Expression of matrix metalloproteinases and tissue inhibitors of metalloproteinases define the migratory characteristics of human monocyte-derived dendritic cells

Abstract

Dendritic cells (DCs) have an essential role in the initiation of immune responses as they deliver antigen/epitope and the appropriate signals to activate naïve T cells and thus start an immune response. In order to fulfil their function, DCs have to patrol different part of the body, thus migrating through the extracellular matrix to sample the local ‘antigenic’ environment. In the present study, we have investigated which enzymes might be involved in this process using the Matrigel trans-well migration assay, an in vitro model of extracellular matrix migration. In this assay we analysed the migratory ability of interleukin-4 (IL-4)/granulocyte macrophage–colony-stimulating factor (GM-CSF)-derived immature DCs as well as mature DCs, induced by tumour necrosis factor-α (TNF-α) and modified vaccinia virus Ankara (MVA). The ‘mature’ DCs showed an increased migration through Matrigel, which was significantly inhibited by inhibitors of matrix metalloproteinases (MMP). We also observed that the dominant MMP involved in this process was MMP-9, and a concomitant decrease of the endogenous tissue inhibitors of metalloproteinases (TIMP)-1 and TIMP-2 was also observed. Collectively these data suggest that the balance between MMP/TIMP determines the net migratory capacity of human DCs. Surprisingly, TIMP-3 was significantly increased in mature DC. Our data thus indicate that MMP and TIMP play a role in the migratory ability of human DCs. Our results also suggest that TIMP-3 expression might represent a new marker of maturation of human DCs.

Introduction

During their life, dendritic cells (DCs) must pass through different stages of maturation, which reflect in well-defined functional stages.¹ These maturative steps are necessary as diverse ‘skills’ are required at different functional stages. Thus, immature DCs must have the ability to avidly capture antigens and, once a second signal (often associated with foreign antigens such as bacteria, viruses or inflammatory signals, collectively defined as ‘danger’) is provided, rapidly change their functional characteristics and become powerful initiators of antigen-specific immune responses.² Obviously, to fulfil their mission, DCs must be able to migrate and patrol almost every tissue in the body and, upon encountering ‘antigen and danger’, quickly report to lymph nodes, the local outposts of the immune system, in order to activate specific T cells and thus the adaptive branch of the immune system.³ The process of cell migration is rather complex and requires multiple factors, including chemo-attractants to dictate the direction of the migration,^4,5 adhesion molecules acting as docking stations,⁶ and proteinases as essential means to ‘open’ the road in the extracellular matrix on the way back to regional lymph nodes.^4,6 The chemo-attractants and adhesion molecules involved in DC migration have been characterized⁶ and include homing receptors (e.g. CD44, CD62L, CD62P ligand) and chemokine receptors (e.g. CXC chemokine receptor-1, -2, -3 and CC chemokine receptor-1 to -7).^5,7,8 Less satisfactory is the definition of the factors that unwrap the extracellular matrix, although proteinases, in particular the matrix metalloproteases (MMP), which play an important role in cell migration,⁹ are probable candidates. There is evidence that MMP, in particular MMP-9, are involved in the migratory abilities of murine epidermidal Langerhans' cells.¹⁰ MMP also play a role in the migratory capabilities of T cells¹¹ and natural killer (NK) cells.¹² The MMP are a family of zinc-binding metalloproteinases that consist of an N-terminal pro-peptide domain, a metalloproteinase domain and a hemopexin-like domain in the C-terminal region.¹³ At present, 20 members of the human MMP family have been characterized, and they have been classified into different subfamilies according to their substrate specificity and cellular location. These include the collagenases, gelatinases, stromelysins and membrane-type MMP.¹⁴ MMPs play an important role in remodelling the extracellular matrix, which is critical in many biological processes such as embryo development, morphogenesis, tissue resorption, reproduction and angiogenesis.¹⁵ During these processes, the activity of MMPs is carefully regulated by tissue inhibitors of matrix metalloproteinases (TIMPs)¹⁶ and disruption of this regulation may play a role in various diseases such as arthritis, inflammatory diseases and cancer cell metastasis.¹⁷ Because previous studies have shown that MMPs are important in cell migration,^10,18–20 we speculated that the mobility of DCs might also be dependent on MMP-mediated degradation of basement membrane proteins.

