A Chlamydia trachomatis strain with a chemically generated amino acid substitution (P370L) in the cthtrA gene shows reduced elementary body production

Chemical mutagenesis allows identification of SNVs that impact on CtHtrA function

Two existing libraries of C. trachomatis mutants were screened for mutations in the cthtrA gene, one based on C. trachomatis serovar L2 (CTL0195) and the other on C. trachomatis serovar D (CT_823). The screening of two libraries was performed to enable enable
higher coverage of the cthtrA gene, while enabling the reduction of possible SNV bias given that the libraries
were prepared using different methods. The C. trachomatis L2 and D serovars share 99.5 % nucleotide identity and 99.4 % amino acid identity
at the cthtrA locus and thus are not expected to exhibit functional differences.

Firstly, we determined that there were 797 C/G to T/A mutations possible in cthtrA; 601 would be non-synonymous, 160 synonymous, and 36 nonsense. We have previously
used a chemical inhibition approach to demonstrate that CtHtrA is critical for the survival of C. trachomatis15], so it was expected that strains harboring null mutations in cthtrA would be non-viable. However, strains containing cthtrA alleles with non-synonymous mutations in functionally important residues could potentially
be used to investigate CtHtrA physiological function(s). The first library screened was a collection of 937
C. trachomatis serovar L2 (434/Bu) mutant strains in which each SNV was mapped by whole genome sequencing.
This library was generated with a mutation frequency ranging from 6 to 25 mutations
per genome following treatment with EMS and ENU 29]. The second library screened was based on C. trachomatis serovar D-LC (CtD-LC) 30] and included ~4600 mutant strains with a mutation frequency of seven SNVs per genome.
Between the two libraries a total of 5537 mutants were screened and a total of 52
cthtrA variants were identified (Additional file 1: Table S1). Eighteen variants had synonymous substitutions and were excluded from
further analysis, while the remaining 35 were analyzed in silico to predict their functional impact based on our previous in silico and biochemical understanding of the CtHtrA amino acid sequence and associated functions 11], 12].

Three CtHtrA mutations have an impact on in vitro proteolysis but not oligomerization

The cthtrA EMS mutations were examined using molecular models of the inactive hexamer and active
24-mer oligomers of CtHtrA, according to 2D and 3D structural motifs known to have a role in CtHtrA protein function 11], 12], 17], 18]. Mutations were identified throughout the entire cthtrA gene, with SNVs identified in the signal peptide, protease domain, PDZ1 domain, and
PDZ2 domain (Additional file 1: Table S1). Six mutations that resulted in amino acid substitutions were selected
for characterization based on in silico prediction of a structural and/or functional impact: E47K (G – A), R55Q (G – A),
A240V (C – T), G268R (G – A), P370L (C – T), and G475E (G – A) (Fig. 1a). The E47K and R55Q mutations were located at the N-terminus of the protease domain,
upstream from loop LA, and had the potential to disrupt the trimeric protease domain
interface by forming a steric clash with an adjacent protease domain loop (Fig. 1b). A240V represented a minor substitution from alanine to valine, two small and hydrophobic
amino acids, however this residue is situated on loop L1 near to the catalytic serine
and any conformational shift that may occur as a result of this mutation, however
minor, would likely impact CtHtrA proteolytic activity (Fig. 1b). G268R was situated in the protease domain active site in loop L2 and potentially
results in a steric clash with loop L2 from an adjacent protease domain, also potentially
disrupting proteolysis (Fig. 1b). P370L was located in the PDZ1 domain at the base of the ‘carboxylate binding loop’
and is likely to disrupt the conformational turn of this loop, which may impact the
binding of the substrate C-terminus to the PDZ1 domain and the subsequent activation
cascade (Fig. 1b). G475E was located in the PDZ2 domain and resulted in a potential steric clash with
a nearby PDZ2 domain loop, as well as an adjacent PDZ1 domain in the oligomeric state,
thus potentially affecting the oligomerization mechanism (Fig. 1b). These six mutations were then generated in vitro in our CtHtrA recombinant protein expression construct using site-directed mutagenesis for
biochemical characterization of the recombinant mutated proteins.

