Interleukin-1 Receptor–Associated Kinase M–Deficient Mice Demonstrate an Improved Host Defense during Gram-negative Pneumonia

Toll-like receptors (TLRs) occupy a central position in the initiation of cellular innate immune responses (1). TLRs can be expressed on the cell surface (TLR1, -2, -4, -5 -6, -10) or in intracellular compartments (TLR3, -7, -8, -9), serving a key role in the early detection of pathogens. Uncontrolled stimulation of TLRs potentially can lead to disproportionate inflammation and tissue injury, such as may occur during sepsis (2). Therefore, TLR signaling is tightly regulated to avoid such detrimental inflammatory responses. Several negative regulators of TLRs have been implicated in preventing excessive TLR signaling, including myeloid differentiation primary-response protein (MyD88), short, ST2, single-immunoglobulin-interleukin (IL)-1 receptor-related-molecule (SIGIRR), toll-interacting protein (TOLLIP), suppressor-of-cytokine signaling (SOCS), A20 and interleukin-1 receptor (IL-1R)–associated kinase (IRAK)-M (3). MyD88 is an adaptor protein essential for signaling via all TLRs, except TLR3 (1,4). In addition, MyD88 mediates intracellular activation after engagement of the type I IL-1 receptor and the IL-18 receptor (5). MyD88 initiates intracellular signaling by recruitment of IRAK-4 and subsequent association and phosphorylation of IRAK-1. IRAK-M inhibits the IRAK-1/IRAK-4 complex and thereby mitigates intracellular responses elicited by all MyD88 dependent receptors (6). Accordingly, IRAK-M-deficient (IRAK-M) macrophages produced higher levels of proinflammatory cyto kines upon stimulation with various


INTRODUCTION
Toll-like receptors (TLRs) occupy a central position in the initiation of cellular innate immune responses (1). TLRs can be expressed on the cell surface (TLR1, -2, -4, -5 -6, -10) or in intracellular compartments (TLR3, -7, -8, -9), serving a key role in the early detection of pathogens. Uncontrolled stimulation of TLRs potentially can lead to dispropor-tionate inflammation and tissue injury, such as may occur during sepsis (2). Therefore, TLR signaling is tightly regulated to avoid such detrimental inflammatory responses. Several negative regulators of TLRs have been implicated in preventing excessive TLR signaling, including myeloid differentiation primary-response protein (MyD88), short, ST2, single-immunoglobulin-interleukin (IL)-1 receptor-related-molecule (SI-GIRR), toll-interacting protein (TOL-LIP), suppressor-of-cytokine signaling (SOCS), A20 and interleukin-1 receptor (IL-1R)-associated kinase (IRAK)-M (3). MyD88 is an adaptor protein essential for signaling via all TLRs, except TLR3 (1,4). In addition, MyD88 mediates intracellular activation after engagement of the type I IL-1 receptor and the IL-18 receptor (5). MyD88 initiates intracellular signaling by recruitment of IRAK-4 and subsequent association and phosphorylation of IRAK-1. IRAK-M inhibits the IRAK-1/IRAK-4 complex and thereby mitigates intracellular responses elicited by all MyD88 dependent receptors (6). Accordingly, IRAK-M-deficient (IRAK-M -/-) macrophages produced higher levels of proinflammatory cytokines upon stimulation with various pathogens, TLR ligands or IL-1β (6). Considering its central position in the regulation of TLR and IL-1/IL-18 signaling, IRAK-M likely plays an important role in the host response to bacterial infection. In particular, enhanced IRAK-M expression has been implicated in the immune suppression frequently observed in patients with sepsis, a condition also called LPS tolerance and characterized by a reduced capacity of immune cells to release proinflammatory cyto kines upon restimulation. Indeed, IRAK-M -/cells did not become as tolerant to LPS upon reexposure to this bacterial component as WT cells (6), whereas our laboratory recently reported that LPS tolerance observed in healthy humans exposed to intravenous LPS and in patients with gram-negative sepsis correlated with enhanced IRAK-M expression in circulating leukocytes (7,8). Most importantly, in mice with polymicrobial abdominal sepsis, enhanced IRAK-M expression in pulmonary macrophages contributed to a diminished capacity of these cells to respond to Pseudomonas aeruginosa ex vivo, which resulted in a strongly impaired host defense response during secondary (following abdominal sepsis) Pseudomonas pneumonia (9).
Current knowledge of the functional role of IRAK-M in the host response to invading bacteria and the pathogenesis of sepsis is highly limited and focused primarily on its contribution to LPS tolerance. Pneumonia is the most common cause of sepsis by far (2,10). We argued that IRAK-M could play a pivotal role in host defense against primary bacterial pneumonia, considering its expression in the two most prominent resident cells in the bronchoalveolar space, that is, macrophages and respiratory epithelial cells (6,9,11,12), and its central place in TLR signaling. Therefore, here we induced gram-negative (using Klebsiella pneumoniae) pneumonia in IRAK-M -/and WT mice, seeking to establish the contribution of this negative TLR regulator in antibacterial defense in the previously healthy host.

