Published online before print April 17, 2007
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
,1
* Department of Medical Microbiology and
Turku Graduate School of Biomedical Sciences, University of Turku, Finland;
Turku Centre for Biotechnology, Turku, Finland; and
Department of Medicine, Helsinki University Central Hospital, Helsinki, Finland
1 Correspondence: Department of Medical Microbiology, University of Turku, Kiinamyllynkatu 13, FI-20520, Turku, Finland. E-mail: pauliina.hartiala{at}utu.fi
ABSTRACT
Lyme borreliosis is a disease, which can affect several organs and cause a variety of symptoms. In some patients, the infection may become chronic, even after antibiotic therapy, and cause persisting damage. Dendritic cells (DC) are involved in the initiation of innate and adaptive immune responses. To study interactions between Borrelia garinii (Bg), one of the causative agents of Lyme borreliosis, and human DC, we used a cDNA microarray to compare the Bg-induced DC transcriptional response with the response induced by LPS. The Bg-induced response consisted of a smaller number of genes than the LPS-induced response. The microarray showed that the ectoenzyme CD38, which has an important role in DC chemotaxis and migration to lymph nodes, was strongly up-regulated by LPS but practically not at all by Bg. This finding was confirmed with quantitative RT-PCR and with flow cytometry at the protein level. In addition, RT-PCR showed that CCR7 expression was 11-fold greater in LPS-stimulated than in Bg-stimulated cells. These findings suggest that Bg may affect crucial DC functions by blocking the up-regulation of important molecules in DC migration to lymph nodes, thus affecting further immune responses in Lyme borreliosis infection.
Key Words: gene expression • Lyme borreliosis
INTRODUCTION
Borrelia garinii (Bg) is a spirochete bacterium of the Borrelia burgdorferi (Bb) sensu lato (B. burgdorferi s.l.) complex. It is the causative agent of Lyme borreliosis, a disease that can affect several organs and cause a variety of symptoms, typically affecting the skin, musculoskeletal system, and nervous system [1 ]. The infection is transmitted to humans via tick bites. Antibiotic treatment usually cures Lyme borreliosis. Some patients recover from infection without specific therapy, and in others, the infection may become chronic, even after antibiotic therapy. The infection can persist for years or even decades, causing a wide variety of symptoms and irreversible damage in the body. The chronic symptoms of the infection have been proposed to be a result of persistent infection or infection-induced autoimmunity [2 ].
Dendritic cells (DC) are in close contact with mucosal surfaces and are among the first cells to meet invading pathogens in the body. After suitable stimuli, such as bacteria, the microbial cell-wall component LPS, and a variety of cytokines [3 ], DC undergo a maturation process and migrate to lymph nodes, where they present foreign antigens to T cells [3 ]. DC also influence the type of T cell response and participate in the activation and recruitment of immature DC, NK cells, macrophages, granulocytes, and B cells through chemokine and cytokine production [3 ]. DC migration to lymph vessels and positioning to lymph nodes are controlled prominently by CCR7 [4 , 5 ], although various other factors affecting DC migration have also been discovered [6 ]. Recently, the ectoenzyme CD38 has been ascribed an important role in DC chemotaxis and migration [7 , 8 ].
The first dermatologic symptom of Lyme borreliosis infection, erythema migrans (EM), is characterized histologically by perivascular infiltrates of lymphocytes, DC, macrophages, and a few plasma cells [9
]. However, only a small number of neutrophils occur in this skin lesion, which is a nontypical finding in bacterial infections [10
]. Inflammatory cells in EM produce proinflammatory cytokines, including TNF-
and IFN-
[9
, 11
]. Antibodies have been shown to be responsible for immune protection against Lyme borreliosis [12
, 13
]. Although antiborrelial antibodies can be detected in most patients with late disease, in some patients, the antibody responses are weak, delayed, or in rare cases, absent [14
15
16
]. Immune response in patients with late Lyme borreliosis is shifted toward the Th1 side [17
].
DC phagocytose Bg Å218 and Bb B31, process borrelia-specific antigens, and activate borrelia-specific T cells [18 , 19 ]. These strains also induce DC maturation [19 ]. DC also secrete IL-8 after borrelial encounter in a manner similar to LPS, leaving the reasons for the sparse neutrophil infiltrate in EM unclarified [19 ]. Langerhans cells (LC), which are cells of the DC lineage, are present in EM and acrodermatitis chronica athrophicans (ACA), the late skin manifestation of Lyme borreliosis [20 , 21 ]. Although the number of LC has been found to be higher in ACA than in normal skin, in EM and ACA, the MHC II expression of LC has been found to be down-regulated compared with normal skin [21 ].