We analysed the migratory characteristics of ‘immature’ interleukin 4 (IL-4)/granulocyte macrophage–colony-stimulating factor (GM-CSF)-derived and ‘mature’ dendritic cells using the Matrigel assay. To induce maturation we used tumour necrosis factor-α (TNF-α), widely used to induce DC maturation,²¹ as well as the modified vaccinia virus Ankara (MVA),²² a viral vector already employed in humans for vaccination purposes to induce protective immune responses.^23,24

The results we report indicate that mature DCs have an increased ability to migrate and that this migration is significantly controlled by the balance of MMP (in particular MMP-9) and TIMP (TIMP-1 and -2). In addition, we observed a further up-regulation of TIMP-3, which may contribute to define a final step of DC maturation.

Materials and Methods

DC isolation

Human DCs were generated according to a well-established procedure.^25,26 Briefly, ≈5−7 × 10⁶ CD14⁺ cells were isolated from 2−7 × 10⁷ peripheral blood monocytes (PBMC) using a magnetic cell separation system (MACS; Milteny Biotech, Surrey, UK). The CD14⁺ cell fraction was then incubated for 5–7 days in the presence of granulocyte–macrophage colony-stimulating factor (GM-CSF) (20 ng/ml) and IL-4 (10 ng/ml) (R&D, Minneapolis, USA). In vitro-generated DCs were, as expected, CD14⁻ CD1b⁺ and expressed high levels of major histocompatibility complex (MHC) class II. In some cases, DCs were further matured using TNF-α (10 ng/ml) (a kind gift from the Centre of Molecular and Macromolecular Studies, Lodz, Poland).

Gelatin zymography

Culture media (10 µl) containing 1 × 10⁶ DCs/ml was added to an equal volume of sample buffer [0·5 m Tris–HCl, pH 6·8, 4% sodium dodecyl sulphate (SDS), 0·005% Bromophenol Blue, 20% glycerol] (Novex, San Diego, CA), incubated at room temperature for 10 min and separated on 10% SDS–polyacrylamide gel electrophoresis (PAGE) gels containing 0·1% gelatin. The gels were renatured in 2·5% Triton-X-100 in water for 30 min at room temperature and then incubated at 37° overnight in developing buffer (50 mm Tris, 0·2 m NaCl, 5 mm CaCl₂, 0·02% Brij 35, pH 7·6). After incubation, gels were stained with 0·25% Coomassie Brilliant Blue R-250 for 4 hr at room temperature and destained in distilled water containing 30% methanol and 10% glacial acetic acid to reveal zones of lysis within the gelatin matrix. Protease inhibitors such as EDTA (5 mm), E64 (50 µm), phenylmethylsulphonyl fluoride (PMSF) (1 µm) or hydroxamic acid (10 µm) were added to the developing buffer to identify the classes of proteases that are responsible for lysis of the gelatin. Hydroxamic acid, compound SE206 (2S, 2R, 6S-3-aza-4-oxa-5-hexyl-2-(methylcarboxamido)-10-paracyclophane-6-N hydroxycarboxamide) is a nanomolar inhibitor of both MMPs and ADAM-TS4/TS5²⁷ and was synthesized at DuPont Pharmaceuticals (Wilmington, USA). rMMP-9 was a kind gift from Prof. Ghislain Opdenakker (University of Leuven, Leuven, Belgium).

Affinity purification of DC-derived metalloproteinase

Twenty millilitres of culture media from DC cultures were neutralized to pH 7·5 and 10 ml was passed over a sulphonic acid (S) column or a quaternary ammonium (Q) column (Sartorius, Goettingen, Germany). The column was washed three times with buffer (10 ml) containing 50 mm Tris (pH 7·5) and 100 mm NaCl. The protein was eluted off both columns with 10 ml of 1% SDS. The flow-through, washes and eluate were analysed for gelatinase activity by zymographic analysis. The flow-through from the S-column (10 ml) was then passed over an affinity resin column containing hydroxamic acid linked to Poros resin²⁷ (a kind gift of Dupont Pharmaceuticals). The column was washed three times with buffer (10 ml) containing 50 mm Tris (pH 7·5) and 100 mm NaCl, and the bound protein was removed with a high-salt solution (10 ml) containing 50 mm Tris (pH 7·5) and 500 mm NaCl. Residual bound protein was removed with 1% SDS (1 ml). The flow-through, washes and eluate were examined for gelatinase activity by zymographic analyses.

