|Home | About | Journals | Submit | Contact Us | Français|
Conceived and designed the experiments: HS ABJ. Performed the experiments: HS. Analyzed the data: HS ABJ. Contributed reagents/materials/analysis tools: HS ABJ. Wrote the paper: HS ABJ.
Neisseria meningitidis is a human-specific pathogen with capacity to cause septic shock and meningitis. It has been hypothesized that invasion of the central nervous system (CNS) is a complication of a bacteremic condition. In this study, we aimed to characterize the invasion route of N. meningitidis to the CNS. Using an intranasally challenged mouse disease model, we found that twenty percent of the mice developed lethal meningitis even though no bacteria could be detected in blood. Upon bacterial infection, epithelial lesions and redistribution of intracellular junction protein N-cadherin were observed at the nasal epithelial mucosa, especially at the olfactory epithelium, which is functionally and anatomically connected to the CNS. Bacteria were detected in the submucosa of the olfactory epithelium, along olfactory nerves in the cribriform plate, at the olfactory bulb and subsequently at the meninges and subarachnoid space. Furthermore, our data suggest that a threshold level of bacteremia is required for the development of meningococcal sepsis. Taken together, N. meningitidis is able to pass directly from nasopharynx to meninges through the olfactory nerve system. This study enhances our understanding how N. meningitidis invades the meninges. The nasal olfactory nerve system may be a novel target for disease prevention that can improve outcome and survival.
Neisseria meningitidis is a human-specific pathogen with the capacity to cause meningitis, which is an important cause of mortality and morbidity. Up to 50% of the survivors sustain neurological sequelae despite advances in antimicrobial therapy . As a commensal bacterium, N. meningitidis asymptomatically colonize to the nasopharyngeal mucosa of up to 10% of the healthy population . Under certain conditions, the bacteria can cross the epithelial barrier, enter and multiply in the bloodstream and disseminate to other organs including the central nervous system (CNS).
Sepsis and meningitis are the major clinical features of meningococcal disease. The high level of bacteremia together with bacterial adhesion to the endothelial cells have been suggested to be an important determinant for meningococcal traversal of the blood-brain barrier (BBB) , . In vitro studies have shown that N. meningitidis interacts with both human vein endothelial and brain microvascular endothelial cells. Both transcellular and paracellular traversal of the host cells have been demonstrated for N. meningitidis , . Pili-mediated microcolonies can recruit the cellular polarity complex upon adhesion and by this way disorganize the junction proteins at the cell-cell interface and allow bacteria to cross the BBB . Moreover, bacteria have developed diverse strategies for efficient survival in the blood stream and the septic condition could induce increased BBB permeability and pleocytosis, which are the hallmarks of meningitis .
Apart from our knowledge of the molecular and cellular mechanisms of meningococcal traversal of the BBB, the disease pathogenesis remains incompletely understood. It is important to note that the clinical symptoms of meningococcal infection are highly diverse. More than 60% of patients develop meningitis without septic shock . In cases of neonatal meningitis, meningococcal sepsis occurs in only 5–20% of patients . The complex nature of the disease indicates that meningococcal invasion of the CNS can be independent of systemic bacterial dissemination in the bloodstream. An alternative non-hemetogenous traversal route to the CNS, occurring in a parallel way with or preceding systemic dissemination, could exist.
N. meningitidis inhabits the mucosa of the human nasopharynx, which is unique for its anatomical connection to the CNS. The posterior nasal cavity is covered by olfactory epithelium, which contains peripheral nerves of the olfactory system. The olfactory nerve axons traverse the cribriform plate, a bone behind the nose, and terminate in the olfactory bulb, which is at the inferior side of the brain. Cerebrospinal fluid (CSF), produced in the choroid plexus, acts as a cushion that protects the CNS from shocks and supports the venous sinuses. The circulation of CSF is aided by the pulsations of the choroid plexus and by the motion of the cilia of ependymal cells. The drainage of CSF from the brain plays an important role in the homeostasis and metabolism of the central nervous system. It has been assumed that CSF is absorbed from the subarachnoid compartment by arachnoid villi and granulations . However, mounting evidences have proved that the lymphatic pathway external to the cranium and spinal cord has an important function in CSF absorption . The association between the CSF and nasal lymph compartments in both humans and other mammalian species has been documented . In vivo studies using animal models have shown that CSF can be removed from the cranium via transportation through the cribriform plate in association with the olfactory nerves. CSF is then absorbed into lymphatic vessels located in the sub-mucosa region of the olfactory and respiratory epithelium , .
