Background
The innate immune system detects and eliminates invading foreign material through non-specific defense mechanisms elicited by, e.g., mononuclear phagocytes (MNP). MNPs originate as monoblasts in the bone marrow, reside as monocytes (MOs) in blood and become, e.g., tissue macrophages (Mϕs) or dendritic cells (DCs) upon extravasation into tissues. The innate immune system is presumably involved when biological response modifiers (BRMs) are utilized in the treatment of diseases such as cancer [
1,
2] and lymphangiomas [
3]. Killed bacterial toxins [
4] along with bacillus Calmette-Guerin (BCG) [
5], β-glucan [
6], interferons [
7] and monoclonal antibodies [
8] are examples of BRMs used in cancer treatment.
Japan has a long standing tradition in using penicillin-killed lyophilized
Streptococcus pyogenes, denominated OK-432 or picibanil, as a biological response modifier (BRM) for treatment of cancer [
1,
2]. Sakamoto
et al. [
1] published in 2001 a meta-analysis showing a 20% 5-year survival improvement with immunochemotherapy, compared to chemotherapy alone, following OK-432 treatment in patients diagnosed with non-small-cell lung cancer. This meta-analysis was based on 1,520 patients enrolled in 11 randomized clinical phase III trials. Furthermore, Oba et al [
2] published in 2007 a meta-analysis including 8009 gastric carcinoma patients from 8 randomized clinical phase III trials and concluded that compared to control conditions, addition of OK-432 treatment increased survival. There are also reports suggesting that patients with other cancers, such as head and neck squamous cell carcinoma (HNSCC) [
9], may benefit from OK-432 treatment. OK-432 has also been used as a maturation factor for DCs cells as part of vaccination therapy of cancer patients [
10].
Lymphangiomas are benign neoplasias of lymphatic origin, often congenital, that may extend around vital structures [
11]. Surgical removal of lymphangiomas has been the standard treatment, but may be technically difficult. Injection of OK-432 into cystic lymphangioma lesions may lead to shrinkage and subsequent cure as first reported by Ogita [
3]. Today, injection with OK-432 is established as a major treatment option for lymphangiomas [
11], and is an important example of the use of BRMs in medicine.
OK-432 presumably exerts its effect by activating the immune system to secrete toxic substances, which in turn eliminate tumor cells [
12]. It is, however, relatively little known about the mechanisms of action of OK-432. The principal cells responding to the drug, the engaged receptors or signal transduction pathways, are to a surprisingly large extent unknown. This should be an important area of study. The efficiency of OK-432 treatment needs to be improved in order to make OK-432 a better cancer treatment drug. Additional drugs, which could improve OK-432's response, as well as studying to what extent other (cancer) drugs potentially interfere with the receptors and signal transduction pathways driving the OK-432 response, should therefore be determined as to optimize and improve treatment.
Furthermore,
S. pyogenes is an important pathogen causing human disease. The diseases caused by
S. pyogenes range from tonsillitis, impetigo, necrotizing fasciitis, and scarlet fever to lethal toxic shock syndrome [
13]. Thus, studying the interactions of OK-432 with the immune system may shed light into the biology of these diseases as well.
One major MNP function is secretion of chemokines, such as monocyte chemoattractant protein (MCP) -1 and macrophage inflammatory protein (MIP) -1α/β [
14]. Chemokines have been shown to be secreted in response to OK-432 stimulation [
15] and have the main effect of attracting new leucocytes to the site from where they are secreted [
16]. It should be of interest, to determine the modality of chemokine secretion from MOs and Mϕs upon OK-432 stimulation. We have therefore found it pertinent to address this issue in the present investigation.
Priming provided by adherence, presumably associated with MNP differentiation from MO to Mϕ, contributes to interleukin secretion following OK-432 MNP stimulation [
17]. Therefore, it is of interest to study to what extent the MNP chemokine response to OK-432 depends on adherence. Molecularly, we have found it relevant to investigate the role of β-integrin receptors, as these receptors participate during differentiation of MNP from MO to Mϕ. Furthermore, these receptors may be functionally blocked by addition of piceatannol (Syk kinase inhibitor) and/or LY294002 (phosphoinositide (PI)-3 kinase inhibitor) [
18‐
22]. OK-432 may stimulate TLR receptors. In addition to OK-432, we have studied chemokine secretion upon stimulation with the known TLR2 agonist LTA and TLR4 agonist LPS. Moreover, we explored the role of the TLR2 co-receptor CD36 [
23] in OK-432 stimulation.
