provides an excellent European urban model due to its diverse environmental elements, such as heavily trafficked roads in densely-built locations, tram and train lines, urban green elements such as parks and tree linings, and an industrial harbour region. The city, including the harbour region, is also well monitored for its air quality by the Flanders Environment Agency (VMM). For 2015, the VMM reported the modelled city centre background for Antwerp, calculating annual pollutant levels ranging from 21-25 pg m'³ for PMw, 11-13 pg m'³ for PM2.5, 31-40 pg m'³for NO₂.

In this study, sampling occurred during the day (09:00 to 17:00) from July until September 2015 (n=42) with real-time air temperatures ranging from 18-33°C. All sampling sites were distributed spatially within Antwerp (not more than 15 km apart). To account for local emission sources, samples were categorized as urban traffic (A, B), urban green (C, D), and industrial (E, F) based on the expected local dominant pollutant and bioaerosol source. Typically, three samples were collected per day, of which each represented a different area (traffic, green and industrial) to account for confounding day-to-day variation. Samples were collected at a median height of 1.6 m using the Coriolis® p air sampler (Bertin Technologies, France), where air was drawn in at a flow rate of 300 L min^-1 (for 40 min; 12 m³ air), creating a vortex by which the particles were deposited into a pyrogen-free, polycarbonate cone containing 15 ml of ultra-pure water. The Coriolis sampler technology has been validated by the Health Protection Agency (HPA; Porton Down, UK) and ISO 14698-1 certified for biological/physical efficiency with a d50 < 0.5pm (i.e. at a particle diameter of 0.5 pm and above, the sampler efficiency is 50% or more). To test for contaminants, both the sampling water and filled cones were regularly tested in all analyses.

After collection, samples were transported on ice back to the laboratory. The volume of every sample was normalized to 15 ml with ultra-pure water to compensate for evaporation losses of the collection liquid during sampling. The samples were vortexed (10 sec), aliquoted accordingly for the various assays, and stored in glass vials at-20°C (typically within 12 hours after sampling) until analysis.

2.2 Endotoxin concentration

Recombinant Factor C assay. All samples were thawed only once and endotoxin quantification was determined in triplicate using the recombinant Factor C (rFC) assay according to manufacturer’s instructions (Lonza Walkersville Inc., MD, USA; lot 0000 416 097). For rFC assay analysis, a blank and a five-point standard curve (10, 5, 1, 0.1, 0.02 Endotoxin Units (EU) ml^-1) was set up in glass vials using endotoxin standards (Escherichia coli O55:B5 lot 0000 441 186; Lonza Walkersville Inc., MD, USA). The samples (neat or diluted) were vortexed (10 sec) and added to a 96-well plate heated to 37°C before the 100 pl mixture of enzyme, buffer, and fluorogenic substrate was added with a multichannel pipette. The plates were incubated at 37°C for 1 h and read (t=0 and t=60 min) in a fluorogenic microplate reader (MX Synergy, BioTek; Gen5 software) at excitation and emission wavelengths of 380 and 440 nm, respectively. Background fluorescence was subtracted, and the logarithmic change in fluorescence was plotted against the logarithmic endotoxin concentration over the range of 0.02 to 10 EU mF¹ (R²>0.98). The endotoxin concentration for a sample was calculated from the arithmetic mean of those dilutions that fell within range of the standard curve and expressed as EU per m³ of air based on the sampling conditions. Samples were analysed at a 1:2 dilution and a single lot (0000 475 024) of rFC was used for all analyses. Endotoxin distribution plots were constructed in GraphPad Prism v6.05 for Windows (GraphPad Software, La Jolla California USA).