Fig. 1. Mutated residues shown on the CtHtrA monomer of the active 24-meric structure. a. The CtHtrA active monomer with black boxes indicating the location of mutated residues.
Green: protease domain, orange: PDZ1 domain, blue: PDZ2 domain. b. Close-up of wild-type residues (orange side-chain) aligned with the mutated residue (green side-chain)

The proteolysis and oligomerization activities of these recombinant proteins were
compared to the wild-type recombinant protein (CtHtrA). All recombinant proteins were able to cleave full-length ?-casein at 37 °C,
with proteins CtHtrA
_E47K
, CtHtrA
_G268R
, and CtHtrA
_R55Q
displaying similar proteolytic activity to the wild-type (Fig. 2a). The rate of proteolysis was slower for the CtHtrA
_A240V
, and CtHtrA
_G475E
proteins compared to the wild type, while the proteolysis activity of the CtHtrA
_P370L
was substantially reduced (Fig. 2a; Additional file 2: Figure S1). Similarly, in the presence of a peptide substrate, the CtHtrA
_A240V
and CtHtrA
_G475E
proteins demonstrated a reduction in proteolytic rate, respectively, compared to the
wild-type, while no proteolytic activity could be detected for the CtHtrA
_P370L
protein (Fig. 2b). The proteolytic rate of the CtHtrA
_G268R
and CtHtrA
_R55Q
proteins were not substantially different from the wild-type. We have previously shown
that the addition of a second peptide based on the 12 C-terminal residues of ?-casein
(Act1) can activate or increase the proteolytic activity of CtHtrA 11], 12]. Thus, the activation of proteolysis by the recombinant proteins was investigated
and a 1.4–2.3-fold increase in the rate of proteolysis was observed for CtHtrA
_E47K
, CtHtrA
_G268R
, CtHtrA
_R55Q
, CtHtrA
_A240V
and CtHtrA
_G475E
. Alternatively, CtHtrA
_P370L
displayed no detectable proteolytic activity against this peptide substrate (Fig. 2b), which may be due to this mutation causing a conformational change to the ‘PDZ1
activation cleft’, impacting the correct binding of Act1. Substrate specificity of
the proteins was also examined using different para-Nitroanilide (pNA)-labeled substrates that differ in length and sequence (P1 – P4).
No differences in substrate specificity were observed for any of the proteins compared
to the wild type, suggesting that the mutations do not appreciably change the conformation
of the catalytic domain (at least using these substrates; Fig. 2c). The CtHtrA
_P370L
protein again displayed no proteolytic activity. CtHtrA has been previously shown to oligomerize from a resting hexamer to an activated
12–/24-mer in the presence of both peptide and protein substrates 11], 12]. To investigate whether any of the recombinant mutated protein disrupted or impaired
the activation of oligomerization, each were incubated with full-length ?-casein at
37 °C and cross-linked with glutaraldehyde. In the presence of substrate, each mutant
formed particles that are consistent with a higher order oligomer of CtHtrA (such
as a 24-mer), with no evidence of the hexameric or trimeric forms that are observed
when the oligomerization mechanism has been disrupted 12], 31]. While this method cannot detect differences in oligomeric structure or the exact
number of monomers present, oligomeric activation to some form of oligomer appears
to be unperturbed for each of the mutants (Fig. 2d) 11]. These data demonstrated that the CtHtrA
_A240V
, CtHtrA
_G475E
, and CtHtrA
_P370L
mutations resulted in an appreciable disruption to the in vitro proteolytic activity
of CtHtrA and therefore isolates with these three mutations were selected from the libraries
for further in vitro analysis of their phenotypic impact on chlamydial growth.