We have determined that in vitro purified adherent MNPs may secrete chemokines MCP-1, MIP-1α and MIP-1β following OK-432 stimulation. MOs deprived of adherence did not respond to OK-432 stimulation as assessed by chemokine secretion. Inhibition of Syk and PI-3 kinase did not block the stimulatory effect of OK-432. MIP-1α/β production in WB and MOs was higher upon stimulation with LTA or LPS while OK-432 stimulation gave a higher MCP-1 secretion in WB and MOs in comparison with LTA or LPS. CD36, and to some extent CD18, participate in the OK-432-modulated MNP chemokine response.
Discussion
We have studied
in vitro OK-432-stimulated mononuclear MNPs chemokine secretion. Purified MOs, PBMCs or WB secreted MCP-1, MIP-1α, and MIP-1β following OK-432 stimulation at a similar rate, except for a substantially lower MIP-1α secretion from WB. This is in line with other investigators, e.g., showing that
S. pyogenes directly activates Mϕs to secrete chemokines [
24]. Cytokine mRNA quantification of MCP-1 transcripts showed that OK-432 stimulation was followed by cytokine mRNA synthesis within hours. There was a correlation between measured cytokine mRNA levels and cytokine-secreted levels suggesting that OK-432 mainly exerts its effects at the pre-transcriptional level. We furthermore separated by adherence most of the MOs from a sample of PBMCs. Upon OK-432 stimulation both these cell fractions secreted MCP-1 at about twice the level of the MO-depleted fraction. These findings support that MNPs are the main source for MCP-1 within PBMCs. OK-432 also elicited a chemokine response in purified MOs from HNSCC patients.
All studied chemokines were secreted without particular stimulation factors at low adherence conditions, whereas MIP-1α/β secretion was completely inhibited by adherence. At low adherence conditions, the secretion response of MIP-1α/β and MCP-1 from PBMCs and MOs was abolished upon OK-432 stimulation. This might indicate that MNPs undergoing adherence fundamentally alter their chemokine secretion responses and supports that MO to Mϕ differentiation interferes with the regulation of chemokine secretion. This is also supported by our observations of MIP-1α-derived OK-432 stimulated secretion being lower in WB compared to PBMCs and MOs. Diapedesis depends on, e.g., integrin receptor engagement co-triggering differentiation of MOs to Mϕs [
18]. Such stimulation may be provided when MOs are isolated by
in vitro adherence. Syk phosphorylation is a necessary part of β
1–3 integrin-generated activation [
19‐
21], whereas PI-3 kinase phosphorylation is required in β
2 integrin-generated activation [
20]. We have presently shown that adherence of MOs stimulated phosphorylation of Syk but not PI-3 kinase, as determined by Western blots. Thus, it appears that possibly both β
1 and α
vβ
3 integrins, but not β
2 integrin, play a role in mediating MO adhesion. Based on Western blots, binding of OK-432 to MOs elicited a higher Syk phosphorylation compared to unstimulated MOs, possibly suggesting an adherence-mediated priming effect. PI-3 kinase phosphorylation occurred only upon stimulation with OK-432. We have shown that addition of either one of these inhibitors to OK-432-stimulated MOs did not change chemokine secretion. On the contrary, with addition of both of these inhibitors, MCP-1 secretion was increased, pointing to the existence of other possible signaling pathways. In addition, presence of both inhibitors throughout the isolation and separation of monocytes blocked the unstimulated secretion of chemokines supporting that β integrins are important for chemokine secretion during the adhesion process. Our observed Syk phosphorylation upon OK-432 binding pointed towards an integrin dependence and resonated with Cuzzola et al [
22], who have shown that the MO interleukin response to Group A Streptococci (GAS) was dependent on the β
2 integrin. We explored whether this applied to the OK-432-mediated activation of MOs by adding a blocking anti-CD18 antibody. MCP-1 production appears to be partially inhibited by the anti-CD18 antibody. Thus, the β
2 integrin may play a role in mediating the effects of OK-432 on MOs. Nevertheless, other receptors may phosphorylate either Syk or PI-3 kinase or both and their nature has yet to be established.
