Background
Lyme disease (LD) is a highly prevalent tick-borne illness caused by the spirochetal bacterium
Borrelia burgdorferi (
Bb) [
1‐
3]. The disease is characterized by a wide array of clinical manifestations which vary in duration and severity among patients. Early clinical manifestations of LD include the characteristic “bullseye” rash known as
erythema migrans and flu-like symptoms, while late manifestations include arthritis, carditis and neurological compromise [
4,
5]. The invading spirochete induces both innate and adaptive immune responses, and it is believed that the innate immune response to
Bb contributes to the development of clinical findings in LD [
6]. The macrophage is a principal cellular element of the innate immune response to the bacterium at sites of infection in both humans and mice [
7‐
9]. Macrophages also play a prominent role in the pathogenesis of murine Lyme carditis, and their recruitment to heart tissue is important in spirochetal clearance [
10]. Macrophages have the phagocytic and signaling machinery necessary to bind, engulf, and degrade
Bb. Binding of
Bb to macrophages is mediated by surface integrins, such as Complement Receptor 3 (CR3) [
11,
12] and α
3 [
13]. Once attached, phagocytosis of
Bb is complex and can occur by either a sinking or coiling mechanism [
14,
15]. Both cases require rearrangements of the actin cytoskeleton to internalize Bb into the endosome, where degradation takes place [
16].
Bb is an extracellular pathogen that needs to be taken up and degraded for significant recognition by the host immune system [
17]. We have defined this process as “phagosomal signaling” [
14]. Spirochete degradation exposes borrelial pathogen-associated molecular patterns (PAMPs), such as lipoproteins and nucleic acids, to endosomal toll-like receptors (TLRs) for recognition, resulting in signaling cascades which induce pro-inflammatory cytokine production [
17‐
19].
Bb does not contain LPS and, therefore, does not engage TLR4. The cell envelope of
Bb contains abundant triacylated lipoproteins [
20], which are known to be recognized by TLR1/2 heterodimers [
21‐
26]. However, the three fatty acid chains in the N-terminus of
Bb lipoproteins, which serve as the TLR2/1 PAMP, are tethered in the outer membrane [
27]. We have shown that this results in minimal recognition of lipoproteins in intact spirochetes at the cell surface [
17,
18,
21,
28]. Instead, principal recognition of
Bb TLR2 ligands occurs within macrophage endosomal structures after the spirochete is phagocytosed and degraded [
17,
28]. Bacterial degradation results in exposure of both lipoprotein ligands and nucleic acids, which are recognized by endosomal TLR2 and TLRs 7, 8 and 9 respectively [
18,
19,
29]. Signaling cascades initiated by engagement of these TLRs utilize the adaptor protein MyD88 [
14,
30], indicating that this adaptor protein is a crucial element in mediating the inflammatory response to
Bb.
A role for MyD88 has been implicated in each of the four general steps associated with phagocytic clearance of bacterial pathogens: uptake, phagosome maturation, degradation and cytokine production. Murine macrophages lacking MyD88 show markedly diminished uptake of several bacterial species, including
Bb [
28,
31‐
35]. In WT macrophages, prior studies have shown that
Bb-induced MyD88 signaling results in increased PI3K activation and when PI3K is inhibited
Bb uptake is decreased [
36]. In addition, formin proteins (FMNL1, mDia1, and Daam1) have been shown to play a critical role in mediating phagocytosis of
Bb [
15,
37]. Whether MyD88 increases activation of these formins, and the role of PI3K signaling in this process, has not been established. Degradation of bacteria is impaired in the absence of MyD88 due to inefficient acidification of phagosomes [
38]. In the context of
Bb infection, lysosome maturation markers are recruited to
Bb-containing phagosomes in macrophages lacking MyD88 [
28]. However, the degree of phagosome maturation and acidification required to expose
Bb ligands from the bacteria cell envelope for recognition has not been studied. Murine macrophages lacking MyD88 also show markedly diminished production of NFκB-triggered pro-inflammatory cytokines, such as TNFα and IL-6, when stimulated with different bacterial species, including
Bb [
28,
31]. Nevertheless, the key host components involved downstream of these MyD88-mediated phagosome signals and their effects have not been well studied in the context of
Bb infection.