In the present study, we aimed to investigate the dissemination route of N. meningitidis during invasive disease. CD46 transgenic mice that were challenged intranasally developed lethal meningitis even though no bacteria were detected in the bloodstream. We demonstrated that N. meningitidis is capable to invade the meninges through the olfactory nerve system. Bacterial colonization was associated with tissue damage and remodeling of N-cadherin expression patterns in the nasal olfactory epithelium.
The human cell surface receptor CD46 transgenic mouse model (CD46+/+) has been demonstrated to mimic many aspects of meningococcal disease in humans , , . In order to study disease pathogenesis and naturally occurring infection routes, we challenged mice intranasally and monitored disease development. CD46+/+ and B6C3F1 mice challenged with PBS were set as uninfected control. At 3 days post-infection, nasal washes were collected from mice to determine bacterial colonization of the nasopharyngeal mucosa. Significantly higher amounts of bacteria were detected in nasal washes of CD46 transgenic mice (2680±1378 CFU/ml) compared to non-transgenic mice (432±210 CFU/ml). Further, bacteria were found in only 10% of control mice, whereas, 40% of CD46 transgenic mice were detected to carry meningococci (data not shown). We repeated the experiment and health status of all infected mice was monitored over a 10 days period. Signs of the infection, such as neck stiffness, tremble and listless, were observed in 20% of infected CD46+/+ mice, which had decreased survival compared to control groups, 80% versus 100% (Fig. 1B). In contrast, all B6C3F1 mice survived and no bacteria were isolated from the CSF samples (data not shown). All mice in uninfected control groups survived and no bacteria were detected in nasal washes (data not shown).
To determine the underlying pathogenesis of the disease, we measured bacterial numbers in blood of infected mice each day post infection for 10 days. Surprisingly, we were unable to detect bacteria in blood samples of all infected mice (data not shown), not even in those CD46+/+ mice with symptoms of lethal disease (Fig. 1A). The detection limit of the method used in our study is 500 CFU/ml of tail vein blood. Thus, we can not exclude lower (<500 CFU/ml blood) levels of bacteria in tail vein blood. Also, increased numbers of bacteria could exist in other areas of the circulatory system. Nevertheless, no bacteria were found in homogenized peripheral organs, such as liver, kidney and spleen (data not shown). However, we could detect bacteria from nasal washes and CSF of the mice that showed disease symptoms, indicating bacterial colonization of the nasopharyngeal mucosa and development of meningitis (Fig. 1A). Bacterial infection of the CNS was further confirmed by in vivo bioluminescent imaging. A bioluminescent FAM20 strain , which behaves like the wild-type strain in growth and infection behavior, was given i.n. to mice. Bioluminescence signals emitted from the bacteria were monitored in vivo and in real time. As shown in Fig. 2A, bacterial signals were not identified in the circulation system. In contrast, a clear bioluminescence signal was detected in the brain region. To further confirm the location of bacteria, brain tissue sections were stained with an antibody against N. meningitidis. Bacteria were detected in cells of the meninges (Fig. 2B, right panel, arrowheads) and in the subarachnoid space (data not shown). No bacterial signal could be detected in uninfected control mice (Fig. 2B, left panel). Thus, meningococcal dissemination to the CNS appears to be independent of bacteremia.
As described above, we could not detect bacteria in the bloodstream after intranasal infection of N. meningitidis. A reason could be that bacteria are cleared rapidly by the host defense system in the blood and consequently cannot reach detectable levels. We therefore tested the kinetics of bacterial growth in whole blood collected from CD46 transgenic mice. At a moderate infection dose, i. e. 104 cfu bacteria/ml, bacterial growth rate was similar in blood and in the control with cell culture medium. However, a three-fold bacterial multiplication in blood was observed after infection with 105 cfu bacteria/ml. At lower infection dose, i. e. 103 cfu/ml, bacteria could not grow in the blood (Fig. 3). Similar results were found when human blood was tested (data not shown). The impact of initial infection dose upon bacterial survival in blood was further investigated by an in vivo assay. Bioluminescent FAM20 bacteria were injected i.p. to CD46+/+ mice and bacterial growth was monitored using an in vivo bioluminescence imaging camera. Infection of mice with 106 cfu resulted in bacterial clearance at 24 h post-infection. In contrast, when the infection dose was increased to 107 cfu/mouse, around 60% of the mice developed sepsis (data not shown). These data indicate that a threshold level of bacteria is required for successful multiplication of N. meningitidis in blood.