OK-432 probably stimulates MOs via TLR2 [
15] and/or TLR4 [
25]. In order to study the role of these receptors, MOs were stimulated in parallel with OK-432, and the known TLR2 or TLR4 agonists, LTA and LPS, respectively. Results showed that MOs respond to OK-432 in a similar fashion to LTA stimulation, thus, arguing for primary TLR2-dependent stimulation of MOs by OK-432.
A potential candidate for co-stimulation of MOs is CD36. This receptor has been found to function as a co-receptor for TLR2 receptor [
26]. We determined a reduced MIP-1β and MCP-1, but an unchanged MIP-1α response upon employing blocking anti-CD36 antibodies. Thus, CD36 may function as an important receptor in the OK-432 stimulation of MOs with respect to chemokine secretion.
Other co-stimulating receptors than the presently shown ones may be important for chemokine production following OK-432 stimulation of MNPs. Examples include CD62L (L-selectin receptor), a mediator of MOs rolling on and interacting with endothelial cells [
27] and CD162 (P-selectin glycoprotein ligand-1), a promoter in the interaction between MOs and platelets [
28]. Future studies are needed to establish their roles in OK-432 stimulation in MOs.
The shown MNP stage-dependence for MIP-1α secretion and adherence-dependent chemokine response to OK-432 supports the hypothesis that chemokine secretion is less favoured when MOs are stimulated by OK-432 in pb compared to extravasal stimulation. This is in line with our previous observation that TNF-α is primarily secreted upon OK-432 stimulation after initiation of MO differentiation [
17]. The present report thus adds more support to the notion that OK-432 is a target-seeking substance. Only when MNPs are stimulated by, e.g., tumor cells or intercellular substances [
29], one observes a chemokine response. The present results argue in favor of chemokine secretion being an important effector function of MNPs during OK-432 stimulation. This is important as to designing future BRM treatments with OK-432. Platinum-based drugs, e.g., negatively interfere more with monocyte than lymphocyte function [
31] and should thus be avoided in combination with OK-432 treatment.
MIP-1α/β and MCP-1 secretion is induced by stimulation of MOs with OK-432. These chemokines are ligands of the receptors CXCR1, CCR1, CCR2 and CCR5, present on polymorphonuclear granulocytes, basophiles, granulocytes, MOs, Mϕs, DCs, NK cells, T, and B cells [
32]. Thus, the shown chemokine secretion after OK-432 stimulation of MOs may co-explain the systemic effects of OK-432 on the immune system. Similar findings were made by Veckman et al [
24] upon stimulating Mϕs with pathogenic live bacteria (serotype T1M1 (IH32030)) isolated from a child with bacteremia. In contrast, we have stimulated monocytes with a heat- and penicillin G-killed, low-virulence SU strain of
S. pyogenes and found very low levels of RANTES (CCL5). Therefore, it is plausible that virulence status/degree, strain, and preparation of bacteria influence chemokine production and MOs may have a different chemokine profile compared to Mϕs. This becomes especially interesting when the emphasis is on OK-432 being a lyophilized, killed
S. pyogenes.
S. pyogenes is an important pathogen causing human disease. The diseases caused by
S. pyogenes range from tonsillitis, impetigo, and scarlet fever to lethal toxic shock syndrome [
13]. The shown mechanisms of MNP activation by OK-432 thus shed light on these diseases as well.
Methods
Reagents and antibodies
OK-432 (Picibanil) (Chugai Pharmaceutical Co. Ltd., Tokyo, Japan) was dissolved in sterile water as a stock solution with a concentration of 1 KE (Klinische Einheit)/ml = 0.1 mg/ml. LTA from
S. pyogenes and LPS from
E.coli 026:B6 (Sigma, St.Louis, MO, USA) were dissolved in PBS as a stock solution of 5 mg/ml (LTA) and 25 μg/ml (LPS) and used at a final concentration of 0.5 μg/ml and 1 μg/ml, respectively. The Syk inhibitor piceatannol (Calbiochem, La Jolla, CA, USA) was dissolved in ethanol as a stock solution with a concentration of 2 mg/ml (8000 μM) and used at a final concentration of 0.5, 1, 30 or 120 μM [
18,
19,
33]. The PI-3 kinase inhibitor, LY294002 (Sigma) was dissolved in sterile water at a concentration of 5 mg/ml (14.5 mM) and used at a final concentration of 50 μM. The PI-3 kinase inhibitor, wortmannin (Sigma) was dissolved in DMSO at a concentration of 1 mg/ml (2.3 mM) and used at a final concentration of 100 nM. Azide-free, blocking mouse anti-human CD36 (HyCult, Uden, The Netherlands) came as a 100 μg/ml stock solution and anti-mouse rat IgG
1 isotype control (R&D Systems Europe Ltd., Abingdon, Great Britain) was dissolved in sterile PBS as a 250 μg/ml stock solution. F(ab) (degraded by pepsin (Sigma)), azide-free, blocking rat anti-human CD18 (Chemicon International, Temecula, CA, USA) was dissolved in sterile PBS as a 100 μg/ml stock solution. Anti-rat IgG
2B isotype control (R&D Systems Europe Ltd.) was dissolved in sterile PBS as a 500 μg/ml stock solution.
