Significance
Streptococcus pneumoniae is an important bacterial pathogen responsible for many serious infections worldwide. Infections are often treated with penicillin antibiotics. This provides a selection pressure for the emergence of resistant strains over time, reducing treatment options and threatening patients. Combining lab evolutionary and comparative genomics approaches, we identify unique loss of function mutations in the Pde1 enzyme that lead to penicillin resistance in the absence of classical resistance determinants. We confirm this effect across clinical isolates and characterize the impact of natural genetic variation on Pde1 function. Characterization of end-stage penicillin resistance genes has, so far, not led to effective mitigations. Here, by characterizing the evolutionary events leading toward resistance, we open different possibilities for interventions against resistant S. pneumoniae.
Abstract
Streptococcus pneumoniae is a major human pathogen and rising resistance to β-lactam antibiotics, such as penicillin, is a significant threat to global public health. Mutations occurring in the penicillin-binding proteins (PBPs) can confer high-level penicillin resistance but other poorly understood genetic factors are also important. Here, we combined strictly controlled laboratory experiments and population analyses to identify a new penicillin resistance pathway that is independent of PBP modification. Initial laboratory selection experiments identified high-frequency pde1 mutations conferring S. pneumoniae penicillin resistance. The importance of variation at the pde1 locus was confirmed in natural and clinical populations in an analysis of >7,200 S. pneumoniae genomes. The pde1 mutations identified by these approaches reduce the hydrolytic activity of the Pde1 enzyme in bacterial cells and thereby elevate levels of cyclic-di-adenosine monophosphate and penicillin resistance. Our results reveal rapid de novo loss of function mutations in pde1 as an evolutionary gateway conferring low-level penicillin resistance. This relatively simple genomic change allows cells to persist in populations on an adaptive evolutionary pathway to acquire further genetic changes and high-level penicillin resistance.
Streptococcus pneumoniae is a major human pathogen and causes over 1 million deaths each year (1, 2). S. pneumoniae infections are commonly treated with β-lactam antibiotics, such as penicillin, amoxicillin, or ampicillin, but an alarmingly high number of strains have developed resistance to this global front-line treatment strategy (3–5). High levels of penicillin resistance in S. pneumoniae are classically associated with dramatic changes to the drug target, which are enzymes named penicillin-binding proteins (PBPs) that have important functions in cell wall synthesis (6, 7). Resistance involves alterations to the underlying pbp genes and the introduction of large sequence blocks of divergent genetic material originating from other Streptococcus species (7, 8). The resulting chimeric, or mosaic, PBP proteins retain their essential cell wall synthetic functions but have lower binding affinity for β-lactams. In this way, the acquisition of mosaic-PBPs has been the paradigm for gaining resistance against penicillin antibiotics in S. pneumoniae since their discovery in the 1970s (6).
Whilst studies of S. pneumoniae penicillin resistance have focused on mosaic-PBPs, the acquisition of these genes is not always sufficient for high levels of resistance, with other genes also identified as important (9–11). For example, the two-component regulatory system CiaRH (12, 13) and mutations in the murMN operon are linked to penicillin resistance (14, 15). As more genomic data becomes available, understanding of the genetics underlying penicillin resistance has dramatically increased (11, 16), with comparative genomic analyses identifying numerous associated genetic variations (7, 17). However, many questions remain about covariation and interactions among associated genetic elements, their clinical significance, and the order of their acquisition. Here, we combined laboratory evolution experiments with comparative genomics of natural populations to find that novel mutations in the gene called pde1 lead to penicillin resistance in S. pneumoniae. We propose pde1 be considered a new penicillin resistance determinant in S. pneumoniae and characterize the effect natural variation in this gene has on the cell.
