Intestinal helminth co-infection is an unrecognised risk factor for increased pneumococcal carriage density and invasive disease

Infection with Streptococcus pneumoniae is the leading cause of death in children and burden of disease is greatest where helminth infections are also common. We investigated the impact of intestinal helminth co-infection on pneumococcal carriage; a risk factor for invasive disease. We used a mouse co-infection model and clinical data to assess the impact of co-infection on carriage density. Co-infection in mice was associated with increased pneumococcal carriage density and dissemination into lungs. Helminth-infected children also exhibited increased carriage density as compared to uninfected children. Anthelmintic treatment may be a cost-effective method of reducing pneumococcal disease burden in lower-income countries.


Results
Intestinal helminth co-infection increases pneumococcal carriage density and leads to significantly enhanced dissemination into lungs. We established a T. muris (mouse whipworm) and S. pneumoniae co-infection model to determine whether STH infection altered pneumococcal carriage density, and/ or dissemination into the lower airways. Mice were infected with T. muris, and 28 days later, when the worms had matured into adults, mice were infected with a carriage-inducing dose of S. pneumoniae. Co-infection with T. muris led to a consistent increase in nasopharyngeal carriage density over time, which may be biologically important, although this did not reach statistical significance (P = 0.062, Fig. 1A). In the lungs, a substantial difference was observed whereby co-infected mice had significantly increased pneumococcal dissemination into their lungs at 1, 3, and 7 days post-infection (Fig. 1B). Co-infection led to a significant increase in the number of macrophages within the nasopharynx compared to mono-infected animals (P = 0.0118, Fig. 1C) and a clear trend towards increased neutrophils (P = 0.0561, Fig. 1D), which was reversed by anti-helminthic treatment (P = 0.0083 and 0.0019 for macrophages and neutrophils respectively). Anthelminthic treatment also led to a small (but non-significant, P = 0.075) reduction in nasopharyngeal bacterial load (Fig. S1).
Intestinal helminth co-infection is a risk factor for increased pneumococcal carriage density in Ecuadorian children. We next investigated whether STH co-infection was also associated with increased pneumococcal carriage density in children. A total of 387 children enrolled in the ECUAVIDA study (age range 54-95 months, median = 60 months) were tested for pneumococcal carriage as measured by lytA PCR per- Mice were infected with 20 T. muris eggs by oral gavage, followed by induction of pneumococcal carriage 28 days later. Co-infected mice had a trend towards increased nasopharyngeal carriage density as compared to pneumococcus only infected animals (A) and this was accompanied by significantly increased pneumococcal dissemination into lungs of co-infected animals (B). Co-infection was accompanied by increased macrophages (C) and neutrophils (D), which was reversed by anthelminthic treatment. S.p. = S. pneumoniae, T.m. = T. muris, MBZ = mebendazole, 6 h = 6 hours, 1d = 1 day. Mean and SEM are indicated, *(P < 0.05) and **(P < 0.01). Broken line indicates limit of detection (50 colonies). www.nature.com/scientificreports/ formed on oropharyngeal swabs, collected as described 9 . Pneumococcal carriage density (as measured by number of lytA copies) was higher in STH-infected children than uninfected children (Fig. 2), although this did not reach significance. A. lumbricoides infections were associated with the largest increase in pneumococcal density (P = 0.029, univariable analysis) of all the factors measured (Table 1). Multivariable analysis was performed with all helminth infections and age, sex and household overcrowding were included a priori 10 . PCV13 vaccination status was not included in the analyses, as only 1.8% of children sampled had received the vaccine.

Discussion
The impact of a single pathogen is likely multiplied in the context of one or more co-infections in complex and challenging ways, particularly for STHs, which have evolved to modulate the host immune response, in order to favour chronicity of infection 4 . The impact of STH co-infection on pneumococcal disease burden and outcome is largely unknown despite sharing prevalence in the same geographical regions of the world. Therefore, we sought to study this using a combination of mouse models and human data. Here, we demonstrate that the murine intestinal helminth, T. muris, enhances pneumococcal (D39) nasopharyngeal carriage density and significantly increases pneumococcal CFU load in the lungs. Although D39 is virulent in mice, it does not cause invasive pneumonia when used in models of carriage, in the absence of other stimuli, instead providing a stable and reproducible model of long-term asymptomatic nasopharyngeal infection. It thus offers an opportunity to investigate bacterial, host and environmental factors which control carriage or that perturb stable nasopharyngeal infection towards invasive disease [11][12][13] . We show that co-infection with T. muris drives host immune cell infiltration with increased macrophages and neutrophils in the nasopharynx of co-infected mice; suggesting a pro-inflammatory environment conducive to increased pneumococcal lung dissemination, as has been shown previously 14 . Interestingly, increases in pneumococcal carriage density and cellular infiltration were both reduced by anthelmintic treatment. Similar mechanisms may also be at play in humans, as we observed increased pneumococcal carriage density in helminth-infected Ecuadorian children, as compared to uninfected children, and previous studies have shown a link between increased carriage density and pneumococcal pneumonia in both mice and humans 11,14-17 . As the lytA qPCR is not as specific for oropharyngeal swabs as it is for nasopharyngeal swabs, we sought to confirm pneumococci with a second PCR for cpsA; there was a positive correlation between pneumococcal DNA copies of both genes as determined by cpsA and lytA qPCR (P < 0.0001). The largest difference in pneumococcal density was recorded in A. lumbricoides infected children, however, it is not possible to study chronic A. lumbricoides infection in mice since the parasite is expelled before the adult worms establish in the gastrointestinal tract 18 , which is why we opted for another roundworm species, T. muris, a close relative of T. trichuira 19 . Our study had a relatively small sample size (68 lytA positive children out of 387 in total), and therefore it is possible that interactions may have been missed where effect sizes were small, for instance in the case of T. trichuira co-infection. Similarly, some of the observations reported using our mouse model fall short of reaching statistical significance (for example, the increase in nasopharyngeal neutrophil count for co-infected versus those infected with pneumococci only). Future studies should seek to further validate these observations as well as providing additional mechanistic understanding of these processes. www.nature.com/scientificreports/ Overall, these data suggest that coordinated deworming programmes alongside pneumococcal vaccination may aid in the reduction of pneumococcal disease in areas of the world where STH infections are also common. Pneumococcal carriage is known to be a risk factor for pneumococcal pneumonia, and non-vaccine therapeutic methods to reduce carriage density could also have a major impact on the prevalence of pneumococcal pneumonia at low economic cost 7 .
In summary, our data suggest that intestinal helminth co-infection is associated with increased levels of pneumococcal carriage and a clear and significant increase in bacterial dissemination to the lower airways. Given that the incidence of pneumococcal disease is highest in areas of the world where STH infections are most common, these data suggest that anthelminthic treatment could be a cost-effective method to reduce pneumococcal disease in lower-income countries.

