Bacterial Factors Required for Transmission of Streptococcus pneumoniae in Mammalian Hosts

Graphical Abstract:

Rowe et al conduct a screen to identify pneumococcal genes required for effective transmission in a ferret model. They establish the fitness landscape of S. pneumoniae genes during mammalian transmission and find that maternal vaccination with the identified transmission factors can block bacterial transmission in the offspring.

Summary:

The capacity of Streptococcus pneumoniae to successfully transmit and colonize new human hosts is a critical aspect of pneumococcal population biology and a prerequisite for invasive disease. However, the bacterial mechanisms underlying this process remain largely unknown. To identify bacterial factors required for transmission, we conducted a high-throughput genetic screen with a TnSeq library of a pneumococcal strain in a ferret transmission model. Key players in both metabolism and transcriptional regulation were identified as required for efficient bacterial transmission. Targeted deletion of the putative C3-degrading protease CppA, iron transporter PiaA, or competence regulatory histidine kinase ComD, significantly decreased transmissibility in a mouse model, further validating the screen. Maternal vaccination with recombinant surface-exposed PiaA and CppA alone or in combination blocked transmission in offspring and were more effective than capsule-based vaccines. These data underscore the possibility of targeting pneumococcal transmission as a means of eliminating invasive disease in the population.

Introduction:

Introduction of the pneumococcal conjugate vaccine greatly reduced the burden of invasive disease by S. pneumoniae; however rates of colonization and pneumonia (McLaughlin et al., 2018) remain largely equivalent due to serotype replacement (Azarian et al., 2018) and limited efficacy of the vaccine at the mucosal surface. Critical to the success of S. pneumoniae is the capacity to initially colonize the human nasopharynx and subsequently transmit and colonize a new host. As such, both colonization and transmission dynamics reflect strong evolutionary pressures on this pathogen within populations and is key for understanding epidemiology. Exploration of colonization factors (Hava and Camilli, 2002; van Opijnen et al., 2016) has provided insight into pneumococcal biology, but despite the acknowledgement that transmission is a fundamental aspect of pneumococcal biology there remains limited understanding of the bacterial and host factors involved compared to our understanding of invasive disease.

Streptococcus pneumoniae (the pneumococcus) is a member of the human nasal microbiome, especially of children (van den Bergh et al., 2012) but can progress to invasive diseases such as otitis media, pneumonia, sepsis and meningitis. Pneumococcal transmission can be inferred from studies of human populations by monitoring nasal colonization dynamics of children (Azarian et al., 2018). Seasonal patterns of pneumococcal disease and colonization patterns support a role of respiratory viruses in promoting pneumococcal transmission, particularly the influenza A virus (Althouse et al., 2017; Grijalva et al., 2014). An infant mouse model of pneumococcal transmission has been developed (Kono et al., 2016; Zafar et al., 2017a; Zafar et al., 2016; Zafar et al., 2017c; Zangari et al., 2017) and has provided valuable insights into the importance of capsule type (Zafar et al., 2017a) and the contribution of pneumolysin (Zafar et al., 2017c) for transmission but are not ideal for large scale genetic screens, as only a single bacteria is transmitted (Kono et al., 2016).

Long used in studies of influenza virus transmission and pathogenesis, the ferret model closely reflects the human respiratory tract (Belser et al., 2018; Maher and DeStefano, 2004). Respiratory infection by influenza virus recapitulates human respiratory symptoms including sneezing and nasal discharge. Ferrets are also permissive for colonization and both contact dependent and airborne transmission of S. pneumoniae (McCullers et al., 2010). While influenza virus co-infection is not required for pneumococcal transmission, it greatly enhances pneumococcal burden in both donor and contact ferrets (McCullers et al., 2010). Such levels of bacterial burden are potentially amenable to large scale genetic screens as bacterial loads in both donor and contact animals are sufficient to allow multifold coverage by a transposon library in a species with approximately 2100 genes. Transposon sequencing (Tn-Seq) (van Opijnen et al., 2009) is a robust technique for determining the relative fitness of bacterial mutants under various conditions both in vitro as well as in vivo. These approaches have been utilized to identify niche-specific genetic requirements during pneumococcal colonization, pneumonia, and bacteremia leading to vaccine candidates effective in these respective infection models (Carter et al., 2014; Mann et al., 2012; van Opijnen et al., 2016; Verhagen et al., 2014). The second advantage to Tn-seq is the high number of unique inserts will allow for determination of the relative contributions of transmission from host to host versus bacterial outgrowth as well as precisely determine the population bottlenecks of how many bacteria/unique transposon inserts transmit from donor to contacts. Here we report a genetic screen for transmission factors in S. pneumoniae. Leveraging a highly saturated Tn-Seq library of >65,000 unique inserts in a transmissible strain of pneumococci in influenza virus co-infected ferrets we were able to recover sufficient mutants from donor and contact ferrets to power statistical predictions of the contribution of pneumococcal genes to transmission. These data revealed critical metabolic and regulatory cues that facilitate bacterial transmission as well as validated vaccine antigens designed to specifically inhibit bacterial transmission.

Results:

Ferret transmission screen:

To enable the most permissive pneumococcal population bottlenecks, both donor and contact ferrets were infected intranasally with A/Sydney/5/97 (H3N2) influenza virus as previously described (McCullers et al., 2010); a strain and dosage designed for maximal recovery of pneumococcal populations from both donor and recipient animals. Three days following influenza virus challenge, one ferret per cage (donor) was infected with a TnSeq library generated in transmissible serotype 19F strain BHN97. Bacterial burden in donors and cagemates (contacts) was determined by daily induced sneezing and nasal lavage collection. Pneumococcal transmission was rapid and robust (Fig S1A). Over 80% of contact ferrets were infected within 24 hours, and all contact ferrets were infected by three days post infection with greater than 85% of contact ferrets reaching a bacterial density of greater than 10⁴ cfu/sneeze. Total bacterial populations in nasal discharge was collected daily for four days, and total nasal bacterial population was collected by retrotracheal lavage at the end of the study. Input and total recovered bacterial populations from both donor and contact animals were prepared for Illumina sequencing as previously described (van Opijnen et al., 2015) and mapped to a high quality closed genome of BHN97 (NCBI PRJNA420094). It should be noted that insertion mutants that failed to initially colonize (Supplementary Table 1) were eliminated from this analysis as the initial colonization event is a prerequisite for subsequent transmission. Transposon inserts in the donor animals ranged from 339 to 5945 unique inserts. Comparison of output libraries of unique inserts from donor ferrets to contact ferrets identified a transmission bottleneck where between 5 and 68 percent of inserts in the donor animal are found in the contacts, with between 103–953 unique inserts present in contact ferrets (Fig S1B). The transmission bottlenecks varied based on donor transposon and bacterial loads (Fig S1C and D). Comparison between the co-housed contacts indicated they share approximately 30% of inserts. These data indicate sufficient transmission was occurring for prediction of S. pneumoniae genes that are required for transmission between hosts.

