Convergent Impact of Vaccination and Antibiotic Pressures on Pneumococcal Populations

Abstract

Streptococcus pneumoniae is a remarkably adaptable and successful human pathogen, playing dual roles of both asymptomatic carriage in the nasopharynx and invasive disease including pneumonia, bacteremia, and meningitis. Efficacious vaccines and effective antibiotic therapies are critical to mitigating morbidity and mortality. However, clinical interventions can be rapidly circumvented by the pneumococcus by its inherent proclivity for genetic exchange. This leads to an underappreciated interplay between vaccine and antibiotic pressures on pneumococcal populations. Circulating populations have undergone dramatic shifts due to the introduction of capsule-based vaccines of increasing valency imparting strong selective pressures. These alterations in population structure have concurrent consequences on the frequency of antibiotic resistance profiles in the population. This review will discuss the interactions of these two selective forces. Understanding and forecasting the drivers of antibiotic resistance and capsule switching are of critical importance for public health, particularly for such a genetically promiscuous pathogen as S. pneumoniae.

Graphical Abstract

Introduction

Antimicrobial resistance is one of the most significant microbial-related threats to human health. Deaths in 2019 caused by multidrug resistant (MDR) bacterial infections were higher than HIV and malaria combined¹. Vaccines are effective against antibiotic-resistant and -sensitive bacteria since they enable immune-mediated bacterial clearance. Thus, for a given bacterial population, antibiotics and vaccines exert two mechanistically distinct evolutionary pressures on survival. Streptococcus pneumoniae (pneumococcus) is a model pathogen to study the interplay between vaccine escape and antimicrobial resistance because of its diversity of capsules that are the vaccine target and the plasticity of its penicillin binding proteins (PBPs) as an antimicrobial target, both of which can intermingle in an arena of heightened ability to undergo genetic exchange.

Antibiotics and vaccines can force any of several genetic strategies to allow a bacterium to circumvent current treatment. Strategies include: 1) serotype switching, selecting for different capsule types that can be rapidly disseminated through genetic recombination²; 2) an increase in non-encapsulated and non-vaccine type pneumococci that are not covered by capsule-based vaccines and which can serve as a genetic reservoir for the exchange of antimicrobial resistance genes^3,4; and 3) allowing for the expansion of non-pneumococcal streptococci in shared ecological niches that could serve as genetic reservoirs for both resistance and novel capsule types^5,6. This review will focus on the interplay of pneumococcal vaccination pressures on antibiotic resistance and discusses current challenges on how to predict the outcome of these interactions over time.

Pneumococcus is a Gram-positive coccus that usually exists as a commensal of the human nasopharyngeal mucosa but is often characterized as a pathogen due to its ability to disseminate from the nasopharynx into the middle ear space, lower respiratory tract, blood and brain^7,8. Community acquired pneumonia (CAP) is the most common manifestation of invasive pneumococcal disease (IPD) especially in those under five and over sixty-five years old⁸. S. pneumoniae is estimated to cause nearly 15% of all deaths in children under five⁹. Together, this makes the pneumococcus the fifth leading cause of death overall and second in children under five¹⁰. Even in the modern era of antibiotics, CAP continues to be a leading cause of death despite being considered one of the most vaccine-preventable diseases by the World Health Organization^11,12. Antibiotic treatment of CAP accounts for 30–50% of all antibiotic use¹³. Pneumococcus has developed resistance to many antibiotics, including beta-lactams, fluroquinolones, and macrolides and is one of the top five pathogens with the highest rate of global deaths attributed to antimicrobial resistance^1,14.

Critical to the success of the pneumococcus as a major human pathogen is the predominant virulence factor, the polysaccharide capsule (CPS), encoded by the cps locus. The CPS, the dominant surface structure covalently attached to the outer surface of the cell wall peptidoglycan, can be up to 400 nm thick, making up more than half of the bacterial volume¹⁵. Over 100 distinct serotypes have been identified, underscoring the diverse biochemical and structural composition of the CPS^16,17. Most serotypes are immunologically distinct with limited cross-reactivity^18,19. As such, immune responses against the CPS are a major selective pressure on the pneumococcal population as protective immunity is typically engendered in a serotype-specific manner following CPS-specific antibody responses.

