Background
Streptococcus pneumoniae is a major cause of serious infections such as sepsis, pneumonia and meningitis. Despite advances in pneumococcal vaccines and effective antimicrobial therapy, the disease burden of invasive pneumococcal disease remains high, especially in resource-poor countries [
1,
2]. Pneumococcal meningitis is a severe form of bacterial meningitis in children and adults [
2‐
5]. The mortality rate ranges from 16 to 37 % in developed countries and up to 51 % in resource-poor areas [
1,
4]. Approximately 30 to 52 % of patients surviving pneumococcal meningitis have disabling long-term neurological sequelae, such as focal neurologic deficits and cognitive slowness [
5‐
7].
Susceptibility to and severity of pneumococcal meningitis are determined by host as well as pathogen characteristics [
8,
9]. Immune status and disruption of the natural barriers of the brain are well-recognized factors influencing host susceptibility [
1]. In recent years, the host’s genetic make-up has been increasingly recognized to determine susceptibility, for instance, due to genetic variation in innate immune receptors (Toll-like receptor 4), Fc gamma (Fc-γ) receptors and complement system [
1,
8]. Also, the make-up of the pathogen is important; pneumococci harbour an array of virulence factors [
10,
11]. The most important of these is the polysaccharide capsule with over 90 distinct serotypes identified. Carriage rates and invasiveness differ for the different serotypes [
12]. The capsule protects the bacteria from opsonophagocytosis and inhibits complement activation [
13]. Other important virulence factors include the cytolytic toxin pneumolysin and several cell-surface proteins, such as pneumococcal surface protein A (PspA) [
10,
11]. The relationship between the bacterium and the host drives pneumococcal genome variation; less than 50 % of pneumococcal genes is present in all strains (the core genome) exemplifying this genome variability [
14]. Both the presence and absence of genetic regions but also single nucleotide variations in the core genome can increase the pathogen’s capacity to cause disease and influence disease severity [
9,
15].
To study pneumococcal virulence, different experimental murine models have been developed [
16‐
18]. Limitations of these murine models include ethical issues, high costs and time needed for experiments; these limitations render mice not suitable for large-scale screening [
17]. The zebrafish (
Danio rerio) has emerged as a powerful vertebrate model to study infectious diseases caused by human pathogens or their related animal pathogens [
19‐
22]. Zebrafish are teleost fish with an innate and adaptive immune system similar to the human immune system [
22‐
24]. The innate immune system is already active at very early stages during zebrafish embryo development, whereas the adaptive immune system is active 4–6 weeks post fertilization [
23]. Another advantage of this model is the unique ability to study host-pathogen interaction in real time because of the transparency of zebrafish embryos and the wide range of available fluorescent tools [
22,
25]. Other advantages include high fecundity of the zebrafish, external development of the embryo and availability of gene-editing tools and tools to manipulate gene expression [
26]. Additionally, this model can be used for medium-throughput screening to identify bacterial mutants with altered virulence or medium-throughput screening of pharmacological compounds [
27,
28]. Recently, this model has been adapted to visualize and study mycobacterial meningitis — a central nervous system infectious disease [
29]. Moreover, it has been shown that zebrafish embryos as well as adult zebrafish are susceptible to pneumococcal infection and develop meningitis [
30,
31].
The aim of our study was to develop a zebrafish embryo infection model of pneumococcal meningitis that allows for real-time analysis of the host-pathogen interaction. To this end, we infected zebrafish embryos with a highly green fluorescent strain of pneumococcus that is still fully virulent [
32]. Visualization of the infection was further improved by using a transgenic fish line (
kdrl:
mCherry) that has red fluorescent blood vessels, in combination with fluorescent far red staining of phagocytic cells with fluorescently labelled anti-L-plastin. Phagocyte dynamics were studied in more detail in a double-labelled
mpx:
GFP/
mpeg1:mCherry zebrafish line with green fluorescent neutrophils and red fluorescent macrophages.
Discussion
We developed and characterized a zebrafish embryo infection model of pneumococcal meningitis allowing real-time investigation of early host-microbe interaction. Meningitis developed both after systemic injection in the caudal vein or local injection in the hindbrain ventricle. Infection with a pneumolysin-deficient pneumococcal mutant strain in the hindbrain ventricle showed attenuated growth in the subarachnoid space and attenuated migration through the brain as compared to the wild-type strain. In the wild-type strain infection, the number of phagocytes reduced quickly after initial accumulation at the site of infection, in contrast to the pneumolysin-deficient mutant infection, where numbers of phagocytic cells kept accumulating in the subarachnoid space. This observation suggested that cytolytic activity, mediated by pneumolysin, may be responsible for the reduction of phagocytic cells. Time-lapse imaging showed that the initial zebrafish phagocytic innate immune response in pneumococcal meningitis mainly consisted of neutrophils, comparable to the human situation [
4].