Fig. 2. In vitro biochemical activity of wild type C. trachomatis L2 and recombinant mutants. a. Cleavage of full-length ?-casein by wild type CtHtrA and recombinant mutants over a 60 min time course. The corresponding SDS-PAGE
gels are provided in Additional file 2: Figure S1. b. Rate of proteolysis of the ?cas1 peptide substrate with and without allosteric activation
(Act1 peptide) by recombinant proteins. Numbers above the bars represent the fold-change
in proteolytic activity following the addition of the Act1 activator. Rate of proteolysis
is measured as ?M MCA min
^?1
?g CtHtrA
^?1
. Error bars represent standard error of the mean (n?=?6). c. Measurement of the proteolytic activity of wild type CtHtrA and mutants in the presence of pNA1-4 peptide substrates. Rate of proteolysis
is measured as pNA 405 nm min
^?1
. Error bars represent standard error of the mean (n?=?6). d. Wild type CtHtrA and the mutants oligomerise to 24-mer in the presence of full-length ?-casein.
The figure shows glutaraldehyde crosslinked protein samples prepared by oligomerisation
assays following separation on 3–8 % TrisAcetate gels prior to silver staining. The
size of the 24-meric oligomer is indicated to the right of the gel and the molecular
weights are indicated on the left (High Molecular Weight Marker, Invitrogen)

A strain with a P370L substitution in cthtrA exhibits decreased or delayed EB production

Previous work by our group suggested that CtHtrA has a crucial role in the replicative phase of C. trachomatis development, as the JO146 CtHtrA inhibitor was lethal when added to the cultures at mid-replication 15]. Accordingly, growth rates were assessed for Chlamydia strains containing the cthtrA variants cthtrA_A240V
, cthtrA_P370L
, cthtrA_G475E
and wild-type C. trachomatis serovar L2 (referred to as for this study CtL2
_wt
) (Additional file 1: Table S2), by determining the time at which infectious EBs were first detected,
with additional times points during the growth cycle. The presence of infectious EBs
was detected between 20 and 48 h post-infection, with each mutant strain producing
fewer EBs than the wild-type (Fig. 3a). The cthtrA_P370L
mutant produced significantly fewer infectious EBs compared to the wild type and other
mutants, displaying a ~20–40-fold reduction (~95 % reduction) in infectious EB progeny
(for example at 32 h post infection, 2.26?×?10
⁷
EBs were detected for CtL2
_wt
while 5.45?×?10
⁵
EBs were detected for cthtrA_P370L
, corresponding to a 41.4-fold reduction in infectious EB production at this time
point).

Fig. 3. Growth and heat shock response phenotypes for C. trachomatis mutants containing cthtrA SNVs. a. One-step growth curve for CtL2
_wt
and mutants over a 48 h time course. McCoy B cells were infected at an MOI of 0.3
and collected over the time course for determination of inclusion-forming units (IFUs).
b. Infectious yield of CtL2
_wt
and mutants at 44 h post infection after 4 h heat shock (42 °C, 5 % CO
₂
; 20–24 h post infection). **** indicates p??0.0001. Error bars represent standard error of the mean (n?=?27)

The C. trachomatis isolate harboring a P370L substitution in cthtrA displays increased
susceptibility to heat shock

The role of bacterial HtrA during heat shock has been widely reported 32]–34], and CtHtrA in particular has been shown to be upregulated during heat stress conditions
16] and critical for heat stress survival when the chemical inhibitor against CtHtrA was used 14]. Therefore, Chlamydia strains with functionally disruptive mutations in cthtrA are likely to be more severely impacted by heat stress. HEp-2 cells were infected
with each strain and subjected to 42 °C heat shock during the replicative phase (20 h
post-infection) for 4 h, prior to restoration to 37 °C for the remainder of the development
cycle. The impact of heat shock treatment on the mutants was determined by calculating
the subsequent infectious yield from cultures harvested at 44 h post infection. The
cthtrA_A240V
mutant had an infectious yield that was comparable to the wild-type, while both the
cthtrA_G475E
and cthtrA_P370L
mutants exhibited a significant reduction in infectious yield following heat shock,
relative to the wild-type (Fig. 3). Notably, the cthtrA_P370L
mutant resulted in a 32.5-fold reduction in infectious EB yield following heat shock
compared to the wild type (p??0.0001).