The objective of these studies is to examine which downstream effects of MyD88 phagosomal signaling potentially enhance clearance of Bb. Using an ex vivo murine macrophage system, we show that MyD88 signaling enhances, but is not required, for bacterial uptake or phagosomal maturation. Through RNA-sequencing analysis, we provide evidence that MyD88 signaling drives transcription of multiple genes involved in phagocytosis and identify potential intermediate proteins that facilitate the association between MyD88 and bacterial uptake. We also demonstrate that internalization of Bb by macrophages induces robust MyD88-independent inflammatory responses via production of chemokines. Our findings highlight the importance of MyD88 in efficient uptake of the Lyme disease spirochete by macrophages and provide potential mechanistic insight into how MyD88 mediates this process.
Methods
Mice
Female 6–8-week-old C57BL/6 J wild type (WT) and C57BL/6 J MyD88
−/− (MyD88
−/−) mice used in these studies were obtained from breeding colonies maintained in the UConn Health (UCH) Center for Comparative Medicine facility according to guidelines set by the UCH Institutional Animal Care and Use Committee (IACUC). Female mice were used exclusively in these studies due to differences in expression of TLR7 between males and females [
39]. Original WT breeding pairs were purchased from The Jackson Laboratory. Original MyD88
−/− breeding pairs were kindly provided by Dr. Egil Lien at the University of Massachusetts with permission from Dr. S. Akira in Osaka, Japan. Disruption of the murine MyD88 gene was confirmed through PCR [
40]. Both WT and MyD88
−/− breeding colonies are maintained on the antibiotic Sulfatrim (sulfomethoxazole [40 mg/mL] + trimethoprim [8 mg/mL]) diluted in water 1:50, which has been previously shown to not impact the degree of
Bb infection [
41]. In preparation for euthanasia, individual mice were exposed to isoflurane to calm the animal prior to sedation by injection with an overdose of an anesthetic cocktail [Ketamine 50–75 mg/kg at 15 mg/ml, Xylazine 5–7.5 mg/kg at 2.5 mg/ml, and Acepromazine 0.5–1.25 mg/kg]. Following euthanasia by cervical dislocation, each animal was confirmed to lack a heartbeat in accordance with the approved IACUC protocol.
Bacterial strains
Low-passage virulent wild-type
Bb strain 297 [
42] or a strain 297 isolate containing a stably-inserted copy of green fluorescent protein (GFP) under the control of the constitutively-expressed
flaB promoter (Bb914) [
43] were maintained in Barbour-Stonner-Kelly (BSK)-II media supplemented with normal rabbit serum and gentamicin (50 μg/μl) [
43]. Cultures were grown at 23 °C for at least 1 week prior to being shifted to 37 °C as previously described [
43]. Spirochetes were centrifuged at 3300 x
g for 20 min at 4 °C and resuspended in either BSK-II for in vivo experiments or DMEM (Gibco, 15,630–080) supplemented with sodium pyruvate (Gibco, 11,360–070) and HEPES (Gibco, 15,630–080) for ex vivo experiments. After resuspension, the cultures were counted by dark-field microscopy using a Petroff-Hausser counting chamber (Hausser Scientific) and diluted accordingly.
Staphylococcus aureus (
Sa) was cultured and fluorescently labeled with FITC as previously described [
44].
BMDM stimulation
Bone marrow-derived macrophages (BMDMs) were isolated from 6 to 8-week-old WT and MyD88
−/− mice as described previously [
18]. Single cell macrophage suspensions were seeded into either 12-well tissue culture-treated plates at a concentration of 1 × 10
6 cells/ml per well or 1 × 10
5 cells/500 μL per well in 8-chamber cell microscopy slides. Cells were then incubated overnight at 37 °C/5% CO
2 to allow cell adherence before experimentation. Cells were incubated for either 0.5, 1, 4 or 6 h at 37 °C/5% CO
2 with live GFP-
Bb or labeled
Sa at multiplicities of infection (MOIs) of either 10 or 100. Stimulation media was DMEM supplemented with 1% sodium pyruvate and 1% HEPES. At the end of the incubation period, culture supernatants were collected and stored at − 80 °C until cytokine analysis. Cells stimulated in chamber slides were processed for confocal microscopy. Cells stimulated in 12-well plates were processed for RNA extraction. All culture media and reagents were confirmed free of LPS contamination (< 10 pg/ml) by Limulus amoebocyte lysate assay quantification (Cambrex, MA).