Nasopharyngeal mucosa is the natural niche for N. meningitidis. More than 50% of the nasal cavity surface in mice is covered by olfactory epithelium , which is a unique part of the nervous system containing olfactory sensory neurons. Bacterial colonization and histopathological alterations in sections of head tissue were therefore studied. Mice infected with meningococci had conspicuous morphological changes and tissue lesions in the olfactory epithelial region. Polymorphonuclear cells were abundant in the olfactory epithelium and in the nasal cavity resulting in an atypical multilayer structure (Fig. 4A, right panels). These characters were not observed in the saline vehicle-treated control mice (Fig. 4A, left panels). Furthermore, epithelial atrophy was observed upon infection, the average thickness of the olfactory epithelium after infection was reduced by more than 50% compared with control mice (Fig. 4B).
As shown above, although bacteria reached the CNS, they could not be detected in blood. Furthermore, tissue damage of the olfactory epithelium was observed upon meningococcal infection. We postulated that the olfactory nervous system, a structural and functional connection between the nasopharynx and the CNS could be a pathway for bacterial dissemination. Three days after infection, mice were sacrificed and the head tissue was split at the midsagittal plane, giving a complete anatomic structure between nasopharyngeal and the CNS. By immunohistochemistry straining of the tissue sections, N. meningitidis was detected at the olfactory epithelium of the nasal cavity (Fig. 5, red signal, arrowhead). Moreover, bacterial signals were also found at the basement membrane (BM) (Fig. 5, arrows) and in lamina propria (LP) (Fig. 5, empty triangles). We could not detect obvious bacterial signals in the olfactory epithelium (OE) (Fig. 5). We used an antibody recognizing olfactory marker protein (OMP) to stain the olfactory nerve system, including olfactory receptor cells in the olfactory epithelium, bundles of axons in lamina propria, the cribriform plate and the olfactory bulb. The olfactory receptor cells are bipolar neurons that form bundled axons at the base of the cells, penetrate the cribriform plate and reach the olfactory bulb of the brain. The above region was therefore studied carefully. Strong bacterial signals (red) were identified in the cribriform plate and at the nerve fiber layer of the olfactory bulb (Fig. 6A). All observed bacterial signals co-localized with the staining of olfactory marker protein, indicating that bacteria disseminated to the CNS through the olfactory nerve (Fig. 6B and C). We did not observe obvious bacterial signals in olfactory bulb tissue (Fig. 6A and C). In control mice, no bacterial signal was detected (Fig. 6D).
By using human brain vascular endothelial cells, Coureuil et al.  showed that pilus-mediated adhesion of N. meningitidis can lead to the formation of ectopic intercellular junction domains at the site of bacteria-cell interaction. This resulted in a depletion of junction proteins at the cell-cell interface, and an opening of the intercellular junctions, which would enable bacteria to cross the BBB and invade the meninges through paracellular gaps. To extend this finding, we next investigated whether a similar molecular mechanism of bacterial invasion exists under a non-hematogenous pathway. The expression of junction proteins (N-cadherin, ZO-1 and β-catenin) in the olfactory epithelium was analyzed by immunofluorescence staining. In uninfected mice, N-cadherin was expressed mainly in apical olfactory epithelium and bundles of olfactory nerve in lamina propria (Fig. 7A, left panels). After meningococcal infection, expression of N-cadherin was dramatically affected (Fig. 7A, right panels). Signal intensity of N-cadherin in the olfactory epithelium (OE) and the lamina propria (LP) was significantly reduced upon meningococcal infection compared to the uninfected control (Fig. 7B). The expression levels of ZO-1 and β-catenin in olfactory epithelium were not obviously affected upon infection (data not shown) indicating that the down-regulation of N-cadherin is infection specific.