Mice
Nine-to 11-wk-old C57BL/6 WT mice were purchased from Harlan Sprague Dawley Inc. (Horst, the Netherlands). IRAK-M -/mice, backcrossed > 10 times to a C57BL/6 genetic background were generated as described (6) and bred in the animal facility of the Academic Medical Center (Amsterdam, the Netherlands). Age and gender matched mice were used in all experiments.

Study Design and Sample Harvesting
The Animal Care and Use Committee of the University of Amsterdam approved all experiments. Pneumonia was induced as described (13,14). Briefly, K. pneumoniae serotype 2 (ATCC 43816; American Type Culture Collection, Manassas, VA, USA) was grown for 3 h to mid-logarithmic phase at 37°C using Tryptic Soy broth (Difco, Detroit, MI, USA). Bacteria were harvested by centrifugation at 1,500g for 15 min, and washed twice in sterile isotonic saline. Bacteria were then resuspended in sterile isotonic saline at a concentration of approximately 10 4 colony-forming units (CFUs)/50 μL, as determined by plating serial 10-fold dilutions on sheep-blood agar plates. Mice were lightly anesthetized by inhalation of isoflurane (Upjohn, Ede, the Netherlands) and bacteria were inoculated intranasally. In most experiments, mice were euthanized at predefined time points (n = 7-9 per group at each time point); sample harvesting and processing, and determinations of bacterial loads and cell counts were done as described (13,14). Briefly, mice were anesthetized with Domitor (Pfizer Animal Health Care, Capelle aan der IJssel, the Netherlands; active ingredient: medetomidine) and Nimatek (Eurovet Animal Health, Bladel, the Netherlands; active ingredient: ketamine) and euthanized by heart puncture. Blood was collected in tubes containing heparin. Lungs and spleen were harvested and homogenized in sterile saline (1:5, weight/vol) using a tissue homogenizer (Biospec Products, Bartlesville, OK, USA). CFUs in organ homogenates and blood were determined from serial dilutions plated on blood agar plates, incubated at 37°C for 16 h before colonies were counted. For bronchoalveolar lavage (BAL), the trachea was exposed through a midline incision and BAL fluid (BALF) was harvested by instilling and retrieving 2 × 0.3 mL of sterile phosphate-buffered saline. Cell counts were determined for each BALF sample in a hemocytometer (Beckman Coulter, Fullerton, CA, USA) and differential cell counts were performed on cytospin preparations stained with Giemsa stain (Diff-Quick; Dade Behring AG, Düdingen, Switzerland). For immunoassays in lung homogenates, (unlavaged) lungs were excised, weighed and homogenized in saline (1:5, weight/ vol). Lung homogenates were diluted 1:2 in lysis buffer containing 300 mmol/L NaCl, 30 mmol/L Tris, 2 mmol/L MgCl2, 2 mmol/L CaCl2, 2 % Triton X-100 and AEBSF (4-[2-aminoethyl]benzenesulfonyl fluoride), EDTA-Na2, pepstatin and leupeptin (all 8 μg/mL; pH 7.4) and incubated on ice for 30 min and spun down. Homogenates were centrifuged at 1,500g at 4°C for 15 min and stored at -20°C until assays were performed. In separate studies, mice were followed for up to 10 d and survival was monitored at least every 12 h (n = 10 per group).

Pathology
Paraffin lung sections were stained with hematoxylin and eosin as described (15), and scored from 0 (absent) to 4 (severe) for the following parameters: interstitial inflammation, endothelialitis, bronchitis, edema, thrombi, pleuritis and percentage of the lung surface demonstrating confluent (diffuse) inflammatory infiltrate by a pathologist blinded for groups. The total lung inflammation score was expressed as the sum of the scores for each parameter, the maximum being 28.