Zebrafish embryos and larvae are becoming increasingly popular to model infectious diseases, including infections of the central nervous system [
19,
21,
26,
29,
46,
47]. The optical clarity of zebrafish embryos and larvae in conjunction with transgenic zebrafish lines, fluorescently labelled bacteria and immunohistochemistry provide unique possibilities for real-time in vivo imaging of infection dynamics in the central nervous system in detail. This approach has led to the successful modelling of tuberculous meningitis and
Streptococcus agalactiae meningitis in zebrafish embryos and better understanding of the molecular and cellular pathogenesis of the disease [
29,
46]. Given the potential of this model to study central nervous system infections, it is therefore recommended and desirable to further adapt the zebrafish embryo to study other forms of bacterial meningitis. Since it has been demonstrated that zebrafish are susceptible to
Listeria monocytogenes,
Streptococcus suis,
Streptococcus iniae and
Escherichia coli infection, zebrafish meningitis models should be developed for these bacteria and opportunities for other meningitis-causing pathogens explored [
48‐
51].
Injection of wild-type
S. pneumoniae in the hindbrain ventricle or caudal vein caused a fulminant dose-dependent infection in zebrafish embryos. Caudal vein injection was associated with more severe disease outcome as compared to hindbrain ventricle injection, suggesting tissue-specific susceptibility to pneumococcal infection. A similar trend was observed in
Staphylococcus aureus-infected zebrafish embryos, where infection of the hindbrain ventricle elicited a stronger immunological response as compared to systemic infection [
47]. A recent study showed that the innate immune response to pneumococcal infection in zebrafish embryos is highly dependent on phagocytic cells (macrophages and neutrophils) [
30]. The difference in phagocyte recruitment upon injection via different routes may explain the difference in survival that we observed. The association between lack of leukocyte response and adverse outcome has been described before: a low cerebrospinal fluid white-cell count was associated with an adverse outcome in patients with pneumococcal meningitis [
5]. Studies in rats showed a relation between large numbers of bacteria in the cerebrospinal fluid load, lack of response of cerebrospinal fluid leukocytes and intracranial complications [
52].
Pneumolysin is a crucial multifunctional virulence factor, which is best known for its cholesterol-dependent cytolytic activity of host cells, but also induces activation of the complement pathway, activation of pro-inflammatory immune cell reactions and induction of apoptosis [
53]. The role of pneumolysin in the pathogenesis of pneumococcal meningitis has been controversial [
54]. Recent studies, however, show that pneumolysin plays an important role in the pathogenesis of pneumococcal meningitis and that unfavourable outcome in meningitis is driven by a combination of bacteria and host-derived toxins [
4,
54‐
57]. In line with these studies, the pneumolysin-deficient mutant was attenuated as compared to the wild-type strain after hindbrain ventricle injection; real-time imaging showed the differences in innate immune response upon infection with these strains. After infection of the hindbrain ventricle with wild-type pneumococci, large numbers of phagocytes migrated to the site of infection. However, as the infection progressed, the numbers of phagocytes diminished over time in the presence of increasing numbers of bacteria. In contrast, after hindbrain ventricle infection with pneumolysin-deficient pneumococci, phagocytes remained present in large numbers while bacteria were cleared over time. These observations may be explained by a biological phenomenon called apoptosis-associated killing of bacteria. Macrophage apoptosis has been described as a mechanism for pneumococcal clearance when other killing mechanisms are exhausted and is initiated by lysosomal membrane permeabilization [
58‐
60]. Induction of this mechanism by pneumococci requires opsonization and is correlated with the intracellular bacterial burden [
59‐
61]. A recent study shows that pneumolysin is necessary for lysosomal membrane permeabilization and thus induction of macrophage apoptosis-associated killing of pneumococci [
62]. In addition, in vitro studies show that infection using pneumolysin-deficient strains resulted in a significant reduction of macrophage apoptosis [
59,
63]. Altogether, these data strongly suggest that the observed differences in host innate immune response between infection with wild-type and pneumolysin-deficient pneumococci may be due to the ability and necessity to activate this macrophage apoptosis-associated killing mechanism.