Studies of DC-microbe interactions have shown that a core population of genes is commonly regulated by various pathogens, including viruses, bacteria, and yeast [22
23
24
]. Inflammatory and innate immunity-related genes (TNF-, CCL3, CCL4, CXCL2) have shown early up-regulation peaks after microbial encounter [25
]. Escherichia coli LPS has been shown to mimic and account for almost the entire bacterial response [23
] and thus, was chosen as the reference stimulus for our studies. Although borrelia lacks LPS [26
], it expresses many lipoproteins with a wide variety of inflammatory and immunogenic effects [27
, 28
]. No gene expression studies about borrelia and DC interactions have been carried out so far and are needed for a complete view of the DC transcriptional response and for investigating the possibility that borrelia manipulates crucial DC functions to its benefit.
The aim of this study was to characterize potential differences between human DC gene expression profiles induced by Bg and those induced by E. coli LPS. Quantitative RT-PCR of selected genes was done to confirm microarray results. We also studied the early cytokine secretion profile of DC induced by Bg and LPS to compare mRNA and protein level findings. We found that the gene encoding CD38, an important factor in DC chemotaxis and migration, was not up-regulated by Bg stimulation at any of the studied time-points, whereas LPS induced its up-regulation at three of the four studied time-points by both comparison methods. This finding was confirmed with RT-PCR and also at the protein level. This led us to study the gene expression of CCR7, which was shown to be more up-regulated in LPS-stimulated DC than Bg-stimulated DC in all studied replicates.
MATERIALS AND METHODS
Bacterial culture
Bg Å218, a Finnish tick isolate, has been described previously elsewhere [29
]. The bacteria were grown in liquid Barbour-Stoenner-Kelly II medium and passaged weekly. Low-passage bacteria (with Passage Number 10 or less) were used. Prior to the experiments, the borreliae were counted in a Neubauer counting chamber, centrifuged at 1400 g for 10 min, and resuspended in DC medium at a concentration of 6 x 107/ml.
In vitro generation of DC
PBMC were isolated from buffy coats of healthy donors (Finnish Red Cross Blood Transfusion Service, Turku, Finland) by Ficoll-paque (Pharmacia, Uppsala, Sweden) density gradient centrifugation. CD14-positive monocytes were isolated with MACS CD14 microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany) according to the manufacturer’s instructions. Cells (1x106 per well) were cultured on 24-well plates (Costar, Cambridge, MA, USA) in IMDM (Gibco-BRL, Grand Island, NY, USA) with phenol red, supplemented with 10% heat-inactivated FCS (HyClone, Logan, UT, USA), 1 mmol/l HEPES, 0.1 mmol/l 2-ME, and 100 mg/ml gentamycin (Biological Industries, Kibbutz beit Haemek, Israel). Recombinant human IL-4 (1000 IU/ml; R&D Systems, Minneapolis, MN, USA) and GM-CSF (375 IU/ml; R&D Systems) were added on Days 1, 3, and 5 to the culture. Bg (6x106 bacteria) or LPS (final concentration 1 µg/ml; from E. coli Serotype O127:B8, Sigma Chemical Co., St. Louis, MO, USA) was added on Day 7. Prior to addition of stimuli, DC were shown repeatedly to be CD1a+, CD14–/low by the FACSCalibur flow cytometer (Becton Dickinson, Franklin Lakes, NJ, USA). Unstimulated DC were used as controls.
RNA preparation and microarray hybridization
DC RNA was extracted before stimulation at 0 h and after 2, 4, 6, or 8 h of Bg or LPS stimulation using the RNeasy mini kit (Qiagen, Valencia, CA, USA). cDNA was synthesized and labeled with fluorescent dyes. The reference sample was labeled with FluoroLinkTM Cy3-deoxy (d)UTP (Amersham Pharmacia Biotech, Uppsala, Sweden) and the samples of interest, with Cy5-dUTP (Amersham Pharmacia Biotech). The reference sample and the sample of interest were mixed in one tube before hybridization.
Hybridization was done using direct and indirect comparison. In direct comparison, LPS-stimulated DC RNA was used as the reference sample, and Bg-stimulated DC RNA was used as the sample of interest in the same array. In the indirect comparison method, unstimulated DC RNA was used as a reference sample and LPS- or Bg-stimulated DC RNA was used as samples of interest.
Hum-16K cDNA microarrays (Finnish DNA Microarray Centre, Turku, Finland), representing 
10,500 genes, were hybridized with cDNA, originating from 20 µg total RNA. The hybridization was done as described previously [30
], and three biological and two technical replicates at each time-point were hybridized independently.