We next evaluated the data to identify insertion mutants able to colonize donor ferrets but were rarely or not recovered from contact ferrets. For each animal, abundance of each mutant strain was quantified by counting the number of corresponding reads at each transposon insert site per gene. For contact animals, read counts were dichotomized indicating whether the animal was infected (read count > 10) or not infected (read count ≤ 10) by each strain. A cutoff of 10 reads was used, because operational taxonomic units with zero counts in the input had up to 10 counts in the donors and spurious read counts of up to 10 were possible. We identified 87 factors (Fig 1), 85% of which are conserved amongst the model strains of pneumococcus TIGR4 and D39, that while able to colonize at least 5 donors to high levels, were absent from all contact animals (Supplementary Table 2). Using density of colonization at each gene in the donor, a model was built to predict transmission based on colonization density. The model was used to compare actual recovery from contact animals to the modeled predictions to identify an additional 118 genes as significantly reduced (estimated transmission probability ≤ 10%, see supplemental methods) but not absent in recipient animals (Supplemental Table 3). Many putative transmission factors were found to be involved in metabolism or metabolite transport (Fig 1), suggesting metabolic sensing is key for transmission. Other classes of genes identified were regulatory, including four two-component systems: ComD (SP_2236) the sensor kinase of the competence cascade, SP_0662 whose cognate response regulator has been suggested to control expression of pilus (Basset et al., 2017), both SP_2000 and SP_2001, and response regulator SP_0156, which have not been implicated in other cellular processes. Additional transcriptional regulators were identified, many controlling the expression of metabolic genes, further supporting a role for metabolic sensing and control during transmission. Three choline binding proteins were identified in the class of factors absent in all contact animals, including a choline binding protein (gene id=peg.403) that is notably absent from laboratory strains TIGR4 and D39. These and other surface exposed factors could serve as adhesins or release factors during inter-host transmission. In addition to gene deletions with defects in transmission, we identified two mutants over-represented in contact animals, being successfully transmitted to 100% of the recipient animals despite not being overrepresented in the donors, suggesting deletion of these factors promote transmission. These factors were a known virulence factor, the pyruvate oxidase spxB (SP_0730)(Echlin et al., 2016; Pericone et al., 2000; Spellerberg et al., 1996) and its positive regulator, spxR (SP_1038) (Ramos-Montanez et al., 2008).

Figure 1. Fitness landscape of S. pneumoniae genes during mammalian transmission.

Transposon insertions displaying reduced transmission in the ferret model (red lines) as determined by absence in recipient animals while present in 5 or greater of 9 donor animals indicating not a colonization defect or enhanced transmission (green lines) as determined by recovery from all 21 recipient animals without being over-represented in donor animals. Gene designations for both BHN97 and TIGR4 are indicated. Predicted gene functionality is indicated by color coding and functional grouping. Genes with a homolog in either TIGR4 or D39 model strains are indicated by a solid border and genes without homology in either TIGR4 or D39 are indicated by a dotted border.

Confirmation of hyper-transmissible mutants:

To confirm the Tn-seq predictions, we generated targeted mutations in our transmissible strain background and tested them in the infant mouse model (Zafar et al., 2016). While our ferret screen likely identified both influenza virus-dependent and independent pneumococcal transmission factors, confirmation in influenza virus-naïve pups will confirm pneumococcal-specific transmission factors in the absence of influenza co-infection. Four-day old pups were infected in a 1:1 donor to contact ratio and sampled daily for ten days by tapping the nares of the pup on an agar plate to identify bacteria present in the anterior nares. Pups were deemed colonized when bacteria were present on two consecutive sampling days. Confirmation of pneumococci was confirmed by random serotyping of the recovered colonies. Entire nasal passages were collected from donor and contact pups ten days post infection to enumerate colonization burden. The wild type strain, BHN97, is able to transmit in the absence of influenza co-infection, with 75–80% of contact pups becoming colonized within 10 days and 50% colonized by day 5 (Fig 2A). The DspxB strain is able to transmit to 100% of contact pups, with 50% being colonized on day 2, highlighting an enhancement in transmission kinetics (Fig 2A), even though the mutant is defective in colonization of donor animals (Fig 2C). Complementation of SpxB resulted in reduction of transmission (Fig 2B) and enhanced colonization of donor animals compared to the deletion strain (Fig 2C). Identification of spxB was unexpected, as this virulence factor is best known for its contribution to the production of millimolar quantities of hydrogen peroxide generated by the pneumococcus and it might be expected that the remaining mutants with a functional copy of spxB would rescue the phenotype of the mutant if secreted hydrogen peroxide was the primary factor underlying altered transmissibility. We hypothesize additional cellular consequences arising from loss of SpxB activity are responsible for the heightened transmission phenotype.

Figure 2. Metabolic factors enhance environmental stability.

A) Confirmation of hyper-transmissibility of a targeted deletion in ΔspxB (SP_0730). Infant mice, when 4 days old, one half of each litter was infected intranasally with 2000 CFU of wild type or mutant pneumococcus. Pups were sampled daily for colonization by taping the nares on an agar plate with days until contact pups are colonized defined as two consecutive positive samples. Each panel represents data from between at least 10–20 contact pups were sampled per strain, from at least four independent litters. B) Complementation of SpxB deletion on transmissibility. C) Bacterial burden in nasal passages of donor pups 10 days post challenge. D) Percent bacterial survival following desiccation and 24 hours incubation at room temperature compared to CFU/mL prior to desiccation. As deletion of SpxB in a serotype 4 strain eliminates capsule expression, a strain where a secondary mutation restores capsule expression was also examined for the TIGR4 strain (SpxB capsule+). E) Percent bacterial survival of SpxB complemented strain following desiccation and 24 hours incubation at room temperature compared to CFU/mL prior to desiccation. F) Percent bacterial survival 24 hours post desiccation following two hour growth in carbohydrate free CY media or G) survival 24 hours post desiccation following two hour growth in metal depleted. Statistics calculated by Mantel-Cox log-rank for transmission and for percent survival comparisons, Mann-Whitney test was used. A P-value < 0.05 considered significant (*) with the respective P-values indicated for each panel.