Pneumococcal capsule is necessary for bloodstream infections. Anti-CPS vaccines effectively prevent IPD by antibiotic-resistant and -sensitive serotypes. The mechanism by which anti-CPS vaccines affect colonization remains unknown. Conversely, antibiotics treat invasive disease independent of serotype, but are not effective against antibiotic-resistant bacteria. By targeting different parts of the pneumococcal population, each of these interventions remain effective in situations where the other could fail^20,21, establishing a synergistic effect to maintain pressure on the pneumococcal population. Since pneumococcal vaccines currently only target up to one quarter of the ~100 known serotypes, post vaccine IPD favors non-vaccine serotypes arising in the human population through serotype replacement or serotype switching via genetic recombination and subsequent expansion of clones outside of vaccine coverage^22–27. Serotype replacement occurs at the population level and is defined as the expansion of non-vaccine serotypes upon elimination of vaccine-targeted serotypes (Fig. 1A)²⁸. For example, 95% of pneumococci carried by PCV13 vaccinated healthy children were found to be of serotypes not covered by the vaccine, and specific non-vaccine strains bloomed as a cause of IPD^29–31. Serotype switching occurs at the single bacterium level and is defined as the genetic exchange of a new cps locus (Fig. 1B)². Such major shifts in population dynamics make forecasting of both serotype prevalence and antibiotic resistance trends particularly challenging.

Figure 1. S. pneumoniae respond to antibiotic and vaccine pressures.

A) Serotype replacement: Deployment of capsule based vaccines and antibiotics leads to elimination of target bearing clones within a human population and expansion non-targeted clones (vaccine escape mutants or antibiotic resistant mutants). B) Serotype switching: Pneumococci and other oral streptococcal species occupy the nasopharynx. Shared natural competence allows for rapid intraspecies and interspecies recombination, often spreading novel virulence factors and antibiotic resistance genes to the pneumococci.

Historical dueling interventions for pneumococcal infections

Vaccines and antibiotics have been integral in our attempts to control IPD for over 50 years; however, genetic promiscuity of the pneumococcus has driven constant re-emergence of antimicrobial resistance and vaccine escape¹⁹. Early vaccines were based on a combination of purified pneumococcal capsular polysaccharides (PPVs)^19,32. The first PPV, deployed in the United States in the 1930’s, targeted only serotypes 1 and 2 and successfully reduced IPD and transmission³³. Soon after, pneumonia caused by other serotypes emerged and dominated clinical cases, resulting in a reformulation of the PPVs in the 1940s³⁴. However, a major limitation of PPVs was lack of an immune response in children under two years of age, a peak age group for mortality from IPD³⁵. Therefore, penicillin gradually overshadowed PCVs as the intervention of choice. Penicillin binds the penicillin binding protein transpeptidase active site, inhibiting the final cross-linking steps of cell wall synthesis. Before 1970, the most common pneumococcal infections in children, including otitis media and CAP, were easily treated with penicillin. This was so effective that it was questioned whether early pneumococcal vaccines developed in the 1930’s and 40’s were even clinically useful³⁶.

Interest in vaccines rebounded in the mid-1970s, when penicillin-resistant isolates emerged and spread globally^37–41. Treatment failures were related to resistant pneumococcal isolates and clinical interest swung back to improving vaccines^42,43. A major milestone for the prevention of pneumococcal disease occurred in 2000 with the introduction of pneumococcal conjugate vaccines (PCV)^44–46. PCVs are composed of multiple polysaccharides conjugated to immunogenic non-pneumococcal proteins, such as diphtheria toxoid. This formulation increases immunogenicity in children under 2 years old by eliciting appropriate T-cell and memory B-cell responses³⁵. A seven-valent PCV (PCV7) that targeted the predominant serotypes circulating in North America (serotypes 4, 6B, 9V, 14, 18C, 19F, and 23F) was developed and dramatic decreases in IPD followed deployment in 2000’s^19,32. PCV7 deployment also reduced transmission from the young to elderly³³. It was thought that widespread vaccination with PCV would result in a concurrent decrease in rates of antibiotic resistance in circulating pneumococci since many of the targeted serotypes had a high predominance of penicillin resistant strains²¹. However, reality proved this assumption short-sighted. After PCV7 introduction, penicillin resistant pneumococcal infections declined temporarily; however, infections by the highly penicillin-resistant serotype 19A, that was not included in PCV7, increased leading to a rebound in penicillin-resistant infections^39,47. Similar trends were identified in the United States, Germany, South America (Brazil, Colombia, and Chile) and Japan¹⁴. In turn, resistant 19A was subsequently eliminated by inclusion in the next generation PCV13, which was expanded to include serotypes 1, 3, 5, 6A, 7F, and 19A⁴⁸. From 2000 to 2015, vaccination was predicted to result in a 51% decline in the number of pediatric pneumococcal deaths⁴⁹.