Microarray data analysis
Hybridized Hum-16K cDNA microarrays were scanned using the ScanArray Express® optical scanner (Perkin Elmer, Wellesley, MA, USA) to determine the fluorescent intensities of Cy3 and Cy5 dyes at each spot. Microarrays were scanned first at a 633-nm and then at a 543-nm wavelength to acquire separate images for Cy3 and Cy5 dyes. The images were combined, and the spots were identified using ScanArray Express® microarray analysis software (Perkin Elmer). Spots were specified using the histogram method. Microarray data were analyzed using Kensington software (InforSense Knowledge Discovery Environment, London, UK). Gene expression levels were determined from the background decreased log-transformed intensity ratio values. Systematic variation in measured intensity values was eliminated using lowess-normalization. A gene was considered expressed differently if at least a twofold difference between the sample of interest and the reference sample was seen in all three replicates, and the difference was also statistically significant at risk level P < 0.05. Statistical significance was computed using the two-sided t-test. Hierarchical clustering was used to visualize gene expression profiles. To classify genes and to divide them into functional groups, different databases (GenBank, Gene, Kegg, and GeneOntology) were searched. Interesting genes were verified by sequencing.
RT-PCR
Quantitative RT-PCR of chosen genes [CD38, Kruppel-like factor 4 (KLF4), nerve growth factor-inducible protein A-binding protein 2 (NAB2), CCL2, matrix metalloproteinase 9 (MMP9), MMP12, MMP19, CXCL10, CCL5] was done to confirm the Hum-16K cDNA array results. Some interesting genes not included in the microarray were also studied (CCR7, STAT6). DC were stimulated for 8 h with Bg or LPS, and RNA was extracted as described above. Probes and primers were designed using Universal Probe Library probes (Roche, Indianapolis, IN, USA) and Primer Express (Applied Biosystems, Foster City, CA, USA; Supplemental Table 1). cDNA was prepared using a Superscript II kit (Gibco-BRL, Life Technologies, Paisley, Scotland) and used as a template for gene expression analyses. The PCR reactions were carried out using TaqMan Universal PCR master mix (Applied Biosystems) with 300 nM oligonucleotide primers (Gibco-BRL, Life Technologies) and 200 nM fluorogenic probe. The TaqMan ABI Prism 7700 sequence detection system (Applied Biosystems) was programmed to have an initial step of 15 min at 95°C following 40 thermal cycles of 15 s at 95°C, finishing with 1 min at 60°C. All measurements were done in duplicate in two separate runs, using samples derived from three individuals. Housekeeping gene EEF1A1 was used as a reference transcript.
Cytokine antibody arrays
Cell-free culture media supernatants of unstimulated, Bg-stimulated, and LPS-stimulated DC were collected after 8 h stimulation. RayBio® human cytokine antibody array C Series 1000 (RayBiotech, Inc., Norcross, GA, USA) containing 120 different cytokines was used to detect cytokine levels in culture medium. The arrays were prepared according to the manufacturer’s instructions. Membranes were exposed to X-ray film (Biomax XAR, Kodak, New Haven, CT, USA) within 30 min of exposure to the substrate. Biotin-conjugated IgG served as a control, and each membrane contained six positive control spots. The image was analyzed with MicroComputer Imaging Device image analysis system M5+ software (InterFocus Imaging Ltd., Linton, UK). Baseline OD was subtracted from total OD. The samples were made comparable by subtracting the OD level of the membrane’s negative control from all studied spots and dividing the OD of the studied spots by the OD level of the membrane’s positive controls.
Flow cytometry and Western blotting of CD38
DC were generated as described above. Bg (6x106 bacteria per 1x106 cells) and E. coli LPS (final concentration, 1 µg/ml) were added on Day 7 of culture. Cells were stained with a PE-conjugated mAb for CD38 (BD Pharmingen, San Diego, CA, USA) before stimulation and after 7, 24, and 48 h of stimulation. Isotype-matched antibodies were used as negative controls. Cells were analyzed using the FACSCalibur flow cytometer (Becton Dickinson) with CellQuest software (Becton Dickinson).
Cell-free culture media supernatants of unstimulated, Bg-stimulated, and LPS-stimulated DC were collected at 24 h of stimulation and subjected to Western blotting. LPS-stimulated DC (1x106; 24 h; CD38 expression was confirmed by flow cytometry) suspended in Laemmli buffer was used as a positive control. Anti-human CD38 mAb (R&D Systems) and HRP-conjugated goat anti-mouse IgG2a (Santa Cruz Biotechnology, Santa Cruz, CA, USA) were used to probe the samples (30 µl undiluted culture medium). Ponceau S staining (Sigma Chemical Co.) confirmed that there were detectable amounts of protein in the samples.