Migration assay

Cell migration was quantified using Transwell inserts (6·5 mm) fitted with polycarbonate filters (5-µm pore size) (Corning Costar, High Wycombe, UK).¹¹

The upper side of the wells were coated (for 1 hr at room temperature) with Matrigel (Collaborative Research Inc., Becton Dickinson, San Diego, CA) diluted in phosphate-buffered saline (PBS) (30 µg/filter).

DCs (5 × 10⁵ in 100 µl of medium) labelled with ⁵¹Cr (Amersham, Bucks., UK) were preincubated with different concentrations of the inhibitor, hydroxamic acid or control at 37° for 30 min and then added to the upper compartment of the insert. The lower compartment contained 500 µl of medium containing 100 ng/ml of monocyte chemotactic protein-3 (MCP-3). Chambers were incubated at 37° in 5% CO₂ for 4 hr. Migrated cells were recovered from the lower compartment and lysed with 100 mm Tris buffer containing 0·1% Triton-X-100. The radioactivity of samples was measured in a gamma-counter and expressed as counts per minute (c.p.m.). Results were expressed as the percentage of migrating cells, as follows: % Of migrating cells =(c.p.m. of lysate from migrated cells ÷c.p.m. of lysate from cells of total input)×100.²⁸

MVA virus infection of DCs

DCs were prepared as described above, and cultured at a density of 1 × 10⁶/ml in a 24-well plate at 37° in an atmosphere of CO₂ with a sonicated preparation of MVA (a kind gift from Professor G. Smith, University of Oxford, Oxford, UK) at a multiplicity of infection (MOI) of 10.²⁹ Supernatants were collected 4 hr postinfection for zymography. In some experiments, cells were collected for RNA extraction. Analysis of the surface expression of human leucocyte antigen (HLA)-DR, CD86 and thyroid peroxidase (TPO) was performed 2 days postinfection.

Western blot analysis

DC expression of MMP-9 was detected by Western blotting. Briefly, DCs (3 × 10⁶ cells/ml) were incubated in serum-free medium for 48 hr at 37°. Supernatant (40 µl) was subjected to SDS–PAGE (10% gel), transferred to nitrocellulose and blocked with 5% fat-free milk in phosphate-buffered saline (PBS) and 0·05% Tween-20. Primary anti-human MMP-9 [antibody (Ab)-1, mouse anti-human immunoglobulin (Ig)G1, Clone 6-6B] (CalBiochem, Darmstadt, Germany) was used at 2 µg/ml, followed by incubation with horseradish peroxidase-conjugated goat anti-mouse IgG (Jackson Immuno Research Laboratory Inc., West Grove, PA). The proteins were detected using an enhanced chemiluminescence detection system (Amersham). The expression of the human TPO was analysed by Western blotting of DCs infected with rMVA-TPO (MVA engineered to express the human TPO gene). Cells were lysed in buffer containing 10 mm Tris (pH 7·4), 10 mm NaCl, 3 mm MgCl2, 0·5% Nonidet P-40 (NP-40) and protease inhibitors. Proteins were separated by SDS–PAGE (10% gel), transferred to nitrocellulose and blocked with 1% bovine serum albumin (BSA) in TBST (25 mm Tris-buffered saline and 0·05% Tween-20). Primary anti-TPO antibody (mouse anti-human IgG1, clone 6G12; (Advanced Immuno Chemical, Claremont Long Beach, CA) was used at 1 µg/ml in TBST containing 0·5% BSA and incubated with the nitrocellulose membrane for 2 hr at room temperature. Membranes were treated as described above. The description of the construction of rMVA-TPO is detailed elsewhere (S. Quaratino et al., unpublished).