Phagocytosis
Alveolar macrophages were harvested from WT and IRAK-M -/mice (n = 8 per strain) by BAL, washed and resuspended in IMDM containing 2 mmol/L L-glutamine and 10% fetal calf serum to a concentration of 25 × 10 4 cells/mL and cultured overnight. UV-irradiated K. pneu moniae were labeled with CFSE and opsonized with 10% normal mouse serum (19). Labeled bacteria were incubated with macrophages for 1 h at 37°C (MOI 100:1) and cells were analyzed using FACSCalibur (Becton Dickinson, Breda, the Netherlands). Phagocytosis index of each sample was calculated: median fluorescence of positive cells × % positive cells.

Statistical Analysis
All values are expressed as mean ± SEM. Comparisons for more than two groups were done with Kruskall-Wallis followed by Dunn multiple comparison tests, and other comparisons with Mann-Whitney U tests. Survival was compared by Kaplan-Meier analysis followed by a log-rank test. Analyses were done using GraphPad Prism version 4.0, GraphPad Software (San Diego, CA, USA). Values of P < 0.05 were considered statistically significant.

IRAK-M Is Induced during Klebsiella Pneumonia
Knowledge of the expression of IRAK-M during primary gram-negative pneumonia is not available. Therefore, we infected WT mice with K. pneumoniae via the airways using an established model of severe pneumonia (13,14) and determined pulmonary IRAK-M expression at mRNA and protein level. IRAK-M mRNA was present at low levels in lungs of uninfected mice and displayed strong increases after infection with K. pneumoniae ( Figure 1).

IRAK-M -/-Mice Demonstrate a Reduced Bacterial Outgrowth and Dissemination during Klebsiella Pneumonia
To obtain a first insight into the functional role of IRAK-M in bacterial pneumonia, we harvested lungs, blood, spleen and livers at predefined time points after infection with K. pneumoniae for quantitative cultures, seeking to collect data representative for local defense at the primary site of infection and subsequent dissemination and representative for the early (3 and 6 h) and late host response (24 and 48 h, that is, just before the first deaths occurred, see below) (Figures 2A-D). In  Figure 3D). Together, these data strongly suggest that IRAK-M impairs antibacterial defense in the lungs upon infection with Klebsiella, which subsequently results in enhanced bacterial dissemination and increased organ injury.

IRAK-M -/-Mice Display Increased Lung Inflammation but Unaltered Neutrophil Influx Early after Infection with Klebsiella via the Airways
Pneumonia results in local inflammation and inflammatory cell recruitment, which is an integral part of the host immune response (20,21). To obtain insight into the role of IRAK-M herein, we performed semiquantitative analyses of lung histology slides prepared from IRAK-M -/and WT mice 3, 6, 24 and 48 h after infection. These analyses revealed a gradually developing histological picture of pneumonia, as reflected by interstitial inflammation followed by pleuritis and endothelialitis, and, in a later stage, bronchitis and edema. Of interest, IRAK-M -/mice demonstrated significantly more lung inflammation early after infection: both at 3 h (P < 0.05) and 6 h (P < 0.01), pathology scores were higher in mice lacking IRAK-M relative to WT mice (Figures 4A-F). Especially the almost doubled inflammation score in  IRAK-M -/mice at 6 h, which, in particular, was caused by the presence of evident bronchitis that, in this phase of the infection, was still absent in all but one WT mice; in addition, IRAK-M -/mice displayed more extensive endothelialitis and edema at early time points, which is remarkable in light of the five-fold lower bacterial load at this time point (which, in theory, provided a less potent proinflammatory stimulus). At later time points after infection, (24 and 48 h) pathology scores had increased considerably in all animals and were no longer any different between mouse strains (Figures 4G-L). Of note, however, at this stage of the infection, IRAK-M -/mice had up to 100-fold fewer bacteria in their lungs, again pointing to an inflammation-enhancing effect of IRAK-M deficiency.
Considering that neutrophils play an important role in innate defense early after Klebsiella airway infection (22,23) and in light of the early benefit of IRAK-M -/mice with regard to pulmonary bacterial loads, we determined neutrophil counts in BAL fluid (BALF) harvested from IRAK-M -/and WT mice 3 or 6 h after infection ( Figure 4M); in addition, we determined MPO concentrations in whole lung homogenates, providing insight into the total neutrophil content of lung tissue ( Figure 4N). These analyses revealed no differences between groups, indicating that the lower pulmonary bacterial loads detected in IRAK-M -/mice 6 h after infection could not be explained by an accelerated recruitment of neutrophils to the primary site of infection. During progressed pneumonia (24 and 48 h after infection), pulmonary MPO concentrations strongly increased in both groups and significantly more so in WT mice (see Figure 4N), probably as a consequence of the much higher bacterial burdens in these animals relative to IRAK-M -/mice.