TLR4 stimulation in HEK293 cells. HEK-Blue™ hTLR4 cells are stably transfected with human TLR4/CD14/MD2 (Invivogen). The cells were cultured in DMEM (Gibco, Life Technologies), supplemented with 10% FCS, normocin, and HEK-Blue Selection (Invivogen) and maintained at 37°C in a humidified incubator containing 5% CO₂. Cells (100 pl) were seeded into 96 well plates at a concentration of 0.5 x 10⁶ cells ml'¹ and incubated for 48 hours until approximately 90% confluent. Media was removed from adherent cells and replaced with 100 μΙ concentrated (x2) DMEM (with no FCS, normocin, or selective marker) which was diluted with 100 μΙ sample/control. A standard curve was generated by stimulating the cells with a six-point serial dilution (0.5 -75 EU ml'¹) of the same endotoxin standard as the rFC assay (Escherichia coli O55:B5, lot 0000 441 186; Lonza Walkersville Inc., MD, USA). After 20 h incubation, 50 μΙsupernatant was transferred in duplicate to a new 96 well plate, upon adding 100 μΙ of freshly prepared substrate solution (final concentration of 1 mg ml'¹ pNPP in 100 mM Tris-HCI, 100 mM NaCI, 5 mM MgCI2, pH 9.5 buffer). After a 20 min incubation at 23°C, the absorbance was measured at 405 nm. From the standard curve and sampling efficiency, the predicted endotoxin concentrations (EU m'³) could be calculated. Cell viability was determined by the CellTiter-Glo 2.0 Assay (Promega).

2.3 Particle count

Particle size distribution and count were determined by Coulter counter analysis (Beckman Coulter Inc.). Samples were diluted 1:4 in Isoton solution and 500 pl of sample analyte was analysed in duplicate using the Coulter Counter containing a 50 pm aperture (with counting limits of 0.5 - 50 pm). Using the Multisizer 3 software, insoluble particles were counted for the coarse fraction (aerodynamic diameters of 2.5 -10 pm), which were then calculated to be expressed as particles m'³.

2.4 Transition metal analysis

From each of the collected air samples, 2 ml was transferred into polypropylene vials and immediately acidified to 2% using highly purified nitric acid so as to prevent metal adsorption. Corresponding preparation and reagent blanks were always included. Samples were stored at 20°C until elemental analysis. High resolution sector field ICP-MS (Element XR, Thermofisher Scientific, Bremen, Germany) was used to determine elemental concentrations in the samples. The following metals were measured in the Low Resolution (LR), silver (Ag), cadmium (Cd), and lead (Pb), while aluminium (Al), chromium (Cr), manganese (Mn), iron (Fe), cobalt (Co), nickel (Ni), copper (Cu), and zinc (Zn) were measured in the medium resolution (MR). Arsenic (As), which is typically more problematic in matrices containing calcium and chloride, was measured in the highest resolution mode (HR). The three mass resolutions correspond to 300, 4000, and 10,000 as defined by the 10% valley, equivalent to 5% peak height. A NIST standard (SRM 1640a) was used as analytical quality control and the recoveries were consistently within 3% of the certified values for all the measured elements. Metal concentrations (ng m'³) of the blank samples (collection liquid in cone; n=5) were all below the detection limit except for very low traces of Ag (<0.013), Mn (<0.061), Cu (<0.086), and Zn (<0.068).

In addition, nine transition metals (As, Cd, Pb, Zn, Cu, As, Ni, Cd) from the Traffic 1 and Metal recycling plant locations were reported by the VMM during the sampling period. These samples were collected (24 h) using a SEQ 47/50 sequential gravimetric sampler (Leckel, Germany) with PM₁₀ inlet and 2.3 m³ h'¹ flow rate, after which ED-XRF analysis was performed (following the EN14902 reference).

2.5 Monitoring of pro-inflammatory response genes at mRNA level

The human monocyte cell line U937 (ATCC® CRL-1593.2™) was cultured in Roswell Park Memorial Institute (RPMI) medium (Gibco, Life Technologies, Grand Island, New York, USA) supplemented with 10% (v/v) heat-inactivated fetal bovine serum (FBS; Hyclone, GE Healthcare, Little Chalfont, UK), streptomycin (100 pg ml^-1) and penicillin (100 units ml^-1; Gibco), and maintained at 37°C in a humidified incubator containing 5% CO₂. Cells were seeded in 12well plates at a concentration of 5 x 10⁵ cells ml'¹ in complete growth medium with phorbol 12myristate 13-acetate (PMA, 100 ng ml'¹; Sigma Aldrich, St. Louis, Michigan, USA) to allow for

