Structural characterization of muropeptides from Chlamydia trachomatis peptidoglycan by mass spectrometry resolves “chlamydial anomaly”

Mathanraj Packiama,1, Brian Weinrickb,c, William R. Jacobs Jr.b,c,2, and Anthony T. Maurellia,2

aDepartment of Microbiology and Immunology, F. Edward Hébert School of Medicine, Uniformed Services University of the Health Sciences, Bethesda, MD 20814; bHoward Hughes Medical Institute, Albert Einstein College of Medicine of Yeshiva University, Bronx, NY 10461; and cDepartment of Microbiology and Immunology, Albert Einstein College of Medicine of Yeshiva University, Bronx, NY 10461

Contributed by William R. Jacobs Jr., July 20, 2015 (sent for review June 18, 2015; reviewed by Harlan D. Caldwell and Joseph P. Dillard) The “chlamydial anomaly,” first coined by James Moulder, describes In a similar fashion, variations in the stem peptide sequence would the inability of researchers to detect or purify peptidoglycan (PG) from alter NOD1 receptor recognition and the host inflammatory re- pathogenic Chlamydiae despite genetic and biochemical evidence and sponse (15). Because two serious inflammatory consequences of antibiotic susceptibility data that suggest its existence. We recently Chlamydia trachomatis infection are pelvic inflammatory disease detected PG in Chlamydia trachomatis by a new metabolic cell wall and trachoma, the exact molecular identity of chlamydial PG has labeling method, however efforts to purify PG from pathogenic Chla- high biological relevance and is essential to identifying the role PG mydiae have remained unsuccessful. Pathogenic chlamydial spe- plays in the induction of inflammation in these disease states. cies are known to activate nucleotide-binding oligomerization Pilhofer et al. (16) recently succeeded in purifying PG sacculi domain-containing protein 2 (NOD2) innate immune receptors from an environmental Protochlamydia species but were unable to by as yet uncharacterized ligands, which are presumed to be PG detect or purify PG from Simkania, a distinct environmental spe- fragments (muramyl di- and tripeptides). We used the NOD2-depen- cies, by standard PG sacculi purification protocols. Because path- dent activation of NF-κBbyC. trachomatis-infected cell lysates as a ogenic Chlamydiae are more similar to Simkania (17) than to biomarker for the presence of PG fragments within specific lysate Protochlamydia in the content of PG biosynthetic pathway genes, fractions. We designed a new method of muropeptide isolation con- Pilhofer et al. speculated that pathogenic Chlamydiae could be sisting of a double filtration step coupled with reverse-phase HPLC devoid of PG sacculi similar to Simkania (16). We recently dem- fractionation of Chlamydia-infected HeLa cell lysates. Fractions onstrated the presence of PG in C. trachomatis by a new metabolic that displayed NOD2 activity were analyzed by electrospray ion- cell wall labeling method (18) and observed that chlamydial PG ization mass spectrometry, confirming the presence of muramyl does not appear to form a sacculus but is present in a ring-like di- and tripeptides in Chlamydia-infected cell lysate fractions. structure at the apparent cell division plane. Given the absence of a Moreover, the mass spectrometry data of large muropeptide frag- PG sacculus, we hypothesized that purification of PG from path- ments provided evidence that transpeptidation and transglycosy- ogenic Chlamydiae required a technique other than the standard lation reactions occur in pathogenic Chlamydiae. These results PG sacculi isolation method to succeed. In addition, previous un- reveal the composition of chlamydial PG and disprove the “glycan- successful searches for PG in pathogenic Chlamydiae focused on less peptidoglycan” hypothesis. Significance chlamydia | peptidoglycan | mass spectrometry | NOD2 receptor | muropeptide The existence of peptidoglycan (PG) in pathogenic Chlamydiae is supported by genetic data and antibiotic susceptibility, but the he existence of peptidoglycan (PG) in pathogenic Chlamy- failure to isolate PG from pathogenic Chlamydiae has led to the Tdiae has long been debated. Although genetic analysis and an- “chlamydial anomaly.” Moreover, the lack of a transglycosylase tibiotic susceptibility indicate the presence of PG in Chlamydia (1–3), domain in some Chlamydia penicillin-binding proteins suggests all attempts to detect and purify PG have been unsuccessful (4–7), that Chlamydiae may possess a “glycanless PG.” We successfully resulting in the paradox known as the “chlamydial anomaly” (8). enriched Chlamydia muropeptides from Chlamydia-infected cell Ghuysen and Goffin (9) hypothesized that Chlamydia might syn- lysates using nucleotide-binding oligomerization domain-con- thesize a “glycanless PG” based on the observation that the Chla- taining protein 2 (NOD2)-dependent NF-κB activation as a bio- mydia genome encodes two high molecular mass penicillin-binding marker for the presence of PG fragments in specific fractions. proteins (PBPs) that are devoid of transglycosylase activity, which is Mass spectrometry analysis indicated the presence of chlamydial essential for elongation of the glycan chain of classical PG. Besides muropeptides and classified chlamydial PG as type A1γ in the doubts about the presence of a glycan backbone in chlamydial PG, Schleifer and Kandler classification. This study disproves the gly- there are ambiguities about the sequence of the stem peptide. Patin canless PG hypothesis and is, to our knowledge, the first structural et al. (10) suggested glycine, L-serine, or L-alanine as the possible confirmation of chlamydial PG in pathogenic species. first amino acid residue of the chlamydial PG stem. Resolution of the chlamydial anomaly by purification and structural characteriza- Author contributions: M.P. and A.T.M. designed research; M.P. and B.W. performed re- tion of chlamydial PG will have a profound impact not only on search; W.R.J. contributed new reagents/analytic tools; M.P., B.W., and A.T.M. analyzed data; and M.P., B.W., W.R.J., and A.T.M. wrote the paper. chlamydial biology but also on shaping our understanding of the host Reviewers: H.D.C., NIH National Institute of Allergy and Infectious Diseases; and J.P.D., innate immune responses. Many pathogenic Chlamydia species ac- University of Wisconsin-Madison School of Medicine and Public Health. tivate the nucleotide-binding oligomerization domain-containing The authors declare no conflict of interest. protein (NOD) family of innate receptors during infection (11–14). 1Present address: Biological Research Laboratory, Eminent Services Corporation, Frederick, The inflammatory potential of chlamydial PG would vary depending MD 21703. on the composition of both the sugar backbone as well as the stem 2To whom correspondence may be addressed. Email: [email protected] or peptide sequence. NOD2 receptors recognize the sugar backbone of [email protected]. “ ” PG, and therefore, a glycanless chlamydial PG may not activate This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. NOD2 receptors and fail to induce an inflammatory response (15). 1073/pnas.1514026112/-/DCSupplemental.