RESULTS
In indirect comparison, Bg induces a weaker transcriptional response than LPS
cDNA microarrays were hybridized with RNA obtained from Bg- and LPS-stimulated and unstimulated DC. The hybridizations were done using indirect and direct comparison methods. Overall, the transcriptional response in DC induced by LPS was greater than that induced by Bg. This was seen in up-regulated and down-regulated genes at almost each time-point. In indirect comparison (unstimulated cells=reference sample, and Bg-stimulated or LPS-stimulated cells=samples of interest), the total number of differentially regulated genes in Bg-stimulated DC was 60 at 2 h, 151 at 4 h, 324 at 6 h, and 214 at 8 h. The respective numbers for LPS-stimulated cells were 396, 357, 380, and 796. Responses induced by Bg differed remarkably from those induced by LPS, and only a minority of genes showed similar regulation (Fig. 1
).
 View larger version (25K): [in a new window] |
Figure 1. Differences in the regulation of genes at different time-points—indirect comparison. Comparison of all differentially regulated genes. The number of specific Bg regulates is shown in the left circles, that of specific LPS regulates in the right circles, and that of commonly regulated genes in the middle.
|
Neutrophil chemoattractants are up-regulated by Bg by both comparison methods
A total of 272 genes was up-regulated by Bg or LPS in direct and indirect comparison. Of these genes, 26 were up-regulated by Bg (Table 1
) and 246 by LPS. Genes encoding CSF-1, CCL20, CXCL1, CXCL2, CXCL7, MMP9, CLDN1, IL-1, and TRAF3 were among genes up-regulated by Bg in both comparison methods. The gene encoding CSF-1 was up-regulated by Bg at three of the four studied time-points. Figure 2A
and 2B
, shows the results of direct and indirect comparison of LPS- and Bg-specific responses.
|
View this table: [in a new window] |
Table 1. Chosen Genes Up-Regulated by Bg in DC by Both Comparison Methodsa
|
 View larger version (20K): [in a new window] |
Figure 2. Comparison of genes up-regulated by Bg and LPS by indirect and direct comparison methods. (A) Genes up-regulated by LPS. The number of genes up-regulated in the direct comparison is shown in the left circles, that of genes up-regulated in the indirect comparison in the right circles, and that of genes up-regulated by both comparisons in the middle. (B) Genes up-regulated by Bg. The number of genes up-regulated in the direct comparison is shown in the left circles, that of genes up-regulated by indirect comparison in the right circles, and that of genes up-regulated by both comparison methods in the middle.
|
, IL-1, IL-1ß, and IL-6), and TNF--related genes [TNF--induced proteins 3 (apoptosis inhibitor) and 6, TNFR superfamily member 5, and TRAF1]. Genes encoding IL-7R, neutrophil chemoattractants [CXCL1 (GRO), CXCL2 (GROß)], DC differentiation and maturation markers (human ADAM19, CD83, SLAMF1), macrophage stimulants and other chemokines [CSF-1, CCL3 (MIP-1), CCL20 (MIP-3)], PG receptor EP4, and adhesion molecules (CD58, tenascin C, ninjurin 1) were also similarly up-regulated. Genes encoding apoptosis inhibitors [BIRC 2, BIRC 3, BCL2A1, PBEF1 (inhibits neutrophil apoptosis), CFLAR] and various genes related to metabolism, signal transduction, transcription, and transport were up-regulated.
 View larger version (42K): [in a new window] |
Figure 3. Interesting genes up-regulated by Bg, LPS, or both. Representative hierarchical clusters of 85 genes. The genes were grouped according to function on the basis of public databases and published studies. Red indicates up-regulation and green, down-regulation. IP-10, IFN-inducible protein 10; GRO, growth-related oncogene . (See Supplemental Table 2 for a list of genes and definitions of terms used.)
|
Bg-specific genes include MMPs and chemokines
Bg-specific genes were found by means of hierarchical clustering. Bg-specific, up-regulated genes include genes encoding three MMPs (MMP9, MMP12, MMP19). Genes encoding the chemokine receptor CXCR4 and chemokines CXCL7 (leukocyte-derived growth factor-platelet basic protein) and CCL2 (MCP-1) were up-regulated as well as genes encoding protein kinase C, 
(many functions, e.g., in BCR/TCR-mediated signaling), and TRAF3. Cell cycle/apoptosis-related genes such as BTG3 (antiproliferative, may interact with CCR4) and CYCS (mitochondrial electron transport, involved in apoptosis initiation) and genes encoding SGCD [cytoskeleton, forms a link between f-actin and extracellular matrix (ECM)], CDC42EP3 (mediates actin cytoskeleton reorganization at the plasma membrane), NFKBIA (inhibits NFKB complex), and NAB2 (transcriptional repressor) were up-regulated.