Confocal microscopy
After stimulation, BMDMs were fixed in 2% paraformaldehyde with 0.05% Triton-X-100 (Fisher, BP151–100) for 10 min. Slide wells were then incubated with 5% bovine serum albumin (BSA) solution in PBS overnight at 4 °C to block non-specific antibody binding. The next day, cells were stained with different combinations of anti-GFP (Thermo Scientific A-21311, 1:100), phalloidin conjugated with Alexa Fluor 647 (Biolegend 424205, 1:20), anti-MyD88 (Santa Cruz 11356, 1:100), anti-TLR2 (eBioscience 14–9021-82, 1:100), anti-TLR7 (R&D MAB7156, 1:100), anti-ASC (Santa Cruz 22514-R, 1:100) and anti-LAMP-1 (eBioscience 14–1071-82, 1:100). A secondary antibody, Alexa Fluor 350, was used to detect anti-MyD88, anti-TLR2, anti-ASC and anti-LAMP-1 (Life Technologies A21093, 1:100). Incubations with primary and secondary antibodies were done for 1 h each at room temperature; slide wells were washed following each incubation three times with PBS supplemented with 0.5% Tween-20, with a final wash in distilled H2O before mounting. After antibody staining, slides were mounted using Vectashield (Vector H-1000) and imaged using a Zeiss 880 confocal microscope. Image processing and analysis were performed using ImageJ (NIH, v1.41b). Colocalization values were determined by first analyzing profile plots in ImageJ (Plug-in: “Plot Profile”) across ten different phagosomes for each cell genotype and then calculating the average difference between the fluorescence intensity curves of the markers of interest (i.e., LAMP-1 and Bb). Binding percentages were calculated by imaging 100–200 cells using a confocal microscope and then measuring the ratio of cells containing at least one surface-bound or internalized spirochete to the total number of cells imaged for each condition, represented as %BMDMs interacting w/Bb. Uptake percentages were calculated by imaging 100–200 cells using a confocal microscope and then measuring the ratio of cells containing at least one internalized spirochete to the total number of cells imaged for each condition, represented as %BMDMs w/internalized Bb.
Western blotting of BMDM supernatants and lysates
Protein lysates were generated from BMDM cell culture lysates and supernatants after Bb stimulation. In these experiments, adenosine triphosphate (ATP) (Sigma, 3A6419-1G) was added to WT BMDMs (already stimulated with Bb for 5 h) 1 h prior to harvest for generation of lysates. Supernatants were treated with an equal volume of methanol and ¼ volume of chloroform, vortexed and spun at 16000 x g for 10 min. After removal of the upper phase, 500 μL of methanol was added to the intermediate phase, which was then vortexed and spun at 16000 x g for 10 min. The pellets were then dried at room temperature, resuspended in 30 μL of 2x Laemmli buffer and incubated in a 37 °C water bath until proteins became soluble. BMDMs were lysed using RIPA buffer at − 80 °C and spun at maximum speed for 10 min. Protein pellets were resuspended in 2x Laemmli buffer. Lysates were boiled at 99 °C for 10 min and run on a 12.5% SDS-PAGE gel at 140 V for 1 h (5 μL per lane, 15 lanes). Proteins were then transferred to nitrocellulose membranes (Bio-Rad 162–0177) at 20 V for 20 min. Membranes were blocked for 1 h in milk block solution and then incubated overnight at 4 °C with primary antibodies for either β-actin (Sigma A5441, 1:2000), IL-1β (R&D AF401NA, 1:800) or caspase-1 (Adipogen AG-20B-0042, 1:1000) diluted in milk block solution. Membranes were then washed 5 times for 5 min each in wash buffer (PBS supplemented with 0.5% Tween-20) and incubated with goat anti-mouse HRP-conjugated IgG (GE NA931) diluted 1:5000 (β-actin and Caspase-1) or 1:1000 (IL-1β) in milk block for 2 h at room temperature. Following additional washes, membranes were incubated in HyGlo spray chemilunescent substrates (Denville Scientific, E2400) for 5 min and imaged on a Biorad ChemiDoc MP imaging system.