11660–11665 | PNAS | September 15, 2015 | vol. 112 | no. 37 www.pnas.org/cgi/doi/10.1073/pnas.1514026112 Downloaded by guest on September 23, 2021 A Larger PG fragments present in the activating fractions provided 0.4 evidence for cross-links formed between PG stem peptide chains and for glycan chains composed of N-acetylmuramic acid (MurNAc) and N-acetylglucosamine (GlcNAc) saccharides. Our molecular 0.3 detection of chlamydial PG fragments resolves the much-debated chlamydial anomaly, disproves the glycanless PG hypothesis re- 0.2 garding pathogenic Chlamydiae, and identifies chlamydial MDP as a ligand for the host innate immune receptor NOD2. 0.1 Results

NF-κB activation (OD 650) 0.0 Chlamydia-Infected HeLa Cell Lysates Specifically Induce NOD2-Dependent NF-κBActivity.Multiple studies have shown that intracellular patho- Mock genic Chlamydiae induce NF-κB signaling by activating NOD2 re- Inf 2 hrsInf 8 hrs Inf 18 hrsInf 28 hrsInf 38 hrs ceptors (11–14). To isolate chlamydial ligand(s) that activate NOD2 Time post-infection (i.e., MDP), HeLa cells were infected with C. trachomatis L2 serovar for 2 h and cell lysates were harvested at various time points *** throughout the developmental cycle (2, 8, 18, 28, and 38 h) and B filtered first through a 0.2-μm filter and then through a 3-kDa 0.8 *** centrifugal filter. When added (extracellularly) to HEK-Blue NOD2 secreted alkaline phosphatase (SEAP) reporter cell lines, the 3-kDa 0.6 NOD2 centrifugal filter flow-through from Chlamydia-infected cells at 18 h *** ** Null2 κ *** postinfection (PI) induced maximum NF- B activity compared with 0.4 earlier or later time points (Fig. 1A). Because peak induction of *** ns ns ns NOD2 activity was found at 18 h PI, all experiments were per- 0.2 formed with cell lysates harvested at 18 h PI. NF-κB induction from infected cell lysates was significantly higher than mock-infected controls (P < 0.005 unpaired t test), and the activity was NOD2-