Environmental Stability:

Transmission can be contact dependent or airborne, requiring the bacteria to be outside of the host, either briefly as it transmits through the air, or for more prolonged periods of time in the environment. Our study showed evidence of airborne transmission, as transposon inserts were present in contacts that were not present in cagemate donor animals, however were present in another donor in the room. However, we were unable to determine the relative contributions of airborne versus direct nose-to-nose contact transmission between cagemate ferrets. S. pneumoniae is capable of survival for prolonged periods in the extracellular environment following dehydration (Walsh and Camilli, 2011). We hypothesized both reduced hydrogen peroxide production and other cellular consequences of glycolytic metabolic alterations by the spxB mutant may impart fitness benefits during dehydration stress. Twentyfour hours post desiccation, the ΔspxB strain displayed dramatically increased environmental stability via retention of viability when compared to parental wild-type (Fig 2D) with complementation of SpxB partially restoring desiccation sensitivity (Fig 2E). The fitness benefit of the spxB deletion was independent of capsule type and strain background with similar phenotypes being observed in the serotype 4 TIGR4 and serotype 2 D39 strain backgrounds (Fig 2D). Tolerance of desiccation stress was not related to metabolically induced alterations in capsule production (Echlin et al., 2016), as both encapsulated and non-encapsulated TIGR4 spxB mutants displayed similar desiccation stress phenotypes (Fig 2D). The enzymatic reaction carried out by SpxB requires the glycolytic product pyruvate for its metabolic activity. As such, a similar desiccation tolerant phenotype might be expected to be imparted by nutrient starvation conditions. Upon shifting cultures of S. pneumoniae into media deficient of a carbon source, we were able to confer a significant desiccation tolerance phenotype to wild type cultures (Fig 2F).

Carbon source limitation is not the only means by which bacterial pathogens are metabolically constrained in the mammalian host. Transition metal bioavailability is also a critical aspect of successful pneumococcal colonization (Turner et al., 2017) with both bacterial and host (Palmer and Skaar, 2016) strategies for metal acquisition and sequestration, respectively. We sought to determine if metal limitation would impart a similar desiccation tolerance phenotype to S. pneumoniae due to the reduced metabolic activity under such metal starvation conditions. Upon transfer to metal-depleted media conditions we observed a similar phenotype to that observed in carbohydrate deprived cells, with a significant increase in desiccation tolerance being observed in cells cultured under metal-limiting conditions compared to cells cultured in standard non-depleted media (Fig 2G). Taken together these data indicate nutrient limiting conditions significantly increase the capacity of pneumococcus to survive desiccation stress and in turn promotes environmental stability and may enhance transmissibility of S. pneumoniae.

Confirmation of transmission defective mutants in murine model:

To confirm factors required for transmission in the ferret screen we generated targeted deletions and tested transmission dynamics. Deletion of the putative C3-degrading protease CppA (SP_1449 homolog), iron transporter PiaA (SP_1032 homolog), or competence regulatory histidine kinase ComD (SP_2236 homolog), significantly decreased transmissibility from 75–80% by wild type BHN97 to 50, 45 and 13% respectively (Fig 3A–C) despite similar colonization burdens in donor pups (Fig 3D). Complementation of ComD and PiaA by insertion into an inert region of the chromosome, or CppA by expression on a plasmid , restored transmission levels to levels statistically indistinguishable from the parental wild-type (Fig 3A–C). We were unable to confirm BlpM (SP_0539 homolog) or LivG (SP_0752 homolog) in our influenza-naïve pups. This could suggest these factors may only be important for transmission in the context of IAV co-infection, the context of the ferret model, or a combination of such factors. These data generally confirm the Tn-Seq predictions of the contribution of pneumococcal genes to mammalian transmission.

Figure 3. Confirmation of genes required for mammalian transmission.

C57/Bl6 mice were mated and one half of each litter of pups when 4 days old was infected intranasally with 2000 CFU wild type or mutant or complemented pneumococcal strain. A) CppA: SP_1449, B) ComD: SP_2236, C) PiaA: SP_1032. Pups were sampled daily for colonization by taping the nares on an agar plate colonization of contact pups defined as two consecutive positive samples. Each panel represents data from between 10–20 contact pups that were sampled per strain, from at least 4 independent litters. D) Bacterial burden in nasal passages of donors 10 days post challenge. For transmission, statistics were calculated by Mantel-Cox log-rank test and bacterial burden compared by Mann-Whitney using Prism 6. A P-value < 0.05 was considered significant (*) with the respective P-values indicated for each comparison indicated in the respective panels, n.s. = non-significant.

Maternal vaccination with transmission factors:

Capsule-based vaccines are extremely effective against invasive pneumococcal disease due to the requirement for capsule during systemic infection. Upon introduction of the conjugate vaccine, pneumococcal populations rapidly undergo a shift towards non-vaccine serotypes that continue to colonize at equivalent rates (Weinberger et al., 2011). We hypothesized that vaccination with antigens based on the pneumococcal transmission factors may result in effective inhibition of bacterial spread between hosts. Recombinant forms of the transmission factors PiaA (Brown et al., 2001) and CppA (Carter et al., 2014) both alone or in combination were utilized to vaccinate female mice, which were subsequently allowed to breed. Both these factors are highly conserved, with PiaA being conserved in a majority of available S. pneumoniae genomes and CppA present in all publicly available pneumococcal genomes. Additionally, these antigens have been shown to be immunogenic in mice and protective against systemic challenge (Carter et al., 2014). As controls, mice were also vaccinated with either alum control or the currently licensed PCV-13 vaccine. ELISAs and Western Blots indicated that vaccination induced antibody responses against both antigens (Fig S2). Neonatal mice from these respective litters were utilized in the murine transmission model. Vaccination with any antigens conferred no significant protection benefit during initial colonization of donor animals (Fig 4A). A high degree of protection from pneumococcal transmission was engendered by vaccination with either PiaA (Fig 4B) or CppA (Fig 4C), or using both proteins (Fig 4D), resulting in a significant delay and lower overall rates of transmission. In contrast, no significant protection against transmission was observed in pups from dams vaccinated with PCV-13 compared to alum controls (Fig 4E). This was of particular interest as serotype 19F is included in the PCV-13 vaccine, indicating lack of efficacy even in cases of homologous challenge. Maternal vaccination can provide protection both through transplacental antibody trafficking and presence of maternal antibody in milk. Cross-fostering experiments where pups of vaccinated dams were nursed by non-vaccinated dams, and pups of non-vaccinated dams were nursed by vaccinated dams suggest the protection primarily comes from transplacental antibody (Fig 4F). We conclude that vaccines specifically tailored to target factors involved in pneumococcal transmission may be a strategy for elimination of S. pneumoniae from populations in a serotype independent manner.