Bg-specific, down-regulated genes included those encoding the adhesion molecule CD31 (PECAM1, counter-receptor of CD38) and SAMHD1 (DC-derived IFNG-induced protein).
LPS specifically up-regulates IFN-inducible genes and immunity-associated genes including CD38
LPS specifically up-regulated the transcription of many IFN-inducible genes (GIP3, IFIT1, IFITM1, IFITM2, IFITM3, EIF2AK2) and IFN-related genes (IRF2, IRF7, ISG20, ISGF3G). Transcription of chemokine and cytokine-encoding genes (IP-10, RANTES, MCP-3, TNFSF10, TNFSF13B), immunity-associated genes (CD38, ADORA2A, BF, IL7R, IL15RA, JUNB), TLR adaptor protein MyD88, and the gene encoding cell adhesion protein CD44 was up-regulated as well as the transcription of genes encoding many transcription factors and metallothioneins MT1H, MT1X, and MT1F. LPS specifically down-regulated the genes encoding IL1R1 and IFNGR1.
CD38 and CCR7 transcription is defective in Bg-stimulated DC
Quantitative RT-PCR was done to confirm the microarray results. CCR7 and STAT6, not included in the Hum-16K array, were also studied. Gene expression was studied at 8 h of stimulation in three biological replicates. RT-PCR results were in line with microarray results for almost all studied genes. CCL2 expression alone showed regulation by RT-PCR analysis different from that by microarray analysis. Genes encoding CD38, CXCL10, CCL5, Klf4, and Stat6 were up-regulated more by LPS than by Bg. The gene encoding CCR7 was also up-regulated more by LPS than by Bg. Genes encoding NAB2, MMP9, MMP12, and MMP19 were up-regulated more by Bg than by LPS (Fig. 4
). The log2 difference between Bg and LPS stimulation for CD38 was 5.8, indicating an up-regulation, which was 56-fold greater in LPS-stimulated cells compared with Bg-stimulated cells. For CCR7, the log2 difference was 3.4, indicating an 11-fold greater up-regulation and for CXCL10, 7.4, indicating a 167-fold greater up-regulation, respectively.
 View larger version (28K): [in a new window] |
Figure 4. RT-PCR results. Three biological replicates were studied at 8 h after stimulation. The results are shown as  comparative threshold cycle (Ct) values for each biological replicate. The Ct value is the normalized quantitative value of the expression level of the target gene obtained by subtracting the Ct value of the housekeeping gene EEF1A1 from the target gene. A difference of 1 in the Ct value represents a fold increase value of 1.
|
, MCP-1, MCP-4, PARC, thymus and activation-regulated chemokine, epidermal growth factor receptor, TIMP-1, TIMP-2, and uPAR. Bg and LPS induced the secretion of eotaxin-2, IL-10, TNF-, GRO, IL-8, and MIP-1, whereas the secretion of TIMP-1 and TIMP-2 was decreased. Bg increased the secretion of MCP-1 specifically, whereas LPS did not induce the secretion of any cytokine specifically. However, eotaxin-2, PARC, uPAR, and TIMP-1 secretion decreased in LPS-stimulated cells less than in Bg-stimulated cells, with the result that the secretion still remained at a significant level (normalized density value x100>50).
 View larger version (18K): [in a new window] |
Figure 5. Cytokine profile of DC stimulated by Bg or LPS. DC culture supernatants were collected before stimulation and at 8 h of stimulation, by Bg or LPS and subjected to the cytokine antibody array. The results include density values of chosen cytokines. Cytokine names are those given by the manufacturer. Results of one individual cytokine array are shown. PARC, Pulmonary and activation-regulated chemokine; PDGF-BB, platelet-derived growth factor-BB; sTNFrRII, soluble TNFR RII; TIMP, tissue inhibitor of MMP; uPAR, urokinase-type plasminogen activator receptor.
|
 View larger version (19K): [in a new window] |
Figure 6. Comparison of CD38 expression in Bg- and LPS-stimulated DC. Monocyte-derived DC were cultured with LPS or Bg and stained with mAb specific for CD38 (shaded histograms) or with matching isotype control (open histograms) at indicated time-points. The data are reported as fluorescence intensity per number of cells and represent three experiments.
|
DISCUSSION
CD38 and CCR7
The transcriptional response of human DC to stimulation by Bg was compared with that by E. coli LPS. The gene encoding CD38 was found not to be up-regulated by Bg, whereas LPS highly increased the expression of the gene. This finding was confirmed further by RT-PCR and flow cytometry. CD38 is an ectoenzyme recently found to be important for DC chemotaxis and migration to lymph nodes. Using RT-PCR, we also studied the gene encoding CCR7, a dominant mediator of DC migration, and found it less up-regulated by Bg than by LPS.
