Research article

Effects on antigen-presenting cells of short-term interaction with the human host defence peptide β-defensin 2

Francesca Morgera, Sabrina Pacor, Luisa Creatti, Nikolinka Antcheva, Lisa Vaccari, Alessandro Tossi


β-Defensins are antimicrobial peptides that exert their host-defence functions at the interface between the host and microbial biota. They display a direct, salt- and medium-sensitive cidal activity, in vitro, against a broad spectrum of bacteria and fungi, and there is increasing evidence that they also play a role in alerting and enhancing cellular components of innate and adaptive immunity. Their interaction with biological membranes plays a central role in both of these types of activities. In the present study, we have investigated the interaction of fluorescently labelled hBD2 (human β-defensin 2) with monocytes, macrophages and iDCs (immature dendritic cells), observing a differential capacity to be rapidly internalized into these cells. Complementary microscopy techniques [TEM (transmission electron microscopy), optical microscopy and IR microspectroscopy] were used to explore the functional and biological implications of these interactions on iDCs. Short-term exposure to the peptide resulted in significant alterations in membrane composition and re-organization of the endomembrane system, with the induction of degranulation. These events may be associated with the antigen-presenting activities or the chemotaxis of iDCs, which appears to occur via both CCR6 (CC chemokine receptor 6)-dependent and -independent mechanisms.

  • β-defensin
  • chemotaxis
  • host defence
  • infrared microspectroscopy (IRMS)
  • innate immunity
  • membrane interaction


Mammalian BDs (β-defensins) are small cationic HDPs (host defence peptides), with a characteristic β-sheet fold, that exert their protective functions at the host–environment interface [1]. They were initially proposed to act as innate immune system effectors by a direct antibiotic-like activity [2], but more recently a complex, sophisticated and possibly more relevant role in infection has emerged, as signal molecules alerting and/or activating cellular components of both innate and adaptive immunity [3,4].

Humans possess two related defensin families with similar folds but with differing disulfide bridging patterns: α-defensins are expressed in neutrophils, certain macrophages and Paneth cells, whereas BDs are mostly produced by epithelial cells lining different organs, such as the epidermis, bronchial tree and genitourinary tract [2]. Although some are produced constitutively, more often they are induced by microbial products or pro-inflammatory cytokines. They generally display a salt-sensitive antimicrobial activity in vitro against a broad spectrum of bacteria, fungi and some enveloped viruses [5]. In vivo, they are likely to be secreted into the phagocytic vacuoles of phagocytes and/or on epithelial surfaces and mucosa, where the ionic strength is relatively low, reaching sufficiently high concentrations for antimicrobial activity.

Defensins are not cytotoxic to host cells at antimicrobial concentrations, and this selectivity may depend on the different compositions of cellular membranes; in the case of bacteria, the greater abundance of negatively charged phospholipids and the lack of cholesterol may favour the membrane as a target. The mechanism by which bacteria are inactivated by defensins is, however, not well understood. Conversely, human BDs alert cellular components of both the innate and adaptive immune systems, and enhance the activities without membrane damage [1]. hBD2 (human BD2) is the best characterized in this respect: it induces the activation and degranulation of mast cells with the release of histamine and prostaglandins, thus favouring the recruitment of neutrophils to the inflammatory site. Moreover, it displays a chemotactic activity for CD8+ T-cells and iDCs [immature DCs (dendritic cells)], as well as TNFα (tumour necrosis factor α)-activated neutrophils and epithelial cells, and triggers a robust production of cytokines from PBMCs (peripheral blood mononuclear cells) [611]. hBD2 also inhibits the classical pathway of the complement system, suggesting a protective role against its uncontrolled activation [12].

The overall picture that is emerging on the role of defensins in vivo is that they act in an articulated and quite complex manner by directly contrasting invading micro-organisms, by recruiting PMNs (polymorphonuclear cells), T-cells and iDCs to the site of infection, and as enhancers or regulators of inflammatory and/or healing responses. It suggests that this class of HDPs may lead to promising compounds for the future development of multifunctional anti-infective agents and/or vaccine adjuvants. Many of the reported biological functions are, however, based on phenomenological observations in vitro, and important questions remain to be answered regarding the underlying mechanisms before the in vivo relevance in host defence can be assessed.

In the present study, several different techniques have been used to investigate the consequences of short-term (30–60 min) interaction between hBD2 and APCs (antigen-presenting cells). Flow cytometry and confocal microscopy were used to determine whether fluorescently labelled hBD2 localized exclusively at the membrane level or was also localized intracellularly in macrophages, monocytes or iDCs. Focusing then on its interaction with iDCs, the functional and biological implications were explored using complementary microscopy techniques. In particular, SR (synchrotron radiation)-IRMS (IR microspectroscopy) was used to observe biochemical changes in macromolecular constituents of hBD2-treated cells. SR-IRMS is a label-free and non-destructive microanalytical tool for the characterization of molecular modes of vibrations, and was employed for typifying cellular samples through correlation with their morphological features. The high brilliance of SR was exploited to achieve cellular and subcellular spatial resolution [13]. Analysis of band shape, position and intensity revealed subtle biochemical changes relating to membrane composition and order [14], or protein and nucleic acid structure [15,16]. Distinct subsets of cells with different responsiveness to peptide treatment were clearly identified, in agreement with flow cytometry, and suggested that the effects of exposure to hBD2 included potential variations in cellular protein content and lipid/phospholipid distributions. These effects were also considered in relation to known activities of the peptide, such as the chemotactic effect on iDCs.