IRAK-M Deficiency Does Not Influence the Capacity of Alveolar Macrophages to Phagocytsoe Klebsiella
Considering the reduced bacterial loads in lungs of IRAK-M -/mice very early after infection, we were interested to examine the impact of IRAK-M deficiency on the capacity of alveolar macrophages to phagocytose Klebsiella in vitro.

IRAK-M Deficiency Enhances Klebsiella-Induced Cytokine and Chemokine Release by Alveolar Macrophages Ex Vivo but Does Not Impact on Mediator Release in Lungs In Vivo
IRAK-M -/bone marrow-derived macrophages have been reported to produce higher levels of proinflammatory cytokines upon incubation with bacteria such as Escherichia coli, Salmonella typhimurium and Listeria monocytogenes (6). To determine the impact of IRAK-M on cytokine and chemokine release by alveolar macrophages exposed to Klebsiella, we incubated primary IRAK-M -/and WT alveolar macrophages for 16 h with either growtharrested K. pneumoniae or LPS derived from K. pneumoniae, and measured TNF-α, IL-6, CXCL1 and CXCL2 in culture supernatants. IRAK-M -/macrophages released increased amounts of all four mediators, although for IL-6 and CXCL2 the difference with WT cells did not reach statistical significance (Figures 6A-D). Next, we determined whether IRAK-M deficiency influences the release of cytokines and chemokines in the lung during Klebsiella pneumonia in vivo. First, we focused on early (3 h and 6 h) mediator release into the bronchoalveolar space, considering that alveolar macrophages are the most prominent resident leukocytes there (Figures 6E-H). Although the concentrations of TNF-α, IL-6, CXCL1 and CXCL2 measured in BALF harvested from IRAK-M -/mice 3 h after infection tended to be higher than those detected in BALF from WT mice, the differences between groups were not significant. At 6 h after infection, TNF-α and CXCL2 levels were even lower in BALF obtained from IRAK-M -/mice as compared to WT mice, indicating that the lower bacterial burdens in lungs of the former mouse strain had a greater impact on mediator release than the absence of IRAK-M. To more closely study the contribution of the Klebsiella load to pulmonary cytokine and chemokine concentrations, we measured TNF-α, IL-6, CXCL1 and CXCL2 in whole lung homogenates harvested 3 to 48 h after infection, spanning the period of a gradually growing bacterial burden (Figures 6I-L). TNF-α was the only mediator that displayed high concentrations early after infection, decreasing thereafter. In contrast, the lung levels of IL-6, CXCL1 and CXCL2 strongly increased as the infection progressed in WT mice, and, from 24 h onward, the levels of these mediators were all lower in IRAK-M -/mice. Together these data suggest that the bacterial load drives the extent of mediator production in the lungs, overruling the possible inhibiting effect of IRAK-M hereon.
Although IRAK-M expression originally was considered to be restricted to macrophages, recent evidence suggests that IRAK-M likely also is expressed by respiratory epithelial cells (11,12). To examine a possible effect of IRAK-M deficiency on the production of antimicrobial proteins produced by the respiratory epithelium, we measured lipocalin 2 and CCL20, which have been implicated in host defense against respiratory tract infection (24,25).  Early after infection (3 h and 6 h) lipocalin 2 and CCL20 concentrations did not differ in BALF obtained from IRAK-M -/and WT mice (Figures 7A, B). In lung homogenates lipocalin 2 and CCL20 levels were lower in IRAK-M -/mice as the infection proceeded (Figures 7D, E). To obtain further evidence that IRAK-M deficiency did not enhance the responsiveness of respiratory epithelial cells in vivo, we measured CXCL5, a CXC chemokine that is produced exclusively by epithelial cells (26,27), in BALF and lung homog enates and found no differences between IRAK-M -/and WT mice (Figures 7C, F). Finally, considering that mucosal immunity in the lung during pneumonia regulated at least in part is by Th17 cytokines (24), we measured IL-17 and IL-22, both of which have been shown to play a protective role in the host response to Klebsiella pneumonia (23,28). However, in both mouse strains IL-17 and IL-22, levels were either low or not above baseline concentrations in BALF obtained 3 h or 6 h after infection (data not shown).