-8differentiation to an adherent macrophage-like stage for 48 h. Thereafter, PMA and nonadherent cells were removed by replacing the RPMI medium (without supplements) for a further incubation of 24 h. Medium was removed and cells were co-incubated for 3 h with a 10xRPMI media (Sigma Aldrich, St. Louis, Michigan, USA) diluted out with the supplements sodium bicarbonate (2 g I'¹), folic acid (1 mg I'¹), GlutaMAX (2 mM) and sample (at a 1:1 ratio), or a positive control (100 ng/well lipopolysaccharide from E. coli 0111:B4; Sigma Aldrich) or the negative control (unsupplemented RMPI). After co-incubation, total RNA was isolated with the RNeasy Mini kit (Qiagen, Hilden, Germany) following the manufacturer’s instructions and stored at -80°C. RNA integrity was verified on a bleach gel as described by Aranda et al., 2012. One pg of RNA (quantified with Qubit 3.0; Thermo Fisher, Waltham, Massachusetts, USA) was used for cDNA synthesis with the ReadyScript® cDNA Synthesis Mix (Sigma Aldrich). The expression of three selected pro-inflammatory markers [interleukin 8 (IL-8 orCXCL8) interleukin 1-beta (IL1 β) and tumor necrosis factor-alpha (TNFa)] and two reference genes (M=0.258, CV=0.09) was quantified by RT-qPCR on a StepOne Plus Real-Time PCR System (v.2.0; Applied Biosystems, Foster City, California, United States). Guanine nucleotide binding protein beta polypeptide 2like 1 (GNB2L1) and cytochrome c-1 (CYC1) were selected as reference genes, following the geNorm analysis in qbase⁺. All primers were tested for efficiencies between 90-110% and their sequences can be found in Table 1.

Table 1: Primer sequences of gene expression from U937 cells.

Gene	Forward primer (5’ - 3’)	Reverse primer (5’ - 3’)	Product
			(nt)
IL-8	TGGCAGCCTTCCTGATTTCT	TTAGCACTCCTTGGCAAAACTG	61
TNFa	TCTTCTCGAACCCCGAGTGA	CCTCTGATGGCACCACCAG	151
IL-1fi	TTGCTCAAGTGTCTGAAGCAGC	CAAGTCATCCTCATTGCCACTG	89
GNB2L1	CACTGTCCAGGATGAGAGCCA	CATACCTTGACCAGCTTGTCCC	111
CYC1	CATGTCCCAGATAGCCAAGGA	CTTGTGCCGCTTTATGGTGTAG	145

Every cDNA sample was used in duplicate, each 20 pL reaction consisting of Power SYBR® Green PCR Master Mix (Applied Biosystems), 0.15 μΜ of each primer, 25 ng of diluted cDNA, and nuclease-free water. The resulting data were analyzed using the qbase⁺ software package for calculating relative expression levels of the cytokines as compared to the reference genes and statistical analysis for comparison amongst groups (One-way ANOVA, corrected for multiple testing).

2.6 Statistical analyses

In GraphPad Prism v6.05 the logarithm-transformed data that were normally distributed were analyzed for significance by one-way ANOVA followed by the Tukey test to correct for multiple comparisons, unless otherwise stated. Welch’s t-test was used for unequal variance. P<0.05 was judged to be statistically significant.

RESULTS AND DISCUSSION

1.1 Atmospheric endotoxin concentration

First, we monitored the airborne endotoxin concentrations (Fig. 1) over the different urban land use classes (all within 15 km distance), categorized as green, traffic, and industrial, based on the expected dominant local emission source (data not shown). As measured by the rFC assay, endotoxin concentrations ranged from 0.65 to 11.72 EU rri³. Although greener urban areas have many microbial sources (from plants, soil, and animals), the green areas studied here did not show significantly higher concentrations of airborne endotoxins (geomeans (GM) 2.79 EU m^-3; 95% confidence interval 1.93-4.04) than the industrial (GM 1.92 EU m^-3; 95% confidence interval: 1.21-3.02) or the urban traffic locations (3.68 EU m^-3; 95% confidence interval 2.595.23). In fact, the traffic locations showed the highest GM (Fig 1).