NF-κB activation (OD 650) 0.0 p mA dependent in HEK-Blue NOD2 cells as NF-κB induction by in- Mock k + fected cell lysates was minimal in HEK-Blue Null2 cells that are Infected coMM Inf + Amp Inf + DCS M Mock + DCS devoid of the NOD2 receptor (Fig. 1B). Heat and RNase A treatments of cell lysates had minimal effect on the infected Fig. 1. Kinetics of Chlamydia NOD2 ligand production during the develop- sample’s ability to induce NF-κB activity (Fig. S1), indicating that mental cycle and enhancement of activation by antibiotics that inhibit PG the NOD2 ligand is heat stable and excluding RNA as a possible synthesis. (A) Chlamydia-infected and mock-infected HeLa cell lysates were NOD2 chlamydial ligand. Treatment with inhibitors of PG bio- harvested at various time points during the developmental cycle, and their synthesis such as D-cycloserine (DCS) and ampicillin significantly ability to induce NOD2-dependent NF-κB activity was assessed in a HEK-Blue enhanced NOD2-dependent activity of infected cell lysates over in- NOD2 reporter cell line carrying a NF-κB–inducible alkaline phosphatase reporter fected but untreated cell lysates (P < 0.005 and P < 0.01, respectively, gene. Peak NOD2-dependent activity was observed for Chlamydia-infected cell unpaired t test), strongly suggesting that the nature of the NOD2 lysates harvested at 18 h PI. (B) The ability of Chlamydia-infected HeLa cell lysates ligand in infected lysates could be a muropeptide(s) (Fig. 1B). harvested at 18 h PI to induce NF-κB activity was tested in HEK-Blue cell lines expressing (gray bars) or devoid of (white bars) NOD2 innate immune receptors Chlamydial MDP and MTP are Present in NOD2-Activating Infected andcarryingaNF-κB–inducible alkaline phosphatase reporter gene. Alkaline phosphatase activity was measured at an OD of 650 nm following addition of Cell Lysates. Mass and Enhanced Product Ion (EPI) scans were performed with mock and infected cell lysates to detect MDP infected and mock-infected cell lysates. Chlamydia-infected cell lysates induced + higher NF-κB activity over mock-infected cell lysates in a NOD2-dependent and MTP (m/z of [M+H] = 494.2 and 666.2, respectively). manner. Treatment with PG synthesis inhibitors ampicillin (Amp, 1 μg/mL) and Commercially available MDP and MTP served as synthetic stan- DCS (30 μg/mL) greatly enhanced the ability of Chlamydia-infected cell lysates to dards. The cell lysate of a midlog phase culture of Shigella flexneri induce NF-κB activity in a NOD2-dependent manner over the corresponding was used as a source of bacterial muropeptides, and an ion of 648.2 + untreated infected cell lysates. **P < 0.01, ***P < 0.005 unpaired t test; ns, m/z, consistent with [M+H] anhydro MTP (anhyrdo N-acetyl ± = nonsignificant. OD650 values are mean SEM; n 3. muramic acid-alanyl-glutamyl-diaminopimelic acid or anhydro MurNAc-Ala-Glu-DAP), was detected. The identity of this ion was confirmed by MS/MS analysis (Fig. S2). Detection of Shigella purification of the developmental forms of Chlamydia—that is, anhydro MTP validated our muropeptide isolation method by cell elementary bodies (EBs) or reticulate bodies (RBs) (4, 16, 19)—as lysate fractionation and served as a positive control in our opposed to sampling intracellular Chlamydia directly from infected search for chlamydial PG fragments. MS/MS spectra of the two MICROBIOLOGY cells. Therefore, we saw the need to develop a new PG isolation synthetic ligands (Fig. S3 A and B) along with Shigella anhydro technique for pathogenic Chlamydiae. MTP (Fig. S2) were used to interpret chlamydial MDP and MTP PG fragments induce NF-κB signaling in infected cells through species (Table 1). We regularly observed an ion of 494.4 m/z, + the cognate cytoplasmic NOD receptors (15, 20, 21), and many consistent with the [M+H] ion of MDP, present exclusively in pathogenic Chlamydia species activate NOD receptors during in- Chlamydia-infected lysates. MS/MS spectral analysis (Fig. 2A)of fection (11–14). The identities of Chlamydia NOD ligands are un- this ion confirmed it to be MDP (MurNAc-Ala-Glu). We also knownbutarepresumedtobePGconstituents—that is, muramyl confirmed the presence of MTP (MurNAc-Ala-Glu-DAP) in in- dipeptide (MDP) and muramyl tripeptide (MTP). Our strategy to fected lysates by MS/MS analysis of an ion of 666.2 m/z (Fig. 2B). purify PG fragments from pathogenic Chlamydiae used a reporter The quantity of MTP was very low in infected samples, which cell line to detect NOD2-dependent activation of NF-κB. By testing made consistent detection of this species difficult by MS/MS. individual chromatographic fractions of C. trachomatis-infected cell Moreover, the sub 3 kDa lysates were very complex, resulting in lysates for the ability to activate NOD2, we identified those that substantial background that obscured MS/MS detection of low- contain immunostimulatory PG fragments. Electrospray ionization abundance MTP species. (ESI) mass spectrometry was used to determine the molecular identity of NOD2 ligands in these fractions. Our results showed HPLC Fractionation of Cell Lysates to Enrich for NOD2-Stimulating that fractions that activated NOD2 were enriched in the canonical Activity. To further purify NOD2 ligands from the highly complex NOD2 ligand MDP, as well as MTP and larger PG constituents. sub 3 kDa lysate, we fractionated the lysates by HPLC. The sub