Peptide synthesis and characterization

hBD2 was synthesized by the Fmoc (fluoren-9-ylmethoxycarbonyl) solid-phase method using a 2-chlorotritylchloride resin (substitution ≤0.2 mmol/g) and, after cleavage, the crude peptide was of sufficiently good quality to be directly oxidatively folded for 24 h in N2-saturated aqueous buffer (0.1 M ammonium acetate, pH 7.8, containing 0.5 M guanidinium HCl and 2 mM EDTA) in the presence of a 100-fold excess of cysteine and a 10-fold excess of cystine with respect to the peptide (described in more detail in the Supplementary Material and methods section at Monitoring by analytical RP-HPLC (reverse-phase HPLC) and ESI–MS (electrospray ionization MS) confirmed completion, and final purification was carried out on a preparative C18 column. CF [5(6)-carboxyfluorescein] labelling at the free N-terminal position of hBD2 (CF–hBD2) was carried out on resin-bound fully side-chain-protected peptide prior to cleavage and folding, whereas that with BY {BODIPY® [boron dipyrromethene (4,4-difluoro-4-bora-3a,4a-diaza-s-indacene)]} (BY–hBD2) was carried out on folded peptide, as this molecule is not stable to cleavage conditions (see the Supplementary Materials and methods section). The complete oxidation and correct labelling was verified by ESI–MS of the peptide, whereas the correct folding was partly confirmed by ESI–MS of the proteolytic digests. Unlabelled peptide concentrations were determined based on the molar absorption coefficients (ϵ280) of tyrosine and cysteine residues, whereas labelled peptides concentrations were based on the absorption coefficients of CF or BY respectively. Peptide stock solutions (~100 μM in Milli-Q water) were determined to be LPS (lipopolysaccharide)-free using the LAL (Limulus amebocyte lysate) assay (Bio-Whittaker).

Cellular uptake of hBD2

Cells were resuspended in complete medium [RPMI 1640 supplemented with 10% NHS (normal human serum), 2 mM L-glutamine, 25 mM Hepes, 100 units/ml penicillin and 100 μg/ml streptomycin] at 1×106 cells/ml and incubated at 37 °C in the presence of CF–hBD2 or BY–hBD2 for various time periods before cytofluorimetric analysis. The fluorescent signal from total bound peptide (on the cellular membrane and/or internalized into cells) was acquired first. Treated cells were then incubated with 0.1% TB (Trypan Blue), which quenched the peptide bound to the external membrane surface only, as it is excluded from the interior of intact cells. Permeabilization of cells by hBD2 was monitored by the PI (propidium iodide) exclusion assay and the PI+ cell population, if any, was excluded from subsequent analysis. Flow cytometric analyses used a 488-nm (argon ion laser) excitation on a Cytomics FC 500 instrument (Beckman Coulter). A minimum of 10000 events/sample were acquired and histograms were analysed using FCS Express software (De Novo Software). The fluorescence of each sample was determined as the MFI (mean fluorescence channel intensity). All experiments were carried out in triplicate and were repeated at least three times. Results are expressed as means±S.E.M. ANOVA was carried out with GraphPad InStat software.

Generation of MDDCs (monocyte-derived DCs) from peripheral blood

Monocytes were prepared from buffy coats obtained from 20 healthy donors, who provided informed consent, and were isolated by Histopaque® density gradient centrifugation. Plastic-adherent cells (monocytes) were maintained at 37 °C in a humidified atmosphere of 5% CO2 in complete medium. iDCs were induced with 25 ng/ml GM-CSF (granulocyte/macrophage colony-stimulating factor) and 0–44 ng/ml IL-4 (interleukin-4) for 7 days [17]. They were then phenotypically characterized using antibodies for HLA-DR, CD11c, DC-SIGN (DC-cell-specific intracellular adhesion molecule-3 grabbing non-integrin), CD83, CD1b, CD14 and CCR6 (CC chemokine receptor 6) as markers to confirm differentiation, as described in detail in the Supplementary Materials and methods section. Although HLA-DR, CD11c and DC-SIGN were persistently highly expressed (80–90% positive in all cell preparations), the percentage of cells positive for membrane-exposed CCR6 was quite variable in different preparations (ranging from 0 to 70%), despite the fact that cytosolic CCR6, as revealed by treatment of cells with ethanol and saponin, was persistently highly expressed (>90% positive cells in all preparations). This indicates that, although the induction conditions were appropriate, a considerable variability occurred among different donors, as also reported in the literature [18]. By altering the concentration of IL-4 in the culture medium it was possible to modulate receptor exposition somewhat [19].

Cytotoxicity and effect on protein content

To take cytotoxic effects of hBD2 into account when assessing its biological activities, monocytes were incubated with increasing peptide concentrations [1–16 μM in sodium phosphate buffer (10 mM Na2HPO4/NaH2PO4, pH 7.4) with 0–150 mM NaCl] for 30 min at 37 °C, washed, resuspended in PBS and treated with 10 μl of PI solution (0.5 mg/ml in PBS) immediately prior flow cytometric analysis.

To assess the effect of hBD2 on protein content, treated and untreated iDCs were diluted to 5×105 cells/ml in PBS, fixed with ice-cold 70% (w/v) ethanol and stored at 4 °C overnight. Cells were washed twice in PBS, centrifuged at 400 g at 4 °C, before incubation with FITC (0.05 μg/ml) in PBS. Measurements were carried out by flow cytometry as described above.

