Strengthening phage resistance of Streptococcus thermophilus by leveraging complementary defense systems

Abstract

CRISPR-Cas and restriction-modification systems represent the core defense arsenal in Streptococcus thermophilus, but their effectiveness is compromised by phages encoding anti-CRISPR proteins (ACRs) and other counter-defense strategies. Here, we explore the defensome of 263 S. thermophilus strains to uncover other anti-phage systems. The defense landscape of S. thermophilus is enriched by 21 accessory defense systems, 13 of which have never been investigated in this species. Experimental validation of 17 systems with 14 phages reveals a range of anti-phage activities, highlighting both broad and narrow specificities across the five viral genera infecting S. thermophilus. Synergies are observed when combining CRISPR immunity with accessory systems. We also assess the fitness cost associated with the chromosomal integration of these systems in their native context and find no impact under laboratory or industrial conditions. These findings underscore the potential of these accessory defense systems to enhance the resistance of S. thermophilus, particularly against ACR-encoding phages.

Fig. 1. S. thermophilus anti-phage defense systems extend beyond CRISPR-Cas and RM systems.

A Prevalence of known defense systems in S. thermophilus (N = 263, RefSeq as of March 2023). These systems are categorized into previously reported and newly identified in this study. The 18 systems experimentally validated in this work are highlighted in bold. The “Borvo-like” system differs from the original Borvo, being annotated as a two-gene system composed of two BovA subunits, rather than one. B Distribution of the number of defense systems per genome. The red dashed line indicates the average number of defense systems per S. thermophilus genome. C Phylogenetic tree of complete S. thermophilus genomes (N = 85), illustrating the distribution of identified defense systems. Six incomplete genomes were included to show systems which were not found in complete genomes (marked with an asterisk). Branch lengths represent the number of amino acid substitutions per site. Source data are provided as a Source Data file.

Fig. 2. CRISPR and RM systems are widespread in S. thermophilus, and phages have several counter defense mechanisms.

Co-occurrence of (A). CRISPR-Cas loci and (B). RM types within the same S. thermophilus genome. C Distribution of the number of RM systems per genome according to the type. D Heatmap of co-occurrence within the same genome of RM type II with different predicted restriction sites. Only strains harboring at least two RM type II systems are shown on the phylogenetic tree (N = 95). RM systems with no predicted restriction site were divided into different unknown categories based on the restriction enzyme annotation (MvaII or AlwI) or subtype (IIG). E Prevalence of predicted anti-CRISPR (ACR) proteins in phages infecting S. thermophilus (N = 191, NCBI June 2023). Experimentally validated ACRs in S. thermophilus are marked with a star. Color coding corresponds to phage genus. F Plot of the ratios of observed to expected number of restriction sites in streptococcal phages for RM type II and III restriction sites (predicted with REBASE). The observed number of sites corresponds to the count of each site within the phage genomes, while the expected number was estimated using a Markov-immediate neighbor dependence statistical model. Ratios < 0.75 indicate underrepresentation and ratios > 1.25 indicate overrepresentation. Source data are provided as a Source Data file.

Fig. 3. Activity of defense systems against phages infecting S. thermophilus and L. cremoris.

A Strategy for assessing the efficiency of defense systems against streptococcal phages from five genera and lactococcal phages from two genera. Anti-phage activity was measured by evaluating the EOP of phages on the host strain containing the defense mechanism cloned on a plasmid compared to the host strain with the empty plasmid. B Comparison of Dodola and Kiwa defense systems efficiency against L. cremoris phage p2 when cloned into low-copy and high-copy vectors. C EOP heatmap of phages on strains carrying each defense system. Darker shades indicate stronger anti-phage protection. An EOP of 10⁻¹ was used as the minimum threshold for anti-phage activity. Three biological replicates were done. Some defense systems were not tested against a few phages, as they are naturally present in corresponding host strains. Two homologs were tested for Gao19, PD-Lambda-1, RosmerTA, and Hachiman. Phages encoding known ACR are highlighted in bold. The top to bottom presentation of the defense systems reflects the high to low prevalence (Fig. 1A). The genetic organization of the defense systems is shown on the right side of the heatmap. A detailed description of their functional domains is available in Supplementary Data 6. D Comparison of anti-phage activity of two Gao19 homologs. Gao19 was found in 35 S. thermophilus strains, of which 2 corresponded to Gao19^ST3 and 9 to Gao19^ST109. Bars show the mean EOP with error bars indicating standard deviation from biological replicates (n = 3). Filled circles indicate the presence of countable plaques, while hollow red circles denote lysis zones where plaques were not counted. Half circles indicate the absence of visible plaques or lysis zones even with undiluted phage lysate, showing that the defense system reduced phage levels below the limit of detection (EOP < 10⁻⁷). AB^R Antibiotic resistance gene, Cm^R Chloramphenicol resistance, DUF Domain of unknown function, EOP Efficiency of plaquing, Ery^R Erythromycin resistance, EV Empty vector, ori Origin of replication, SIR Sirtuin, TM Transmembrane domain. Source data are provided as a Source Data file.