Fluorescence-activated cell sorter (FACS) analysis

Human DC cells were collected in ice-cold PBS [supplemented with 1% fetal calf serum (FCS) and 0·05% azide] and stained using mouse fluorescein-labelled phycoerythrin (PE)-conjugated anti-human DR CD86 monoclonal antibodies (mAbs) (Pharmingen, San Diego, CA). Mouse anti-human isotype-control staining confirmed specificity. Cells were also stained using a mouse anti-human TPO mAb (a kind gift of Dr Jean Ruf, Universitè d’Aix, Marseille, France).³⁰ The counterstain was a biotinylated rat anti-mouse IgG (Southern Biotechnology, Birmingham, AL), followed by fluorescein isothiocyanate (FITC)-conjugated streptavidin (Southern Biotechnology). Cells were then analysed on a FACS analyser (FACStar Plus; Becton-Dickinson).

RNA purification and reverse transcription–polymerase chain reaction (RT–PCR)

Total RNA from DCs, rMVA-TPO-infected DCs and TNF-α-treated DCs were prepared by using the RNasy Mini Kit (Qiagen, Hilden, Germany). The concentration and purity of the RNA obtained were determined spectrophotometrically. Reverse transcription of the RNA and PCR amplification (RT–PCR) were performed as previously described,³¹ using oligonucleotide primers complementary to the human tissue metalloproteinase inhibitors TIMP-1, TIMP-2³² and TIMP-3,³³ and to β-actin.³⁴ Each DNA reaction mixture containing 15 mm MgCl₂ and 50 pmol of each of the two oligo-primers was subjected to 25 cycles of amplification in a thermal cycler (PTC-200 DNA Engine; MJ Research, Watertown, MA) as follows: 30 seconds at 94°, 30 seconds at 55°, and 45 seconds at 72°. A final 5-min extension was performed at 72° to ensure formation of fully duplexed DNA. Amplifications were conducted on the same sample of reverse-transcribed RNA (1:20 RT volume), under the conditions described by Mohler & Butler,³⁵ to allow a semiquantitative estimate to be made.

Contamination by genomic DNA was shown to be negligible by performing PCR on the RNAs, but omitting the RT step. The PCR products were then migrated on a 2% agarose gel with appropriate molecular-weight markers.

Statistical analysis

The Student's two-tailed t-test for triplicate samples was applied for analysis of DC migration across the extracellular matrix, as well as for the analysis of DC migration upon different treatments (MVA or TNF-α).

Results

A gelatinase is secreted by DCs

MMPs have been shown to play an important role in cell migration through the extracellular matrix.¹¹ Therefore, we hypothesized that DCs may secrete MMPs to facilitate their migration through tissue. To identify DC-secreted MMPs we employed zymography as a highly sensitive and functional assay for analysing proteolytic activity. Zymography is widely used for research on extracellular matrix-degrading enzymes, in particular MMPs.³⁶ Human DCs were generated in vitro from peripheral blood monocytes, as described previously.^21,26 Supernatant from DC cultures and media alone were analysed for MMP activity by gelatin or casein zymography. Gelatin zymography revealed the presence of gelatinase activity represented by a single band with an approximate molecular weight (MW) of 90000 (Fig. 1a), while no activity was detected in casein zymography (data not shown), suggesting that MMP-3, -7 and -10 (caseinases) are not secreted by DCs.

Figure 1.

Gelatin zymography of dendritic cell (DC) supernatants. (a) DCs were prepared as described in the Materials and methods. CD14⁺ cells were cultured in the presence of granulocyte–macrophage colony-stimulating factor (GM-CSF) and interleukin-4 (IL-4) for 7 days. DCs were washed three times and incubated for a further 4 hr in RPMI after which the supernatant was harvested and analysed by sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS–PAGE) gelatin zymography (lane 1). The arrowhead indicates the DC-derived gelatinase. Lane 2 indicates the zymographic activity of medium only. (b) SDS–PAGE gelatin zymograms were incubated for 18 hr at 37° in the absence (M) or in the presence of the following different protease inhibitors: E-64 at 50 µm; phenylmethylsulphonyl fluoride (PMSF) at 1 µm; EDTA at 5 mm; and hydroxamic acid (HA) at 10 µm. Complete inhibition of the DC-induced gelatinase was observed in presence of EDTA and HA.