TEM (transmission electron microscopy) analysis of iDC ultrastructure

After centrifuging, the cell pellet was fixed in 3% (w/v) glutaraldehyde in 0.1 M cacodylate buffer (pH 7.3) at 37 °C for 2–3 h. The cells were post-fixed for 1 h at 4 °C with 1% (w/v) osmium tetroxide in 0.1 M cacodylate buffer, dehydrated in series of ethanol concentrations and then embedded in epoxy resin. Ultra-thin sections were examined after post-staining with 5% (w/v) uranyl acetate in 30% (v/v) ethanol for 20 min and Reynolds lead citrate for 5 min on a Philips EM 208 transmission electron microscope.


Single whole-cell and intracellular analyses were carried out at the IR beamline SISSI (Synchrotron Infrared Source for Spectroscopy and Imaging) [20] at the Elettra Synchrotron Light Laboratory in Trieste, Italy. hBD2-treated and untreated iDCs were deposited on calcium fluoride IR-transparent windows and cell adhesion was obtained with a 30 min incubation at 37 °C. Cells were then washed with PBS and fixed in 4% (w/v) formalin in PBS for 20 min at room temperature (20 °C), a type of fixation that does not interfere significantly with IR absorbances and maintains fundamental biochemical and structural cellular aspects [21]. Washed samples were dried at an ambient temperature and then stored under vacuum until FTIR (Fourier-transform IR) measurements. Spectral data were collected in transmission mode (×15 condenser objective) using a Bruker Vertex 70 interferometer coupled with the Hyperion 3000 Visible-IR microscope, equipped with a N2-cooled mercury/cadmium/telluride detector.

Single cells were selected visually and whole-cell IR spectra were collected with at a 50 μm×50 μm spatial resolution (minimum of 30 cells/sample), in order to have information on the general state and homogeneity of the sample. Each spectrum was the average of 256 scans with 4 cm−1 spectral resolution, and were processed and analysed with OPUS 6.5 software (Bruker Optics), applying a compensation algorithm correcting for atmospheric water vapour and CO2 spectral contributions. Cluster analysis was then applied in the 3600–1100 cm−1, 3020–2800 cm−1 and 1760–1475 cm−1 spectral ranges (Euclidean distances, Ward's classification algorithm), by processing vector-normalized first derivatives of FTIR cell spectra to enhance spectral band resolution and minimize baseline variations (Savitzky–Golay algorithm, nine smoothing points). Selected iDCs were subsequently subjected to higher-resolution IR mapping, with a 5 μm×5 μm spatial resolution using a grid that matched the whole-cell area and collecting an FTIR spectrum at each grid point, achieving a good compromise between S/N (signal-to-noise) in the range of interest (3600–1000 cm−1) and adequate spatial resolution for resolving biomolecular macro-domains. A background spectrum was recorded every five or six map points in order to correct for the SR current decay during the course of the experiment. The maps were analysed using the univariate technique of functional group mapping. The integrated bands used to generate maps were the Amide I (1707–1590 cm−1) for proteins, the carbonyl ester band (1760–1720 cm−1) for phospholipids and the acyl chain bands (3000–2800 cm−1) for lipids in general.

Chemotaxis assays

iDC migration was assessed in vitro using a Transwell chamber with 8-μm pore polycarbonate filters (Corning-Costar). Cells (1×106 cell/ml) were stained with FastDil dye (10 μM) for 20 min at 37 °C, thoroughly washed and resuspended in chemotaxis medium [RPMI 1640 containing 1% (w/v) BSA] and a 100 μl of suspension was added to the upper compartment. Peptides (1 μM) or positive control [12.5 nM recombinant human MIP-3α (macrophage inflammatory protein-3α); Chemicon International] were then placed in the lower compartment. After incubation for 90 min, both migrated and transmigrating cells were quantified fluorimetrically at 530 nm on a Packard FluoroCount instrument. Results are expressed as a percentage with respect to spontaneous motility of cells (migration+transmigration) in chemotaxis medium without chemokines. In some experiments, cells were pre-treated with an anti-CCR6 antibody for 30 min to specifically inhibit receptor-mediated chemotaxis.

Confocal microscopy

APCs were allowed to adhere on to glass slides and were then treated with CF–hBD2 (1 μM) for 60 min; after washing, the cells were fixed with 3% (w/v) paraformaldehyde at room temperature for 20 min. Fixed cells were than incubated for 5 min with 0.1 M glycine and 0.02% sodium azide in PBS. Permeabilization was performed with 0.01% Triton X-100 for 5 min, and actin was counter-stained with 3.8 μM phalloidin–TRITC (tetramethylrhodamine β-isothiocyanate) (Sigma–Aldrich) for 15 min. After washing, cells were mounted with Mowiol (4–88) (Polysciences) and kept in the dark at −20 °C until microscope analysis. Cells were examined with a Nikon Eclipse C1si instrument equipped with two detectors at a standard confocal detection unit, and images were elaborated using EZ-C1 software (version 3.30).


Prior to performing analyses of hBD2 binding to and internalization into APCs, we verified that it was not cytotoxic to cells of the mononuclear/phagocytic lineage (CD14+) at the concentration used by monitoring membrane permeabilization to PI and vitality as assessed using the MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium bromide] test. No significant toxicity was observed up to 4 μM peptide, under any of the medium conditions used, for exposure times ranging from 0.5 to 72 h (results not shown).