Packiam et al. PNAS | September 15, 2015 | vol. 112 | no. 37 | 11661 Downloaded by guest on September 23, 2021 Table 1. Ions detected in MS/MS analysis of NOD2-activating samples + + Sample M [M+H] [M+H] -H2O Fragment ions

Synthetic MDP (isoglutamine) 492.2 493.2 475.3 347.2, 329.3, 273.2 Synthetic MTP 665.3 666.3 648.2 463.2, 347.2, 320.1, 191.2, 173.3 Shigella lysate anhydro MTP 647.2 648.2 Not detected 463.2, 320.1, 191.2, 173.3 Chlamydia lysate MDP 493.2 494.2 476.2 347.2, 329.2 Chlamydia lysate MTP 665.3 666.3 648.2 463.2, 320.1, 191.2

3 kDa lysates were fractionated on a reverse-phase C18 col- (mass-to-charge ratios from greater than 600 up to 1,800) that umn, and 1.2-mL fractions were collected each minute (designated could be part of polymerized PG. The criteria to identify these F1 to F20) while UV absorbance was monitored at 215 nm. The larger fragments included presence solely in infected lysates UV chromatograms of the eluted fractions from the infected cell and sensitivity to ampicillin treatment. We identified two lysates looked similar to those of the mock-infected cell lysates; the prominent doubly charged ions of 653.0 and 763.9 m/z (eluting chromatogram of the infected lysate is shown in Fig. S4. The ability at 8.8 min) (Fig. S5A), the abundance of which was sensitive to of each fraction (F1 to F20) to induce NOD2-dependent NF-κB ampicillin, a known inhibitor of transpeptidation and carboxy- activity was tested on HEK-Blue NOD2 SEAP reporter cells. peptidation reactions (Fig. S5B). Because these fragments could be The fraction from infected lysates collected in the second minute generated by mechanical shearing, in-source fragmentation, or di- (F2) specifically induced NOD2 activity compared with the corre- gestion by bacterial or eukaryotic enzymes, their nature is difficult to sponding mock-infected F2 fraction (Fig. 3). Shigella-infected HeLa predict. To verify the identity of the fragments as PG constituents cell lysates used as a positive control showed an F2 fraction that and solve their structure, we identified peaks in the MS/MS spectra retained most of the NOD2 activity, similar to the Chlamydia- separated by mass differences that are characteristic of fragmen- infected fraction. MDP and MTP synthetic muropeptides used as tation of the stem peptide of PG (129 amu, loss of glutamate; positive controls also eluted from the C18 column in the F2 frac- 172 amu, loss of DAP; and 71 amu, loss of alanine). The MS/MS tion (Fig. S4 A and B). These results suggest that NOD2-activating analysis of the doubly charged ion of 653.0 m/z was interpreted to ligand(s) found in the Chlamydia F2 fraction have C18 column yield a structure that provides evidence for transpeptidation and interacting properties (i.e., they are highly hydrophilic and have carboxypeptidation reactions in chlamydial PG, as the structure limited affinity for the hydrophobic C18 column) similar to those of includes two DAP residues linked by an alanine residue (Fig. 4 A the activating fraction from Shigella-infected cells and synthetic and B). Interestingly, the first residue in the peptide of this fragment MDP and MTP. was glycine. The chlamydial MurC enzyme, which ligates the first amino acid of the stem peptide to MurNAc, has been shown to lack Structural Evidence for Transpeptidation and Transglycosylation selectivity for alanine over glycine or serine (23). Moreover, chla- Reactions in Formation of Chlamydial PG. The NOD2 activa- mydial MurE and MurF (which add subsequent amino acids to the tion-based screening assay yielded a fraction that coeluted with growing stem peptide) are able to use glycine-containing peptides as the NOD2 ligands MDP and MTP, which were identified in substrate (10, 24). These observations suggest that C. trachomatis Chlamydia-infected lysates. However, these muropeptides could PG may contain a mixture of alanine and glycine as the first amino potentially originate from early PG precursors such as Park’s acid in the stem peptide. MS/MS analysis of the second most nucleotide, lipid I, or lipid II. To determine if these chlamydial prominent ion, 763.9 m/z, which shared many of the fragments seen muramyl peptides could be constituents of polymerized PG, we from the 653.0 m/z ion, revealed the presence of two MurNAc used the information-dependent acquisition (IDA) method of sugars joined by a GlcNAc sugar (Fig. S6 A and B). The mass spectrometry (22) to identify larger muropeptide fragments chlamydial PG is classified as A1γ type PG (25) based on the