CD38 is an ectoenzyme expressed on many lymphoid cells. It has been attributed to many roles in the immune system [32 ]. It is involved in the regulation of calcium release and the entry of extracellular calcium to cells as a result of catalysis of formation of calcium-mobilizing metabolites [33 ]. It also can sustain adhesion and rolling of lymphocytes through interaction with its counter-receptor CD31 on endothelial cells [34 ]. CD38 is down-regulated during the differentiation of immature monocyte-derived DC and again, expressed upon maturation [35 ]. It was considered to be part of the common cellular response to infection in the study by Jenner and Young [24 ], and it is up-regulated in DC by various microbes, including LPS-expressing bacteria as well as viruses, yeasts, and LPS-lacking bacteria. Recent studies have revealed novel functions for CD38 in DC. In murine models, CD38 is involved in chemotaxis and transendothelial migration of polymorphonuclear leukocytes and DC, and this function requires its enzymatic activities [8 , 36 ]. CD38-deficient DC are recruited inefficiently from the skin to local lymph nodes after antigenic stimulation [8 ]. This results in poor priming of T cells and impaired induction of humoral immune responses. CD38 is needed for the chemotaxis of immature and mature DC to CCL2, CCL19, CCL21, and CXCL12 [8 ]. A recent study showed that mAb and other reagents interfering with CD38-mediated signals lead to powerful inhibition of human DC migration [7 ]. Our results showed at mRNA and protein levels that the expression of CD38 is weak in Bg-stimulated DC. The low surface expression of CD38 is not caused by shedding of the protein, as no CD38 could be detected in the culture media of unstimulated, LPS-stimulated, or Bg-stimulated cells by Western blotting (data not shown). The microarray results showed that CD31 (PECAM1), which is the counter-receptor of CD38, was also down-regulated at 6 and 8 h in Bg-stimulated cells, but its expression was unchanged in LPS-stimulated cells.
For further study of factors associated with DC migration, we investigated CCR7 expression using quantitative RT-PCR. CCR7 is a chemokine receptor and a dominant mediator in the mobilization of DC to lymph nodes via lymphatics. CCR7 ligands, CCL19 and CCL21, are expressed by the lymphatic endothelium and/or within lymph nodes by stromal cells, endothelial cells, and DC themselves [5 ]. These cells participate in the migration of DC to lymph nodes from peripheral tissues. Previously, it has been shown that defective CCR7 expression and thus, impaired DC migration play a role in the pathogenesis of visceral leishmaniasis [37 ]. Although CCR7 has been given the title as the dominant mediator of DC migration, there is controversial information available about the importance of CCR7 in DC migration. A recent study by Velan et al. [38 ] showed that DC pulsed with Yersinia pestis showed decreased migration toward CCL19 in an in vitro assay and in an in vivo assay but still showed up-regulated CCR7 expression. In a study concerning the role of CD47 in DC migration, CD47–/– mature DC showed normal CCR7 expression but impaired migration to CCL19 in an vitro assay [39 ]. We found that CCR7 was down-regulated in Bg-stimulated DC compared with LPS-stimulated DC.
The transcriptional core responses in DC are similar after Bg and LPS stimulation
The gene expression studies were done using direct and indirect comparison methods. Our earlier results have shown that the surface expression of DC terminal maturation markers peaks at 15 h after Bg stimulation (P. Hartiala, J. Hytönen, J. Suhonem, M. K. Viljanen, unpublished data). With these data and earlier microarray studies of DC-microbe interactions as a point of departure, we decided to study gene expression profiles at early time-points (2, 4, 6, and 8 h) after stimulation [23
, 25
]. The RT-PCR results were in line with the microarray results for most genes, and different results were obtained only for CCL2 expression. These results confirm the reliability of the Hum-16K cDNA array.