Peptide binding and internalization into APCs

MDMs (macrophages derived from blood circulating monocytes) were incubated with 1 μM CF–hBD2 at 37 °C for increasing periods of time up to 24 h and were analysed by flow cytometry to confirm the ability of hBD2 to bind to the cellular membranes. MFI of cells rapidly increased, indicating that hBD2 bound rapidly to the cell surface, with more than 80% of cells having bound CF–hBD2 after 10 min and all cells by 60 min (Figure 1A), and total fluorescence remained fairly constant up to 24 h. PMA-derived macrophages behaved in a similar manner (results not shown).

Figure 1 Peptide binding and internalization into monocytes, MDMs and iDCs upon short-term exposure to hBD2

(A) Different cell types incubated with 1 μM CF–hBD2 at 37 °C for 60 min with (open bars) or without (closed bars) TB quenching. Values represent the means±S.E.M. of three independent experiments. MFI from autofluorescence was less than five for all cell types. (B) A representative confocal fluorescence microscopy image of a macrophage cell treated with CF-hBD2 for 60 min. Cortical actin (stained with phalloidin–TRITC, red) and CF-labelled peptide (green) are shown. (C) A flow cytometric dot plot displaying forward scatter (FS) and side scatter (SS) parameters for iDCs. The R1 and R2 subpopulations are gated separately as indicated by the boxes. (D) Differential interaction of CF–hBD2 with the two cellular subpopulations, where the unshaded peak (MFI ≈101) is the R2-gated population and the shaded peak (MFI ≈5×102) is the R1-gated population. (E) Protein content was determined by measuring FITC fluorescence in untreated and hBD2-treated R1 and R2 subpopulations.

Treating samples with 0.1% of the impermeant quencher TB immediately prior to flow cytofluorimetric analysis allowed selective quenching of the peptide externally bound to the surface and to confirm the rapid internalization of CF–hBD2 into macrophages (compare open bars with closed bars in Figure 1A). At 60 min, over 95% of the cells were positive for the peptide, with 60% of the signal persisting after TB quenching. Confocal microscopy of the macrophages confirmed that, at 60 min, most of the peptide fluorescence was in the cytoplasm and distributed in a punctuate manner (Figure 1B).

Blood-derived monocytes also bound CF–hBD2, but to a lesser extent than macrophages (Figure 1A). Overall, a 5–6- fold increase in fluorescence and 40% of internalized peptide was observed for monocytes, against an 18–20-fold increase in fluorescence and 60% of internalized peptide observed for macrophages. iDCs showed a considerable capacity to bind peptide (Figure 1A, closed bar), with a very strong (40–50-fold) increase in fluorescence immediately after contact with 1 μM CF–hBD2. Fluorescence due to internalized peptide at 60 min exposure was approximately 35% of total, so that overall a comparable amount of peptide was internalized as with macrophages (Figure 1A, open bars).

Increasing the concentration of CF–hBD2 resulted in a linear increase in the accumulation of the peptide into macrophages, as indicated by a highly significant correlation between concentration and MFI (seven points; R2=0.99, P<0.0001). Furthermore, experiments were repeated with hBD2 linked to a different uncharged fluorophore (BY), without any discernable differences in behaviour, indicating that the presence of the fluorophore did not greatly affect binding or internalization (results not shown).

Flow cytometric and TEM analyses of the hBD2 interaction with iDCs

The effects of hBD2 interaction with iDCs were investigated further by correlating data from flow cytometry with TEM and optical and FTIR microspectroscopies (see below) under standardized conditions (1 μM peptide for 30–60 min of exposure in complete medium). iDCs were phenotypically characterized for specific co-stimulatory molecules and chemokine receptors in order to confirm the cell differentiation into immature (CCR6+) dendritic cells (DC-SIGN+) (Supplementary Figure S2 at Flow cytometric analyses indicated the presence of two subpopulations, R1 and R2 (Figure 1C), 70±10% and 30±10% of the total respectively. R1 cells were morphologically more heterogeneous, with higher SS (side scatter) and FS (forward scatter) values, indicating a greater size and internal complexity than R2, which probably consisted of viable but incompletely differentiated cells. Phenotypical analysis confirmed a lower content of iDC markers for R2 (Supplementary Figure S2), which also showed a lower capacity to internalize labelled hBD2 (Figure 1D). Naïve and hBD2-exposed cells (unlabelled peptide) of both subpopulations were treated with FITC to monitor variations in protein content in terms of variations in MFI. A statistically significant decrease (10±5%; P<0.001) in total protein was observed only for the R1 subpopulation after peptide treatment with respect to untreated controls (Figure 1E).

Control and treated iDC samples were then investigated by TEM (Figure 2). Their morphology resembles that of CD11c+ DCs, characterized by microvillous projections of the plasma membrane, multi-lobulated nuclei and abundant vacuoles; unexposed cells were also characterized by the presence of numerous cytoplasmic vesicles (Figure 2A) of approximately 200 nm in diameter, with a dark electron-dense appearance, tightly packed at the level of the plasma membrane and also distributed in the cytoplasm. On exposure to hBD2, these appeared to release their content, while maintaining their size and integrity (Figure 2B). At a higher resolution, vesicle membranes in peptide-treated cells presented discontinuities of 10–20 nm, which are consistent with the formation of some kind of large pore or channel.

Figure 2 TEM images of hBD2-treated iDCs

TEM images for untreated (A) and peptide-treated (B) cells are shown at increasing resolution [×1800 (1.8K×), ×11000 (11K×) and ×56000 (56K×) respectively], highlighting the presence of electron-dense vesicles in the cytoplasm of control iDCs (A, panel 1–3) and less electron-dense vesicles upon peptide treatment (B, panel 1–3). Vesicle size was approximately 200 nm. These images were selected from a large set of electron scans showing similar features.