A A B C 494.5 463.2 4741 2.6e4 4600 OH 2.5e4 OH Parent Formula: C19H31N3O12 4400 2.4e4 A [M+H]+ = 494.2 4200 O + O 2.3e4 B [M+H] -H2O = 476.2 4000 HO 2.2e4 HO 3800 NH 2.1e4 + NH C [+ 2H] O OH 3600 2.0e4 O OH 463.2 1.9e4 3400 O O H3C C 1.8e4 O O 3200 H3C C H3C NH 1.7e4 D 3000 Parent Formula: C26H43N5O15 + 1.6e4 A [M+H] = 666.3 329.3 H3C NH 2800 O C B [M+H]+ - H O = 648.2 1.5e4 363.4 2 2600 NH 1.4e4 O C 347.2 C -H O 2 2400 [+ 2H]+ [+ 2H]+ 1.3e4 329.2 D E D O E NH 2200 320.1 C 191.2 Intensity, cps 1.2e4 Intensity, cps 191.2 501.4 H 2000 OH N 1.1e4 O NH2 C C 1.0e4 1800 365.2 216.2 C O C C 9000.0 OH O 1600 O 152.3 349.1 O OH 347.4 C HO 8000.0 1400 D B 7000.0 198.1 244.3 OH 1200 189.27 .091 210.2 319.9 336.1 428.2 514.2 6000.0 476.4 B 1000 5000.0 136.0 166.1 279.0 324.0 648.2 800 A 4000.0 600 160.3 233.2 257.2 210.29 .603 373.1 397.1 409.2 483.2 496.0 556.9 666.0 3000.0 270.2 352.75 .473 313.3 458.5 400 2000.0 235.2 198.7 217.1 251.3 379.3 180.1 4.272 332.00 .403 310.2 338.4 419.4 432.6 448.4 1000.0 252.74 .162 343.2 356.0 396.4 414.4 464.4 485.8 200 213.8 229.2 371.4 0.0 0 180 200 220 240 260 280 300 320 340 360 380 400 420 440 460 480 500 50 100 150 200 250 300 350 400 450 500 550 600 650 700 m/z, Da m/z, Da

Fig. 2. Molecular detection of chlamydial MDP and MTP in Chlamydia-infected cell lysates. Mass and product ion scans were performed on 3 kDa flow- through of Chlamydia-infected HeLa cell lysates specifically looking for ions with an m/z of 494.4 and 666.2. (A) Tandem mass spectrum of the molecular ion + [M+H] = 494.4 m/z. The spectral interpretation of the daughter ions generated is provided in the Inset; fragmentation is indicated by dashed lines. + (B) Tandem mass spectrum of the molecular ion [M+H] = 666.2 m/z. The spectral interpretation of the daughter ions generated is provided in the Inset; fragmentation is indicated by dashed lines.

11662 | www.pnas.org/cgi/doi/10.1073/pnas.1514026112 Packiam et al. Downloaded by guest on September 23, 2021 0.8 failure of previous attempts to isolate chlamydial PG could possibly Mock be due to the kinetics of PG production during the developmental Infected cycle. Most previous studies focused on the isolation of PG sacculi 0.6 from purified EBs or RBs at early or late stages of the develop- mental cycle. The current study indicated that PG production is 0.4 minimal or absent at the start as well as at the conclusion of the *** developmental cycle and peak production occurs 18 h PI. These 0.2 findings are consistent with the results of our labeling studies (18), in which PG ring formation was observed at 18 h PI, localized only κB activation (OD 650) to the apparent cell division plane, and did not form a sacculus

NF- 0.0 F1 F2 F3 F4 F5 MDP TriDAP around the bacterium. Our success in detecting chlamydial muropeptides in this study was largely due to our use of infected cell Fig. 3. NOD2 stimulating activity of F2 fraction of Chlamydia-infected cells col- lysates (instead of purified bacteria), selection of the time point with lected from a reverse-phase C18 HPLC column. HPLC fractionation was performed, maximal muropeptide abundance (18 h PI), fractionating the lysates and 1.2-mL fractions were collected every minute. Mock-infected cell lysates were from potential PG degrading host and bacterial enzymes (3 kDa also fractionated, and NOD2 activities induced by them were considered as κ cutoff), and the greatly increased sensitivity of our muropeptide background. NOD2-induced NF- B activity (measured at OD650) for the various detection approach compared with conventional methods. fractions collected from Chlamydia-infected (gray bar) or mock-infected lysates We used NOD2 receptor activation by Chlamydia-infected cell (open bar) was tested in HEK-Blue NOD2 cell lines as in Fig. 1. MDP (positive) and lysates as a reporter for the presence of muropeptide ligands Tri-DAP (L-Ala-γ-D-Glu-mDAP) (negative) control ligands were used at 1 μg/mL. in lysate fractions. The typical ligands for NOD2 activation are ***P < 0.005 unpaired t test. OD650 values are mean ± SEM; n = 3. muropeptides, but instances of viral RNA-induced NOD2 activa- tion have been reported (26). RNase A and heat treatment of cell molecular identity of the sugars, stem peptides, and the cross- lysates had little effect on NOD2 activation and thus ruled out RNA and heat-labile proteins as possible NOD2 ligands. We were able to links detected in our study. The detection of these muropep- confirm the presence of MDP and MTP in the infected cell lysate tides exclusively in the infected cell lysate fraction conclusively fractions that retained NOD2 activity. The identification of MTP shows that pathogenic Chlamydiae possess a polymerized PG allowed us to determine that DAP is the third residue of the stem containing both cross-linked stem peptides and glycan chains peptide. Although initial annotation of the Chlamydia genome composed of MurNAc/GlcNAc residues. showed only an incomplete pathway for the synthesis of DAP (3, 6), subsequent studies from our laboratory showed that Chlamydia can Discussion synthesize DAP by use of a previously unidentified aminotrans- Although we recently demonstrated the presence of PG rings in ferase that completes the pathway (27). Thus, our structural data pathogenic Chlamydiae by a metabolic cell wall labeling method, confirm that Chlamydia uses DAP as a component of the PG stem the molecular identity of chlamydial PG remained unknown. The peptide. Although we did not include a PG digestion step in our