Pde1 is involved in bacterial second messenger signaling by reducing the cyclic di-AMP (di-adenosine monophosphate, c-di-AMP) concentration in the cell. Since its discovery over a decade ago (18), changes in cellular c-di-AMP concentrations have been linked to a variety of nonantibiotic-related phenotypes and stress responses in S. pneumoniae (19–22). The c-di-AMP molecule is generated from two Adenosine Tri Phosphate (ATP) molecules by the cyclase CdaA (DacA, SPD_1392, see Fig. 1A). To reduce the c-di-AMP concentration in the cell, the Pde1 phosphodiesterase enzyme (SPD_2032) cleaves c-di-AMP into the linear molecule phosphoadenylyl adenosine (5′-pApA) (19). This protein is an ortholog of GdpP found in Staphylococcus aureus (18) or other Bacillota (formally Firmicutes) bacteria like Bacillus subtilis (23) and Enterococcus faecalis (24, 25). The Pde1 protein contains two N-terminal transmembrane helices, a PAS (Per-Arnt-Sim) sensory domain, and a degenerated GGDEF domain (Fig. 1B). Yet, for its in vitro c-di-AMP hydrolytic function, only the DHH and DHHA1 domains are required (19).
Reduction of Pde1 function in S. pneumoniae has been previously linked to survival in various conditions. These include a higher susceptibility to DNA damage, defective ion homeostasis, reduction in competence, as well as to lower tolerance to heat, acidic, and osmotic stress (19–21). Pde1 is also required for full virulence in mouse and rabbit infection models (19, 26–28). In addition to Pde1, the S. pneumoniae genome encodes a second phosphodiesterase termed Pde2 (SPD_1153, also called PapP, 29). Similarly to Pde1, Pde2 hydrolyzes c-di-AMP but also uses the linear molecule 5′-pApA as substrate, in each case forming AMP and thereby returning the signaling molecule back to the metabolic nucleotide pool (Fig. 1A, 19). More recently, cells lacking Pde2, but not Pde1, stimulated induction of interferon β and led to hyperactivation of the macrophage responses during infection. This suggests that Pde1 and Pde2 have distinct functions in the cell despite sharing similar c-di-AMP phosphodiesterase activities (19, 30).
In this study, we identified an association between the second messenger c-di-AMP and increased levels of penicillin resistance against cell wall targeting antibiotics in S. pneumoniae. Laboratory evolution experiments and genome sequencing revealed that loss of function mutations in the pde1 gene conferred low-level ampicillin resistance. To study the clinical significance of these data, we extended this observation to natural populations and analyzed >7,200 S. pneumoniae genomes. This identified five common mutations in pde1 associated with penicillin resistance across these populations. To gain a mechanistic understanding of how loss of function changes in Pde1 lead to penicillin resistance, we tested each of the substitution variants identified and revealed that they all reduce Pde1 hydrolytic activity, thereby increasing the concentration of c-di-AMP in the cell. Our results suggest loss of Pde1 function may act as an evolutionary gateway toward increased penicillin resistance in S. pneumoniae, allowing cells to survive at low antibiotic concentrations to later thrive in these conditions after acquiring additional genetic changes and complementary resistance mechanisms.
Results
Loss of Pde1 c-di-AMP Phosphodiesterase Function Increases Ampicillin Resistance.
To identify novel genetic loci that support penicillin resistance, we took a laboratory evolution approach. For this, we exposed S. pneumoniae cells to ampicillin, a member of the penicillin group of antibiotics. We generated ampicillin slope plates, selecting for spontaneous resistance mutants along a gradient that ranged from 0 to a maximum of 62.5 ng mL−1 ampicillin. The maximum concentration of 62.5 ng mL−1 was chosen as it represents the “susceptible” to “intermediate” penicillin resistance boundary by international clinical standards (SI Appendix, Fig. S1A, 31).
This laboratory evolution approach generated 20 independent mutants. All strains showed increased ampicillin resistance and no mutations in any pbp genes or previously known resistance determinants (SI Appendix, Fig. S1B). A comparative genomics analysis of these strains revealed that 40% of isolates contained mutations in the same gene (SI Appendix, Fig. S1B). This gene encodes the c-di-AMP phosphodiesterase pde1 (spd_2032). The identified mutations were almost exclusively pde1 nonsense or frameshift mutations (6 in total), with only one nonsynonymous mutation resulting in a H345Y substitution (SI Appendix, Fig. S1B). This substitution was located in the DHH domain, which is the region of Pde1 responsible for c-di-AMP hydrolysis and is predicted to be important for Mn2+ cofactor binding (23). This suggested that the H345Y substitution also results in either the reduction or loss of Pde1 hydrolytic activity. Combined, these findings suggest that loss of Pde1 function is responsible for the increased ampicillin resistance phenotype. To confirm this observation, we generated a pde1 deletion mutant and tested its resistance to ampicillin using a viability assay. Consistent with our expectation, deletion of pde1 increased the ampicillin resistance in our viability assays (Fig. 1C).