The non-invasive SR-IRMS method was used to detect variations in functional groups present in cellular macromolecules without the use of exogenous probes. iDCs, either untreated or treated with hBD2 for 30–60 min, were fixed on IR-transparent calcium fluoride windows and first analysed by acquiring spectra of single whole cells. Up to 60 cells from each sample were analysed in order to gain comprehensive information on a statistically relevant number of samples. The vibrational pattern of these specimens is rather simpler than one would expect, considering the number of contributing biological molecules [22]. However, as can be seen from a representative IR spectrum from an iDC (Figure 3), key absorptions can be specifically assigned.

Figure 3 Representative IR spectrum from iDCs and band assignment

The most important cellular vibrational bands are indicated by arrows and assigned (ν, stretching modes; as, asymmetric; and s, symmetric; [22]). The position of each absorption band and primary cell constituents to which the absorption modes are conventionally associated are: ν(C-H) at ~3010 cm−1, olefins from unsaturated lipid acyl chains; νas,s(C-H) at ~2960 and ~2875 cm−1, methyl from saturated lipid acyl chains; νas,s(C-H) at ~2920 and ~2852 cm−1, methylene from saturated lipid acyl chains; ν(C=O) at ~1737 cm−1, esters from phospholipids; Amide I [80% ν(C=O), 10% ν(C-N) and 10% N-H deformation] at ~1655 cm−1, mainly from protein peptide bonds. Protein side chains and carbonyl stretching of nucleic acids may also occur in this region; Amide II [40% (C=O) and 60% N-H deformation] at ~1543 cm−1, protein main chain; νas(P=O) at ~1240 cm−1, phosphate group from nucleic acids; νs(P=O) at ~1088 cm−1, phosphate group from nucleic acids; band from 1300 to 900 cm−1, complex network of vibrations from carbohydrates [ν(C-O-C), (C-O-P), (C-O), (P-O-P)]. The attribution considers both the spectroscopic variables and the relative mass ratio of the most fundamental constituents of a eukaryotic cell: proteins (50% of the total cellular dry mass), nucleic acids (15%), carbohydrates (15%), lipids (10%) and minor constituents (10%) [35].

Whole-cell spectra were clustered into groups using a multivariate statistical algorithm that maximized intra-class similarities and inter-class differences [23], performed on the first derivatives of spectra in order to enhance spectral resolution and minimize baseline variations. This was applied over the entire IR spectrum, from 3600 to 1100 cm−1, for both control (Ct) and treated (T) cells. It fully succeeded in distinguishing two subpopulations, which on visual inspection correlated with substantial morphological differences (Figure 4). The minor population (Ct-R2; ~30%) appeared smaller, with a lack of pseudopodia and a heterochromatic nucleus, consistent with the R2 population observed in flow cytometric experiments, whereas the main population (Ct-R1; ~70%) was formed by larger cells, with an extended cytoplasm and paler euchromatic nucleus, consistent with the R1 subpopulation. The first derivatives of spectra for each of the R1 and R2 subpopulations differed principally in the bands centred at 1713 and 1600 cm−1 (Figure 4B) respectively, associated with carbonyl stretching vibrations of nucleic acids and to variation in DNA base-pairing [16].

Figure 4 FTIR single-cell analysis of untreated iDCs

(A) Dendogram resulting from cluster analysis (Euclidean distances, Ward's algorithm) on the first derivatives of spectra of untreated iDCs (Ct, control) performed in the 1755–1475 cm−1 range, showing the discrimination between R1 (Ct-R1) and R2 (Ct-R2) cells. (B) The averaged first derivative of spectra for Ct-R1 (black line) and Ct-R2 (grey line). The spectral line thickness is proportional to the S.D. from the mean.

Peptide-treated cells also consisted of two subpopulations with the same relative abundances (T-R2 and T-R1; results not shown), but while the treated R2 population did not cluster separately with untreated R2 cells when applying the classification algorithm over the two populations taken together (Figure 5A), the R1 population did (T-R1 and CT-R1 clusters were well separated in the dendrogram).

Figure 5 FTIR single-cell analysis of untreated and treated iDCs

(A) Dendrogram resulting from cluster analysis (Euclidean distances, Ward's algorithm) on the first derivatives of spectra of untreated (Ct) and treated (T) iDCs performed in the 1755–1475 cm−1 range. Ct-R1 and T-R1 are discriminated into two distinct classes, whereas a clear class distinction could not be deduced for Ct-R2 and T-R2. (B) IR mean spectra representing carbonyl esters and the Amide I region of untreated (black line) and treated (grey line) iDCs (upper panel) and their subtraction spectrum (lower panel). (C) IR mean spectra representing the acyl chain region of untreated (black line) and treated (grey line) iDCs (upper panel) and their subtraction spectrum (lower panel). Abs, absorbance; a.u., arbitrary units.

Spectral differences between the treated and untreated R1 cells were observed principally in the spectral ranges 3020 to 2800 cm−1 and 1755 to 1475 cm−1 (Figures 5B and 5C), relating to lipid/phospholipid bands as well as nucleic acid and protein bands, according to literature assignments (see the legend to Figure 3). These indicate significant alterations in the general lipid composition of R1 on exposure to hBD2, the variation being well above the sensitivity of the method (~0.005 arbitrary unit), and greater than the S.D. of the averaged spectra. In particular, the reduced intensity for the band at 1736 cm−1 (Figure 5B) was consistent with a decrement in phospholipid concentration relative to other signals, whereas variations in the 3000–2800 cm−1 region indicated an increase in CH3 with respect to CH2 moieties in lipids [24]. The band at 3012 cm−1 (Figure 5C), on the other hand, was not altered, ruling out significant variation in acyl chain saturation. Furthermore, a moderate shift to higher wave numbers of both symmetric and asymmetric C-H stretching bands (at 2852 and 2921 cm−1 respectively) was indicative of an alteration in lipid order [14].