A B OH O HO A O O O 653.0 5.1e5 5.0e5 HO NH NH 4.8e5 O 169.1 L 4.6e5 O O O H3C C 4.4e5 NH 241.1 K 4.2e5

4.0e5 O C

3.8e5 NH 298.1 J Parent Formula: C53H83N11O28 2+ 3.6e5 O [M+2H] :1321.5 H2O =- 1303.5 + 2 = C 1305.5 / 2 = 652.8 A 3.4e5 Ala DAP Glu OH O 3.2e5 C HO 3.0e5 E C NH 427.2 I O 2.8e5 860.5

O 2.6e5 MICROBIOLOGY Glc Lac Gly Glu DAP 803.4 O 2.4e5 C H2N C Intensity, cps NAc HO OH 2.2e5 K J O

HN C O 2.0e5 +2H 617.3 G 222.14 .892 G 1.8e5 C HO O -H2O H3C N C O 617.4 599.3 H H 1.6e5 L 488.4 H H 599.4 HN C N O C N C +2H 1.4e5 H 644.0 +2H +2H CH2 1.2e5 C 688.3 F O 1074.4 B 169.0 387.2 745.2 +2H 1.0e5 989.4 C H3C 728.4 861.5 989.6 1171.4 860.4 E -HO = 971.4 D 8.0e4 230.9 I D 2 F 971.5 6.0e4 213.1 316.1 332.55 .373 541.1 B 427.3 688.4 745.98 .478 4.0e4 195.2 269.9 502.2 577.4 606.5 763.4 1074.4 141.0 312.72 .904 489.4 678.2 905.8 1056.7 2.0e4 310.0 563.56 .006 714.4 1100.9 1176.8

0.0 100 150 200 250 300 350 400 450 500 550 600 650 700 750 800 850 900 950 1000 1050 1100 1150 1200 m/z, Da

Fig. 4. Molecular detection of a muropeptide fragment that serves as evidence for transpeptidation in chlamydial PG. A high-throughput IDA method of mass spectrometry was performed on a NOD2-activating (F2) fraction, and an ion with an m/z of 653.0 was selected for tandem MS/MS analysis. (A) Structural interpretation of the major ions of the MS/MS spectrum is given in B. Fragmentation is indicated by dashed lines.