3.2 Biological recognition of endotoxin through human TLR4/MD-2 complex

Samples were tested for their biological recognition of endotoxin to human TLR4 using the HEK293 hTLR4 reporter cell. A standard curve was generated from the same endotoxin stock as used in the rFC assay to predict the endotoxin units per ml (EU ml^-1). In terms of this TLR4 bio-assay, the biological recognition of endotoxin differed significantly between the land use classes. Furthermore, the traffic areas studied here showed the highest response. Moreover, the methods for endotoxin quantification differed significantly for the green locations, where recognition of endotoxin by TLR4 stimulation was significantly lower than the rFC-determined endotoxin concentrations (p=0.0023). Subsequently, the TLR4 bioassay showed the green areas to have the lowest concentrations of endotoxins, significantly lower than the traffic sites.

3.3 Relationship between the rFC assay and TLR4 recognition.

Although endotoxin has typically been quantified with the LAL assay, we choose the rFC assay to analyse these environmental samples since its simplified reaction pathway offers reduced likelihood of interference from other PM components, such as the false activation by yeast glucans. Despite this, the rFC assay still displayed higher endotoxin concentrations than the TLR4 assay for 34 out of 41 samples (Fig 2B), suggesting the difference between Factor-C based assays and the TLR4 bioassay to be attributed to more than yeast glucans as suggested by Peters et al., 2012. The relationship between these assays was investigated using linear

-10regression analysis and initially showed a moderate correlation (R²=0.55, n=41, p<0.001) (Fig 2A). However, when both land use class and the interaction effect between the latter and rFC measurements (i.e. change in gradients) were included in the model, the fit significantly improved (R²=0.73, n=41, p<0.0001). The land use class of the collected air sample therefore seemed to affect the reactivity of the quantified endotoxin to the human host receptor, with endotoxin from the traffic areas being most reactive and the green areas showing a generally low response ratio (Fig 2B).

While the HEK hTLR4 reporter assay offered a more accurate model for predicting the inflammatory potential of PM associated endotoxin - compared to the rFC assay which is used to quantify endotoxin based on its activation of Factor C, an immune glycoprotein of the horseshoe crab - the bioassay is not without its limitations. While TLR4 is not sensitive to fungal glucans (as with the LAL assay), fungal overgrowth of an environmental sample during coincubation is a problem for the cell line viability, and certain samples may thus need to be excluded. While this problem may be minimized by a shorter incubation period, from 20 h to 6 h, it is an important consideration to keep in mind for future studies.

3.4 Relation of endotoxin to particle concentration

To investigate the difference in TLR4 recognition of endotoxin from the urban land use classes, we turned our attention to the concentration and composition of particles from these environments. Since endotoxins are predominantly associated with the coarse (2.5-10 pm) PM fraction, we first investigated the concentration of these particles in the different land use classes using a Coulter counter, and secondly determined if the biological reactivity of endotoxin may be correlated to PM particle count.

The traffic areas showed a trend to contain the highest number of coarse particles per cubic meter of air (GM: 5.56 x 10⁴) compared to the industrial (GM: 4.83 x 10⁴) and urban green locations (GM: 3.87 x 10⁴), however, not significantly so (p=0.1899; data not shown). Furthermore, the biological recognition of endotoxin was only significantly correlated to the coarse particle count from the traffic locations (R²=0.51, n=17, p=0.0014) (data not shown). This correlation could not be significantly found for green and industrial locations.

3.5 Atmospheric transition metals concentrations

Transition metals are important atmospheric contaminants which may not only confound responses to airborne microbial endotoxins through oxidative stress, but transition metals such as nickel and cobalt are known to directly activate TLR4 (Oblak et al., 2015; Rachmawati et al., 2013). We subsequently quantified 12 major transition metals in each sample by ICP-MS and grouped the data per urban land use class described above. The concentrations of metals collected over 40 minutes with the Coriolis were found to be comparable with the 24 hour

-11averaged concentrations reported by the VMM for the corresponding days and locations (Traffic and the metal recycling plant; Table 2).