By targeting the population of circulating pneumococci, vaccines also have indirect effects on non-vaccinated individuals. In a human cohort in Lao PDR with moderate PCV13 vaccination rate, there was an observed 36% reduction in vaccine serotype carriage through indirect effects for up to five years⁵⁰. Similar effects have also been identified in infants from the United States⁵¹. In infants too young to be vaccinated, carriage of vaccine serotypes decreased for only two years following PCV introduction through indirect effects⁵⁰. Additionally, in select countries, there has been a reported 5% decline in IPD for each 10% increase in vaccination coverage^52,53, highlighting the extent of the vaccine indirect effects on a population level.

Because beta-lactam resistance is due to pbp mutations that decrease the affinity of the antibiotic for the binding protein, high levels of beta-lactam therapy, specifically with amoxicillin and penicillin G, can remain efficacious against pneumococcal infections in some body sites such as the lung where antibiotic penetration is excellent (in contrast to meningitis)^54,55. However, if the isolate exhibits high level beta-lactam resistance, therapy with ceftriaxone or vancomycin may be recommended^54,56. Beta-lactams administered in conjunction with macrolides may provide synergistic therapy against the pneumococcus due to the two different mechanisms of action⁵⁷. As IPD caused by non-PCV13 serotypes emerged⁵⁸, new strains were not only resistant to penicillin but also erythromycin, cotrimoxazole, and tetracycline arose. Multidrug resistance surged compared to the pre-PCV era⁵⁹. In 2019, antibiotic-resistant pneumococci caused nearly one million infections in the United States, and has been classified as a “serious threat” by the Centers of Disease Control and Prevention⁶⁰. Thus, the historical record indicates that the composition of the circulating pneumococcal population undergoes profound alterations as a result of both selective pressures of vaccines and antibiotics.

Genetic mechanisms underlying vaccine and antibiotic escape

Pneumococci are naturally competent and highly recombinogenic. They utilize multiple mechanisms of horizontal gene transfer and recombination including transformation, mobile genetic elements, and bacteriophages⁶¹. During antibiotic stress, pneumococci increase activation of competence systems that facilitate exogenous DNA uptake⁶². One of the best characterized hotspots for recombination in the genome is the cps locus, a genomic cassette that is characterized as an integrative conjugative element (ICE), allowing for accelerated diversification and recombination (Fig. 2A)⁶³. Important for the proclivity of pneumococcal CPS switching is the observation that the cps locus is flanked by highly conserved genes dexB and aliA that serve as homologous hotspots for recombination^2,64. During recombination of the cps locus, genetic content located outside of the cps locus can also be exchanged resulting in gDNA upstream and downstream of the locus being integrated concurrently. The cps locus is around ~20kB, and recombination fragments can range from the predicted ~20kB to ~56.5kB^2,65,66. Such recombination events often include both the cps locus and adjacent pbp1a or pbp2x (Fig. 2B). In one study, 75% of capsule switches from serotype 9V to 14 contained an average recombination block size of 26.5kB in length, spanning the entire cps locus and frequently encompassing the neighboring pbp1a and pbp2x genes^67,68. While this additional contiguous gDNA most frequently includes pbp1a and pbp2x^2,69, genes farther downstream of the cps locus, such as pbp2b, may also be included. Mutated PBPs 1a, 2b, and 2x, which show decreased penicillin binding by changes in polarity, charge distribution, or flexibility of the PBP active site⁷⁰, are among the most commonly identified PBPs in beta-lactam resistant pneumococcal isolates^71,72. These are mosaic genes comprised of recombined blocks of DNA that are hotspots for causative mutations⁷³. Their proximity to the capsule locus underlies the linkage between penicillin resistance and serotype switching. Recombination may facilitate exchange of resistance to antibiotics also. pbp1a, pbp2b, pbp2x, and folA (encoding a dihydrofolate reductase leading to resistance against trimethoprim⁷⁴) have been identified as recombination hotspots across multiple serotypes⁷⁵.