The Pde1 enzyme is known to degrade the c-di-AMP second messenger in S. pneumoniae, hydrolyzing it to 5′-pApA before a secondary hydrolysis step cleaves this intermediate into two AMP molecules (Fig. 1A, 19). Therefore, reducing Pde1 function increases c-di-AMP concentrations by lowering the hydrolysis rate in the cell (19). We confirmed this for our Δpde1 strain using an ELISA assay to measure the cytosolic c-di-AMP concentration. This assay showed that deletion of pde1 increases c-di-AMP concentrations by approximately twofold (Fig. 1D). S. pneumoniae cells are typically surrounded by a polysaccharide capsule and this is often removed in lab-based molecular studies. Although uncommon, this can have an impact on cell growth in certain conditions. Indeed, we found deleting pde1 reduced cell size by ≈15% in our wild-type (WT) strain without the capsule (SI Appendix, Fig. S2), although this phenotype was lost when the Δpde1 mutation was introduced to the encapsulated strain background (SI Appendix, Fig. S3). We also tested the Δpde1 mutant growth rate and found that it was not significantly altered, consistent with previously published observations (SI Appendix, Fig. S2) (26). We conclude that deletion of pde1 increases ampicillin resistance and has no effect on growth rate or cell morphology when the polysaccharide capsule is present.
Loss of Pde2 c-di-AMP Phosphodiesterase Function Increases Penicillin Resistance but Also Causes a Reduction in Cell Size.
The S. pneumoniae genome encodes a second enzyme capable of hydrolyzing c-di-AMP, Pde2, and previous studies have shown that loss-of-function of either enzyme increases the c-di-AMP pool in the cell to a similar degree (Fig. 1D) (19). Because of this, we reasoned that pde2 loss of function might also increase resistance to the penicillin class of antibiotics. As predicted, a Δpde2 mutant led to an increase in ampicillin resistance (Fig. 1C). Next, we tested a Δpde1 Δpde2 double mutant, which gave a similar level of ampicillin resistance as Δpde1 but had fourfold higher levels of c-di-AMP compared to WT (Fig. 1 C and D). In both cases, the increased ampicillin resistance came at the cost of reducing both growth rate and cell size (SI Appendix, Fig. S2). To quantify the morphological changes, we measured the morphology of >2,000 Δpde2 and the Δpde1 Δpde2 double mutant cells. These were both significantly smaller than WT controls (SI Appendix, Figs. S2 and S3), showing a 30 to 40% reduction in cell size and a 20% reduction in both length and width (SI Appendix, Fig. S2). This reduction in size was less pronounced in an encapsulated strain background (15 to 20%), but this still represents a significant deviation from WT (SI Appendix, Fig. S3). We conclude that loss of pde2 also increases penicillin resistance but the S. pneumoniae cell pays a greater fitness cost, evidenced by the reduction in overall cell size and growth rate in strains lacking capsule (SI Appendix, Figs. S2, S3, and S5).
Maintaining High Levels of c-di-AMP Is Required for S. pneumoniae Survival at Sub-MIC (Minimum Inhibitory Concentration) Ampicillin Concentrations.
To measure the effect ampicillin has on c-di-AMP levels in S. pneumoniae cells, we challenged WT, Δpde1, and Δpde2 strains with ampicillin concentrations lower than the MIC. Using an ELISA assay to quantify the c-di-AMP pool. This revealed sub-MIC concentrations of ampicillin cause a significant reduction in c-di-AMP concentration in WT cells (Fig. 1E). This reduction was mitigated by the removal of either phosphodiesterase enzyme Pde1 or Pde2 (Fig. 1E). This suggests the consequence of removing Pde1 or Pde2 from the cell is both an increase in the base c-di-AMP concentration and also a dampening of the c-di-AMP reduction effect when the cells are challenged with sub-lethal ampicillin concentrations.