Patients and controls
Patients admitted to surgery for head and neck squamous cell carcinoma (HNSCC) gave written consent before participating in the study. All patients were males. Patient 1 was 60 years and had tonsillar carcinoma (T3N0M0G1). Patient 2 had carcinoma of the tongue (T2N2M0G2). Patient 3 had carcinoma of the basal tongue (T2N2M0G2). Patient 4 had laryngeal carcinoma (T4N1M0). All patients donated blood approximately 2 weeks post-operatively as is appropriate in the event of starting treatment with OK-432. Whole blood from controls was obtained from randomly selected, healthy donors at Haukeland University Hospital. As selection was done randomly, age and sex of donors were not taken into consideration, i.e. not recorded. The study was approved by the Regional Committee for Medical Ethics.
Whole blood, PBMC, and monocyte preparation
Whole blood was diluted 1:1 with Ultraculture (Lonza, Basel, Switzerland), allocated to a 24-well plate (Nunc A/S, Roskilde, Denmark) and upon experiment completion, the supernatant was collected after the whole blood/Ultraculture mixture was centrifuged. PBMCs were separated by density gradient centrifugation with Lymphoprep® (Nycomed, Oslo, Norway) as density gradient medium in LeukoSep tubes (Greiner-Bio One GmbH, Frickenhausen, Germany). MOs were, unless otherwise stated, isolated from blood by gradient centrifugation followed by adherence to plastic. The PBMC yield of 7 ml blood (6 × 105 cells/well) was allocated to a 24-well plate (Nunc) with RPMI 1640 (Lonza) supplemented with amphotericin B (2.5 μg/ml) and glucose (2% v/v) (both Sigma), HEPES (1% v/v), L-glutamine (1% v/v), penicillin (100 IU/ml), streptomycin (100 μg/ml), sodium bicarbonate (2.7% v/v), sodium pyruvate (1% v/v)(all from Lonza) and 20% AB serum to a total volume of 0.5 ml/well. After 40 minutes pre-incubation, the adherent MOs were purified by washing, and then cultured in complete RPMI 1640 (Lonza)/20% AB serum with 0.5 ml/well. The method yields in general cultures which contain at least 75% MOs, as determined by nonspecific esterase stain (NSES), and more than 90% viable cells, as tested by tryphan blue stain. In some experiments, MOs were also isolated using the negative isolation kit (Dynal Biotech, Oslo, Norway) according to the manufacturer's protocol. Briefly, PBMCs, isolated by gradient centrifugation using Lymphoprep-containing LeukoSep tubes, were resuspended in PBS/0.1% bovine serum albumin (BSA) (Sigma) and exposed to an equal mixture of blocking reagent and antibody mix for 10 min at 4°C. After 10 min, PBMCs were washed, added fresh PBS/0.1% BSA, added to washed beads, and incubated for 15 min with gentle tilting and rotation at 4°C. After incubation, resuspended rosettes were pipetted and placed in the Dynal MPC magnet for 5 min. The supernatant was collected and placed in the Dynal MPC magnet for an additional 2 min. The collected MO-pure supernatant (95% purity – NSES) was washed, cells counted, and added to a 24-well plate.