Figure 2. Homologous gene transfer of the pneumococcal cps locus often includes the upstream and downstream pbp genes.

A) Homologous recombination of the cps locus between encapsulated pneumococci occurs at the highly conserved genes dexB and aliA, allowing the cell to express a new capsule serotype. B) Schematic of the pbp genes flanking the cps locus in S. pneumoniae strain TIGR4 (not drawn to scale). By alignment, the genes encoded within the ~6200 bp upstream of dexB are largely conserved between strains TIGR4, D39, and BHN97, (drawn to scale). These adjacent loci can be included in genetic recombination events at the cps locus. Alignment generated with clinker¹⁴⁶.

Using data obtained from the Global Pneumococcus Sequencing Project, the majority of sequenced pneumococci that exhibit penicillin high or intermediate resistance belong to serotype 19A (Fig. 3A) while penicillin sensitive serotypes are more diverse (Fig. 3D). Breaking down this group of 19A serotypes into sequence types, it is dominated by multidrug resistant sequence type 320 (Fig. 3B) (summarized in⁷⁶). The serotype 19A isolates that are penicillin sensitive are widely distributed across multiple sequence types (Fig. 3E), while conversely serotypes that belong to sequence type 320, regardless of penicillin resistance status, are restricted to serotypes 19F and 19A (Fig. 3C). This example highlights the high level of genetic diversity within serogroup 19 and the clonal success of the serotype 19A multidrug resistant sequence type 320 in causing IPD^76,77.

Figure 3. Penicillin resistant serotype 19A is dominated by the sequence type 320.

Data from the Global Pneumococcal Sequencing Project was used to analyze the serotype and genetic profile of penicillin resistant isolates. A) Isolates characterized as non-meningitis penicillin intermediate or resistant were broken down by in silico serotype. B) Penicillin intermediate and resistant isolates were further characterized by sequence type (ST). C) All 320 sequence types were analyzed by in silico serotype. D) Penicillin sensitive isolates broken down into in silico serotypes. E) Penicillin sensitive 19A analyzed by sequence type (ST). When designated, only the serotypes or sequence types making up >1% of the isolates are displayed in the legend.

Physiological mechanisms underlying vaccine and antibiotic escape

Capsule switching:

Horizontal gene transfer of accessory genes, particularly metabolic genes, outside the of the cps locus is also critical for understanding constraints that govern CPS switching and replacement. Although the rate of a given CPS switch combination is based in part on donor/recipient frequency in the population⁷⁸, serotypes that most commonly switch typically share similar biochemical composition and cps locus gene homology⁷⁹. For instance, if expression of a new CPS requires new metabolic pathways, the metabolic genes, whether located in proximity to or distant from the cps locus, must be transferred along with the cps cassette. Vaccine induced metabolic shift (VIMS) is defined as the apparent transfer of vaccine-targeted metabolic types to a non-vaccine serotype through genetic recombination. This has been identified and well documented following the introduction of PCV7 through deep sequencing approaches^80,81. Non-PCV7 serotypes 15B/C and 19A acquired new metabolic types sharing 84.6% and 90.0% of metabolic loci, respectively, with vaccine serotype 9V⁸¹. Ultimately, CPS switching is likely facilitated by epistatic interactions between cps genes and central metabolic pathways. Metabolic pathways may impose limitations on which bacteria have the biochemical information required to make a specific CPS type. This may explain why CPS types that share certain residues experience less diverse antigenic shifts and remain constricted to one particular serogroup⁸². It is possible to identify multiple accessory genetic loci that drive the competitive successes of certain serotypes⁸³. For example, biosynthetic genes required for production of vaccine-type capsules 14, 19F, and 23F are the same as the genes required for the non-vaccine type capsules of 15B/C, 19A, and 23B, resulting in stabilization of the presence of these accessory genes⁸⁰. Frequency of accessory genes can also be maintained via transfer of mobile genetic elements between pneumococcal strains with divergent core genomes, but similar accessory genomes, allowing for further diversification of strains⁸³. This presents one potential mechanism for persistence of antibiotic resistance genes such as mel/mef, tetM, and ermB within the pneumococcal population^68,80. Evolution of pneumococcal accessory genomes is thought to be due to negative frequency-dependent selection, whereby alleles are most beneficial when they are rarer in the population. This model encourages enrichment of antigenic loci such as CPS biosynthesis genes due to their immunogenicity and can extend to non-capsular genes such as bacteriocin production to promote survival within their niche in the presence of other colonizing bacterial species⁸⁴.