Chemical distribution maps for selected treated and untreated R1 cells were then recorded with a spatial resolution of 5 μm. The intracellular chemical distributions for proteins, lipids and phospholipids in a selected naïve cell are shown in Figure 6(A) (upper row), along with the optical image of the cell for comparison, and the same information is provided for a selected hBD2-treated cell in Figure 6(B) (upper row). Protein signal mapping was according to the Amide I band intensities at 1590–1705 cm−1, lipids according to the acyl chain stretching modes at 2800–3000 cm−1 and phospholipids according to the carbonyl ester absorptions at 1720–1760 cm−1. The most appreciable difference upon peptide treatment occurred for the intracellular distribution of phospholipids, with a relative decrease in the carbonyl ester IR signal, with respect to other signals, such that it fell below the detection limit in the cytoplasmic region.

Figure 6 FTIR maps of hBD2-treated iDCs

Selected untreated cell from the Ct-R1 population (A) and peptide-treated cell from the T-R1 population (B). The optical microscope images on the left correspond to the same cell for which the distribution maps of major cellular macromolecule absorptions was obtained (upper rows). In these maps the colour corresponds to the signal intensity. In the RGB images (lower rows), the red channel (left-hand panel) relates to the protein Amide I signal and the green channel (middle panel) relates to the phospholipid carbonyl ester groups, the signal being proportional to colour intensity. The blue channel (right-hand panel) indicates their superposition showing areas of co-localization, and was obtained using the RG2B co-localization plugin of the ImageJ open source software ( P, protein Amide I band; Ph, phospholipid carbonyl ester band; L, lipid acyl chain band.

Chemotactic activity of hBD2 on iDCs

Chemotactic activity of hBD2 on iDCs was evaluated via the modified Boyden chamber method, using the same iDCs as for the other experiments, collected from numerous different donors. In this respect, our present study differs from those of others that either use CD34+-derived iDCs or HEK (human embryonic kidney)-293 transfected cells stably expressing human CCR6 [10]. Each preparation was extensively phenotyped before use. Moreover in our model, quantification of chemotaxis was carried out using a fluorimetric method that is probably more accurate than cell counting and more sensitive as it measures both migrated and transmigrating cells.

iDCs were derived from monocytes from 13 different donors and the levels of surface CCR6 were varied by using different amounts of IL-4 [25]. The chemotactic activity exerted by 10 nM of the cognate ligand MIP-3α correlated linearly with surface CCR6 expression (P=0.007; Figure 7B), although with a relatively poor correlation coefficient (R2=0.7), which may be ascribed to a wide variation in chemotaxis due to donor differences [18]. Pre-treatment of cells with an anti-CCR6 antibody resulted in efficient neutralization of chemotaxis (Figure 7A).

Figure 7 Chemotactic activity of hBD2 compared with MIP-3α

The percentage of migrating cells for hBD2- or MIP-3α-stimulated cells with respect to spontaneous cell migration are shown. iDCs were divided into three groups: those not presenting surface CCR6 (<10%, denoted iDC/R); those scarcely presenting surface CCR6 (10–60%, denoted iDC/Rlow); and those strongly presenting surface CCR6 (>60%, denoted iDC/Rhigh). For MIP-3α-stimulated cells, the effect of pre-treatment with a neutralizing anti-CCR6 antibody (+Abneut) is also shown. Linear regression of the percentage of migration against the percentage of cells presenting surface CCR6 for (B) MIP3α-stimulated cells and (C) hBD2-stimulated cells. Plots were obtained with the GraphPad Instat package. The solid line represents the regression, whereas the broken lines represent the confidence limits.

The chemotactic effect of hBD2 was determined at 1 μM peptide (following indications in the literature [6]), but in this case no correlation with the presence of surface CCR6 was observed (R=0.0006, P=0.9986; Figure 7C). The chemotaxis index showed an average increase of ~30% irrespective of the presence of surface CCR6 (Figure 7A). Furthermore, the anti-CCR6 antibody neutralized the chemotactic effect of hBD2 only for iDC populations displaying a high level of surface receptor (chemotaxis was reduced to 20%; results not shown).


The effect of short-term exposure to hBD2 was evaluated in macrophages and iDCs, as well as their monocyte precursors. In particular, binding and internalization of fluorescently labelled hBD2 (CF–hBD2 or BY–hBD2) was followed with the aim of determining whether the peptide acts essentially at the membrane level or if some peptide is internalized and may thus also act on intracellular targets.

A simple protocol involving quenching of accessible fluorescein with the impermeant dye TB [26] to separate contributions from surface-bound and internalized peptide clearly indicated that CF–hBD2 entered rapidly and efficiently into macrophage cells (Figure 1). This was confirmed by confocal microscopy, which showed punctuate peptide fluorescence both cortically and in the cytoplasm. The fact that peptide fluorescence was unvaried at 24 h suggests that an endosomal pathway may not be involved, as fluorescein is inactivated in lysosomal conditions.