Table 2: A comparison of the median transition metal concentrations (ng m^-3), followed by the range (min - max) for the traffic 1 and metal recycling plant locations measured on the same days by the Coriolis (40 min) and Leckel SEQ 47/50 sampler from the VMM (24 hr average). Values in bold are significantly higher (two-tailed, paired t-test, p<0.05).

Traffic 1 (n= 9)	Metal recycling plant (n=6)
Coriolis	VMM	Coriolis	VMM
Mn	17.1	(5.93-31.5)	9.0	(5.2-17.6)	15.0	(5.54-34.8)	15.05	(11.2-22.1)
Cr	3.4	(0.5 - 5.07)	3.5	(2.1 - 10.6)	1.16	(0.55-4.12)	5.05	(2.0-10.2)
Pb	16.4	(7.87 - 109)	12.3	(4.3-23.1)	169	(32.1 - 15208)	122	(11.1 -932)
Zn	59.1	(29.8 - 132)	31.4	(15.5-122)	37.3	(23.7- 1071)	75.6	(15.4-159)
Cu	41.1	(18.9-82.5)	20.7	(9.9-38.3)	35.3	(11.9- 1032)	33.4	(5.0-183)
As	0.28	(BMQL-4)	BMQL	(BMQL-0.2)	6.40	(0.02-280)	7.55	(0.4-46.9)
Ni	3.35	(0.85 -28.3)	BMQL	(BMQL-2)	11.8	(1.40-67.6)	2.45	(0.4-9.8)
Cd	0.20	(0.03 - 1.55)	0.30	(0.1 -3.5)

From Fig 3, there are clear differences in metal concentrations between the traffic, urban green, and metal recycling locations. The two locations representing the urban green land use class showed similar and relatively low concentrations for all the corresponding transition metals (Table 3). In contrast, the traffic locations had significantly elevated concentrations of Fe, Cr, Zn, and Cu than the other land use classes (Kruskal-Wallis test with Dunn’s test for multiple comparisons). Furthermore, in comparison to Traffic site 1 (vehicular traffic), Traffic site 2 (which had vehicular and tram traffic) had significantly higher concentrations of Cr, Mn, and Fe, which are relevant indicators for railway abrasion. Lastly, the two industrial locations showed to be the most distinct from each other in terms of their atmospheric metal profiles. The metal recycling plant (non-ferrous based) displayed an entirely different profile with overall high and variable metal concentrations, dominated by cadmium (Cd), cobalt (Co), silver (Ag), lead (Pb), and arsenic (As) - metals commonly recycled at this plant and often measured to be in exceedance of the European target values by the VMM. Further information of the correlation of the twelve metals for the specific locations may be found Table 3.

-12Table 3: Median transition metal concentrations (ng m^-3), followed by the range (min - max) reached within the locations in Antwerp for the sampling period July - September 2015. Values in bold and grey represent the highest and lowest values measured, respectively, for the particular metal over the six locations.

WSF8C GREEN MDUSTRfAi.

Trams 1

Traffic 2

^ark Campus

(Wets! plant

Fa

874

(197-1545}

1SS2

(852-27830)

to

(54.3-455}

w

¢7.5-326)

IDS

(59.1 - 622}

482

(214-1533}

Ai

<338

{37.1-343)

218

(78..5 -.-:80:

$2 to

{18.9-203)

{30,7 - 223)

:52.5

{19,5 - 236)

304:

¢76.4-400)

Mrs-

17.1

ί &.93 - 3 US)

28.6

¢15)1-65.1}

3·ί-·'7:

U-4S )

•6M

¢1.33-15.7}

8.37

(3.1S - 47,3)

iS.G

¢5.34--34.8)

Or

(CSST/}

7.63

¢4.87-14.5}

Ctori

(0. ΰδ-1.42)

,D.S4

{PMQL 1.45}

0.76

5.13 -1.58}

Lie

tp.SS -

Zrt

SS.1

(29.8-132)

71.2

(30.2-91,5)

22,1

{9.74-41.0}

15.4

{7,36 59:3)

24.6

{12.0-66.9}

373

¢23.7-1071}

Cw

41.1

(18,9-82:5}

48.4

{23.3-73.1}

14:0::

¢.15-33.9}

i: .-

{2.49:· 17.5}

S.67

(4.S9 - 29.4!.