Experimental conditions
In different experimental series, MOs (1 × 105 cells/well) were cultured either with piceatannol (0.5, 1, 30 or 120 μM – final concentration), LY294002 (50 μM – final concentration), wortmannin (100 nM – final concentration) or anti-CD36 (2.5 μg/ml – final concentration) or anti-mouse IgG1 isotype control (2.5 μg/ml – final concentration) or pepsin-degraded (Fab) anti-CD18 (12.5 μg/ml – final concentration) or anti-rat IgG2B isotype control (12.5 μg/ml – final concentration) for 30 min before OK-432 (0.01 KE/ml – final concentration) was added and after 24 hours the supernatants were collected and stored at -80°C for further analysis.
In some experiments, total RNA was isolated immediately after harvesting MOs (24 hours post-stimulation), using the RNeasy Minikit's (Qiagen, Hilden, Germany) protocol for isolation of total RNA from animal cells grown in monolayer, as recommended by the manufacturer. Briefly, adherent MOs were added 350 μl lysis buffer (buffer RLT containing beta-mercaptoethanol (Sigma)), homogenized by vortexing for 30 s, and the homogenate was applied to an Rneasy spin column after addition of ethanol in order to bind total RNA to the membrane. Contaminants were removed by washing with buffers RW1 and RPE, and total RNA was eluted using water. In addition, total RNA was isolated from MOs of normal blood donors and subsequently pooled for use in generating the standard curve.
Reverse transcription
TaqMan RT reagents (Applied Biosystems, Foster City, CA, USA) were used to reverse transcribe total RNA from OK-432-stimulated MOs obtained from normal blood donors. Final concentration of the RT reaction was 1 × TaqMan RT buffer, 5.5 mM MgCl 2, 2 mM dNTP mixture (500 μM of each dNTP), 2.5 μM random hexamers, 0.4 units/μl RNase inhibitor, 2.5 units/μl Multiscribe RT, 5–25 ng RNA, and DEPC water to a total volume of 50 μl. The RT thermal cycling was performed on a Mastercycler gradient thermal cycler (Eppendorf, Hamburg, Germany) under the following conditions: primer incubation for 10 min at 25°C, reverse transcription for 30 minutes at 48°C, and inactivation of reverse transcriptase for 5 minutes at 95°C.
Primer and probe design
Primer and probes for the TaqMan assays were designed using the Primer Express software (Applied Biosystems). Intron-spanning primers were used to prevent amplification of genomic DNA. The candidate genes were MCP-1 [GenBank: NM_002982] (70 bp-amplicon); primers: (F) 5'-CTCTCGCCTCCAGCATGAA-3', (R) 5'-GGAATGAAGGTGGCTGCTATGA-3', probe: 5'-CCGCCCTTCTGTGCCTGCTGC-3'and the housekeeping gene used was CD68 [GeneBank: BC015557] (67 bp-amplicon); primers: (F) 5'-CCCCACGCAGCAAAGTG-3', (R) 5'-CCAGGGGTGCTTGGAGATCT-3, probe: 5'-TCTCGGCTCAGAATGCATCCCTTCG-3'). Designed primers and probes were purchased from MedProbe (Oslo, Norway).
Real-time quantitative RT-PCR analysis
The ABI PRISM 7700 Sequence Detection System (Applied Biosystems) with 96-well plates was used to perform real-time quantitative RT-PCR. Reactions were carried out in a total volume of 25 μl containing 12.5 μl of 2 × TaqMan Universal PCR Master Mix (Applied Biosystems), 300 nM (0.25 μl) forward primer, 300 nM (0.25 μl) reverse primer, 200 nM (1 μl) TaqMan probe, 8 μl water, and 3 μl cDNA template. Thermocycler conditions were as follows: incubation for 2 minutes at 50°C, followed by incubation for 10 minutes at 95°C and 40 cycles of two-step (denaturing for 15 seconds at 95°C followed by annealing/extension for 1 minute at 60°C) PCR. As a reference gene, CD68 was used to normalize the gene expression of MCP-1. To generate a standard curve for relative gene expression determination, cDNA synthesized from pooled total RNA of MOs obtained from healthy blood donors was used. The standard curves generated for the cDNA analyzed had slopes with a mean of -3.3.