Cytokine analysis
The Cytokine Bead Array Mouse Inflammation kit (BD Biosciences 552364) was used according to manufacturer’s instructions for simultaneous measurement of IL-6, IL-10, CCL2, IFNγ, TNFα, and IL-12p70 in supernatants from stimulated BMDMs. General statistical analysis was performed using GraphPad Prism 4.0 (GraphPad Software, San Diego, CA), using an unpaired Student t test. For each experiment, both the standard deviation and the standard error of the mean were calculated. P-values of < 0.05 were considered significant.
Identification of differentially expressed genes by RNA-Seq
Total RNA was extracted from three biological replicates of WT and MyD88
−/− BMDMs, either unstimulated or stimulated with
Bb at MOI 10:1 or MOI 100:1 for 6 h. Following stimulation, RNA was isolated using the Macherey-Nagel total RNA isolation kit (Takara, 740955) and was used as input for the NuGen Ovation RNA-seq V1 kit. cDNA output was analyzed for correct size distribution with an Experion Standard Sensitivity RNA chip and quantified using a Qubit Fluorometer. Sequencing libraries were produced using the NuGen Encore NGS Library I kit. Libraries were multiplexed and sequenced at The Jackson Laboratory for Genomic Medicine Sequencing Core with an Illumina HiSeq 2500 as 2X50bp pair end reads. RNA-Seq reads from each individual library were mapped with Tophat2 RNA-Seq spliced reads mapper (version 2.0.5) [
45] to mouse genome build mm9 with parameter settings adjusted to suit strand-specific pair-end RNA-Seq reads. The mapping result bam files were used as input to the HTSeq high-throughput sequencing data analysis package [
46] to quantify the read counts mapped to all genes in UCSC mm9 mouse gene annotation set. The expression levels of genes represented as mapped read counts were normalized using the DESeq2 RNA-Seq analysis package (function:
estimateSizeFactor) [
47]. Genes were considered expressed if the number of reads was above the 25th percentile for the normalized data set. For quality control, only replicates with Pearson correlation coefficient above 0.9 on their FPKM values were considered (Figure S
1). Expressed genes were then further analyzed for differential gene expression using the DEseq2 package with FDR cutoff: 0.1. Differential gene expression was calculated in WT BMDMs stimulated 10:1 with
Bb relative to unstimulated WT BMDMs and MyD88
−/− BMDMs stimulated 100:1 with
Bb relative to unstimulated MyD88
−/− BMDMs. Differentially expressed genes (DEGs) were classified as either up-regulated or down-regulated based on the log2 of the fold change compared to the unstimulated control, which was calculated in R statistical software using package “DESeq2”. Determined DEGs were then separated into five groups based on their expression profiles; WT (all DEGs in WT BMDMs), MyD88
−/− (all DEGs in MyD88
−/− BMDMs), MyD88-dependent (all DEGs in WT but not MyD88
−/− BMDMs), MyD88-independent (all DEGs in both WT and MyD88
−/− BMDMs), and MyD88-privative (all DEGs in MyD88
−/− but not in WT BMDMs).
Identification of enriched transcription binding sites and master regulator analysis
Transcription factor binding sites in promoters of differentially-expressed genes were analyzed using known DNA-binding motifs described in the TRANSFAC library [
48], release 2017.2, available in the GeneXplain software (
http://genexplain.com). Binding site enrichment analysis for each one of our sets of DEGs was carried out as part of a GeneXplain dedicated workflow. The background consisted of 300 mouse house-keeping genes and the TRANSFAC mouse Positional Weight Matrices PWM (motifs) for binding site prediction with
p-value < 0.001 score cutoff. Promoters were extracted by the workflow with a length of 600 bp (− 500 to + 100) and an enrichment fold of 1.0.
Master regulatory molecules were searched for in signal transduction pathways upstream of the identified transcription factors. The GeneXplain workflow available for this analysis was used in conjunction with the GeneWays database. Parameters set included a maximum radius of 10 steps upstream of the transcription factor nodes, the DEG lists from the respective group as context genes and a z-score cutoff of 1.0. All transcription factors and master regulators used in the network analysis had confirmed expression in respective conditions using the total gene expression lists from the RNA-sequencing data set.