The amount of bacteria and the concentration of LPS were chosen according to earlier studies, where 6 x 106 Bg per 0.5 x 106 DC induced DC maturation in a similar manner to 1 µg/ml of E. coli LPS [19
]. The transcriptional response induced by LPS was greater at all time-points studied. This was detected by both comparison methods. Most differentially expressed genes in indirect comparison were specific for LPS or Bg, and the number of jointly regulated genes was limited. The jointly regulated genes included, as expected, endocytosis-associated genes, genes coding for basic inflammatory mediators, chemokines, adhesion molecules, and many apoptosis inhibitors. The results for the jointly regulated genes suggest that the core responses of DC at the transcriptional level are similar after Bg and LPS stimulation. The similar core responses also indicate that the LPS and Bg doses were comparable. These results are in line with previous findings about the transcriptional response of DC and other inflammatory cells to different microbes and their components [22
23
24
]. The meta-analysis by Jenner and Young [24
] clustered 32 different gene expression studies of host-pathogen interactions and defined a common host response occurring in all studied cell types (e.g., DC, macrophages, PBMC) despite stimulus. The genes, which are part of the common host response and were similarly regulated in our study, include genes coding for, e.g., CXCL1, CXCL2, CSF-1, CCL3, CCL20, TNF-, IL-1, IL-1ß, IL-6, BIRC 2, BIRC 3, BCL2A1, PBEF1, CFLAR. Genes encoding DC maturation markers were similarly regulated. This indicates that the core responses in different functional groups induced by Bg are similar to that induced by other microbes.
Differences in TLR-dependent and -independent signaling may account for differences in differentially regulated gene numbers
DC express TLRs, TLR1–4, TLR6, and TLR8 in vitro [40
]. Lipoproteins from Bb activate inflammatory cells through TLR2 and TLR1 [28
, 41
], whereas LPS activates the cells through TLR4 [42
]. A recent study reported that Bb can activate and induce MMPs and inflammatory mediators in host cells in a TLR-independent manner by binding to integrin 3ß1 [43
], suggesting alternative pathways for borrelia-pathogen interactions. The differences in TLR signaling and other host-pathogen signaling pathways may explain some gene expression differences induced by the different stimuli.
Different TLRs have different downstream signaling pathways. TLR2 signaling leads to NF-
B activation, which is believed to require the adapters MyD88 and MyD88 adaptor-like (Mal) protein [44
]. TLR4-mediated NF-B signaling is believed to require the adapters MyD88, Mal, Toll-IL-1R-related adaptor protein inducing interferon (TRIF), and TRIF-related adaptor molecule [44
]. In our microarray results, the adaptor protein MyD88 was up-regulated by LPS by both comparison methods, at least at two time-points but not at all by Bg. This finding supports the existence of alternative signaling pathways for Bg. The gene encoding IRF7 was up-regulated by LPS at all studied time-points, whereas Bg stimulation did not affect its transcription. IRF3 and IRF5 expressions were not induced by either stimulus. Different IRFs, most notably IRF3, IRF5, and IRF7, have been shown to be related to different adaptor proteins and therefore, to different TLR signaling pathways [44
]. We found no significant differences in signal transduction genes. As mentioned previously, TLR2 and TLR4 activation leads to NF-B activation through different pathways. The regulation of signal transduction differs from that of many other genes in that the proteins associated with a certain pathway are already located inside the cell, and their activation is mainly modulated through phosphorylation. Transcriptional regulation has no important function, and its magnitude is hard to predict. This probably explains why no great differences are seen in gene expression patterns. However, differences in TLR downstream signaling pathways and other alternative signaling pathways between LPS and Bg could account for differences in gene numbers and gene expression patterns as a result of regulation of different adaptor proteins and cofactors.
Early cytokine secretion profiles cannot be predicted from gene expression data alone
We studied cytokine secretion profiles to find out how the gene expression results are reflected at the protein level. The constitutive expression of cytokines consisted of a broad number of cytokines. There were some cytokines on the cytokine array, which were not included on the Hum-16K microarray chip. Thus, all results could not be compared. The secretion of IL-8 was induced by both stimuli as described previously [19
]. This was not seen in the microarray results, because of the selection criteria and individual variation between samples: In one of the three studied replicates, IL-8 was not up-regulated significantly at any time-point. The immune inhibitor IL-10 was strongly induced by both stimuli, whereas inflammatory cytokines IL-1 and IL-1ß were not induced compared with constitutive expression, and IL-6 and TNF- were only slightly induced. These proinflammatory cytokines were up-regulated significantly on the microarray by both stimuli, whereas IL-10 was not included on the Hum-16K microarray. The secretion of TIMP-1 and TIMP-2 was decreased significantly by both stimuli, suggesting increased MMP activity because of less inhibition by these factors. Certain MMPs have been shown previously to be important in DC chemotaxis and migration to inflammatory cites [45
, 46
].