DISCUSSION
Pneumonia represents a persistent and pervasive public health problem. In the United States, respiratory tract infections cause more disease and death than any other infection, and, in the last 50 years, mortality due to pneumonia has not changed significantly (29). Rapid recognition of pathogens that reach the lower respiratory tract is essential for an effective host defense against invading microorganisms (21). IRAK-M inhibits the intracellular IRAK-1 to IRAK-4 signaling complex, and, thereby, all MyD88 dependent receptors (6). A recent study implicated MyD88 as an important mediator of a protective immune response during Klebsiella pneumonia (30). Here, we show that IRAK-M strongly impairs host defense during pneumonia caused by K. pneumoniae, as reflected by reduced bacterial growth and dissemination and diminished distant organ injury in mice lacking IRAK-M. In accordance, IRAK-M -/mice showed a reduced mortality (40% versus 80% in WT mice), although the difference between groups did not reach statistical significance (P = 0.07). Conceivably, larger experimental groups and/or a slightly less severe model using a lower infectious dose would reveal a statistically significant survival advantage for IRAK-M -/mice.
A recent study implicated IRAK-M in the reduced resistance to secondary pneumonia caused by P. aeruginosa in mice suffering from polymicrobial peritonitis (9). In this investigation (9), nonlethal septic peritonitis was induced by cecal ligation and puncture to reproduce the immunecompromised state that frequently accompanies sepsis and that is considered to render the host more vulnerable to secondary (nosocomial) infections (2,31,32). Mice subjected to cecal ligation and puncture and subsequently infected with Pseudomonas via the airways showed a strongly enhanced bacterial growth and mortality due to the respiratory tract infection when compared with mice that underwent sham surgery. This vulnerable phenotype corresponded with enhanced expression of IRAK-M in lung macrophages harvested from mice with polymicrobial peritonitis and a strongly reduced capacity of these cells to release TNF-α upon incubation with P. aeruginosa. In con-trast, lung macrophages obtained from IRAK-M -/mice with sepsis released more TNF-α upon exposure to P. aeruginosa, which was associated with a markedly improved host defense against this nosocomial pathogen (9). As such, this previous study expanded knowledge on the role of IRAK-M in the immune suppression associated with sepsis (6)(7)(8)33). Thus far, the involvement of IRAK-M in the innate immune response to primary bacterial pneumonia in the previously healthy host remained unexplored.
Here, we used an established model of primary pneumonia caused by a common respiratory pathogen in which the bacterial load grows over time at the primary site of infection, followed by dissemination to distant body sites and sepsis, thereby allowing investigation of normal innate defense mechanisms in the respiratory tract in a clinically relevant setting (13,14). Previous studies have documented that the complete absence of MyD88 results in an impaired immune response to K. pneumoniae pneumonia (30). Together with the present results, these data suggest that the activity of the MyD88-IRAK-1-IRAK-4 signaling pathway, missing in MyD88 -/mice and less inhibited in IRAK-M -/mice, is a crucial determinant of antibacterial defense during Klebsiella pneumonia. Likely, the influence of IRAK-M on host defense during pneumonia involves its action on multiple MyD88 dependent receptors. Indeed, MyD88 dependent receptors contributing to host defense against Klebsiella pneumonia include TLR4 (34,35) and TLR9 (36); the potential roles of other MyD88 dependent TLRs and the IL-18 receptor have not been studied thus far, whereas IL-1 did not play a role of significance (37). Of interest, IRAK-M -/mice demonstrated reduced bacterial loads already very early after infection, suggesting an effect on Klebsiella by resident cells. Of note, however, we did not detect a difference in the capacity of IRAK-M -/and WT alveolar macrophages to phagocytose Klebsiella ex vivo.