Gene ontology (GO) enrichment analysis
A Gene Ontology (GO) enrichment analysis was performed for the different sets of DEGs, transcription factors, and master regulators using the TRANSPATH [
49] database through GeneXplain software. Input sets were the DEGs, transcription factors, or master regulators from either the MyD88-dependent, MyD88-independent, or MyD88-privative groups. Focus was directed to the GO biological processes output. GO biological processes related to
Bb uptake, inflammation, and chemotaxis were identified by first reviewing previous studies for any genes involved in response to
Bb relating to these phenotypes. Enrichment analysis was performed on these genes to identify GO biological processes that hit at least 60% of the genes on the list, generating a list of relevant GO biological processes. Then an intersection was performed between the list of GO biological processes identified using our DEG, transcription factor, or master regulator lists, and the GO biological processes identified from the relevant genes. Heat maps of expressed genes hits in each biological process were done in R statistical software using package “ggplots”.
Network reconstruction and network analysis
Networks were constructed joining the three identified layers on the networks: DEGs, transcription factors, and master regulators. The subnetworks were extracted from identified master regulators of interest. From the MyD88-dependent master regulator group, effort was directed on linking MyD88 with transcription factors that had binding sites in the promoter regions of the MyD88-dependent DEGs enriched in uptake biological processes. These transcription factors were identified using the TRANSPATH database with the enriched DEGs of interest as input. The output list of transcription factors was intersected with the list of transcription factors that were only expressed in WT BMDMs. Networks were assembled and analyzed using Cytoscape software [
50]. To extract the desired subnetworks, we used OCSANA [
51] within the BiNOM plugin [
52] in Cytoscape 2.8.3. MyD88 was considered as a source node and transcription factors from the intersected list as target nodes. For MyD88-privative chemotaxis subnetwork construction the same analysis pipeline was applied. MyD88-privative master regulators significantly enriched in chemotaxis were used as source nodes and MyD88-privative transcription factors enriched in chemotaxis were used as targets.
Discussion
Previous studies by our group have emphasized that uptake and degradation of
Bb by phagocytic cells, including monocytes and macrophages, are critical in eliciting the inflammatory response to the bacterium [
14,
17‐
19,
29]. The findings from these studies, as well as others [
28], show that the adaptor protein MyD88 plays a critical role in bacterial uptake and phagosomal signaling in macrophages. In the current study, we provide further evidence that the macrophage is a key driver of inflammation, even in the absence of MyD88. We also show that while MyD88 has a significant impact on spirochetal uptake, phagosome maturation and bacterial degradation are not affected (Fig.
1). Of particular novelty, phagosomal signaling cascades induced by
Bb ligands in macrophages trigger a number of inflammatory and chemotactic pathways (Figs.
4 and S
5). Moreover, the inflammatory processes are mediated by different regulatory proteins depending on whether MyD88 is present or absent, while induction of several chemotactic processes occurs independently of MyD88. In-depth analysis of these signaling cascades allowed us to identify previously underappreciated MyD88-dependent transcription factors which could lead to enhanced spirochetal uptake and clearance (Fig.
5).
To better understand the contribution of MyD88 to spirochete binding, uptake, degradation and signaling by macrophages, we used an ex vivo model. Murine macrophages lacking MyD88 show a phagocytic defect when stimulated with
Bb ex vivo compared to WT macrophages. This defect in uptake has been previously demonstrated in macrophage stimulation experiments with other bacteria strains [
31,
34,
35,
38]. Our results reveal that binding of
Bb is not affected and that this phagocytic defect is not dependent on length of stimulation (i.e. time dependent) and is only slightly rescued by increasing the MOI. Thus, in the absence of MyD88, macrophages are still capable of binding and taking up the LD spirochete, but MyD88 signaling enhances the efficiency of
Bb phagocytosis by macrophages. We also show here that when stimulated ex vivo, the macrophage response to
Bb is driven by the signaling cascades induced by MyD88 as a result of bacterial ligands engaging TLR2 and TLR7 receptors in the phagosome (Figs.
2 and
4). Recognition in the phagosome is driven by degradation of bacteria, since more TLR2, TLR7 and MyD88 marker intensity were observed colocalizing with degraded spirochetes. Our results in Fig.