Packiam et al. PNAS | September 15, 2015 | vol. 112 | no. 37 | 11663 Downloaded by guest on September 23, 2021 isolation methodology, we detected free muropeptides (i.e., MDP and DCS treatments enhance the inflammatory potential of and MTP) in NOD2-activating cell lysates. We suggest that host Chlamydia lysates, it would be interesting to determine the cor- and bacterial PG degrading enzymes present during the process of relation of inflammatory markers in Chlamydia-infected patients lysate preparation would be capable of releasing free muropeptides. who are treated with cell wall synthesis-inhibiting antibiotics. The In our previous study (18), we showed evidence that suggested protein synthesis inhibitors azithromycin and doxycycline are the that both transglycosylation and transpeptidation occur in patho- preferred drugs to treat Chlamydia genital infections. By con- genic chlamydial PG. The labeled annular PG structures are sen- trast, patients infected with Neisseria gonorrheae are generally sitive to lysozyme treatment and removal of the peptide-specific treated with third-generation cephalosporins, which are cell wall- label by lysozyme, which targets the 1,4-beta-linkages between acting antibiotics. An estimated 50–70% of individuals who are MurNAc and GlcNAc, suggesting that the labeled peptide is, in infected with N. gonorrheae are coinfected with Chlamydia fact, incorporated into PG. We also demonstrated differential (34–36). Thus, antigonococcal treatment with PG synthesis- labeling using two distinct probes [ethynyl-D-alanyl-D-alanine targeting drugs may exacerbate Chlamydia PG-induced inflam- (EDA-DA) and D-alanyl-ethynyl-D-alanine (DA-EDA)] in mation in coinfected patients. the presence or absence of ampicillin, an inhibitor of trans- We cannot exclude the possibility of PG modification, especially peptidation and carboxypeptidation. These observations showed in light of the DCS-induced enhancement of NOD2 activity. It is that the di-peptide probes are incorporated into PG and are unclear whether pathogenic Chlamydiae have modified PG similar suitable substrates for transpeptidation/carboxypeptidation re- to that of environmental Chlamydiae, which have additional actions. Here, we used mass spectrometry to confirm that 129 and 203 mass units on virtually all of the PG monomers transglycosylation and transpeptidation reactions contribute to (16). However, we could find no evidence of these modified PG synthesis of chlamydial PG. Larger muropeptide fragments that fragments in pathogenic Chlamydia-infected cell lysates. Chlamydia contain two MurNAc sugars and a GlcNAc sugar, along with a muropeptide isolation from infected HeLa cell lysates suggests that DAP residue linked by alanine with another DAP residue, serve NOD2 ligands can translocate from the inclusion membrane and as evidence that the muropeptides we detected polymerize to activate the cytoplasmic NOD receptor in infected cells. Future form a glycan chain composed of MurNAc/GlcNAc residues with studies will characterize the mechanism of chlamydial PG fragment cross-linking between peptide chains. translocation and activation of NOD receptors as well as the nature Although MS/MS data do not reveal the stereoisomeric state of inflammation induced by chlamydial PG and its importance in of amino acids in the stem peptide, we were able to deduce Chlamydia pathogenesis. the chlamydial PG structure to be GlcNAc-MurNAc with a In summary, our study is the first, to our knowledge, to detect L-Ala/Gly-D-Glu-meso-DAP-D-Ala stem peptide. This conclusion pathogenic chlamydial muropeptides by mass spectrometry. Chla- is based on the following observations. MDP detection by NOD2 mydial PG fragments specifically activate NOD2 receptors and receptor is highly stereospecific to L-Ala and D-Glu isomers in the induce inflammation that is enhanced by treatment with cell wall- dipeptide and excludes any reaction to the D-D or L-L analogs acting antibiotics. The chlamydial PG was determined to be of type (28). Patin et al. showed that C. trachomatis has MurE and MurF A1γ chemotype and the muropeptides were unmodified. The ligase activity and uses UDP-MurNAc-L-Ala/Gly-D-Glu (24) detection of chlamydial PG disproved the glycanless PG hy- and UDP-MurNAc-L-Ala/Gly-D-Glu-meso-DAP (10) as substrates, pothesis and resolved the long-standing chlamydial anomaly. respectively. Variability in the first position amino acid of the PG stem peptide has been reported in other bacteria (29–31), and our Materials and Methods results suggest the coexistence of alanine- and glycine-containing Preparation of Cell Lysates. HeLa cells were mock-infected or infected with forms. Our previous studies (18) support the D-stereoisomer C. trachomatis serovar L2 (37) at a multiplicity of infection of 1 for 2 h at 37 °C chemistry for the fourth and fifth alanines of the stem peptide. in 5% (vol/vol) CO2 with rocking. Infection medium was replaced with DMEM Hence, the data in our present study plus the above cited re- and heat-inactivated FBS [10% (vol/vol)], and cells were incubated up to 38 h. ports lead us to conclude that chlamydial PG is of a conven- Cells were harvested with glass beads and resuspended in DMEM. The har- γ vested HeLa cells were sonicated using an Ultrasonic processor (GE 100) at 40 A tional Gram-negative PG structure, the same PG type (A1 ) with 1-min pulses repeated five times. The cell lysates were centrifuged at found in Escherichia coli. 4,000 × g for 5 min at 37 °C to remove cell debris. The supernatant was pro- A limitation of our muropeptide isolation method coupled cessed further for PG isolation by a double filtration method followed by with the NOD2 screening method is that it selects for muro- fractionation of the filtrates by HPLC. peptide fragments that are NOD2 agonists. Because the size cutoff we used for muropeptide isolation was 3 kDa, PG cross- Muropeptide Isolation Method–Double Filtration Step. Cell lysates were filter linked fragments greater than 3 kDa could not be studied. sterilized through 0.2-μm filters (Nalgene 5240020). Filter-sterilized cell Other limitations include that the search for the presence of lysates were further passed through 3-kDa centrifugal filters (Amicon novel modified PGs that are not NOD2 ligands could not be UFC900324) at 4,000 × g for 1 h at 37 °C to obtain flow-through fractions performed easily and that the IDA analysis only selected ions of that would contain PG fragments (the expected size of muropeptides that high abundance. If novel modified PG exists at a low level, then activate NOD2 receptors is <2 kDa). Flow-through fractions from the 3-kDa it becomes very difficult to identify such PG fragments without centrifugal filters were heat inactivated at 95 °C for 6 min and tested for a selective screening method. NOD2 activity using a HEK NOD reporter cell line (38). The flow-through One particularly interesting finding was that ampicillin, which was either analyzed by ESI mass spectrometry for molecular identification inhibits PBPs and directly blocks PG cross-linking, and DCS, which of muropeptides or further fractionated through reverse-phase HPLC to enrich the NOD2 ligands. inhibits alanine racemase and D-alanine:D-alanine ligase and in- directly blocks PG cross-linking, enhanced Chlamydia-induced NOD2 activity. In addition to the accumulation of PG pre- HPLC Purification. Flow-through fractions were concentrated 10-fold and cursors, we also considered muropeptide modification as a pos- fractionated by HPLC. For HPLC analysis, we ran an isocratic method using 5% sible explanation for this increased activity. DCS treatment of acetonitrile in 95% water containing 0.12% trifluoroacetic acid as eluent for 20 min on a Vydac ODS/C18 column (4.6 mm × 150 mm; 5 μm, 300 Å particle Mycobacterium tuberculosis enhances glycolation of muropep- beads W.R. Grace 218TP5415) at 30 °C using a flow rate of 1.2 mL/min. UV tides mediated by NamH (32). Glycolated muropeptides from detection was carried out at 215 nm. M. tuberculosis are highly potent in activating NOD2 receptors compared with their unmodified muropeptides (33). However, NF-κB Reporter Assay. HEK cells expressing the NOD2 receptor and carrying the the Chlamydia genome does not encode a NamH homolog, and NF-κB SEAP reporter gene (Invivogen) were used according to the manufac- we found no evidence of muropeptide glycolyation in DCS-treated turer’s instructions to assess the NF-κB stimulatory property of the various cell Chlamydia-infected cell lysates. The molecular basis of enhanced lysate fractions and eluted fractions collected from the reverse-phase C18 NOD2 activation by ampicillin and DCS-treated Chlamydia-infected HPLC column. Briefly, 20 μL of cell lysate fractions from both Chlamydia- cell lysates is currently under investigation. Because ampicillin infected as well as mock-infected HeLa cells were added to 5 × 104 HEK-Blue