In order to determine which class of protease this DC-derived gelatinase belonged to, the zymographic analysis was carried out in the presence or absence of various protease inhibitors, including PMSF (a general inhibitor of serine proteases), E-64 (a cysteine protease inhibitor), EDTA (a general metalloproteinase inhibitor) and hydroxamic acid, which has been shown to inhibit many MMPs.²⁷ Whilst incubation with PMSF or E-64 did not result in any loss of activity, complete loss of activity was detected in the presence of EDTA or hydroxamic acid, confirming that the DC-derived gelatinase belongs to the metalloproteinase family (Fig. 1b).

Characterization and function of the DC-derived gelatinase

In order to identify the DC-derived gelatinase we employed conventional as well as affinity chromatography. In the first purification step, 10 ml of supernatant collected from DC cultures was passed over a filter column containing a sulphonic acid group (S-column) or a quaternary ammonium group (Q-column) (Fig. 2a). The DC-gelatinase bound to the Q-column but passed through the S-column in neutral buffer, suggesting that the DC-induced gelatinase is an acidic protease. The flow-through of the S-column was then passed over a column containing agarose linked to hydroxamic acid. The affinity resin absorbed 100% of the gelatinase activity, as there was none detected in the flow-through (Fig. 2b). The gelatinase was then eluted off the hydroxamic acid affinity resin with buffer containing 50 mm Tris, 500 mm NaCl (pH 7·5), and SDS was used to remove the residual activity. Following the purification, the eluate was analysed for total protein by SDS–PAGE and silver staining, but no detectable levels of protein were seen (data not shown). This data suggests that the DC-derived metalloproteinase is highly active at very low concentrations. MMP-9 is constitutively expressed in macrophages, T cells and neutrophils,³⁷ and has a similar MW (90000) to DC-gelatinase, therefore we aimed to determine whether the gelatinase detected in the supernatant of DCs was MMP-9. The affinity-purified DC-derived gelatinase was analysed by zymography using different concentrations of recombinant MMP-9. The zymography showed that the DC-derived gelatinase had a MW similar to that of rMMP-9 (Fig. 3a). That the DC-induced gelatinase was indeed MMP-9 was further confirmed by Western blotting (Fig. 3b). To test if the DC-induced gelatinase contributes to their migration, DCs were layered on Matrigel-coated filters in a transwell migration assay. Matrigel, used as a subendothelial matrix equivalent, resembles the basal lamina, as it comprises laminin and collagen type IV.³⁸ Migration tests were performed with DCs only or in the presence of hydroxamic acid. As illustrated in (Fig. 4) DC migration through Matrigel was inhibited by hydroxamic acid in a dose-dependent manner. Typically, inhibition of migration by ≈40% occurred with 10 µm hydroxamic acid. The hydroxamic acid does not affect DC viability at the concentration used (data not shown).

Figure 2.

Gelatin zymography of affinity-purified dendritic cell (DC)-derived matrix metalloproteinase (MMP). (a) DC supernatants were subject to acid (S) and basic (Q) columns, as described in the Materials and methods. The DC-induced gelatinase was detected almost entirely in the flow-through (Ft) fraction of the S-column, as only a very small amount was found in either the washout (W) or the eluate (E). However, the DC-derived gelatinase bound to the Q-column, suggesting that the DC S-induced gelatinase is an acidic protease. (b) The flow-through of the S-column was then passed over a column containing agarose linked to hydroxamic acid (HA). The affinity resin absorbed 100% of the gelatinase activity, as there was none detected in the flow-through. These results are representative of three different experiments.

Figure 3.

Dendritic cell (DC)-derived gelatinase corresponds to matrix metalloproteinase 9 (MMP-9). (a) The gelatinase activity eluted off the hydroxamate-affinity resin was evaluated in the presence of increasing concentrations of recombinant (r)MMP-9 in a sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS–PAGE) gelatin zymograph. (b) Western blot detection of rMMP-9 in DC supernatant.

Figure 4.