Given the complex nature of the meningococcal disease, it has long been hypothesized that N. meningitidis could enter into the CNS via a non-haemetogenous invasion pathway . By using a humanized mouse disease model, we show here that N. meningitidis can invade the meninges through the olfactory nerve system. Induction of meningitis was confirmed by bacterial viable count assay of CSF samples, in vivo bioluminescence imaging and immunofluorescence staining of the brain tissue. An alternative explanation for our findings is that bacterial colonization of the olfactory nerve occurred as a second event following CNS penetration by other pathways. We find this highly unlikely since we could not detect bacteria in blood or peripheral organs. Upon bacterial infection, tissue lesions and remodeling of the intracellular junction protein N-cadherin was observed in nasal mucosa, especially at the olfactory epithelium, which functionally and anatomically connects the CNS and nasopharynx. Structural damages could trigger bacterial entry into the CNS. Indeed, bacteria were detected not only at the surface of olfactory epithelial cells, but also in the sub-mucosa region. Bacteria were further identified along the olfactory nerves in the cribriform plate, at the peripheries of the olfactory bulb and at the meninges and subarachnoid space. Our results highlight the important role of olfactory nervous system in meningococcal infection and might explain the presence of meningeal infection in patients with negative blood cultures.
Olfactory nerves together with the lymphatic system in nasal submucosa build up an anatomical and functional connection between respiratory epithelium and the CNS. Many neurotropic pathogens make use of this physiological route to invade the CNS. In line with this study, Marra  demonstrated that Streptococcus pneumoniae can spread to the CNS and cause meningitis without a bacteremic state. Furthermore, it has been suggested that the interaction of pneumococci with neuronal gangliosides is essential for direct bacterial entry into the CNS along olfactory nerves . Certain viruses, such as herpes simplex virus and influenza A virus, are able to reach the brain through the olfactory epithelium, along the nerve fibers of the olfactory, vagal and trigeminal , , . In a murine disease model, Naegleria fowleri can invade the CNS through olfactory nerves during the early stages of primary amebic meningoencephalitis (PAM). The amebas could be observed in the submucosal nerve plexus, olfactory nerves within the cribriform plate, and the olfactory bulb as early as 24 h post-infection .
The CNS is covered by meninges and is highly protected from most microorganisms and toxins that are circulating in the bloodstream because of the protective properties of the blood-brain barrier (BBB), which is a structural and functional barrier formed by the microvascular endothelium, astrocytes and pericytes. Intercellular tight junctions together with a low rate of pinocytosis make the BBB impermeable for large and hydrophilic molecules as well as for most pathogens , . Adhesion of meningococci to human brain vascular endothelial cells was shown to cause remodeling of cell-cell junction proteins and subsequent formation of gaps between infected cells, allowing bacteria to cross the BBB . In line with the in vitro studies, we found that bacterial colonization plays an important role in the development of meningitis in vivo, since invasive disease was only observed in colonized mice. Moreover, expression levels of N-cadherin in the olfactory mucosa were attenuated upon meningococcal infection. Meningococcal colonization led to conspicuous epithelial damage and tissue atrophy especially at the olfactory epithelial region. These findings are in accordance with a recently published study by Schubert-Unkmeir et al. , who revealed that cellular inflammatory mediators, for example matrix metalloproteinases, degrade intercellular junction proteins. This results in endothelial cell disruption and hence facilitates bacterial invasion. It was therefore speculated that infection-induced tissue damage together with remodeling of the host cell cytoskeleton and junction proteins could be a general mechanism used by N. meningitidis during tissue invasion.
Membrane cofactor protein CD46 is used as a cell receptor by a number of important pathogenic bacteria and viruses, including N. meningitidis . In this study, meningococci bound to the olfactory mucosa even though CD46 is expressed on the basolateral surface of the epithelium . Our data indicates that CD46 might not be involved in mediating the initial bacterial adhesion. Indeed, other cell receptors or strategies have been shown to participate in bacteria-host cell interaction. Carcinoembryonic antigen-related cell adhesion molecules (CEACAMs) can function as a receptor to mediate intimate binding of meningococci to epithelial cells through Opa proteins . Type IV pili component PilV can trigger cell membrane reorganization and by this way enhances bacterial cohesion . It might be hypothesized that the observed epithelial damage and decreased expression of junction protein allows for bacterial binding to CD46. How these events facilitate bacterial invasion will need further investigation. CD46 has also been considered as a regulator of the complement system by facilitating factor I-mediated inactivation of the activated complement proteins C3b and C4b . Many pathogens might recruit extracellular CD46 as a strategy to survive host complement defense as it has been demonstrated that both N. meningitidis and S. pyogenes can induce systemic infection in CD46 transgenic mice , , .