Bacterial meningitis develops when bacteria enter and survive in the bloodstream, interact with the BBB and penetrate the central nervous system [
4]. In pneumococcal meningitis, crossing of the BBB by
S. pneumoniae is thought to occur by intracellular or intercellular translocation, although the exact mechanisms remain unclear [
4]. The BBB is formed by endothelial cells with tight junctions, astrocytes and pericytes, and the main function is to protect the central nervous system from microorganisms and toxins that are circulating in the blood [
64]. Previous studies showed that zebrafish have a functional BBB similar to that of mammals, and are therefore suitable for studying mechanisms involved in the disruption and penetration of the BBB [
43‐
45]. A recent study by Kim et al. demonstrated that infection with
S. agalactiae in zebrafish induces the Snail1 host transcription factor, which downregulates tight junctions, and disrupts the BBB [
65]. In order to investigate whether the zebrafish embryo model can be used to study pneumococcal crossing of the BBB, we infected
Tg(kdrl:mCherry)
s896 zebrafish embryos that express red fluorescence in the blood vessels with green fluorescent pneumococci. Wild-type pneumococci injected in the caudal vein migrated out of the blood vessels and caused meningitis in zebrafish embryos before as well as after the formation of the BBB. Histopathological analysis confirmed these findings and showed bacteria in the subarachnoid space and brain parenchyma in both 2 and 4 dpf zebrafish embryos infected systemically. This is in line with data from adult zebrafish, where intraperitoneal injection of pneumococci causes bacteraemia and subsequent meningitis [
31]. These findings suggest that the zebrafish embryo model is suitable to elucidate the mechanism by which pneumococci cross the BBB in meningitis.
In our analysis, we also detected clogging of the blood vessels by pneumococci in the bloodstream-infected zebrafish embryos, with bacteria localized outside and in proximity to these affected vessels. In addition to the aforementioned mechanisms by which pneumococci can infiltrate the central nervous system, mechanical disruption of vascular endothelium by pneumococci could possibly be another mechanism by which pneumococci leave the bloodstream and invade the brain. Furthermore, clogging of the blood vessels may cause interruption of the blood flow and subsequent cerebral infarction in zebrafish embryos. In patients with pneumococcal meningitis, cerebral infarction has been described as a common complication [
66‐
70]. Whereas the exact mechanism remains to be elucidated, previous studies show that severe infection can activate the coagulation pathway and diffuse intravasal coagulation may contribute to the pathogenesis of cerebral infarction [
69,
71].
Although there are many advantages of the zebrafish as an infection model, there are also some limitations. First, most human pathogens are adapted to cause infection at 37 °C, whereas the ideal temperature for zebrafish is around 28 °C. The difference in temperature might influence the natural disease course of human pathogens in these animals. Translation from the zebrafish model to the human infectious disease might therefore not always be possible. Second, monoclonal antibodies directed to surface antigens of cells of the zebrafish immune system are scarce [
21,
72]. Finally, the immune cells of the adaptive immune response that have been assumed to play a role in the innate immune response to pneumococcal infection show a different pattern in zebrafish as compared to mice and human [
31,
73]. Moreover, there is evidence that zebrafish have a tissue-restricted expression of Toll-like receptors and the repertoire of components of the zebrafish innate immune system seems to be more diverse that in mice or humans [
74,
75]. Despite these differences, the zebrafish embryo model has been proven very useful to study several human pathogens, e.g.
Mycobacterium tuberculosis, and has provided important new insights in the pathogenesis of tuberculosis [
76]. Also, our findings with respect to the pathogenesis of pneumococcal meningitis appear in line with those found in other animal models. Therefore, the zebrafish remains a powerful model organism to study infectious diseases.
Abbreviations
BBB, blood-brain barrier; BSA, bovine serum albumin; CFU, colony-forming units; Fc-γ, fc gamma; GFP, green fluorescent protein; HlpA, histone-like protein; hpf, hours post fertilization; hpi, hours post injection; kdrl, kinase insert domain receptor like; mpeg, macrophage expressed gene; mpx, myeloperoxidase; NGS, normal goat serum; PBS, phosphate-buffered saline; PBTx, Triton X-100 in phosphate-buffered saline; ply, pneumolysin; PspA, pneumococcal surface protein A; PTU, 1-phenyl 2-thiourea; sfGFP, superfolder green fluorescent protein; Tg, transgenic
Acknowledgements
The authors would like to thank Gunny van den Brink-Stempvoort, Lisanne van Leeuwen, Ben Nelemans, Manuel Schmitz, Wim Schouten and Theo Verboom for technical assistance. The anti-L-plastin antibody was a kind gift from Professor Paul Martin (Bristol University, UK).