Hydroxamic acid (HA) can inhibit dendritic cell (DC) migration across the extracellular matrix. DCs (5 × 10⁵ in a 100-µl volume) were placed in the upper well of a Matrigel migration chamber (control) together with different concentrations of HA, a matrix metalloproteinase (MMP) inhibitor (kindly donated by Dupont Pharmaceuticals).²⁷ Average control migration was 15% of the total DC input. At 10 µm, HA inhibited 40% of the DC migration. DCs showed 100% viability after incubation with 10 µm HA. Graph bars show a significant difference between the inhibited DCs and the control. These results are representative of three different experiments.

Maturation of DCs further increases the production of MMP-9

The DCs so far studied represent an immature evolutionary stage. To further analyse the functional significance of MMP-9 release, we studied how ‘danger’ stimuli, which induce final DC maturation, would affect MMP-9 release. For this purpose immature DC were treated with TNF-α (10 ng/ml) or infected with recombinant MVA (at an MOI of 10). We used MVA as this viral vector has been used in humans for vaccination purposes.^23,24 To test the efficiency of infection and validate MVA with an immunologically active insert, we used an MVA that expressed the human thyroid peroxidase gene (rMVA-TPO). Analysis of the transduced DC revealed surface expression of the TPO molecule, as detected by both FACS immunostaining and Western blot (Fig. 5a, 5b). Upon MVA transduction, DCs underwent a further maturation, as shown by the over-expression of both HLA-DR and CD86 molecules (Fig. 5c), as compared to DCs matured in the presence of TNF-α, a known stimulus for DC maturation.³⁹

Figure 5.

Modified vaccinia virus Ankara (MVA) can successfully infect dendritic cells (DCs) and induce maturation. (a) DCs were infected for 48 hr with MVA-TPO [MVA engineered to express the human thyroid peroxidase (TPO) gene] [at a multiplicity of infection (MOI) of 10] and expression of human TPO was detected by fluorescence-activated cell sorter (FACS) immunostaining using a mouse anti-human TPO monoclonal antibody (mAb)³⁰ (thick black line) as compared to an isotype control (dotted line) or unstained cells (grey line). (b) TPO expression induced by DCs infected with the rMVA-TPO, and not by the wild-type MVA, was detected by Western blot, using a mouse anti-human TPO mAb. (c) Immature DCs were treated with tumour necrosis factor-α (TNF-α) (10 ng/ml) or infected with MVA (MOI of 10). After 2 days, DCs were stained for CD86 and human leucocyte antigen (HLA)-DR, the expression of which increases during DC maturation, using phycoerythrin (PE)-conjugated mouse anti-human mAbs. As expected, TNF-α-treated DCs up-regulated the expression of both HLA-DR and CD86 compared to the immature DCs, but a more dramatic up-regulation of both HLA-DR and CD86 was observed when DCs were infected with rMVA-TPO or MVA alone (data not shown). The grey line represents cells stained using the isotype-control antibodies. These results are representative of three different experiments.

After 4 hr, DCs were challenged with TNF-α or rMVA-TPO and the supernatants were collected from DC cultures and analysed for zymographic activity. As cells need to be maintained in serum-free medium to be tested in zymography, we performed this experiment after 4 hr of stimulation, to exclude cell death or stress induced by the deprivation of serum. Stimulation of DCs with rMVA-TPO or TNF-α yielded an increase of gelatinase activity of 2·07- and 3·20-fold, respectively, compared to the immature DCs (Fig. 6a), suggesting that the final maturative evolution of DCs correlate with the release of MMP-9 in vitro and probably in vivo. Similar results were obtained when DCs were infected with MVA alone (data not shown).

Figure 6.

Matrix metalloproteinase (MMP) production and migratory activity in immature and mature dendritic cells (DCs). An equal number of DCs were stimulated with modified vaccinia virus Ankara (MVA) at a multiplicity of infection (MOI) of 10, with tumour necrosis factor-α (TNF-α) at 10 ng/ml, or with media only. (a) Supernatants were collected after 4 hr of stimulation and the MMP-induced activity was analysed by gelatin zymography. The integrated density of each specific signal was measured using the Bio-Image system. The fold of induction was calculated as the ratio of MMP measured in cell supernatant from stimulated (MVA or TNF-α) and unstimulated DCs. (b) The migratory activity of DCs was analysed on a Matrige l-coated transwell. Results are expressed as percentage of migrating cells, calculated as follows: % of migrating cells  = (c.p.m. of lysate from migrated cells ÷c.p.m. of lysate from cells of total input) × 100.²⁸ A significant increase in the migration of DCs was detected upon treatment with MVA and TNF-α. c.p.m., counts per minute.