Fig. 4. Combining defense systems enhances phage resistance.

A CR1-immune S. thermophilus DGCC7710 was transformed with the pTRKL2 vector carrying either Gabija, Hachiman, or Thoeris. Strains expressing combinations of two defense systems were compared to strains carrying individual systems. Spot tests illustrate resistance profiles against phage D5691 (encoding AcrIIA6, which inhibits CR1) and phage 2972 (lacking known ACR). B Liquid culture assays comparing CR1 + Gabija combinations against phages 2972 and D5691 at initial MOIs ranging from 0.0005 to 50. Control strains include wild-type DGCC7710 (WT) and CR1-immune DGCC7710 (CR1BIM) carrying the empty pTRKL2 vector. The area under the curve (AUC) was calculated for each condition, and the relative AUC was determined by normalizing to the non-infected control. Synergy between defense systems is indicated by a black asterisk, and additive effects are indicated by a red asterisk. C Relative AUC values for combinations of CR1 with either Thoeris or Hachiman against phages 2972 and D5691. The term “System” refers to Thoeris or Hachiman as indicated. Synergy and additive effects are indicated by an asterisk color in black and red, respectively. D Quantification of escape phages recovered after liquid assays at MOI 0.05 with phages 2972 and D5691 for each individual and combined defense system. Filled circles indicate the presence of countable plaques, while hollow red circles signify lysis, but plaques were not counted. E Schematic illustration of pyramiding (combining two defense systems in the same strain) versus mixing (co-culture of two strains, each expressing a different system). F Comparison of pyramiding versus mixing strategies for three combinations: CR1 + Gabija, Dodola + Gabija, and CR1 + Thoeris. For all experiments, lines or bars represent the mean value of biological replicates (n = 3), with shaded areas indicating confidence intervals and error bars indicating standard deviation. The statistical significance of the synergy scores was evaluated using a two-sided one-sample t test with Benjamini-Hochberg correction for multiple testing. A positive score significantly different from zero (p-value < 0.05) indicated synergy, while a score not significantly different from zero indicated an additive effect.

Fig. 5. Fitness impact of integrating an additional defense system or island into the chromosome of S. thermophilus DGCC7710.

A Schematic illustration of the insertion of a defense system into the chromosome of S. thermophilus through natural competence. A recombination template, consisting of the defense system flanked by 1 kb regions that are homologous to the upstream and downstream sequences in the recipient strain (e.g., DGCC7710), was transformed. Phages were used to select bacteria that integrated the defense system. B, C Assessment of the fitness cost (bacterial growth and milk acidification monitoring) following chromosomal integration of PD-Lambda-1 (B) or a defense island comprising Hachiman and RM II (C) at their native locus in DGCC7710. For bacterial growth graphs, each line represents the mean of biological replicates (n = 3) with shaded areas indicating confidence intervals. In milk acidification assays with phages, a phage cocktail was used at an initial MOI 0.0005. Each line shows the mean values, with shaded areas representing confidence intervals calculated from either three technical replicates (for PD-Lambda-1, Hachiman clone 2.4, and Hachiman clone 2.1 knockout conditions) or two biological replicates, each with three technical replicates (for Hachiman clone 2.1 and DGCC7710 WT). D Genomic context of selected defense systems integrated at non-native loci in DGCC7710. Thoeris originates from a short contig lacking conserved neighboring genes, preventing its precise localization within the DGCC7710 genome. E Evaluation of defense efficiency (EOP) and fitness cost (bacterial growth monitoring) for systems integrated at non-native loci. All experiments were performed in biological replicates (n = 3). Bars and curves indicate mean values, with standard deviations shown as error bars and confidence intervals as shaded areas. In the EOP plots, filled circles represent conditions with countable plaques, while hollow red circles indicate lysis zones without distinct plaques. EOP Efficiency of Plaquing, WT Wild Type, OD Optical Density, RM Restriction-modification system, R-S-M Restriction-Specificity-Methylase subunits of RM type I system. Source data are provided as a Source Data file.