Capsule replacement:

Successful CPS replacement events are also selected through non-genetic mechanisms such as the relative fitness of pneumococcal strains during carriage. Studies have shown that pneumococci that are more successfully carried are more heavily encapsulated and have an association with specific polysaccharide structures^85,86. These capsules are more resistant to neutrophil-mediated killing and are thus more successful colonizers and may be a predictor of serotype replacement⁸⁶. Consistent with this data, the surface charge of CPS also been shown to contribute to escape from host immunity. Negatively charged capsules decrease the electrostatic interactions between mucus and neutrophils, which are also negatively charged. Thus, serotypes with more negatively charged capsules may be more prevalent due to their ability to evade host responses and be a potential contributor to serotype prevalence. This mechanism could have been one of the selective forces behind serotype replacement following implementation of the PCV7 vaccine⁸⁵. This underscores the complexity and challenges of accurate serotype replacement forecasting in responses to CPS-based vaccination strategies as both population structure as well as impact on host-pathogen interactions can influence the ultimate rise of new serotypes in a population.

Antibiotic resistance:

Metabolic constraints also impact acquisition of antibiotic resistance traits, particularly as non-vaccine serotypes now more frequently interact with vaccine-type pneumococci⁸¹. Serotype 24F has been increasingly identified in surveillance studies after PCV13 introduction³. In the 1980s, prior to PCV13, IPD caused by serotype 24F was infrequent⁸⁷. Today, this is an invasive serotype with high rates of meningitis reported, with some lineages of the serotype harboring multi-drug resistance³. The coexistence of multiple pneumococcal strains, including antibiotic resistant and sensitive strains, within a single human host further complicates the diversification of the pneumococci through genetic recombination. Deep sequencing of the pneumococcal population within healthy individuals revealed the evolutionary dynamics of the pneumococci through antibiotic pressures over the course of three years⁸⁸. Major findings included the negative association between antibiotic resistance and colonization with multiple pneumococcal strains and the bottleneck presented with antibiotic treatment that allow for the rise of subsequent colonization of antibiotic resistant strains⁸⁸.

In silico mathematical modeling has demonstrated that elimination of vaccine-targeted serotypes can increase the prevalence of non-vaccine serotypes due to the removal of cross-immunity or resource competition within the shared niche^81,89–92. This concept has been further expanded, investigating the changes in the metabolic profiles of the pneumococcal population required to synthesize new CPS types. The selection of metabolic types with new serotypes upon vaccination highlights the hypothesis that metabolic types contain successful genetic repertoires that allow the pneumococcus to successfully CPS switch in potentially nutrient-limited and competitive niches⁹³. These models have the potential to predict the influence of new vaccines or antibiotic therapies on the pneumococcal population, allowing for the development of rational interventions that limit adverse selection for resistant populations.