Figure 4. Maternal vaccination with transmission factors blocks pneumococcal transmission in non-vaccinated offspring.

Dams were vaccinated 2 weeks prior to mating, and every subsequent two weeks with recombinant protein vaccines based on transmission factors PiaA and CppA or 1:50 human dose PCV-13, and alum controls. One half of each litter was infected intranasally with 2000 CFU pneumococcal strain BHN97. A) Bacterial burden in donor nasal passages 10 days post infection. B-E) Days until contact pups become colonized (two consecutive positive samples) B) dam vaccinated with rPiaA, C) dam vaccinated with rCppA, D) dam vaccinated with both rCppA and rPiaA, E) dam vaccinated with PCV-13. F) Cross fostering, days until contact pups become colonized, placental= dam vaccinated with rCppA, foster dam unvaccinated. Lactation=dam unvaccinated, foster dam vaccinated with rCppA. For transmission, statistics were calculated by Mantel-Cox log-rank test and bacterial burden compared by Mann-Whitney using Prism 6. A P-value < 0.05 was considered significant (*) with the respective P-values indicated for each comparison indicated in the respective panels, n.s. = non-significant.

Discussion:

These data represent a comprehensive genetic screen to identify bacterial factors required for the mammalian transmission of S. pneumoniae. These data suggest that under conditions of metabolic limitation the pneumococcus demonstrate heightened environmental stability. This may result in the bacteria becoming increasingly transmissible, a phenotype mimicked via deletion of the pyruvate oxidase SpxB. The advantage for transmissibility of loss of SpxB would suggest that this factor should have been lost to evolution. However, SpxB is also important for colonization, indicating that striking a balance between colonization and transmissibility is vital for pneumococcal biology. This screening methodology allowed for identification of numerous pneumococcal genes required for successful transmission between mammalian hosts. Our identification of ComD as a transmission factor complements findings of the importance of pneumococcal competence in colonization (Marks et al., 2012; Shen et al., 2019; Zheng et al., 2017) in addition to its role in genetic exchange. We also identified three two-component regulatory systems previously not implicated in transmission as well as iron acquisition and complement evasion surface proteins. A number of hypothetical and proteins of unknown function were also identified, indicating a number of important functional aspects of transmission remain to be characterized. Utilization of surface exposed transmission factors as vaccine antigens proved extremely efficacious at preventing transmission independent of donor colonization burden. Rationally designed combination vaccines could prove to be especially effective at blocking both transmission and invasive disease. These data indicate that vaccines targeting transmission may prove an efficacious strategy for the elimination of S. pneumoniae from populations.

STAR Methods:

CONTACT FOR REAGENT AND RESOURCE SHARING:

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Jason Rosch (Jason.rosch@stjude.org).

EXPERIMENTAL MODEL AND SUBJECT DETAILS:

Ethics Statement:

All experiments involving animals were performed with prior approval of and in accordance with guidelines of the St. Jude Institutional Animal Care and Use Committee. The St Jude laboratory animal facilities have been fully accredited by the American Association for Accreditation of Laboratory Animal Care. Laboratory animals are maintained in accordance with the applicable portions of the Animal Welfare Act and the guidelines prescribed in the DHHS publication, Guide for the Care and Use of Laboratory Animals.

Ferret infection:

Wild type outbred ferrets (Mustela putorius furo) were obtained from Triple F Farms. Ferrets were 9 week old (adolescent) males and were castrated just prior to arrival at our facility. During study they were housed 3 to 4 per cage (Allentown, stainless steel 3 tier/6 unit ferret/rabbit cage) with food (Purina, LabDiet 5L14) and water given ad libitum and kept on a 12 hour light/dark cycle. All infections and collections were performed in the light phase. Ferrets were infected intranasally under 4% isoflurane sedation, with 10⁵ TCID₅₀/mL A/Sydney/5/1997 (H3N2) influenza virus in a volume of 1mL PBS (0.5 mL per nostril)(McCullers et al., 2010). Influenza virus was grown in the allantoic fluid of 10–11 day embryonated chicken eggs and titered on MDCK (Manin Darby Canine Kidney) cells by infection with 100μL 10-fold serial dilutions of sample and incubated at 37°C for 72   hours. Following incubation, viral titers were determined by hemagglutination assay using 0.5% turkey red blood cells and analyzed by the method of Reed and Munch(Reed LJ, 1938).

Three days post viral challenge one ferret per cage (donor) was infected with 10⁷ cfu BHN97 transposon library grown in ThyB (see below) in static conditions at 37°C +5% CO₂ to midlog, washed and adjusted to desitred concentration for an inoculum of 10⁷ CFU/animal in a volume of 0.6mL PBS (0.3 mL per nostril). Each day post bacterial infection 1 mL nasal washes (0.5 mL per nostril of PBS) were collected following ketamine sedation of ferrets from donor and cagemate (contact) ferrets for four days, and additionally at day four ferrets were euthanized with an overdose of pentobarbital sodium and phenytoin sodium solution and post terminal retrotracheal lavage collected in 2mL PBS. Aliquots were removed for bacterial and viral burden determination, and the remainder was plated on selective media.