Loss of pde1 and pde2 Function Increases Resistance to Cell Wall–Targeting Antibiotics.
To examine the more widespread clinical significance of our ampicillin resistance measurements, we confirmed that our Δpde1 and Δpde2 strains were also more resistant to penicillin G (SI Appendix, Fig. S5). Penicillin G is used for standardized MIC testing (31) and is often used to treat S. pneumoniae infections in the clinic (3, 4). We found Δpde1 increased the penicillin G MIC by approximately twofold (from 6 to 12 ng mL−1), while Δpde2 increased the MIC to a slightly lower concentration of 8 ng mL−1 (SI Appendix, Fig. S5C). To test whether these increases were specific to the penicillin class of antibiotics or more broadly applicable to cell wall-targeting antibiotics, we repeated our MIC testing using vancomycin, a potent inhibitor of cell wall synthesis (SI Appendix, Fig. S5). We also tested both erythromycin and chloramphenicol, which act as inhibitors of protein synthesis (SI Appendix, Fig. S5). In these assays, we observed an increase in vancomycin resistance but no change in erythromycin or chloramphenicol resistance levels (SI Appendix, Fig. S5C). We conclude that increases in antibiotic resistance in the absence of the Pde1 and Pde2 function are specific to cell wall–targeting antibiotics.
In the S. pneumoniae genome, the pde1 gene is found in an operon with rplI (spd_2031) with a short overlap between the two open reading frames, while pde2 lies upstream of rpmE2 [spd_1154, (32)]. To confirm the ampicillin resistance phenotype of both Δpde1 and Δpde2 strains was specific to each gene and not a product of downstream genetic effects, we repeated our ampicillin resistance and cell growth assays using ΔrplI and ΔrpmE2 strains. These assays revealed no difference in resistance, no change in growth rate, and no change in cell morphology compared to WT (SI Appendix, Fig. S6). As an additional control, we tested the impact each of the antibiotic resistance cassettes used for strain construction had on our strains and found that none of them influenced ampicillin resistance (SI Appendix, Fig. S6D). These results confirm the ampicillin resistance phenotype is specific to disruption of the pde1 and pde2 loci.
Genetic Variation in pde1 Is Associated with Increased Penicillin Resistance in Circulating S. pneumoniae Populations.
To broaden the significance of our laboratory evolution experiments, we aimed to understand the role pde1 mutations play in penicillin resistance in circulating human carriage and clinical S. pneumoniae populations. We hypothesized that genes involved in c-di-AMP metabolism (cdaA, pde1, and pde2—Fig. 1A) would contain significant genetic variation associated with penicillin resistance in these isolates. A total of >7,200 S. pneumoniae genomes, available through the PubMLST database (33), were split into susceptible (MIC ≤ 62.5 ng mL−1), intermediate (62.5 ng mL−1 < MIC ≤ 2,000 ng mL−1), and resistant (MIC > 2,000 ng mL−1) groups, according to their reported penicillin MICs (Fig. 2). The amino acid sequences of Pde1, Pde2, and CdaA were analyzed for sequence variation in this population, normalized by the number of isolates in each resistance group, and adjusted for protein length. For nonsynonymous substitution mutations, as well as insertions and deletions (indels), we found that Pde1 and Pde2 had similar levels of sequence variation, and both were more variable than the c-di-AMP cyclase CdaA, which is consistent with reports that cdaA is essential (Fig. 3, 19, 34). Furthermore, mutations found in pde1 were enriched in resistant strains, suggesting a link between sequence variation at this locus and penicillin resistance (Fig. 3A). However, more strikingly, we found a fourfold higher accumulation of nonsense mutations in pde1 within the penicillin-resistant group (Fig. 3B). Of these, ≈90% of the mutations clustered very early in the pde1 reading frame, predicted to truncate the protein within its N-terminal trans-membrane domains (Fig. 3D). As these mutations result in a premature stop of translation, Pde1 function would be abolished in these cells. In contrast, nonsense mutations were rarely detected in Pde2 or CdaA (Fig. 3B and SI Appendix, Fig. S7), consistent with the high fitness cost in cell viability, growth, and cell division these mutations would likely bring (19–21, 26–28)…