The cthtrA
_P370L
inclusions are smaller in size

The impact of the mutations on the chlamydial inclusion morphology was examined using
immunocytochemistry and confocal laser scanning microscopy. Cultures were examined
at 24 h post-infection (log phase) and 40 h post-infection (stationary phase; Fig. 4). At 24 h post-infection, the inclusion size of each mutant appeared smaller compared
to the wild-type, while in the cthtrA_P370L
strain there appeared to be fewer chlamydial cells within the inclusions. At 40 h
post-infection, the inclusion sizes appeared to be more comparable to the wild-type,
if slightly smaller. The inclusion sizes were measured to allow statistical comparison
of the mutants against the wild-type, confirming that the mutant inclusion sizes were
appreciably smaller compared to the wild-type at 24 h post-infection (Fig. 5). The difference was less pronounced at 40 h post-infection, where the cthtrA_G475E
inclusion sizes were not significantly different compared to the wild-type. Alternatively,
in the cthtrA_A240V
and, to a greater extent, cthtrA_P370L
strains, inclusion sizes were significantly smaller than the wild-type (p??0.0001; Fig. 5).

Fig. 4. Confocal microscopy images of CtL2
_wt
and mutants at 24 h and 44 h post infection. The Chlamydia isolates (CtL2
_wt
and mutant) are shown to the left of the panels and the time point is shown above
the panels. The second and fourth images for each isolate are enlarged representations
of single inclusions. The image colours are, green: LPS (FITC anti-chlamydial LPS);
red: host cells (Evans blue). The scale bars (bottom right) indicate 50 ?m and 25 ?m
for the enlarged images

Fig. 5. Inclusion sizes of CtL2
_wt
and mutants at 24 h and 44 h post infection. Inclusion sizes were measured from independent
coverslips. ****indicates p??0.0001. Error bars represent standard error of the mean (n?=?27)

Genomics and lateral gene transfer to isolate chlamydial SNVs for further characterization
resulted in a ctl0738
_null
but not a cthtrA
_P370L
isogenic strain

While these observed phenotypes are consistent with our previous observations for
CtHtrA function, it is likely that the additional SNVs present in these mutant genomes
also contribute to the phenotype. As a result, the genome sequences of the cthtrA_A240V
, cthtrA_G475E
, and cthtrA_P370L
mutants were determined by whole genome sequencing (WGS). When compared to the CtL2
_wt
genome sequence, the mutants consisted SNVs at nine loci that were consistent in all
three isolates (cthtrA_A240V
, cthtrA_G475E
, and cthtrA_P370L
): ctl0103, nusA, ctl0518, clpC-1, clpC-2, rpoB, pykF, and pmpC, which were confirmed to originate from the CtL2
_rif
strain used to generate the library (Additional file 1: Table S2) and were thus not expected to contribute to the observed phenotypes. Alternatively,
we observed 13 unique SNVs in the cthtrA_A240V
(including the A240V C – T transition at position 247526 in cthtrA), 19 unique SNVs in the cthtrA_P370L
isolate (including the P370L C – T transition at position 247916 in cthtrA), and eight unique SNVs in the cthtrA_G475E
isolate (including the G475E G – A transition at position 248231 in cthtrA; Additional file 1: Table S3), which could potentially contribute to the observed phenotypes.

Given that the isolate with the cthtrA_P370L
mutation displayed the most marked phenotypes and was severely impaired during CtHtrA recombinant protein in vitro protease assays, it was reasoned that this isolate
will be the most informative for determining the physiological function of CtHtrA
and was selected for further genetic characterization. Of the 19 unique SNVs in the
genome of the cthtrA_P370L
mutant isolate, 11 were non-synonymous, five were synonymous, and three were located
in intergenic regions. Notably, there was a null mutation in ctl0738, a putative DNA methyltransferase. Of the eleven non-synonymous mutations, three
resulted in a change to a similar amino acid (i.e. polar, hydrophobic etc.) and are
therefore unlikely to have a functional effect (these were SNVs found in ctl0493, metG, and ctl0220). Consequently, a total of eight SNVs in the following cthtrA_P370L
isolate were identified as potentially significant, found on the following locus:
recB, murC, incA, ydhO, ctl0738 (putative DNA methyltransferase), ctl0791 (putative membrane protein), ctl0885 (conserved hypothetical protein), and cthtrA_P370L
(Fig. 6, Additional file 1: Table S3).