3 indicate that in the context of
Bb infection, MyD88 is not required for phagosome maturation, evidenced by the recruitment of LAMP-1 to
Bb-containing phagosomes in MyD88
−/− BMDMs. This is in contrast to Blander et al., who published that MyD88
−/− BMDMs infected with
Sa or
E. coli did not colocalize with either Lysotracker or LAMP-1 to the same degree as WT BMDMs [
38]. One possible explanation for the different findings with our study is that the recruitment of LAMP-1 is delayed in MyD88
−/− BMDMs, given that Blander et al. measured phagosome maturation at an earlier time point than in our studies. This explanation is supported by Yates et al. who showed slightly delayed acidification in MyD88
−/− BMDMs stimulated with TLR2 or TLR4 ligands for 40 min [
56]. The fragility of the
Bb membranes also suggests that perhaps less acidification of the phagosome is needed to expose
Bb PAMPs.
Reduced uptake of bacteria in macrophages lacking MyD88 is a phenotypic trait that has been extensively detailed [
28,
31‐
35], but not well understood. To better understand the relationship between MyD88 signaling and phagocytosis, we used a computational systems biology approach. In prior studies, addition of TLR3 ligands to
Bb stimulation of MyD88
−/− BMDMs significantly rescues uptake [
36], suggesting that in the absence of MyD88 TRIF signaling can activate pathways that result in similar actin rearrangement in the cell. This signaling was shown to be mediated through PI3K [
36], but interestingly PI3K was not differentially expressed in our macrophage stimulation. However, our network analysis from RNA-sequencing data identified DEGs that are up-regulated downstream of common phagocytosis effector proteins (Fig.
5).
Rhoq, activated by
Cdc42 codes for TC10, a protein involved in generating long filopodia protrusions [
70]. The gene
Cyfip1, which encodes a part of the WAVE complex that regulates actin polymerization [
71], was also up-regulated according to the network through
Rac1 protein interactions. The WAVE complex has higher involvement with lamellipodia formations [
72]. It is likely that MyD88 controls transcription factors that upregulate these genes to promote phagocytosis through formation of coiling pseudopods, which are more similar to lamellipodia, rather than through straight filopodia protrusions. In addition to MyD88, studies indicating that TLR2 can utilize TRIF have also been completed, but this interaction only appears to contribute to the inflammatory response rather than spirochete uptake [
18,
73]. More recently, the leukotriene LTB
4 has been shown to promote phagocytosis of
Bb by macrophages [
74], but in our BMDM sequencing data we did not find differential expression of
Ltb4 or its receptor
Ltb4r1. This could possibly be due to the later time point we selected for sequencing. It has also been shown that spleen tyrosine kinase (Syk) has an important role in phagocytosis of
Bb via integrin binding [
75]. The Syk gene (
Syk) is significantly up-regulated in an MyD88-dependent manner, suggesting that MyD88 drives over-expression of
Syk to increase phosphorylation and activation of proteins involved in generating actin branches. However, in our GO analysis
Syk was not one of the 164 MyD88-dependent genes that enriched to uptake biological processes, and the transcription factor Zic1, which has binding sites in the promoter regions of a significant number of these genes, is not predicted to bind in the promoter region of
Syk. Zic1 was of particular interest to us because it appeared downstream of MyD88 in our network analysis (Fig.
5) and is controlled by the intermediate protein ApoE. Mice lacking ApoE have increased bacterial burdens when infected with
Bb [
69], suggesting that ApoE signaling plays a role in cell remodeling processes necessary to enhance uptake. In addition, a link between
Bb phagocytosis and cholesterol has been postulated by Hawley et al. who showed that CR3, a known phagocytic receptor for
Bb, is recruited to lipid rafts with the co-receptor CD14 [
76]. Thus, it is possible that MyD88 upregulates ApoE to enhance lipid rafts on the macrophage membrane, which can potentiate signaling to enhance uptake and provide scaffolding for proteins involved in actin remodeling.
Our computational analysis also supported that there are non-canonical sources of inflammation in MyD88
−/− mice. Our results suggest that there is possibly another receptor recruited to the phagosome that initiates chemokine production upon recognition of a
Bb ligand. Another mechanism for triggering chemokine production may be that the TLR receptors are utilizing another adaptor protein to transmit signals out of the phagosome, as postulated by Petnicki-Ocwieja et al. [
73]. Network analysis of DEGs from macrophages identified multiple master regulators that could be controlling production of these chemokines, but further investigation is needed to determine if these master regulators are in fact active in macrophages containing
Bb. Additional studies to test whether acidification of the phagosome is required for
Bb-induced chemokine production will also give insight into which ligand-receptor interaction induces this response.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.