35,3

¢11,9-1032)

As

D.28

ail

{8MQL-0.50}

0.12

{8MQL 0.53}

0.04

ίοΜάΐ-ά.ο^

6.33

(0.02-280}

3.35

10 85 · 28:3)

.6,32

13.25 ’ 37.1J

2.31

{SMQi. - 6.62;

ΰ.63 ’417)·

:1.59

¢0.27 -12.6}

11.8

¢1,40-67.6}

1¾

1S.4

(7.87 109)

26.3

¢10.3-83.2}

14s)

(1.11-170)

531

{0.03 -16.8}

·:·.·$>:

.(1:03 - 65,7)

IfiS

(22,1 -15203)

Cd

{0.03 -1,55}

U.18

(0Ό7- 3./-2)

0.13

{0.04-1,14}

i-.G?

:{3MQ:,-B.5i

¢0.01 -1:53}

2,21

(0.28-62.3}

Ag

0.-28

{0,02-0,25}

0.10

¢.04 -0.721

0,08

{0,02-0,25}

0.04

{0,01 - 0:26)

0.0s

{0.01-0.27}:

0.42

¢0.08-9:61}

0.23

¢.11 0.995

0.36

¢0.18-0,55-

0.11

¢¢3-0.30}

Alfr<

¢.05-0.49}

i to. 12

¢0.05 - 0. 781

0.43

(0.28-5.44)

3.6 Immune response of collected urban air samples

After quantifying microbial endotoxins concentration (Fig. 1A), insoluble particles (data not shown) and transition metals (Fig. 3) in the closely-located urban land use classes, we profiled the relative inflammatory responses of the different samples. Messenger RNA levels of the major pro-inflammatory markers IL-8, IL-1 β, and TNFa were measured with real-time qPCR after a three hour co-incubation of a sample with the model human monocytic U937 cells. Monitoring of mRNA levels, instead of protein levels, was preferred here in order to detect early-response signals and to avoid cross-reaction with unknown contaminants in these environmental samples.

In general, the urban traffic and industrial areas showed a heightened and varied proinflammatory response for the three markers, in contrast to the urban green area samples which showed relatively low and stable responses (Fig 4). The two industrial locations differed clearly for the IL-1 β and TNFa mRNA responses, with the harbour location showing a significantly higher immune response for TNFa mRNA, while no difference was seen for IL-8. Despite both being industrial sites, this suggests the activation of different inflammatory pathways based on the unique composition of the samples, for which the metals profiles were previously seen to differ significantly (Fig 3). Consequently, for all the locations, IL-Ιβ and TNFa are better correlated to each other (R²= 0.84, n=42, p<0.0001) than to IL-8 (R²= 0.61; R²= 0.41, p<0.0001, respectively).

3.7 Associations between source composition and immune response

Consequently, we explored the correlations between the endotoxins, metals and inflammation markers. Typically, isolated LPS is very strongly correlated to the IL-8 mRNA response in the U937 cell model (data not shown). However, this is more complicated for environmental samples where many other components are capable of contributing to the IL-8 response and where different forms and thus potencies of LPS are present. Here, endotoxin concentrations as determined by the rFC assay were only weakly correlated with the IL-8 response (R²=0.2, p=0.003). Although, when taking the land use classes into account, we found a positive correlation originating for the industrial (R²=0.15, p=0.1775) and traffic locations (R²=0.35, p=0.017). Despite a wide range of endotoxin concentrations found in the green locations, the IL8 response remained relatively low (below 2.9 relative expression, Fig 5). However, endotoxin concentrations as measured by the rFC assay - specifically in the green locations - were observed (section 3.1) to be less capable of stimulating TLR4 (Fig 1 A), likely also explaining the lack of activation of the pro-inflammatory response compared to the other locations. Subsequently, by rather using the bioassay of TLR4 stimulation as a proxy for bioactive endotoxin concentration, the correlation significantly improved for the prediction of the IL-8 response (R² =0.38, p<0.0001), no longer with significant effects or interactions from the areas. In fact, of the measured factors, IL-8 was best correlated to the TLR4 bioassay, in addition to Mn, Fe and Cr (r=0.64, 0.63, 0.57; p<0.0001).