Neonatal mouse transmission:

Wild type adult C57/Bl6 mice (Mus musculus) were obtained from Jackson labs and were housed one male and one female per cage (Allentown filter top rodent cage) with corncob bedding (Anderson’s bedding) and with food (Purina, LabDiet 5013) and water given ad libitum and kept on a 12 hour light/dark cycle. All infections and collections were performed in the light phase. Four days after pups were born, all pups (both male and female) were toeclipped for identification, one half of the litter (the donors) was infected intranasally with approximately 2000 cfu BHN97 or mutant strain grown in ThyB (see below) in static conditions at 37°C +5% CO₂ to midlog and diluted in PBS to desired inoculum, in 3μL PBS without sedation. The rest of the litter was designated contacts and was not infected. Each day for 10 days post infection of the donors the nares of each pup was tapped 20 times on an TSA/ blood agar plate supplemented with 20μg/mL neomycin, and spread with a sterile loop for CFU enumeration. Following two consecutive positive samples a contact pup was determined to be colonized. On day 14 post infection of donors all pups were euthanized by CO₂ asphyxiation followed by cervical dislocation. Heads were defleshed, bottom jaw and brain removed and entire skull was homogenized with a 5mL syringe plunger and a 100μm cell strainer. Sample was collected in 750μL of PBS, diluted and plated for CFU enumeration. 10–20 contact pups were sampled per strain or vaccination condition, from at least 4 litters.

Maternal vaccination:

Adult female C57/Bl6 mice housed as above, were vaccinated intrapareteneally with 1:50 human dose (10μL vaccine+90μL PBS) PCV-13 (Pfizer) or 10μg recombinant protein: rCppA, or rPiaA, (prepared as described below) conjugated to 130μg alum, or 130μg Alum in a volume of 100μL PBS two weeks prior to mating and then boosted every two weeks for the duration of the study. Pups were infected with wild type BHN97 and monitored as above. For analysis of circulating antibody response by western and ELISA, two additional adult male mice were vaccinated with each antigen and boosted twice. One week following the final boost the mice were anesthetized with 4% isoflorane and bleed by retro-orbital route and maximum blood volume was collected. Mice were then euthanized by CO₂ asphyxiation followed by cervical dislocation.

METHOD DETAILS

Bacterial growth:

TSA/blood agar plates were prepared from 40 mg/L tryptic soy agar (EMD Millipore, GranuCult, item number 105458) in distilled water, autoclaved 45 minutes. After cooling to 55°C 3% defibrinated sheep blood was added (Lampire biological, item number 7239001) and poured into 100×15mm round petri dishes. Media was supplemented with 20μg/mL neomycin for all animal derived samples to reduce contamination with environmental staphylococci endemic to our animal colonies. Media for TnSeq samples were additionally supplemented with 200μg/mL spectinomycin to select for transposon. Deletion mutants were selected on TSA/blood agar with 1μg/mL erythromycin. Chromosomal complementation mutants were selected on TSA/ blood agar supplemented with 1μg/mL erythromycin and 150μg/mL spectinomycin. Plasmid complementation of CppA was selected on TSA/blood agar supplemented with 1μg/mL erythromycin and 400μg/mL kanamycin.

ThyB was prepared from 30g/L Todd Hewitt (BD item number 249240) with 2g/L yeast extract (BD item number 212750) in distilled water and autoclaved 45 minutes.

ThyB metal deplete was made by mixing ThyB with 15g Chelex resin (BioRad item number 14201253) overnight followed by filtration to remove resin.

CY media was prepared as follows:

Supplement:

Prepare the supplement “3 in 1” salts by adding 50 g MgCl2 6H2O, 0.25 g CalCl2 anhydrous and 0.1 ml (0.1 M) Manganese sulfate 4 H2O in 500 ml distilled water (dH2O), mix well and autoclave. Prepare 20% Glucose, 50% Sucrose, 2mg/ml Adenosine and 2mg/ml Uridine 500 ml each respectively, and filter to sterilize, store at 4°C. Combine all 5 components at the following ratio: 60 ml “3 in 1” salts, 120ml 20% Glucose, 6 ml 50% Sucrose, 120ml 2mg/ml Adenosine, 120 ml 2mg/ml Uridine, mix in a beaker and filter sterilize, label as Supplement, store at 4°C.

Adams Solutions.

Prepare Adams I by combining the following chemicals: 30 mg Nicotinic Acid (Niacin stored at 40C), 35 mg Pyridoxine HCl (B6), 120 mg Ca-Pantothenate (stored at 40C), 32 mg Thiamine-HCl, 14 mg Riboflavin, and 0.06 ml Biotin (0.5mg/ml stock). Add d H2O to 200ml, then add 1–5 drops of 10N NaOH to dissolve chemicals, filter sterilized and store in foiled bottles at 4°C.

Prepare Adams II by adding the following chemicals: 50 mg FeSO47H2O, 50 mg CuSO4, 50 mgZnSO47H2O, 20 mg MnCl2 and 1ml HCl (concentrate), up to 100ml d H2O, filter to sterilize, store at 4°C.

Prepare Adams III by adding the following 5 components: 800 mg Asparagine, 80 mg Choline Chloride, 64 ml Adams I, 16 ml Adams II and 0.64 ml CaCl2 (1% stock), to 400 ml d H2O. Filter sterilize solutions and store in foiled bottle at 4°C.

Buffers

Prepare 1M KH2PO4 and 1M K2HPO4 (autoclaved) as the stocks, mix 26.5 ml 1M KH2PO4 and 473 1 M K2HPO4 and stir well, do not titrate, filter to sterilize, 4°C.

PreC.

Prepare PreC by mixing the following chemicals: 4.83 g Sodium Acetate (Anhydrous), 20 g Difco Casamino Acids/technical, 20mg/L-Tryptophan, and 200 g/LCysteine HCl, dissolve in 800 d H2O, adjust pH to 7.4–7.6 by adding 10 N NaOH, stir well for 60 minutes, fill up to 4 liter dH2O, mix well, aliquot 400 ml portions in 500 ml flasks, autoclave for 30 minutes, and store at 4°C.

C+Y.

Add 0.5g glutamine to 500 ml dH2O, filter to sterilize and store at 4°C. Add 2 g pyruvic acid (stored at 4°C) to 100 ml dH2O, filter to sterilize , and store at 4°C. Solve 5g yeast to 100ml dH2O (25g in 500ml), and autoclave (filter to sterilize if necessary). Add 6 of the following solutions to 400 ml PreC: 13 ml Supplement, 10 ml Glutamine, 10 ml Adams III, 5 ml Pyruvate, 15 ml K-Phosphate buffer, and 9 ml Yeast. Filter sterilize and store at 4°C.

CY sugar deplete media was made as above but with the omission of glucose, sucrose and yeast extract.