Western blot analysis
MOs from 7 ml blood were isolated by gradient centrifugation in Lymphoprep-containing LeukoSep tubes, followed by the monocyte negative isolation kit (Dynal Biotech), allowed to adhere for 40 minutes, after which piceatannol or LY294002 or diluent control was added for 30 min, followed by incubation with OK-432 for 15 min. Upon experiment completion, cells were washed once with cold PBS, added lysis buffer (50 mM Tris-HCl, pH 7.4, 1% NP40, 0.25% sodium deoxycholate, 150 mM NaCl, 1 mM EDTA, 1 mM PMSF, 1 mM sodium orthovanadate, 1 mM NaF, and 1× complete proteinase inhibitor (Roche, Nutley, NJ, USA)) and scraped gently using a cell scraper. The cell lysate was clarified by centrifugation at 16,000 g for one minute at 4°C. Protein levels were measured using the Bio-Rad BCA Protein Assay (Bio-Rad, Hercules, CA, USA). Samples were added 1× loading buffer and the lysate was boiled for 5 minutes. Samples (6 μg/lane) and a Precision Plus Protein Kaleidoscope standard (Bio-Rad) were electrophoresed on a 10% SDS polyacrylamide gel for 60 minutes at 200 V and transferred to nitrocellulose membranes (Bio-Rad) in electroblotting buffer (25 mM Tris base, 0.2 M glycine, and 20% methanol) for 60 minutes at 100 V. Membrane was blocked with 5% BSA in TBS-Tween for one hour at RT with gentle tilting. The blots were exposed either to a polyclonal antibody against phosphorylated Syk or PI3 kinase p85/p55 or total Syk or total PI3 kinase p85 (all: Cell Signaling, Beverly, MA, USA) (all: 1:1000 in 5% BSA in TBS-Tween) overnight at 4°C with gentle shaking. After washing, an HRP-linked goat anti-rabbit IgG (H + L) antibody (Bio-Rad) (1:2000) was added for one hour at room temperature with gentle tilting. Proteins were detected using luminol/enhancer/hydrogen peroxide – based ECL and visualized on the Molecular Imager ChemiDoc XRS System (Bio-Rad).
Multiplex cytokine analysis
Chemokines in supernatants were detected using the Luminex immunobead technology. A 25-plex or 5-plex chemokine kit (all from Invitrogen/Biosource, Carlsbad, CA, USA), were used to analyze cytokine levels in supernatants from successfully completed experiments. In short, antibody-coupled beads were incubated with target analyte after which they were incubated with biotinylated detection antibody before finally being incubated with streptavidin-phycoerythrin. Standard sensitivity was 10 pg/ml (MIP-1α, MIP-1β). Samples were then read by the Bio-Plex array reader (Invitrogen/Biosource), using Luminex fluorescent bead-based technology (Luminex Corporation Austin, TX, USA).
MCP-1 protein analysis
The contents of MCP-1 in supernatants were determined by enzyme-linked immunosorbant assay kit (ELISA) manufactured by R&D Systems (R&D Systems). All procedures were performed according to the specifications of the manufacturer. In short, 96-well microtiter plates (Costar Corning, NY, USA) were coated overnight at room temperature with monoclonal mouse anti-human MCP-1 capture antibodies. Diluted samples and recombinant human MCP-1 standard were added and incubated for 2 h at room temperature followed by addition of biotinylated polyclonal goat anti-human MCP-1. The plates were incubated for 20 minutes at room temperature with streptavidin-conjugated horseradish peroxidase. Tetra-methyl-benzidine (TMB) (Sigma) and H2O2 were used as substrate. Absorbency values were measured at 450 nm using Softmax Pro version 4.0 on an Emax Precision microtiter plate reader (Molecular Devices, Sunnyvale, CA, USA). The lower detection level was 15.62 pg/ml. Chemokine levels obtained with media controls (after 24 h) were substracted from those obtained with OK-432-, LTA-, LPS-stimulation. All media controls tested directly after monocyte purification had nondetectable chemokine levels.
Statistical analyses
The statistical program package Statistical Package for the Social Sciences (Ver. 16 SPSS, Chicago, Ill., USA) was used. Chemokine levels and expression were compared by the Wilcoxon signed-rank test (two-tailed). Differences were considered significant at p < 0.05.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
CO performed the Western blot analysis, Multiplex analysis, all the experiments and data analysis. HS did the RT-PCR experiments and analysis. KB operated the Luminex machine. JO provided OK-432. CO and HJA were responsible for the concept and design of the experiments and primarily writing the manuscript. All authors have read and approved the manuscript.