Fig. 6. Circular representation of the reference L2/434/Bu genome (1.04 Mbp) showing the position
of SNVs found on protein coding genes and the annotation of those genes. Each mutant
strain is represented by a single ring layer (from inner to outer: cthtrA_P3470L
, cthtrA_G475E
, cthtra_A240V
). Blue labels correspond to synonymous SNVs and black labels indicate non-synonymous
SNVs. Figure generated using BRIG 37]

In an effort to separate the cthtrA_P370L
SNV from these remaining SNVs, a lateral gene transfer approach was utilized, by co-infecting
host cells with the rifampicin resistant cthtrA_P370L
stain and a spectinomycin resistant CtL2
_wt
strain to generate recombinant isolates 21]. Positive recombinants were selected by plaque purification in the presence of both
antibiotics. PCR and Sanger sequencing was conducted on 56 plaque-purified double
resistant recombinant strains for the cthtrA_P370L
SNV and ctl738_null
(DNA methyltransferase), in addition to the other SNVs identified as potentially significant.
The cthtrA_P370L
SNV was not detected in any of the 56 plaque purified recombinant isolates, while
the ctl0738_null
SNV was variously distributed among the recombinants. One recombinant isolate (ctl0738_null
) contained the ctl0738 SNV and none of the other seven SNVs, allowing the potential for characterization
of the DNA methyltransferase null mutation in the absence of the cthtrA_P370L
and remaining SNVs.

Characterization of the ctl0738
_null
mutation suggests that the cthtrA
_P370L
SNV is the major contributor to the reduced infectious progeny phenotype

The lateral gene transfer experiments did not enable the generation of an isogenic
cthtrA_P370L
mutant isolate. The SNV that could be contributing to the phenotype observed in the
cthtrA_P370L
strain was the null mutation in the DNA methyltransferase, ctl0738. This mutation was able to be transferred to the genome of C. trachomatis L2 spectinomycin-resistant strain (CtL2
_spc
) in the absence of the eight other EMS mutations that were considered potentially
significant on the ctHtrA_P370L
isolate genome. Therefore, further analysis of the isogenic strain containing the
DNA methyltransferase null (ctl0738_null
) was conducted to examine the role of this mutation for the phenotypes observed in
the cthtrA_P370L
strain. Growth curve and heat shock experiments were conducted using the wild type
strains (CtL2
_wt
and CtL2
_spc
), the original cthtrA_P370L
mutant strain, and the ctl0738_null
strain. The cthtrA_P370L
strain has the same severely impacted reduced infectious progeny yield (8–25-fold
reduction in infectious progeny; Fig. 7a) and the ctl0738_null
strain showed infectious progeny production similar to the wild type. Notably, by
including additional time-points at the beginning of the replicative phase (16–22 h
post infection), it was observed that both the cthtrA_P370L
and ctl0738_null
strains displayed a delayed start to the log growth phase implying slowed RB to EB
conversion. Somewhat unexpectedly, the heat stress phenotype was similar in the cthtrA_P370L
and ctl0738_null
strains with a significant reduction in infectious progeny (CtL2
_wt
: 4.5?×?10
⁷
EBs, cthtrA_P370L
: 1.2?×?10
⁶
EBs; ctl0738_null
: 1.4?×?10
⁶
EBs; p??0.0001), corresponding with a 30–40-fold reduction in infectious progeny after
heat stress relative to CtL2
_wt
(Fig. 7b).

Fig. 7. Cell culture analysis of growth and heat shock phenotypes for the cthtrA_P370L
and ctl0738_null
mutants. a. One-step growth curve for wild-type C. trachomatis L2 and mutants over an 88 h time course. McCoy B cells were infected at an MOI of
0.3 and collected over the time course for determination of inclusion-forming units.
b. Viable infectious yield of wild-type C. trachomatis L2 and mutants at 44 h post infection after 4 h heat shock (42 °C, 5 % CO
₂
; 20–24 h post infection). ****indicates p??0.0001. Error bars represent standard error of the mean (n–27)