Construction of mutants in BHN97:

Allelic replacement deletion mutants were made by replacement of the gene of interest with an erythromycin cassette by splicing by overlap extension PCR. A region 1.5–2kb upstream and downstream of the gene of interest was amplified by PCR using Takara HotStart polymerase according to manufacturer instructions from BHN97 genomic DNA using the primers in supplemental table 4 with an overhang corresponding to the beginning and/or end of the erythromycin cassette. The upstream and downstream fragments were mixed with the erythromycin cassette and amplified with Takara polymerase and upstream forward and downstream reverse primer. The resultant PCR product was gel purified using Qiagen MinElute kit (item number 28606) according to manufacturer’s instructions. BHN97 was transformed with the purified PCR product in CY media using both CSP-1 and CSP-2.

Chromosomal complements were made by insertion of the gene and 150–200 bases upstream containing the promoter, followed by a spectinomycin cassette into a region downstream of amiF similar to as described in (Guiral et al., 2006) except a phage is inserted into the exact chromosomal region described therein in BHN97, therefore insert regions were changed slightly, as to not disrupt the downstream gene. A region of approximately 1.5 kb downstream of amiF was amplified with primers BHN97 insert UP FWD and BHN97 insert DOWN-Spec (see supplemental table 4) as described above, and then mixed with spectinomycin cassette and amplified. The gene plus upstream primer region was amplified with an overlap of the spectinomycin cassette at the 3’ end and an overlap of the region containing amiF at the 5’ end (see supplemental table 4 for primers). The region containing amiF was amplified with primers BHN97 insert amiF FWD and BHN97 insert amiF REV (see supplemental table 4). All three fragments were mixed and amplified with primers BHN97 insert amiF REV and BHN97 insert UP FWD. The resulting PCR product was gel purified and transformed into the deletion strain as described above. Except for complementation of comD, where due to deletion of comD the strain was no longer competent, and therefore the complementation construct was transformed into BHN97 and then the deletion construct was transformed in to the resultant strain. Complementation of cppA transmission phenotype was not able to be accomplished via chromosomal complementation. Therefore an overexpression construct was made in pABG5(Granok et al., 2000) by amplification of cppA from the BHN97 genome using primers 5’ BHN97 CppA EcoRI and 3’ BHN97 CppA PstI and digested with EcoRI and PstI (NEB) according to manufacturers instructions, pABG5 was also digested with EcoRI and PstI. Plasmid and insert were ligated overnight at 14°C i n a thermocycler and transformed into One Shot TOP10 E. coli according to manufacturer’s instructions. Transformants were selected on LB agar (BD, BP1425–500) supplemented with 50μg/mL kanamycin. Following confirmation by Sanger sequencing the plasmid was transformed into the cppA deletion strain as described above. Maintenance of plasmid was ensured by addition of 400μ g/mL kanamycin to any broth during culture of this strain.

SpxB mutants in strain D39 and TIGR4 were previously constructed(Echlin et al., 2016).

Preparation of BHN97 TnSeq library:

TnSeq library was prepared in strain BHN97 as previously described(van Opijnen et al., 2014, 2015). Briefly in vitro transposition was performed using purified BHN97 genomic DNA, plasmid pMagellan6 as source of transposon and purified MarC9 protein as transposase. DNA was purified by ethanol precipitation. Transposition junctions were repaired with T4 DNA polymerase(NEB) and E.coli DNA ligase(NEB). Ligated product was transformed into strain

BHN97 as described above in construction of mutants. Transformations were plated on selective media and grown overnight at 37°C+5% CO 2. All growth was collected into ThyB media, and glycerol was added to a final concentration of 20%, and libraries were stored at - 80°C. Six libraries were combined and expanded in 1 00mL of ThyB until at midlog growth. Glycerol was added to a final concentration of 20% and 1mL aliquots frozen for use in all experiments. For each ferret infection one 1mL aliquot was added to 9mL ThyB and grown to midlog.

Preparation of output libraries:

All of the sneezed material or retrotracheal lavage was plated on 3–5 plates of selective media and grown at 37°C +5% CO₂ overnight. 5mL PBS was added to each plate and all growth resuspended and collected. Bacteria were pelleted by centrifugation, supernatants discarded and pellets stored at −80°C. Genomic DNA was extrac ted from bacterial pellets using the Blood and Tissue Kit (Qiagen) according to manufacturer’s instructions for Gram-positive bacteria. TnSeq libraries were prepared as previously described (van Opijnen et al., 2009; van Opijnen and Camilli, 2010), briefly, genomic DNA was digested with MmeI (NEB) and cleaned up with phenol chloroform extraction. Then adapters were ligated onto the digested DNA, and PCR amplified with Q5 polymerase (NEB). PCR products were gel extracted. Sequencing was performed on Illumina HiSeq platform.

Desiccation resistance:

Bacteria were grown in ThyB to midlog, and 1mL aliquots transferred to 1.7mL microcentrifuge tubes and centrifuged to remove all residual media. Tubes were spun open for 90 minutes in a SpeedVac until pellet was dry. Tubes were closed and stored in the dark at room temperature. 24 hours post desiccation bacterial pellets were resuspended in 100μL PBS and plated for viability. For desiccation resistance in sugar depleted media, pneumococci were grown in CY until midlog and then shifted to CY or CY lacking glucose, sucrose and yeast extract and allowed to grow for two hours, then desiccation protocol followed. Metal depleted media was made by depletion of ThyB using Chelex resin. Midlog bacteria grown in ThyB were shifted to metal deplete or replete media and exposed for two hours, followed by the desiccation protocol. All were done with at least 4 technical replicates, and three biological replicates. Percent survival was calculated for each technical replicate by dividing the post desiccation CFU/mL with the pre-desiccation CFU/mL of the culture and then multiplied by 100 to give percent survival. Desiccation resistance of the SpxB complemented strain was done using a different vacuum pump and desiccation proceeded more rapidly.

Expression and Purification of rCppA and rPiaA:

rCppA and rPiaA were generated by the protein production facility at St.Jude Children’s Research Hospital.