The cytokine profile was examined already at 8 h of stimulation, only highlighting the early secretion events. Overall, the cytokine profiles induced by Bg and LPS were similar, unlike the results for the transcriptional level. Our results show that transcriptional responses at early time-points are not directly comparable with early protein levels and that IL-8 and IL-10 are the most abundantly secreted cytokines at 8 h of stimulation. Only CCL2, IL-6, and TNF- behave similarly on the microarray and cytokine protein array. These differences between mRNA and protein levels can be explained by temporal and spatial differences in transcriptional events and cytokine secretion. In a study by Vizzardelli and others [47
], LPS-stimulated mouse DC showed temporally different peaks for secreted cytokines; e.g., TNF- secretion peaked at 4 h after stimulation, and CXCL2 secretion peaked at 24 h after stimulation. Thus, no conclusions about cytokine secretion can be made from gene expression studies alone.
Additional interesting genes
MMPs are a family of zinc proteases degrading ECM components. They also have other substrates, such as some growth factors, cytokines, and chemokines, including MCP1–4 [48
, 49
]. Several studies have shown that human monocyte-derived DC produce MMP9 and MMP2 [46
, 50
, 51
], but neither of these MMPs was considered to be part of the common cellular response to infection or not even the common DC response to infection in a review article clustering various microarray studies of cell-microbe interactions [24
]. Bg has been shown to induce the production of MMP1 and MMP9 in human monocytes [52
], and MMP9 is up-regulated in EM skin lesions of patients with acute Lyme borreliosis [53
]. In our study, transcription of genes encoding MMP9 and -12 was up-regulated specifically by Bg at two time-points and the gene encoding MMP19, at one time-point. The gene encoding MMP9 was included in the genes up-regulated specifically by Bg in both comparison methods at 8 h of stimulation. This is in line with previous findings and supports the important function of MMP9 in Lyme borreliosis pathogenesis.
As mentioned earlier, a sparse neutrophil infiltrate occurs in the EM skin lesion [10
]. One hypothesis has been that DC do not attract neutrophils effectively to the site of infection. In our microarray analysis, neutrophil chemoattractants CXCL1 and CXCL2 were up-regulated by Bg even more than by LPS. However, CXCL1 (GRO) secretion in the cytokine array was not increased after Bg or LPS stimulation at 8 h, probably owing to the early time-point. Conversely, IL-8 was constitutively secreted, and the secretion was increased after Bg and LPS stimulation. As a conclusion, our results do not explain the sparse neutrophil infiltrate seen in EM.
CONCLUSIONS
Our results show that the genes encoding two important factors, CD38 and CCR7, needed in DC chemotaxis and migration to lymph nodes, are not up-regulated in DC by Bg stimulation compared with LPS stimulation. Bg-stimulated DC also showed low CD38 surface expression at all studied time-points. The abnormal behavior of two important DC migration factors suggests impaired migration of DC after Bg encounter. By impairing the migratory capacity of DC, Bg could weaken the humoral immune response directed against it. This could account for some of the immune abnormalities seen in Lyme borreliosis. LC, DC of the epidermis, are present in EM and ACA, the early and late skin manifestations of Lyme borreliosis, but the MHC II expression of these cells is decreased [21 ]. Bb can also be isolated from both of these skin lesions [54 ]. Our results suggest that the migration of DC after borrelial encounter may be impaired. This could lead to poor antigen presentation in the lymphoid organs and further immune evasion.
In conclusion, our results indicate a novel immune evasion mechanism through pathogen interference with DC CD38 expression. These results are extremely interesting in the light of the pathogenesis of Lyme borreliosis. Future in vitro and in vivo migration studies are needed to determine the role of CD38 and CCR7 in the pathogenesis of this disease. Furthermore, the molecular mechanisms and the extent of this phenomenon relating to other B. burgdorferi s.l. genospecies, other borrelia species, and possibly other pathogens causing chronic infections need to be studied.
ACKNOWLEDGEMENTS
This study was supported by the Academy of Finland (Microbes and Man Project 8102596). We thank Marju Niskala and Taina Kirjonen for excellent technical assistance, Perttu Terho for assistance with the flow cytometer, and Olli Lassila for his constructive comments about the manuscript.
Received November 30, 2006; revised February 28, 2007; accepted March 19, 2007.
REFERENCES
interferon production by peripheral blood mononuclear cells of patients with Lyme borreliosis Infect. Immun. 64,3620-3623[Abstract]3ß1 J. Immunol. 177,657-664
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