Although nickel and cobalt have previously been shown to directly activate hTLR4/MD-2, their effect in this study was negligible in comparison to the surprisingly strong correlation of iron with the stimulation of TLR4 in the HEK293 cell line (R²=0.50; n=41; p<0.0001), with no significant interaction effect of the considered monitoring locations/land use classes. Subsequently, we found that by adding Fe concentrations together with rFC assay as factors in the linear regression model predicting TLR4 stimulation (HEK assay), the model was significantly improved (R²=0.77, n=41; p<0.0001).

The IL-1 β response was best correlated with Mn (r=0.55; p=0.0002), especially in the traffic locations - Traffic 1 (r=0.88, p<0.0001) and Traffic 2 (r=0.94, p<0.0001) - and also the TLR4 bioassay (r=0.52, p=0.0005). Furthermore, the metals Mn, Fe and Cr were often very strongly correlated with each other (data not shown). Particle concentration was also best correlated to Mn for the various locations (data not shown). Lastly, the TNFa response was better associated to the TLR4 bioassay (r=0.5, p=0.0007).

3.8 Influence of iron on TLR4 stimulation

Although iron may originate from different sources in the various land use classes, location had no significant effect on the correlation of iron with the stimulation of TLR4. However compelling this association, studies have repeatedly shown that iron is not capable of independently stimulating the TLR4/MD2 complex, unlike nickel and cobalt (Oblak et al., 2015; Rachmawati et al., 2013). However, the ability of Fe to act synergistically with LPS to stimulate the TLR4/MD2 complex has not yet been investigated. In this study, HEK-Blue hTLR4 cells treated with 5 EU ml^-1 LPS and increasing concentrations of FeCI₂ showed a significant increase in hTLR4/MD2 stimulation at iron concentrations of 0.1 and 0.2 mM (60, 64%), while the iron itself (in absence of LPS) did not stimulate a response significantly higher than the negative control (Fig 6). Dosing LPS with another divalent cation, MgCI₂ (0.1 pM -1mM), showed no significant effect on LPS stimulation from the negative control (Fig. 7). Cell viability was tested with the CellTiter-Glo 2.0 assay and a drop in viability was observed when the iron concentrations (in the absence and presence of LPS) exceeded 0.5 mM (Fig. 8), as similarly seen in the study of Rachmawati etal., 2013. A corresponding drop in TLR4 stimulation was observed beyond 0.5 mM FeCI₂ with LPS. Fortunately, the environmental samples did not reach these concentrations. Furthermore, since alkaline phosphatases depend on Zn²⁺ and Mg²⁺ divalent cations for their enzymatic activity, we established that the increasing concentrations of Fe²⁺ had no significant effect on the activity of the secreted embryonic alkaline phosphatase (SEAP) reporter enzyme in this study. Interestingly in a study by Becker et al., 2005, iron concentrations quantified from coarse PM fractions correlated well with the IL-6 release in alveolar macrophages (from normal individuals through bronchial brushings), while no correlation was found in ultrafine particles. Although endotoxin was not quantified in the study, it is interesting to note that the inflammatory potential of iron was best correlated in the coarse PM fraction where endotoxin is predominantly found.

CONCLUSION

In this study, we not only quantified the concentration of airborne endotoxin for different urban land use classes, but we also determined the ability of endotoxin - as part of PM - to be recognised by host cells in vitro through TLR4. This TLR4 interaction was found to be a better indicator for the IL-8 response than endotoxin concentrations measured by the rFC assay. In the urban locations studied here, the traffic locations were found to be significantly higher in bioactive endotoxin than the industrial and green locations, despite all areas having similar rFCdetermined endotoxin concentrations.