CppA was expressed and purified as described in (Carter et al., 2014). The coding sequence for CppA was amplified from TIGR4 and cloned into the pET28b cloning vector in BL21-DE3 cells. Cultures were grown to OD600 = 0.5 and induced with 0.07 mM IPTG overnight at 23 C.Bacterial pellets were lysed with Bugbuster (Novogen) reagent according to manufacturers protocols. CppA was purified on His- Selected Nickel Affinity Gel following manufacturers protocol for native conditions. Protein was dialyzed using Pierce Slide-A-Lyzer dialysis cassette overnight with sterile PBS. Dialyzed protein was stored at −80 C in a 10% glycerol solution until further use

PiaA was expressed and purified as previously described(Brown et al., 2001). High-level expression of His₆-PiaA was achieved by the addition of isopropyl-β-d-thiogalactoside (IPTG) to a final concentration of 2 mM, and the cultures were incubated for a further 4 h. The cells were harvested by centrifugation at 6,000 × g for 10 min and resuspended in lysis buffer (50 mM sodium phosphate, pH 8.0; 2 M NaCl; 40 mM imidazole). The cells were lysed in a French pressure cell (SLM Aminco, Inc.) at 12,000 lb/in², and the lysates were centrifuged at 100,000 × g for 1 h. Then, 20 mM β-mercaptoethanol was added to the resultant supernatants, which were loaded onto 2-ml nickel-nitrilotriacetic acid resin columns (ProBond; Invitrogen) previously equilibrated with five column volumes of lysis buffer. The columns were washed with 10 column volumes of 10 mM sodium phosphate, 20 mM imidazole, and 1 M NaCl (pH 6.0), and the proteins were eluted with a 30-ml gradient of 0 to 500 mM imidazole in 10 mM sodium phosphate (pH 6.0). Fractions of 3 ml were collected and analyzed by SDS-PAGE to identify fractions containing abundant purified protein. The selected fractions were dialyzed extensively against 10 mM sodium phosphate (pH 7.0) to remove the imidazole. The purified His₆-PiaA protein was then resuspended in 50 mM sodium phosphate (pH 7.0), glycerol was added to a final concentration of 50%, and the proteins stored at −15°C. Purity of protein was determined to be >95% by visualization on a 10% Bis-Tris gel stained with Simply Blue Safestain (Invitrogen).

Fractionation and Western Blot:

Sera from vaccinated animals was analyzed against cell wall and cell membrane fractions from BHN97, ΔCppA, and ΔPiaA.

Cell wall fraction was prepared by:

100 mL cultures of each bacteria were grown in ThyB to OD₆₂₀~0.6. Bacteria were pelleted by centrifugation, washed in 50mM Tris-HCl, 1mM EDTA pH 8.0 supplemented with 1x HALT protease inhibitor (Thermo) and pelleted by centrifugation. Pellets were resuspended in 1mL 50mM Tris-HCl pH 8.1, 1mM EDTA pH 8.1 plus 20% (w/v) sucrose, 10mg/mL chicken egg white lysozyme and 250 U mutanolysin, and incubated 2 hours at 37°C with shaking. Protoplasts were pelleted at 14,000xg 10 minutes. Supernatant (cell wall fraction) was removed and stored at −20.

Membranes were prepped by:

Resuspending protoplasts (pellet from cell wall prep) in 10mM Tris-HCl pH 8.1, 50mM MgCl₂ and 10mM glucose, supplemented with 1x HALT protease inhibitor and lysed with 0.1mm zirconia/silica beads in a FastPrep 24 system (MP Biologicals) for 3 20 second pulses with 1 minute rests on ice between pulses. Beads and intact cells were pelleted at 6000xg 10 minutes. Supernatant containing membranes was transferred to ultracentrifuge tube and ultracentrifuged 30 minutes at 45000xg at 4°C. Supernatant containin g cytoplasmic proteins was removed. Pellet was resuspended in 10mM Tris-HCl pH 8.1, 20mM MgCl₂ and 50mM NaCl and ultracentrifuged 45 minutes at 100,000xg at 4°C. Su pernatant was discarded and pellet was resuspended in 100μL 10mM Tris-HCl pH 8.1, 20mM MgCl₂ and 50mM NaCl.

Western blot:

100μL sample was combined with 30 L 4x sample buffer (NuPAGE, Invitrogen) and boiled 10 minutes. 15 L was loaded into 15 well 4–12% Bis-Tris precast gel, and run 1 hour 45 minutes at 80V in NuPAGE running buffer. Gels were transferred to nitrocellulose membranes at 30V for 90 minutes in NuPAGE transfer buffer supplemented with 20% methanol. Membranes were blocked 1 hour at room temperature in 4% non-fat dry milk in PBS supplemented with 0.1% Tween-20 (PBST). Membranes were incubated with sera from vaccinated mice at a 1:5000 dilution in 4% non-fat dry milk in PBST overnight at 4°C. Blots were washed 3× 10 minutes with PBST. Membranes were incubated with secondary antibody, goat anti-mouse IgG-HRP (Invitrogen), 1:5000 in 4% non-fat dry milk in PBST 3 hours at room temperature. Blots were washed 3× 10 minutes in PBST and 1× 5 minutes in PBS. 2mL each SuperSignal West Dura Extended Duration Substrate (Thermo Scientific) reagent was added to each blot and incubated for 5 minutes at room temperature. Blots were imaged on a ChemiDoc MP (BioRad) using ImageLab 5.0 software, automatic exposure settings for Chemiluminescence, high specificity, optimizing for bright bands.

Purified protein blots were performed as above except 500ng each protein was used as sample.

ELISA:

Sera from vaccinated animals was analyzed by ELISA. Bacterial strains BHN97, ΔCppA, and ΔPiaA were grown in CY until midlog. Each well of a 96 well high binding ELISA plate (NUNC #430341) was coated with 10⁶ CFU in carbonate-bicarbonate buffer reconstituted from tablet (Sigma C3041). Bacteria were pelleted to bottom of plate by centrifugation and supernatant removed. Plates were air dried overnight. Plates were blocked in 10% heat inactivated fetal bovine serum (FBS) in PBS for two hours. Serum from vaccinated mice was serially diluted 1:2 starting with a 1:50 dilution in 10%FBS in PBS and added to wells, and incubated for one hour at room temperature. For normalization between strains, polyclonal rabbit sera against LytA (gifted by Tuomanen Lab) was initially diluted 1:300 and subsequently diluted 1:2 in 10% FBS. Plates were washed 5x with tris buffered saline (TBS). Secondary antibody (Southern Biotech #1030–04 – anti-mouse & 4030–04 – anti-rabbit) was diluted 1:2000 in blocking buffer and incubated 1 hour at room temp. Plates were washed 5 times with TBS. Substrate (Sigma #P7998) was added for 30 minutes and OD₄₀₅ read in 96 well plate reader. Normalization of differential coating by wild type and mutant strains was accomplished by dividing all intensities for each strain by the ratio of the intensity of anti-LytA in wild type versus the mutant strain. Purified protein ELISA was performed as above, except wells were coated with 1205ng purified protein per well overnight at 4°C in coating buffer.