Several non-CD46 expressing mice models have been evaluated to study the earlier stage of meningococcal infection by an intranasal infection route. Infant NIH mice developed both sepsis and meningitis, however, the requirement for lung colonization to precede bacteremia limit the use of this model to study meningococcal pathogenesis. Furthermore, the impact of the premature immune system to the disease development is not clear . When adult Swiss Webster mice were studied, even though a long-term mucosal colonization was maintained, no invasive disease could be observed . Invasive disease can be observed in adult C57BL and Swiss Webster mice following intraperitoneal infection, however, enhancement substances with unknown effects on the host immune response are essential . We have shown that CD46 transgenic mice are susceptible to meningococcal infection. Comparable disease presentations, such as sepsis, meningitis and upper respiratory mucosal colonization, have been observed in adult mice after i.p. or i.v. challenge without using any enhancement substances , . In this study we showed the adult CD46 transgenic mice can also develop invasive meningococcal disease following intranasal infection. Meningitis can be induced even though no bacteria in the blood or peripheral organs can be detected. It is likely that a threshold level of bacteremia is required for the development of sepsis as demonstrated in this study by both in vivo and in vitro assays. Strain-specific capacity to survive in the blood stream could be another reason since JB515, a serogroup W-135 strain used in the previous work, did result in bacteremia . Nevertheless, species-specific differences of the complement system between mouse and human should be considered when conclusions are drawn from animal models.
Taken together, the present study provides compelling evidence that N. meningitidis is able to invade the CNS and induce meningitis through the olfactory nerve system in the absence of bloodstream dissemination. Further studies of the bacterial and cellular components involved in this novel invasion route could lead not only to a complete understanding of the pathophysiology of the disease, but also more efficient therapeutic strategies.
Mice experiments described in the present study were conducted at the animal facility of Wenner-Grens Institute, Stockholm University. Animal care and experiments were conducted adhering to the institution's guidelines for animal husbandry. All protocols were approved by the Swedish Ethical Committee on Animal Experiments (Approval ID: N380/08).
N. meningitidis FAM20 belongs to serogroup C, MLST sequence type 11. The bioluminescent FAM20 strain, FAM20::Lux, has been described previously  and behaves like the wild-type strain in growth and infection behavior. Bacteria were grown at 37°C in a 5% CO2 atmosphere on GC agar (Difco) supplemented with Kelloggs .
The hCD46Ge transgenic mouse line (CD46+/+) was created using B6C3F1 hybrids. It harbors the complete human CD46 gene and expresses CD46 in a human-like pattern . The F1 generation of C57BL/6 and C3H/Hen mice, i.e. B6C3F1, was used as non-transgenic control in this study.
N. meningitidis grown on GC agar was suspended in phosphate-buffered saline (PBS). Each mouse was challenged intranasally (i.n.) with 107 bacteria in 10 µl PBS once per day for two days. Prior to challenge, mice were treated with antibiotics as previously described . The health status of the mice was closely monitored for 10 days and mice were sacrificed as soon as the end point was reached. CD46+/+ and B6C3F1 mice challenged with PBS alone were set as uninfected control. Experiments were performed with 6–8 week old mice (n=10–12 mice per group) and repeated three times.
After bacterial challenge, 5 µl of blood sample was collected from the tail vein each day during the whole experiment. The blood was diluted in 245 µl of GC liquid, and 100 µl was plated on a GC agar plate. The remaining sample was used for further plating of serial dilutions. Plates were incubated over night at 37°C in a 5% CO2 atmosphere. The number of colony forming units (CFU) was counted next day. The limit of detection is 500 CFU/ml blood.
Cerebrospinal fluid were collected either at the end of the experiment or at the point when the disease signs were observed. Mice were sacrificed by inhalation of isoflurane (Forene, Abbott) and the skin of the back neck was disinfected with ethanol. The skull was exposed through a midline incision and the cisterna magna was punctured and approximately 1 to 2 µl of CSF was collected from each mouse. The CSF samples, which were immediately suspended in 100 µl of GC liquid, were examined for the absence of red blood cells before spreading on the GC agar plates. The number of bacterial CFU was determined after an overnight incubation at 37°C in a 5% CO2 atmosphere.