Uptake into macrophages occurred without significant permeabilization in complete medium under physiological salt concentrations. This is unlike the cidal activity of hBD2 towards bacterial cells, which is instead fairly sensitive to the presence of salt and medium [5]. Blood-derived monocytes were also able to bind and internalize CF–hBD2, but uptake was significantly lower, also taking their smaller size into account. iDCs internalized a comparable amount of peptide to macrophages, but displayed a higher capacity to bind it externally. Our present findings thus indicate that three different types of APC display a differential capacity to rapidly bind and internalize hBD2, a process that is not accompanied by membrane permeabilization, and that macrophages showed the highest relative capacity to internalize the peptide, whereas iDCs bound the largest amount on the surface. It remains to be ascertained whether internalization is mediated by a receptor, an endocytotic process or some form of self-promoted uptake through the membrane.

Given the efficient binding and internalization of hBD2 into iDCs, we carried out a further more detailed investigation of how this could affect different cellular parameters, correlating results from flow cytometry, TEM, and optical and FTIR microscopies, under standardized conditions, using cells from several different healthy donors as the model.

Flow cytometric analysis provided information on the morphological and functional differences between naïve and hBD2-treated iDCs from different donors and clearly indicated the presence of two subpopulations, termed R1 and R2. The former was morphologically more heterogeneous and was formed by larger cells with a higher content of cytoplasmic organelles and vesicles. The less-populated R2 probably consisted of a different subset of viable cells, as revealed by iDC marker analysis. These different populations also gave rise to different chemical IR fingerprints, as revealed by SR-IRMS (see below).

Exposure to hBD2 followed by FITC, a marker for protein content, indicated that only the R1 population, which significantly bound the peptide, showed a moderate but statistically significant decrease in protein content. Taken together, our results suggest that only the fully differentiated R1 subpopulation was responsive to short-term exposure to hBD2 treatment in a manner that maintained its immature state, but possibly activated an exocytosis or degranulation event without compromising cell vitality. Transmission electron micrographs of untreated iDCs showed the presence of dark electron-dense cytoplasmic vesicles and on peptide treatment these appeared to release their contents into the cytoplasm (Figure 2). It has been reported that hBD2 can cause the release of pruritogenic mediators from mast cells in a process that involves interaction with a G-protein-coupled receptor, mobilization of cellular calcium and calcium-mediated degranulation [7], probably through vesicle fusion. We report that hBD2 can efficiently penetrate into iDCs and have directly visualized the formation of defined pores or channels in granules, which could account for an osmotic swelling and favour expulsion of the intra-granular contents via an alternative swelling/expulsion mechanism. The biological significance of this degranulation, and what role it plays in host defence, requires further characterization of the granules and their contents, which are as yet not well described in the literature.

To obtain further information on the chemical changes occurring in peptide-treated cells that could accompany the internalization and degranulation events mentioned above, samples were investigated by SR-IRMS and the data were analysed both by multivariate statistical methods (for individual whole cells) and by applying the univariate technique for single-cell intracellular functional group mapping. We succeeded in establishing a correspondence between spectral differences and morphological features of the cells and confirmed the presence of two subgroups matching the subpopulations observed using flow cytometry. Spectral differences mainly occurred in the 1755–1475 cm−1 region in bands related to carbonyl stretching vibrations associated with DNA (indicated by the arrows in Figure 4B), a type of variation that has been assigned to the changes in chromatin structure required for transcription and linked to cell differentiation in haemopoietic cells [27], thus it is in accordance with a differentiated R1 population and a less-differentiated R2 population.

Cluster analysis of whole-cell spectra confirmed further that only the R1 population responded to short-term treatment with hBD2 and displayed significant biochemical alterations, so that whole-cell spectra no longer clustered with untreated controls (Figure 5A). In particular, exposure to hBD2 induced significant alterations in the general lipid composition, suggesting a critical role of membranes in peptide-induced cellular processes. Variations in the 1750–1600 and 3020–2800 cm−1 spectral regions indicate: (i) an increased lipid disorder, insufficient to affect the membrane integrity as cells remain PI, but which could be consistent with an increased membrane fluidity [14]; and (ii) variations in the CH3/CH2 ratio but no evidence of an alteration in the degree of lipid unsaturation, consistent with a relative increase in cholesterol- and sphingomyelin-enriched regions with respect to phospholipids [24]. IRMS thus suggests that short-term exposure to non-toxic concentrations of hBD2 causes significant alterations to the membranes of iDCs, an effect that had not been reported previously. This might also be related to the observed degranulation effects, as also suggested by high-resolution SR-IRMS mapping.

Mapped cells from the R1 subgroup were chosen on the basis of cluster analysis of whole-cell spectra. Protein, lipids and phospholipid signals behaved somewhat differently upon treatment. For both treated and naïve cells, the highest protein distribution located at the nucleus, consistent with the greater cell thickness and presence of histone proteins in this region. Similarly, chemical maps of lipids and phospholipids exhibit particularly high concentrations at the nuclear and perinuclear level, probably deriving from the membranes of the nuclear envelope and of the cellular organelles. However, although the overall distribution of proteins and lipids did not alter markedly in treated with respect to the naïve cells, the phospholipid map showed a significant relative decrease in the phospholipid carbonyl ester signal in the cytoplasmic region, as can be appreciated better in the single-colour RGB maps (Figures 6A and 6B, lower rows). This effect was observed for all treated cells that were mapped, regardless of their derivation (buffy coats from three different donors), and supports a general rearrangement of the endomembrane system upon short exposure to hBD2.