To establish that the above findings are reflected in the migration capacity of DCs, we assessed the in vitro migratory activity of the matured DCs. DCs infected with MVA, treated with TNF-α or control DCs were layered on the Matrigel-coated filter in a transwell migration assay. As speculated, MVA-infected DC increased their migratory ability by 40% compared to immature DCs, while treatment with TNF-α at 10 ng/ml also enhanced migration, but to a lesser extent (20%) than the virally infected DCs (Fig. 6b).

Expression of TIMPS in DCs

Production of TIMPs can determine the net proteolytic activity of MMPs⁴⁰ and therefore we wanted to establish whether the increase in migratory capacity of ‘maturing’ DCs was the result of an imbalance between MMPs and TIMPs. We studied the expression of TIMPs by semiquantitative RT–PCR.³² RNA introduced in the PCR reaction was normalized on the housekeeping gene, β-actin. PCR products were scanned using a densitometer and the integrated density of each specific signal was measured using the Bio-Image system. Expression of TIMP-1 was significantly down-modulated in DCs infected with rMVA-TPO, whilst a negligible difference was observed in TNF-α-treated DCs (Fig. 7). TIMP-2 was strongly down-regulated in both rMVA-TPO- and TNF-α-treated DCs (to 47% and 36% of that detected in the immature DCs). In contrast, expression of TIMP-3 mRNA was clearly up-regulated in rMVA-TPO-infected DCs (225%) and to a lesser extent in the TNF-α-treated DCs (139%), but quantification at the protein level was not possible as no TIMP-3 enzyme-linked immunosorbent assay (ELISA) is presently available. Taken together, these results indicate that MVA infection, as well as TNF-α challenge, modifies the balance between MMPs and TIMPs induced in human DCs.

Figure 7.

Expression of tissue inhibitors of metalloproteinases (TIMPs) in immature and mature dendritic cells (DCs). A semiquantitative reverse transcription–polymerase chain reaction (RT–PCR) was performed on the same set of RNAs isolated from immature DCs, DCs infected with recombinant modified vaccinia virus Ankara (MVA)-thyroid peroxidase (TPO) or DCs treated with tumour necrosis factor-α (TNF-α) (10 ng/ml). Transcripts were quantified from these blots, and β-actin counts were used to normalize RNA used in the PCR reaction. The integrated intensity of each specific signal was measured using the Bio-Image system and normalized to signal determined for control DCs. All data presented were obtained from two sets of experiments that gave consistent and reproducible results.

Discussion

DCs patrol almost every place in our body, and this means that DCs must have the capacity to migrate through the extracellular matrix of different tissues. Upon encounter with foreign antigens, DCs rapidly migrate from a non-lymphoid tissue to the regional lymph nodes, where they activate naive T cells to start the immune response.⁴¹ Inflammatory cytokines, such as TNF-α and IL-1, as well as bacterial products and viruses, represent a danger signal⁴² and are known to stimulate the maturation and migration of DCs from resident tissues to the lymph nodes.^3,43 As the process of DC migration through the tissues would probably require degradation of the extracellular matrix, we hypothesized that matrix-degrading proteinases may be secreted by human DCs, as previously observed in T cells, NK cells, murine Langerhans' cells and tumour cells.^10–12,19 In this study, we have shown that DCs secrete MMP-9, and that the level of secretion increases upon DC maturation.

Using hydroxamic acid, a powerful broad-spectrum inhibitor of MMPs, we significantly inhibited the migratory capabilities of DCs, further confirming the importance of MMPs in the migratory process. However, as the hydroxamic acid is a broad-spectrum MMP inhibitor that can inhibit many different metalloproteinases, we cannot rule out the possibility that other members of MMPs are involved. We did analyse DC-derived media for caseinase activity by casein zymography, but detected no activity, suggesting that MMP-3, -7 and -10 (caseinases) are not involved in the process.