Methods
All experimental protocols were approved by either the University of Liverpool Ethical and Animal Welfare Committee or by the ethics committees of the Hospital Pedro Vicente Maldonado and the Universidad San Francisco de Quito.

Mice. All experimental protocols were performed in accordance with the Home Office Scientific Procedures
Act (1986), Project Licence P86De83DA, and the University of Liverpool Ethical and Animal Welfare Committee, which approved the study. All procedures were carried out in compliance with ARRIVE guidelines. All procedures were carried out on MF1 mice 6-8 weeks or older. Mice were acclimatised for one-week prior to use and had free access to food and water in individually-ventilated cages.
Bacteria. S. pneumoniae Serotype 2 strain D39 (NCTC 7466) was used for all in vivo experiments. Bacteria were cultured as described previously 11 . Table 1. Linear regression analysis of factors associated with S. pneumoniae density in children aged 54-95 months who are carriers. Analysis was restricted to positive carriers (no. of carriers/total for each category is indicated in summary statistic column) and results are reported as linear regression coefficients and 95% CI. All helminth infections alongside age and sex were then used in a multivariable regression model to estimate adjusted coefficients and identify independent associations. a Coeficient was log10 converted, e.g. those with A. lumbricoides infection have a pneumococcal density of 0.3 log10 higher than those with no active infection. b The first component that accounted for 30% of variation was divided into tertiles to represent low, middle and high socioeconomic status as described by Menzies et al. (2014) 22  www.nature.com/scientificreports/ Parasites. The Edinburgh (E) strain of T. muris was used for all experiments. Parasite maintenance was carried out at the University of Manchester, as described previously 20 .

Mouse model of pneumococcal/helminth co-infection. Mice were infected by oral gavage with 20 T.
muris embryonated eggs in 200µL ddH 2 O. Control mice were sham infected with 200µL ddH 2 O. Twenty-eight days later, mice were lightly anaesthetised with a mixture of O 2 and isofluorane and infected intranasally with 1 × 10 5 colony-forming units (CFU) S. pneumoniae D39 in 10µL PBS, as previously described 11 . In those animals treated with the anthelmintic, mebendazole, the drug was administered by oral gavage once daily (100 mg/kg) for three days, immediately after S. pneumoniae infection. Control mice (S.p. only and S.p. + T.m. groups) were sham treated with 200µL PBS by oral gavage. At pre-selected time intervals following intranasal infection, mice were humanely culled, nasopharynx and lungs were removed and blood was collected via cardiac puncture. A viable count of bacteria was determined at each interval by serial dilution of tissue homogenates onto blood agar, as previously described 11 . Worm burden was measured as described previously 20 . Single cell suspension from the nasopharynx were stained as follows: CD45-FITC, GR1-PE-CY7 (RB6-8C5, BD Biosciences), CD11b-PE (M1/70, Biolegend) and F4/80-Pacific blue (T45-2342, BD Biosciences). Acquisition was carried out using a FACSCanto flow cytometer (BD Biosciences) and analysis performed using Flowjo X (Treestar). Fluorescent minus one controls for each of the included antibodies were used to validate results. . Carriage density by qPCR was determined by partial amplification of the lytA gene of S. pneumoniae, as described previously 21 . Carriage density was also assessed using the cspA gene and there was a positive correlation between pneumococcal DNA copies of both genes as determined by cpsA and lytA qPCR (P < 0.0001, Fig. S2). Samples were only classified as positive for pneumococci if both lytA and cpsA qPCR signals were > 10 DNA copies (< 40 cycles). 49 swabs were determined to be negative for pneumococci based upon dissimilarity between the lytA and cpsA qPCR results. Overall, 68 swabs were determined as pneumococci positive. Anthelmintic treatment was administered to children who tested positive for helminth infection for ethical reasons.
Statistical analysis. For the murine experiments, statistical analysis was carried out using GraphPad Prism 7. Mann-Whitney U test and ANOVA with Kruskall-Wallis post-test were performed when comparing two and three or more experimental groups, respectively. Error bars represent SEM. Statistical significance is denoted by *(P < 0.05) and **(P < 0.01).
For clinical data, statistical analysis was performed using IBM SPSS version 25. Bacterial density data were log 10 transformed and linear regression was used to assess relationships between potential risk factors and colonisation density. Analysis was restricted to positive carriers and results were reported as linear regression coefficients and 95% confidence interval (CI). Covariates significant at P < 0.1 alongside age, sex and household crowding (selected a priori) 10 were then used in a multivariable regression model to estimate adjusted coefficients and identify independent associations.