Vaccine impact varies with different antibiotics

While the variations in efficacy of vaccines and beta-lactams on pneumococcal populations has been repeatedly observed, this is not the case for all antibiotics. Given the prevalence of beta-lactam resistant pneumococci, other classes of antibiotics such as fluoroquinolones have been used preferentially to treat respiratory tract infections. However, increased use has not been followed by increased resistance. PCV7 targeted serotypes 6B, 9V, 14, and 23F are known to be resistant to fluoroquinolones⁴⁷. Despite increased use, fluoroquinolone resistance in pneumococci has remained at only 1–2%^94,95. It is possible that additional barriers for acquiring fluoroquinolone resistance contribute to this sustained low prevalence level. Two mechanisms are proposed for the development of fluoroquinolone resistance: a genetic mechanism targeting DNA replication genes, and a mechanism involving active efflux of fluoroquinolones. Fluoroquinolones inhibit DNA replication by binding to either topoisomerase IV (encoded by parC) or DNA gyrase (encoded by gyrA)⁹⁶. Single mutations in the genes encoding either of these proteins incrementally change the MIC and result in a stepwise increased resistance⁹⁷. However, no specific genotypes or clades have been shown to strongly correlate or predict the subsequent development of on-target resistance to fluoroquinolones⁹⁸. A second, but less studied, mechanism of fluoroquinolone resistance involves active efflux of fluoroquinolone compounds. Multiple efflux pumps may indirectly contribute to decreased susceptibility by reducing the intracellular concentrations of fluoroquinolones to a sublethal level^99–101. Current reports estimate that 0–11% of fluoroquinolone-resistant pneumococci are a result of homologous gene transfer of these two complex resistance mechanisms^5,102, and there is no evidence of clonal expansion of these strains^5,103.

Since the early 2000s, pneumococcal isolates with both the ermB and mefE/mel resistance mechanisms have emerged, accounting for over 50% of macrolide-resistant isolates in some regions of the United States¹⁰⁴. This dual mechanism is encoded by the Tn2010 mobile element, a member of the Tn916 family of transposases¹⁰⁵, which emerged from the multidrug resistant strain ST320 upon widespread use of PCV7¹⁰⁶. This strain was a result of capsule switching from the serotype 19F strain to a 19A serotype¹⁰⁷, and has been disseminated into other strains of pneumococci by recombination events¹⁰⁸. Overall, macrolide-resistant IPD has decreased with the introduction of the PCV13 vaccine, and macrolide-resistant serotypes not covered by PCV have only slightly increased¹⁰⁹.

Similarly, other forms of drug resistance have not undergone major expansion in pneumococcal populations. Tetracycline resistance is commonly encoded by the tet(M) gene, encoding a protein that prevents tetracycline binding to bacterial 30S ribosome subunit. tetS and the mosaic tet(M/S), often carried on Tn916 and IS1216, are occasionally identified in other Streptococcus species, but rarely the pneumococcus^110,111. Recently it has been identified that Tn916 with tet(M/S) or tet(M) disseminated locally through southern Africa in a pneumococcal strain that underwent recent capsule switching¹¹², highlighting the impact of inter-species genetic recombination on antibiotic resistance. Mutations in the genes encoding dihydropteroate synthetase and dihydrofolate reductase, targeted by sulfamethoxazole and trimethoprim respectively¹¹³, allow for pneumococcal resistance to cotrimoxazole¹⁴. High levels of resistance to cotrimoxazole have remained wide-spread amongst pneumococci despite vaccines¹¹⁴. The distinct trajectories for the incidence of these various resistance mechanisms underscores the lack of generalizable rules across antibiotic classes for how vaccine selection will impact antibiotic resistance patterns.

In silico models have been used to further understand the evolution of antibiotic resistance post-vaccination and assess interventions that may reduce this. Clearance-accelerating vaccines, that is, vaccines that reduce the duration of pneumococcal carriage, are projected to have greater potential to slow the rate of antibiotic resistance since they would negate any fitness advantage that antibiotic-resistant strains gain from selective pressures due to antibiotic use¹¹⁵. However, decreases in antibiotic resistance evolution also depend on the mechanism by which pneumococci acquire resistance to antibiotics; for example, the degree to which in-host competition is beneficial for either antibiotic-resistance or -sensitive strains will also determine how effective interventions like clearance-accelerating vaccines will be¹¹⁵.

Roles of genetic reservoirs: Nontypeable pneumococci and oral streptococci

Widespread PCV vaccination has increased pressure on specific serotypes, promoting a global expansion of non-targeted serotypes. An extreme result of this constant pressure has been the increasing recognition of non-encapsulated strains in carriage and importantly, as a cause of disease^116,117. Non-encapsulated strains are known as nontypeable pneumococci (NTPn). NTPn fall into two distinct categories: isolates that do not carry CPS synthesis genes¹¹⁸, and isolates that do not produce a CPS but carry down-regulated or defective CPS synthesis genes¹¹⁹. Oftentimes, other Streptococcus species such as Streptococcus pseudopneumoniae are incorrectly classified as atypical pneumococci, getting incorrectly classified with NTPn¹²⁰. NTPn isolates have the highest observed recombination rate, consistent with the hypothesis that a capsule serves as a physical barrier to DNA uptake⁷⁵. NTPn strains have larger genomes than encapsulated pneumococci, and these genomes can harbor novel ICEs and regions of DNA encoding >100 unique clusters of orthologous groups (COG) promoting cell adhesion, immune evasion, and surface macromolecules¹²¹.