The attenuated growth of bacteria in the lungs was accompanied by an enhanced inflammatory response at the tissue level early after infection (3 h and 6 h), as determined by semiquantitative scores of specific histological alterations characteristic for bacterial pneumonia. Although IRAK-M inhibits cytokine and chemokine production by macrophages stimulated by bacteria or purified TLR ligands in vitro ( Figure 6) (6,9), IRAK-M deficiency had little if any effect on the pulmonary levels of these mediators during airway infection in vivo; during late stages of the infection, cytokine and chemokine levels were even lower in the lungs of IRAK-M -/mice. The strongly reduced bacterial loads in IRAK-M -/mice, providing a diminished proinflammatory stimulus to immune cells in the lung, afford a likely explanation for these lower mediator levels in spite of the absence of the proximal TLR inhibitor IRAK-M. These findings differ from those in IRAK-M -/mice with polymicrobial peritonitis suffering from secondary Pseudomonas pneumonia, which showed increased lung levels of several cytokines (9). Apparently, the impact of IRAK-M on cytokine production is more prominent in the already compromised than in the previously healthy host. This may, also, at least in part, explain the fact that IRAK-M deficiency was associated with enhanced neutrophil recruitment to the lungs during secondary pneumonia following peritonitis (9), whereas it did not influence neutrophil influx during primary airway infection (this study). In the current investigation, IRAK-M deficiency did not enhance the production of either lipocalin 2 or CCL20, both antimicrobial proteins produced by the respiratory epithelium implicated in host defense against respiratory tract infection (24,25). In accordance, CXCL5, a chemo kine released exclusively by respiratory epithelial cells (26,27), was not altered in IRAK-M -/mice. Of note, IRAK-M -/type II alveolar epithelial cells were reported to release modestly increased amounts of CXCL1 and CXCL2 upon incubation with LPS in vitro (38). Nonetheless, our current data suggest that during Klebsiella pneumonia in vivo IRAK-M deficiency does not affect proinflammatory mediator release by respiratory epithelial cells to a significant extent.
In a very recent investigation, IRAK-M -/mice displayed an exaggerated inflammatory response in their lungs upon infection with influenza A virus (38).
In contrast to what we found here, the extent of lung inflammation in IRAK-M -/mice was prolonged over the course of 6 d and associated with an impaired antiviral response and increased lethality (38). Hence, in influenza, pneumonia IRAK-M serves to prevent detrimental inflammation thereby facilitating viral clearance. Clearly, this role is different from the function of IRAK-M during Klebsiella pneumonia, where the early inhibition of the lung inflammatory response resulted in enhanced bacterial growth and adversely affected outcome. Previous studies confirm the "double-edged sword" character of innate immunity during lung infection (39,40), wherein not only aberrant activation of leukocytes and epithelial cells, but also of endothelial cells resulting in vascular hyperpermeability, may contribute to collateral damage as a consequence of innate immune activation (41,42).
WT mice purchased from a commercial supplier (and not littermates) were used as controls. As such, differences between IRAK-M -/and WT mice could have been caused in part by differences in composition of the gut microbiome. Additional studies are warranted to examine whether IRAK-M deficiency influences the gut microbiome.
IRAK-M is a negative regulator of TLR signaling that occupies a crucial role in the activation of macrophages upon the first encounter of the immune system with pathogens. Here, we show that IRAK-M becomes upregulated rapidly in the lungs of mice infected with K. pneumoniae via the airways. Physiologically positioned to inhibit potentially damaging lung inflammation, IRAK-M dampens the early antibacterial response during Klebsiella pneumonia as reflected by lower bacterial counts in IRAK-M -/mice from 6 h after infection onward.

CONCLUSION
Our study is the first to demonstrate a function for IRAK-M in host defense against respiratory tract infection in the previously healthy host. Preliminary experiments performed by our laboratory indicate that IRAK-M also impairs host defense during gram-positive pneumonia (data not shown). Together these data identify IRAK-M as a possible therapeutic target in the treatment of bacterial respiratory tract infection.

ACKNOWLEDGMENTS
We thank Joost Daalhuisen, Marieke ten Brink and Regina de Beer for expert technical assistance. JJ Hoogerwerf, GJW van der Windt, DC Blok and AJ Hoogendijk were supported by institutional research grants from the Academic Medical Center, Amsterdam. RA Flavell is an Investigator of the Howard Hughes Medical Institute.

DISCLOSURE
The authors declare they have no competing interests as defined by Molecular Medicine, or other interest that might be perceived to influence the results and discussion reported in this paper.