Our data, which shows that a DC-secreted metalloproteinase may be important in the migration of DCs through the extracellular matrix, is consistent with the findings of other investigators. In particular it has been demonstrated that murine epidermal Langerhans' cells require MMP-9 to migrate and that this MMP-9 is also essential for their maturation.¹⁰ Moreover, MMP-2 and MMP-9 are considered to be essential for migration of NK cells and T cells.^11,12 In our assay, hydroxamic acid at a concentration of 10 µm decreased DC migration by 40% and higher concentrations did not appear to further inhibit the migration, suggesting that other proteases are also involved in this process, as suggested by others.²⁸

Many growth factors and cytokines, such as TNF-α, lipopolysaccharide (LPS) and viral proteins,^44,45 which are also important in the maturation of DCs, have been found to affect MMP-9 expression in a variety of cell lines.

Because of the potential in vivo significance of our study we also used MVA to induce maturation of DCs. This viral vector is widely used in vaccination programmes^23,24 and our studies indicate that MVA is not only effective as an antigen-delivery system, but also acts as a simulator of DC maturation and probably migration to the lymph nodes.

It is important to stress that in contrast to the wild-type vaccinia virus, the MVA is highly attenuated and induces DC maturation (Fig. 5), explaining why this vector is such a powerful adjuvant. In this study we found that virus-infected DCs or treatment with TNF-α increased the DC-derived gelatinase activity. MVA infection of DCs increased their migratory capacity, which is consistent with the increase in the gelatinase activity. MMP activity is controlled by their endogenous inhibitors (TIMPs),⁴⁰ and the balance between MMPs and TIMPs determined the net proteolytic activity, which constitutes an indication of tissue integrity.¹⁷ In this study we found that MVA infection of DCs perturbed the MMP/TIMP balance, i.e. enhanced secretion of MMP decreased the expression of TIMP-1 and TIMP-2, but increased that of TIMP-3. These data indicate that alteration of the MMP/TIMP balance may influence the ability of DCs to migrate through basement membrane and characterize the final maturation of DCs. Similar results were also observed in the TNF-α-treated DCs, although MVA infection showed a stronger induction of TIMP-3 transcript. Increased levels of TIMP-3 did not inhibit protease activity, as demonstrated by an in vitro migration assay. However, TIMP-3 is not as potent in blocking MMP-9 as TIMP-1 or TIMP-2 (M. Tortorella, unpublished). The function of TIMP-3 is not yet well defined; however, recent studies using deficient mice showed its role in lung development⁴⁶ and mammary gland involution.⁴⁷ Remarkably, TIMP-3 might also influence the shedding of very important surface molecules involved in leucocyte migration. In particular, it has been reported that TIMP-3 inhibits the shedding of CD62L, the lymph node homing molecule⁴⁸ as well as other important receptors such as IL-6.⁴⁹ TIMP-3 might thus inhibits members of the ADAM metalloproteinases.^50,51 Indeed, these latter metalloproteinases are potentially involved in the shedding of numerous surface proteins involved in immunological functions.⁵² Moreover, our data indicate that up-regulation of the TIMP-3 transcript occurs during the maturative steps in DCs and it is intriguing to relate this to the induction of the metalloprotease disintegrin MADDAM, observed in mature DCs both in vitro and in vivo.⁵³ Thus, the overall portfolio of MMP and TIMPs appears to characterize the different stages of DC maturation. Studying the potential role of TIMP-3 in the functional classification of mature DCs might shed light on the role of this inhibitor in immunomediated pathologies.

If TIMP-3 plays a key role in the latter stage of DC maturation it is obvious that it might represent an ideal candidate for controlling the function of mature DCs. The same is potentially applicable to the inhibition of MMP, as a reduced migratory activity of mature DCs could dramatically hamper the activity of DCs as initiators of immune responses. More importantly, MMP inhibitors have been considered to control the spread of metastasis⁵⁴ and thus this family of drugs could potentially be implemented in the dampening of unwanted immune responses, such as in autoimmunity. In conclusion, in this study we described that MVA is a powerful activator of DC maturation, that migration of human DCs through the extracellular matrix is significantly controlled by the balance between MMP and TIMP release, that TIMP-3 is selectively up-regulated during the final maturation step of DCs, and that targeting MMP and TIMPs might be an ideal therapeutic target for modulation of DC function.