NTPn are often more resistant to antibiotics such as penicillin, tetracycline, and erythromycin^75,116,122. About 90% of isolated NTPn strains sampled across Japan after PCV introduction were shown to be multidrug-resistant¹²³. 2019 brought the first reported case of pneumonia caused by non-encapsulated, MDR pneumococcus in an immunocompromised patient¹²⁴. Another study found that approximately 10% of otitis media and rhinosinusitis in children in Japan was caused by NTPn¹²⁵. The biology and pathogenicity of infection in the absence of CPS in NTPns remains an area of active investigation⁴. In some NTPn strains, the cps locus is replaced with genes that encode surface proteins that aid in colonization and otitis media¹¹⁷. While the non-encapsulated isolates are typically restricted to mucosal infections and rarely cause invasive disease, they serve as an important threat due to their potential to undergo homologous gene transfer with encapsulated, vaccine-targeted strains, dispersing novel COGs.

Pneumococci readily take up and incorporate DNA not only from other pneumococci but also from the large population of related streptococci that inhabit the same nasopharyngeal niche in the human host (Fig. 1B)¹²⁶. Resistance to beta-lactams in pneumococcus is typically not mediated by de novo mutation but rather arises by stepwise recombination events with other oral Streptococcus species, such as viridans and mitis group, that lead to mosaic PBPs¹²⁷. Interspecies recombination is facilitated through the fratricidal behavior between pneumococci and related oral streptococci¹²⁸. In fact, there is evidence that pneumococcal strains that commonly colonize the human nasopharynx harbor pbp genes that cluster with those genes in co-colonizing S. mitis strains, suggesting that horizontal gene transfer events may readily occur between these species in a human host¹²⁹. For example, the emergence of serotypes 19A/D was a direct result of recombination that began with importing the entire cps locus from S. mitis, integration of transposases allowing for further genetic transfer, and finally inactivating mutations within the cps locus¹³⁰. Genetic analysis strongly suggests that these recombination events are unidirectional, wherein genes from oral streptococci species are transferred to S. pneumoniae, resulting in polymorphisms and acquisition of genetic elements that confer beta-lactam resistance¹³⁰.

Co-carriage of these bacterial species in the upper respiratory tract and nasopharynx promotes successful genetic exchange¹³¹, and strains of pneumococci that have a longer duration of carriage are more likely to accumulate resistance mutations as they may be exposed to antibiotics in the time they are carried¹³². These mosaic pbp genes can then disseminate within pneumococcal populations via homologous recombination of the cps due to their close proximity within the genome^6,133. It is thought that beta-lactam resistance obtained via recombination may result in fewer fitness costs than de novo pbp mutations, because of the benefit of co-transferred metabolic genes⁶⁹. Further evidence supports that many de novo pathways that increase resistance to beta-lactams are associated with greater significant fitness cost in vivo compared with resistance acquired via horizontal transfer⁶⁹. This underscores the importance of recombination for the evasion of both antibiotics and CPS based immunity.