Nasal washes were collected either at the end of the experiment or at the point when the disease signs were observed. Mice were sacrificed by inhalation of isoflurane and the skin of the front neck was disinfected with ethanol. The trachea of mouse was then isolated, 200 µl of PBS was rinsed from the lower end of the trachea and collected from the external narines. The nasal washes were immediately checked for the absence of red blood cells and spread on the GC agar plates containing kanamycin (50 µg/ml) in order to inhibit the overgrow of the normal flora. The number of bacterial CFU was determined after an overnight incubation at 37°C in a 5% CO2 atmosphere.
Mice were anaesthetized with isoflurane and imaged for a maximum of 2 min using an IVIS imaging system (Xenogen Corporation/Caliper Life Sciences) according to the manufacturer's instructions. The dark fur on the back and the head of the mice was shaved away in order to increase the sensitivity of signal detection.
Mice were sacrificed and perfused using 4% formaldehyde in PBS. The head was split in the midsagittal plane, fixed in 4% paraformaldehyde, decalcified in 10% fomic acid and embedded in paraffin. Sequential tissue sections obtained from three mice were analyzed and images from one mouse are presented. For light microscopy, tissue sections were stained with hematoxylin and eosin. Hematoxylin stains negatively charged nuclei dark blue and eosin stains other tissue structures pink. The olfactory epithelium is thereby visualized as a region with a high density of heavily hematoxylin-stained cells. In order to investigate epithelial atrophy, six fields of epithelial layer without obvious neutrophils influx, which could affects the epithelial integrity, were randomly selected and the thickness of the epithelium was analysed by a Carl Zeiss Axio Vision 2.05 image processing and analysis system (Zeiss). The average thickness of the olfactory epithelium is presented.
The primary antibodies applied in immunofluorescence staining are: Goat anti-Olfactory Marker Protein (1100, Waco Chemicals, USA); Mouse anti-N-cadherin (150, Sigma-Aldrich); Goat anti-ZO-1 (1100, Santa Cruz Biotechnology); Goat anti-β-catenin (1100, Santa Cruz Biotechnology) and Rabbit anti-N. meningitidis (1100, USBiological). Before staining, tissue sections were blocked with Image-iT™ FX Signal Enhancer (Invitrogen) for 30 min at room temperature (RT), and subsequently incubated with primary antibody diluted in PBS for 1 h at RT. After three washes in PBS for 5 min and staining with Alexa Fluor 488-, 594-conjugated secondary antibody (1100 diluted in PBS, Invitrogen) for 1 h at RT, slides were washed and mounted in DAPI containing Prolong gold anti-fade reagent (Invitrogen). Controls for unspecific binding by both primary and secondary antibodies were included in each staining. Image acquisition was performed by a Carl Zeiss Axio Vision 2.05 image processing and analysis system (Zeiss).
Whole blood was obtained from healthy volunteers without previous immunization with anti-meningococcal vaccine or collected from the orbital vein of the mice. Heparin (10 units/ml blood) was added to prevent coagulation. N. meningitidis suspensions were prepared in DMEM cell culture medium (Invitrogen), and mixed with the whole blood at a ratio of 11 to final concentrations of 103 to 105 bacteria/ml blood. The bacterial solution was then incubated at 37°C in a 5% CO2 atmosphere. At different time points, bacteria were serially diluted in GC liquid and spread on GC agar plates to enumerate the surviving colonies after an overnight incubation.
The nonparametric Mann-Whitney test was used to analyze differences between groups. All tests were two-tailed and data were considered significant if P value was less than 0.05.
Competing Interests: The authors have declared that no competing interests exist.
Funding: This work was supported by the Swedish Research Council (Grant nr 2006-2911, 2006-6304, 2006-4112, 2006-5073, 2007-3369, 2008-2572, 2008-3367 and 2009-5679), The Swedish Cancer Society, Knut och Alice Wallenbergs Foundation, Torsten and Ragnar Soderbergs Foundation, Swedish Society for Medicine, Magnus Bergvalls Foundation, Olga Jeanssons Foundation, Lars Hiertas Minne Foundation, Harald och Greta Jeanssons Foundation, Petrus och Augusta Hedlunds Foundation, Anna Brita och Bo Castegrens Minne Foundation, Tore Nilsons Foundation, Åke Wiberg Foundation. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.