Flow cytometric, optical microscopy and IRMS results thus all converge to indicate significant changes in cellular features and membrane properties of iDCs upon short-term exposure to hBD2. These effects, and in particular alterations in the membrane system, could be linked to altered cell motility, consistent with the reported chemotactic activity of hBD2 on iDCs [6]. It had initially been reported that this activity was due to the interaction of hBD2 with CCR6 [6,8], but this was later called into question [10,28], so it might be due to other or additional processes, possibly also involving the observed alterations in membrane characteristics. We were thus led to investigate the chemotactic activity of hBD2 in comparison with that of MIP3α, the cognate chemokine for CCR6.

Under our conditions, on average, our results seemed to confirm those first reported by Yang et al. [6], indicating that 1 μM hBD2 induced a comparable chemotaxis with that induced by 10 nM MIP-3α (Figure 7). However, although the chemotactic activity of MIP-3α correlated well with the presence of surface CCR6, that of hBD2 was independent of the presence of surface receptor. Furthermore, although the anti-CCR6 antibody always abrogated the effect of MIP-3α, it had a neutralizing effect with hBD2 only for iDC populations displaying a high level of the surface receptor. We have also reported previously this type of behaviour for a structurally simpler artificial BD [29], and this leads us to suppose that defensin-induced chemotaxis may be partly due to activation of CCR6, but that other mechanisms are also involved and, in fact, predominate if CCR6 is poorly exposed on the surface.

In fact, hBD2 is reported to chemoattract mast cells via another G-protein-coupled receptor [30], and recently still other chemokine receptors, such as CCR2, have been reported to participate in chemotaxis by hBD2 [28]. Blood CD11c+ DCs are reported to express this receptor [31]. On the basis of our present results, we suggest that iDCs may, however, not necessarily be activated by an agonist–receptor-type interaction, but rather in a non-canonical manner involving either interaction with the membrane surrounding the receptor or an alteration of properties such as membrane composition/fluidity. These considerations may help explain apparently conflicting reports in the literature [10,3133], and also the fact that several different BDs, with quite diverse primary structures and even linearized forms or truncated fragments with compromised secondary structure, all display apparently similar chemotactic behaviour on iDCs [1,6,28,29,3134].


Taken together, our results lead to the following conclusions: (i) hBD2 is able to interact rapidly with APCs, leading to differential binding and cellular uptake; (ii) iDCs are most efficient in surface binding and macrophages in internalizing peptide; (iii) sub-cytotoxic concentrations of hBD2 trigger a rapid release of cytoplasmic vesicle content in iDCs; (iv) short-term exposure to hBD2 also seems to trigger a generalized lipid rearrangement in iDCs consistent with endomembrane re-organization and a possible increase in plasma-membrane fluidity; and (v) hBD2-induced membrane variations may possibly correlate with increased cellular motility; accumulation of the peptide in the proximity of membrane-located receptors may favour a receptor-mediated activation of chemotaxis, whereas internalization of the peptide and/or its effects on membrane properties might trigger this process in alternative ways, which need to be defined. This suggests a more complex mode of chemotactic action than was supposed previously.

It is tempting to speculate that the membrane interaction of molecules such as hBD2 on iDCs, at non-toxic concentrations, may lead to accumulation within membrane patches around chemotactic receptors, thus stimulating their activities, while, at the same time, internalization of the peptide and the effects on internal mechanisms could help predispose the cell towards movement by altering its membrane characteristics or for other functions related to the defensive role of iDCs. In this context, the relevance of vesicle degranulation in iDCs, upon short-term exposure to hBD2, to their role in immunity requires further clarification.


Francesca Morgera and Alessandro Tossi wrote the manuscript. Lisa Vaccari and Francesca Morgera planned and carried out all the SR-FTIR experiments and contributed to the relevant parts of the manuscript. Sabrina Pacor and Luisa Creatti planned and carried out the confocal microscopy, flow-cytometry, electron microscopy and chemotaxis experiments, and contributed to the relevant parts of the manuscript. Nikolinka Antcheva carried out all of the peptide synthesis, purification and characterization. Alessandro Tossi and Sabrina Pacor critically assessed and analysed the data.


This work was supported by the Friuli Venezia Giulia (FVG) LR26 [regional grant R3A2] and Italian National Grant PRIN 2007 [grant number 2007K9RFLS]. F. M. and L. C. were supported by Ph.D. grants from Elettra Synchrotron Light Laboratory and Trieste University respectively.

Abbreviations: APC, antigen-presenting cell; BD, β-defensin; BY, BODIPY® [boron dipyrromethene (4,4-difluoro-4-bora-3a,4a-diaza-s-indacene)]; CCR, CC chemokine receptor; CF, 5(6)-carboxyfluorescein; DC, dendritic cell; DC-SIGN, DC-specific intracellular adhesion molecule-3 grabbing non-integrin; ESI–MS, electrospray ionization MS; FTIR, Fourier-transform IR; hBD2, human BD2; HDP, host defence peptide; iDC, immature DC; IL-4, interleukin-4; IRMS, IR microspectroscopy; MDM, macrophage derived from blood circulating monocytes; MFI, mean fluorescence channel intensity; MIP-3α, macrophage inflammatory protein-3α; PBMC, peripheral blood mononuclear cell; PI, propidium iodide; SR, synchrotron radiation; TB, Trypan Blue; TEM, transmission electron microscopy; TNFα, tumour necrosis factor α; TRITC, tetramethylrhodamine β-isothiocyanate


View Abstract