Vaccine vs antibiotic dynamics differ for other pathogens

It is tempting to speculate that the interplay of vaccines and antibiotic resistance for other naturally competent pathogens would follow the pattern of linkage seen for pneumococci. Haemophilus influenzae is another human respiratory pathogen with a polysaccharide CPS that serves as the serotyping antigen. A conjugate vaccine against the encapsulated H. influenzae serotype b (Hib) was introduced in the early 2000s and saw a rapid decline in Hib infections, including meningitis. Similar to the pneumococcus, H. influenzae has serotypes outside of vaccine coverage as well as a non-typeable population (NTHi). However, although recombination resulting in the loss of cps or a cps switch event can occur in vitro, it is rarely observed in humans¹³⁴. Furthermore, vaccination against H. influenzae has not resulted in an increase in prevalence of beta-lactamase producing strains or other drug resistance mechanisms¹³⁵. We suggest that, in contrast to the pneumococcus where serotype switching is readily observed and can impact antibiotic resistance due to proximity of the cps locus to pbp loci, the lack of proximity for resistance genes in relation to the CPS coding region for H. influenzae decouple serotype switching from beta-lactam resistance. Genetic analysis of multidrug resistant strains of H. influenzae suggest they arise via horizontal gene transfer by conjugative transfer of ICEs or transformation of genetic material obtained from reservoirs in other species¹³⁶. One exception may be resistance mediated by PBP3¹³⁷. Recombination of PBP3, also referred to as ftsI, can result in mosaics PBPs and increased beta-lactam resistance in H. influenzae^138,139, similar to the mosaic PBPs conferring beta-lactam resistance observed in pneumococcus. Serotype B meningococci cause a disproportional number of cases in infants and prolonged epidemics leading to aggressive and invasive infections caused by this serotype^140,141. Neisseria meningitidis is another encapsulated respiratory pathogen consisting of multiple serotypes that can undergo capsule switching¹⁴². N. meningitidis conjugate vaccines target serotypes A, C, W, and Y; capsule switching could allow for serotype B expansion. These studies underscore the importance of genetic reservoirs as a wide spread source of antibiotic resistance in naturally competent pathogens, a process independent of clinical interventions driving population structures. Therefore, the trajectories that pneumococci have taken in response to vaccination and antimicrobial pressures may serve as a warning for these other respiratory pathogens.

Concluding thoughts

Despite advances in global surveillance of pneumococcal infections post-PCV, it is still difficult to predict future trends in serotype replacement and antibiotic resistance. Expanding vaccine valency coupled with continued and expanded antibiotic coverage impart strong selective pressures resulting in a highly dynamic population structure (Fig. 4). Thus, it is anticipated that even with the introduction of the newest 20-valent PCV in 2023, subsequent increase of drug resistant strains will continue to emerge in the ensuing vacuum. To stay ahead of these population perturbations and their subsequent potential resistance profiles, there has been a focus on developing predictive models using genomic datasets upon selection by various hypothetical vaccines^{83,84,143,144}. This knowledge can be used to design rational vaccines that can theoretically minimize antibiotic resistance and IPD in high risk populations¹⁴³.

Figure 4. Vaccines and antibiotics alternate impact on pneumococcal populations.

Following introduction of an initial conjugate vaccine (red syringe), vaccine serotypes (red line) are rapidly replaced in the population by non-vaccine serotypes (green line). Expansion of vaccine valency to include relevant serotypes (green and blue syringes) continues this trend of serotype replacement (blue line). Concurrent with vaccine pressure, antibiotics exert pressure on the population escaping vaccine coverage (pill boxes) and encourages capsule switching by homologous gene transfer of contiguous cap and pbp genes. In parallel, non-vaccine and non-typable pneumococci rise in the population via serotype replacement (black line).

Rather than continue the battle between PCVs and pneumococcus, there has been an increased interest in developing a universal pneumococcal vaccine to engender serotype independent protection against both mucosal and invasive infection¹⁴⁵. The ideal universal vaccine could be a protein-based or whole-cell vaccine, targeting proteins that are conserved across encapsulated and non-encapsulated pneumococci. However, there are some areas of further research that must be done to develop optimal protein based or whole-cell vaccines. Success of these advances requires: understanding the expression levels of surface proteins and how they differ across pneumococcal strains to identify the best targets for a protein-based vaccine; how anti-protein or whole cell antibody efficacy changes between colonization and infection, when capsule expression levels differ and thus alter antibody binding to surface macromolecules; and including protein targets that are immunogenic from NTPn. A universal vaccine would theoretically reduce the rate of all serotypes of multi-drug resistant pneumococci and also eliminate NTPn as an important reservoir for genetic recombination. However, even with a successful universal pneumococcal vaccine, the oral streptococcal species that drive the generation of mosaic PBPs would continue to serve as an important reservoir for antibiotic resistance development. While serotypes targeted by vaccines are rapidly reduced in prevalence, the continued utilization of antibiotics and the rise of strains outside vaccine coverage underscore the importance of continued surveillance to anticipate the characteristics of the next wave of pneumococcal disease in response to ever expanding vaccine coverage.