Multiple Peptidoglycan Modification Networks Modulate Helicobacter pylori's Cell Shape, Motility, and Colonization Potential

Helical cell shape of the gastric pathogen Helicobacter pylori has been suggested to promote virulence through viscosity-dependent enhancement of swimming velocity. However, H. pylori csd1 mutants, which are curved but lack helical twist, show normal velocity in viscous polymer solutions and the reason for their deficiency in stomach colonization has remained unclear. Characterization of new rod shaped mutants identified Csd4, a DL-carboxypeptidase of peptidoglycan (PG) tripeptide monomers and Csd5, a putative scaffolding protein. Morphological and biochemical studies indicated Csd4 tripeptide cleavage and Csd1 crosslinking relaxation modify the PG sacculus through independent networks that coordinately generate helical shape. csd4 mutants show attenuation of stomach colonization, but no change in proinflammatory cytokine induction, despite four-fold higher levels of Nod1-agonist tripeptides in the PG sacculus. Motility analysis of similarly shaped mutants bearing distinct alterations in PG modifications revealed deficits associated with shape, but only in gel-like media and not viscous solutions. As gastric mucus displays viscoelastic gel-like properties, our results suggest enhanced penetration of the mucus barrier underlies the fitness advantage conferred by H. pylori's characteristic shape.


Supplemental Materials and Methods
Genetic manipulations. PCR SOEing [1] was used to create knockout alleles as described [2].
The csd4 point mutant was generated by amplifying csd4 with flanking primers containing XhoI and EcoRI restriction sequences and cloning the gene into a Bluescript vector (pBluescript II SK+, Invitrogen). This plasmid, pLKS2, served as the template for PCR-based site-directed mutagenesis using a QuikChange Site-Directed Mutagenesis kit (Stratagene). The presence of only the desired csd4 point mutation (A665C) in the resulting plasmid, pLKS12, was confirmed by sequencing. Complementation at the rdxA locus was achieved by inserting the deleted gene into pLC292, a plasmid containing rdxA flanking sequences [3]. pLKS24 was generated by inserting a PCR product containing csd4 flanked by XbaI and XhoI restriction sequences into pLC292. pLKS27 was constructed by subcloning the XbaI-SpeI fragment containing csd5 from pLKS23, a TOPO vector containing csd5, into pLC292.
The first 20 amino acids of Csd4 are predicted to encode an N-terminal transmembrane domain. In order to create a Csd4 E. coli expression vector, we decided to clone only the portion downstream of the transmembrane domain (aa . The csd4 gene (bp 61 through 140 bp downstream of the stop codon) was amplified by PCR using primers containing a 5' NcoI site and a 3' XhoI site. The primers were designed based on the sequence of HP1075, the csd4 gene in H. pylori 26695, as the G27 sequence was not available at the time this plasmid was made.
Two sequence differences relative to the G27 sequence were introduced by the forward primer: a63g, a silent mutation, and c65g, resulting in a T22M mutation with respect to the G27 amino acid sequence. The PCR product was cloned into the NcoI and XhoI sites of pET15-HE (generously provided by Barry Stoddard, FHCRC), resulting in pLKS1. This plasmid allows expression of N-terminally truncated Csd4 with an N-terminal His•Tag behind a thrombin cleavage site.
Plasmids were prepared using Qiagen kits and H. pylori genomic DNA isolated using the Wizard Genomic DNA kit (Promega). All restriction enzymes and high-fidelity polymerases used for PCR SOEing were obtained from New England Biolabs or Invitrogen. Sequencing was performed by the FHCRC Genomics Shared Resource and sequences analyzed using Sequencher (Gene Codes).
PCR products and plasmids were introduced into the chromosome using natural transformation as described [4] and the presence of the correct allele affirmed by PCR. Clones were additionally checked for urease activity, flagella-based motility and the absence of undesired point mutations in the coding sequences. Single clones were used for phenotypic characterization and infection experiments. Selectable alleles were transferred between H. pylori strains by transforming genomic DNA using the same procedures. Cocktail (Roche)), and stored at -20°C. Cells were thawed and then lysed by three additional freeze-thaw cycles (dry ice/methanol bath), followed by three sonication cycles at 10% power (10 sec total, 1 sec on/1 sec off) (Fisher Scientific Sonic Dismembrator Model 500). Cell debris was removed by centrifugation (10,000×g, 10 min).
Ni-NTA agarose (Invitrogen) was prepared by rinsing once with 5 volumes of water and then twice with 5 volumes of binding buffer (50 mM sodium phosphate pH 8, 500 mM NaCl, 10% glycerol, 10 mM MgCl 2 ). Binding was performed in batch by adding 5 volumes of cell lysate to prepared Ni-NTA agarose at 4°C for 1 hr with gentle mixing. This mixture was poured into a column and washed four times with 10 volumes of wash buffer (binding buffer plus 20 mM imidazole). Csd4 was eluted with 10 volumes of elution buffer (binding buffer plus 250 mM imidazole). Elution fractions containing approximately 95% pure Csd4 (as judged visually by Coomassie staining of SDS-PAGE) at reasonably high concentration (>0.5 mg/mL by Bradford assay) were pooled. The glycerol concentration of the pool was increased to 20% prior to storage at -20°C.
Statistical analysis of cell shape distributions. Kolmogorov-Smirnov (KS) statistics were calculated to assay the differences between the double mutant and single mutant (or wild-type) distributions of side curvature generated by CellTool. In order to determine a biologicallyplausible null distribution for the range of KS statistics expected to be observed by chance due to biological variability, a bootstrapped distribution of KS values was calculated for the wild-type and single-mutant strains as follows: the side-curvature values from each strain were resampled (with replacement) into two separate samples 10 5 times and the KS statistic computed for each pair of samples. For each comparison of a double mutant to a single mutant (or wild-type), the measured KS distance between the two distributions was compared to the null distribution calculated for the single mutant (or wild-type), generating a one-tailed p-value. As the distributions of side-curvature values are quite different for the different strains, it was most appropriate to calculate separate null distributions for each single mutant strain.
Bioinformatic analyses. Csd4/5 homologues were identified by performing a BLASTP search of the NCBI non-redundant protein sequences database using HPG27 Csd4 and Csd5 as queries.
Hits with a maximum E-value of 0.001 were considered potential homologues, but we focused subsequent analyses on hits aligning across at least 75% of the query sequence and showing >50% similarity (determined by MatGAT 3.0 [5]). All of these homologues had BLASTP Evalues < 1E -40 . Altogether our Csd4 BLAST search yielded 108 hits with E-values < 0.001, 87 of which met our 50% similarity cut-off. Our Csd5 BLAST search yielded 39 hits with E-values < 0.001, 29 of which met our 50% similarity cut-off.
To generate the phylogenetic trees in Figure S1 we included only species/strains for which the genome was listed as "complete" by NCBI. We retained all 10 complete H. pylori genomes, but chose only one representative strain for the 24 other species/subspecies with Csd4 homologues. Dataset compilation and editing, multisequence alignment (through a built-in Clustal W algorithm), and construction of neighbor-joining trees were all accomplished using MEGA 5.05 [6].