QUANTIFICATION AND STATISTICAL ANALYSIS

Unless otherwise specified, all statistics were calculated with Graphpad Prism 6. Mantel-Cox was used for colonization. Mann-Whitney was used for other comparisions.

Processing of TnSeq data and bottleneck calculations:

The Illumina sequencing read quality was assessed by FastQC (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/). To remove potential phage phiX contamination, the reads were aligned against the phage phiX genome by Bowtie2 (Langmead and Salzberg, 2012) and the unmapped reads were kept for downstream analyses. The adapters and transposon sequences were removed by Trimmomatic (Bolger et al., 2014). The cleaned reads were demultiplexed by FastX-Toolkit(http://hannonlab.cshl.edu/fastx_toolkit/index.html). The reads for each sample were subsequently aligned against the Streptococcus pneumoniae BHN97 genome sequence by Bowtie2 (parameters: −q −p 1 −S −n 0 −e 70 −l 28 --nomaqround −y −k 1 −a −m 1 –best).

Each sample from each animal was sequenced individually. Then all the samples from each timepoint in a single animal were combined (see Supplemental table 5 for number of samples collected per animal). This gave us the number and location of unique inserts in each animal. Nine groups of animals were analyzed, with a total of 9 donor animals and 21 contact animals.

Bottleneck calculations were determined by dividing the number of unique inserts shared by donor and cagemate contact to the number of inserts present in the donor and multiplying by 100 to get a percent. This calculation was done for each cage.

The number of insertions per gene were enumerated and compared between groups by custom R scripts (available at https://github.com/jiangweiyao/FerretTransmission). That data was then used to build the biostatistics model described below to identify transmission factors.

Details of biostatistics model:

Nine cages were analyzed, with a total of 9 donors and 21 contacts. For each animal, the abundance of each mutant strain was quantified by counting the number of corresponding reads obtained by next-generation sequencing. For donor animals, the infection burden for each strain was defined as log₁₀(read count +1). For all animals, the read counts were dichotomized indicate whether the animal was infected (read count > 10) or not infected (read count ≤ 10) by each strain. A cutoff of 10 reads was used, because operational taxonomic units with zero counts in the input had up to 10 counts in the donors. Thus, we recognized that spurious read counts of up to 10 were possible. Descriptive statistics of infection status of donors and contacts were computed. Also, descriptive statistics of transmission from donor to contact animals were computed in terms of the number of each donor’s contacts that became infected. For each donor, the observed transmission rate was computed as the proportion of its contacts that became infected. A final estimate of the probability of transmission was computed as the average of the transmission rates across infected donors. The analyses were performed using version 1.1.13 of the lme4 package for R software (Windows version 3.3.3; www.r-project.org) with script available at https://github.com/jiangweiyao/FerretTransmission.

Quantification of transmission in the infant mouse model:

Contact pups were monitored daily for bacterial acquisition by taping the nares onto an agar plate. Pups were determined to be colonized following two consecutive positive cultures to eliminate incidental contamination from bedding. 10–20 contact pups were used for each bacterial mutant strain from at least 4 independent litters. Comparison to transmission by wild type bacterial strain was calculated with Mantel-Cox in GraphPad Prism6.

Comparison of Desiccation tolerance:

All were done with at least 4 technical replicates, and three biological replicates. Technical replicate bacterial counts were averaged and percent survival determined from CFU of input. Mann-Whitney in GraphPad Prism6 was used to make comparisons.

DATA AND SOFTWARE AVAILABILITY

Data and Resource Availability

All sequence files and scripts used in this study are freely available in public databases as described in the supplementary methods.

BHN97 genome available at https://www.ncbi.nlm.nih.gov/bioproject/420094

Raw TnSeq output available at: https://www.ncbi.nlm.nih.gov/bioproject/497898

R scripts for analysis available at: https://github.com/jiangweiyao/FerretTransmission

Supplementary Material

3

Table 1: Ferret colonization factors (related to Figure 1)

4

Table 2: Transmission factors absolutely required for transmission, in zero contact pups (related to Figure 1)

5

Table 3: Transmission factors reduced in transmission to contact pups (related to Figure 1)

6

Table 4: Primers used in strain construction (related to STAR methods-construction of mutants in BHN97)

7

Table 5: Ferret housing summary during study (related to STAR methods)

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Goat anti-mouse IgG-HRP	Invitrogen	31436
Goat anti-mouse IgG-AP	SouthernBiotech	1030–04
Goat anti-rabbit IgG-AP	SouthernBiotech	4030–04
Bacterial and Virus Strains
S. pneumoniae strains D39 and TIGR4	Elaine Tuomanen’s lab
S. pneumoniae strain BHN97	(McCullers et al., 2010)
Influenza A Virus A/Sydney/5/1997 (H3N2)	WHO CC
Biological Samples
Chemicals, Peptides, and Recombinant Proteins
rCppA	(Carter et al., 2014)
rPiaA	(Brown et al., 2001; Carter et al., 2014)
Critical Commercial Assays
Deposited Data
TnSeq reads	This study	PRJNA497898
BHN97 genome	This study	PRJNA420094
Experimental Models: Cell Lines
Experimental Models: Organisms/Strains
Ferrets, 9 week castrated males	Triple F Farms
C57/Bl6 mice, adult, male and female	Jackson labs
Oligonucleotides
See supplemental table 4	This study
Recombinant DNA
Software and Algorithms
TnSeq insert comparison	This study	https://github.com/jiangweiyao/FerretTransmission
ferret microbiome analysis	This study	https://github.com/jiangweiyao/FerretTransmission
Prism6	GraphPad
Other

Highlights:

• A pneumococcal TnSeq library was screened in a ferret transmission model

• The fitness landscape of S. pneumoniae genes during mammalian transmission established

• Metabolic factors enhance pneumococcal environmental stability

• Vaccinating dams with identified factors blocks pneumococcal transmission in offspring