-15We subsequently turned our attention to the composition of PM to explain the disparity between the rFC assay and TLR4 stimulation for the different urban land use classes. We initially hypothesized that Co and Ni could improve the regression model since they were previously shown to directly activate the TLR4/MD2 complex (Oblak et al., 2015; Rachmawati et al., 2013;

Schmidt et al., 2010). Surprisingly, the effects of Co and Ni were negligible in comparison to that of Fe, improving the model significantly by explaining 77% of the variation of the TLR4 stimulation and no longer dependent on the urban land use classes. The effect of iron proved to be more than a correlation, since dosing endotoxin with iron chloride led to an increase in TLR4 stimulation in the HEK293 cell line while iron itself was unable to stimulate a response.

Iron is typically one of the most abundant transition metals found in urban PM, originating from crustal matter, iron industries, and traffic and railway related sources. In terms of health effects, iron is a redox-active metal capable of causing oxidative stress by generating reactive oxygen species (ROS) through Fenton reactions. This study showed another way in which iron may 15 contribute to the inflammatory response and confirms the importance of PM composition from different urban land use classes on their resulting inflammatory potential.

-16REFERENCES

Aranda PS, LaJoie DM, Jorcyk CL: Bleach gel: a simple agarose gel for analyzing RNA quality. Electrophoresis 2012, 33:366-369.

Becker S, Dailey LA, Soukup JM, Grambow SC, Devlin RB, Huang Y-CT: Seasonal variations in air pollution particle-induced inflammatory mediator release and oxidative stress. Environmental health perspectives 2005, 113:1032.

Oblak A, Pohar J, Jerala R: MD-2 determinants of nickel and cobalt-mediated activation of human TLR4. PloS One 2015, 10:e0120583.

Peters M, Fritz P, Bufe A: A bioassay for determination of lipopolysaccharide in environmental samples. Innate immunity 2012, 18:694-699.

Rachmawati D, Bontkes HJ, Verstege Ml, Muris J, von Blomberg BME, Scheper RJ, van Hoogstraten IM: Transition metal sensing by Toll-like receptor-4: next to nickel, cobalt and palladium are potent human dendritic cell stimulators. Contact Dermatitis 2013, 68:331-338.

Schmidt M, Raghavan B, Muller V, Vogl T, Fejer G, Tchaptchet S, Keck S, Kalis C, Nielsen PJ, Galanos C: Crucial role for human Toll-like receptor 4 in the development of contact allergy to nickel. Nature immunology 2010, 11:814-819.

Claims (8)

1. A composition comprising iron (ferric or ferrous ions) and at least one vaccine adjuvant.

2. The composition as defined in claim 1; wherein said at least one vaccine adjuvant is selected from the list comprising: TLR4 agonists, LPS (lipopolysaccharide), Monophosphoryl lipid A, TLR2 agonists (Pam2Cys or Pam3Cys), TLR3 agonists, Poly l:C, TLR5 agonists (flagellin-Ag complexes), TLR7/8 agonists (imiquimod, resiquimod), TLR9 agonists (CpG DNA), a PRR (Pattern Recognition Receptor) ligand, a NOD-like (nucleotide oligomerization domain) receptor ligand, a lectin receptor, alum, MF95, saponin, Quil-Aand QS21.

3. The composition as defined in anyone of claims 1 or 2; wherein said iron (ferric or ferrous ions) is selected from the list comprising: FeCI2, FeCI3, Fe3O4, Fe2O3, FePO4, FeSO4, and Fe(OH)2.

4. A composition comprising FeCI2 and Monophosphoryl lipid A.

5. A composition as defined in anyone of claims 1 to 4 for use in human or veterinary medicine.

6. Use of a composition as defined in anyone of claims 1 to 4 as an adjuvant in vaccination or immunomodulation.

7. A vaccine or immunomodulatory product comprising an antigen and a composition as defined in anyone of claims 1 to 4.

8. Use of iron (ferric or ferrous ions) and a vaccine adjuvant in the preparation of a vaccine or immunomodulatory product.