CRISPR-Cas provides limited phage immunity to a prevalent gut bacterium in gnotobiotic mice

Abstract

Many bacteria and archaea harbor the adaptive CRISPR-Cas system, which stores small nucleotide fragments from previous invasions of nucleic acids via viruses or plasmids. This molecular archive blocks further invaders carrying identical or similar nucleotide sequences. However, few of these systems have been confirmed experimentally to be active in gut bacteria. Here, we demonstrate experimentally that the type I-C CRISPR-Cas system of the prevalent gut bacterium Eggerthella lenta can specifically target and cleave foreign DNA in vitro by using a plasmid transformation assay. We also show that the CRISPR-Cas system acquires new immunities (spacers) from the genome of a virulent E. lenta phage using traditional phage assays in vitro but also in vivo using gnotobiotic (GB) mice. Both high phage titer and an increased number of spacer acquisition events were observed when E. lenta was exposed to a low multiplicity of infection in vitro, and three phage genes were found to contain protospacer hotspots. Fewer new spacer acquisitions were detected in vivo than in vitro. Longitudinal analysis of phage-bacteria interactions showed sustained coexistence in the gut of GB mice, with phage abundance being approximately one log higher than the bacteria. Our findings show that while the type I-C CRISPR-Cas system is active in vitro and in vivo, a highly virulent phage in vitro was still able to co-exist with its bacterial host in vivo. Taken altogether, our results suggest that the CRISPR-Cas defense system of E. lenta provides only partial immunity in the gut.

Fig. 1. Timeline of the gnotobiotic mouse model.

A Showing the lifespan of the mice included in the study. The mice were initially bred and housed in a germ-free isolator (light blue arrow) until age of 5 weeks when they were transferred to IVCs (dark blue arrow) for individual group caging followed by two weeks of acclimatization (gray arrow) prior intervention at age 7 weeks. Feces (brown cross) were sampled from each individual mouse before and after inoculation (yellow triangle) with phages and/or bacteria. The baseline mice were euthanized and sampled at age of 3 weeks. B Listing of the experimental groups, their abbreviation, and the inoculated bacterium and/or phage. SM buffer was used as saline solution.

Fig. 2. The order and structure of the type I-C CRISPR-Cas system found in E. lenta DSM 15644.

R repeat, S spacer, TR terminal repeat.

Fig. 3. Overview of cell density and bacterial and phage abundance.

A Growth curve of E. lenta DSM 15644 during infection with phage PMBT5 at four different multiplicities of infections (MOI) and a control with no phages added were performed with biological triplicates (n = 3). Bacterial growth was measured (absorbance at OD_600nm) at several time points for 144 h. The bacterial (B) and phage abundance (C) was measured by qPCR (technical duplicates of the biological triplicates, n = 6). Primers designed to specifically target the genomes of E. lenta DSM 15644 (cas1 gene) and phage PMBT5 (gene coding for a tail-associated lysin) were used to measure total gene copies found in the cultures. A minimum threshold of 10 gene copies was applied. The error bars show the standard deviation within each MOI.

Fig. 4. Overview of spacer acquisitions in the in vitro settings.

A Expanded CRISPR arrays generated by PCR on whole culture populations in selected samples (Fig. S6 for all samples) representing two replicates of all four MOI after 48 h and 24 h of incubation of E. lenta DSM 15644 exposed to phage PMBT5. DNA ladder is a 100-bp scale. With the degenerate primers, the expanded CRISPR array with one spacer “+1” was expected to yield a PCR product at ~110 bp (Fig. S1). No expanded CRISPR arrays were observed in samples with no added phages (after 48 h incubation) or with MilliQ water added. The PCR-product at ~40 bp likely represented primer dimers. B The annotated genome of phage PMBT5 highlights the genes that are presented in (C) with a bar plot showing the number of reads/spacers that matched to phage genes with biological triplicate (n = 3) at MOI 10, 1, 0.01, and 0.01. Three genes appeared as hotspots of spacer acquisitions and were coding for a portal protein (gp04⁺), a SHOCT domain-containing protein (gp12⁺⁺), a replication initiator protein (gp39⁺⁺⁺). These are marked by boxes with red dashed lines. A few genes were targeted at different positions within the same gene. D Graph illustrating a tendency of an inverse relation between MOI and cell density (OD_600nm) of the average number of reads/spacer acquisitions in E. lenta DSM 15644 exposed to phage PMBT5 (n = 3). The error bars show the standard deviation within each MOI.

Fig. 5. Bar plot showing colony forming units per µg DNA (CFU/µg DNA) in a logarithmic scale of transformed E. lenta DSM 15644 cells with plasmid pNZ123 and derivatives that provides chloramphenicol resistance.

E. lenta DSM 15644 was transformed with pNZ123 (WT) and two derivatives containing each the same two protospacers but a different PAM (pNZ123::GGG-S2-GGG-S1, pNZ123::TTC-S2-TTC-S1, pNZ123::WT). Absence of plasmid transformation indicates interference activity of the type I-C CRISPR-Cas system. Transformation assays were, respectively, replicated 2, 2, 4, and 4 times with 3 technical replicates. The error bars show the standard deviation.

Fig. 6. The bacterial and phage abundance measured by qPCR in feces samples at different time points (day –1, 0, 1, 1.5, 2, 3, 4, 5, 12, 19, and 26) of four biological replicates (n = 4).

Where day –1 were feces samples from GB mice sacrificed at the age of 3 weeks, day 0 were feces samples from GB mice when transferred from isolator to individual ventilated cages at another housing facility, and day 1 were just before culture inoculation. Primers designed to specifically target the genomes of E. lenta DSM 15644 (cas1 gene) and phage PMBT5 (gene coding for a tail-associated lysin) were used to measure total gene copies found in the feces samples. A minimum threshold of 10 gene copies was applied. The error bars show the standard deviation within each treatment group at the given day.

Fig. 7. Overview of spacer acquisitions in the in vivo settings.

A An agarose gel showing expanded CRISPR arrays generated by PCR on whole populations in selected samples representing the four EL + Phage mice (Mouse ID; 9, 10, 11, 12) from day 5, 12, 19, and 26, as well as from controls at arrival (Day 1) and baseline mice (Mouse ID: 1, 2, 3, 4). A 100-bp DNA ladder was used to estimate PCR product size. With the degenerate primers, the acquisition of one spacer “+1” was expected to yield a PCR product at ~110 bp (Fig. S1) and then ~70 bp for additional spacers. The PCR-product at ~40 bp likely represented primer dimers. B The annotated phage genome of PMBT5 highlights the one gene coding for a DNA gyrase inhibitor (gp34) that is presented in (C) with a line plot showing average number of reads/spacers over time (n = 4) that matched the phage genome. The error bars show the standard deviation within each treatment group at the given day.