11664 | www.pnas.org/cgi/doi/10.1073/pnas.1514026112 Packiam et al. Downloaded by guest on September 23, 2021 NOD2 or Null2 cells in 96-well plates and incubated for 24 h at 37 °C (total re- source. Chromatographic separation and mass spectrometry conditions are action volume, 200 μL per well). For SEAP detection, 20 μL of supernatant from described in SI Materials and Methods. lysate fraction-treated wells was added to 180 μL of QUANTI-blue substrate (Invivogen) in another 96-well microtiter plate. Supernatants from untreated or ACKNOWLEDGMENTS. We thank Dr. Sean Moran and Mike Flora from the uninfected cells were used as negative controls. The reaction was incubated at Biomedical Instrumentation Core facility for performing the LCMS and HPLC 37 °C overnight, and SEAP activity was assessed by reading OD at 650 nm. experiments, respectively. We thank George Liechti for helpful discussions of the manuscript. This work was supported by National Institute of Allergy and Infectious Diseases Grant R56 AI044033. W.R.J. acknowledges generous LCMS. Liquid chromatography and mass spectrometry (LCMS) experiments support from NIH Centers for AIDS Research Grant AI-051519 at the Albert were performed on an Agilent 1200 Series liquid chromatography system Einstein College of Medicine. This work was also supported by NIH Grant coupled to an AB Sciex Q-Trap 4000 mass spectrometer with a Turbo V ESI AI26170 and Bill and Melinda Gates Foundation Grant OPP1033104.

1. Moulder JW, Novosel DL, Officer JE (1963) Inhibition of the growth of agents of the 20. Girardin SE, et al. (2003) Nod1 detects a unique muropeptide from gram-negative psittacosis group by D-cycloserine and its specific reversal by D-alanine. J Bacteriol 85: bacterial peptidoglycan. Science 300(5625):1584–1587. 707–711. 21. Girardin SE, et al. (2003) Nod2 is a general sensor of peptidoglycan through muramyl 2. Tamura A, Manire GP (1968) Effect of penicillin on the multiplication of me- dipeptide (MDP) detection. J Biol Chem 278(11):8869–8872. ningopneumonitis organisms (Chlamydia psittaci). J Bacteriol 96(4):875–880. 22. Decaestecker TN, et al. (2004) Information-dependent acquisition-mediated LC-MS/MS 3. Stephens RS, et al. (1998) Genome sequence of an obligate intracellular pathogen of screening procedure with semiquantitative potential. Anal Chem 76(21):6365–6373. – humans: Chlamydia trachomatis. Science 282(5389):754 759. 23. Hesse L, et al. (2003) Functional and biochemical analysis of Chlamydia trachomatis 4. Fox A, et al. (1990) Muramic acid is not detectable in Chlamydia psittaci or Chla- MurC, an enzyme displaying UDP-N-acetylmuramate:amino acid ligase activity. mydia trachomatis by gas chromatography-mass spectrometry. Infect Immun 58(3): J Bacteriol 185(22):6507–6512. 835–837. 24. Patin D, Bostock J, Blanot D, Mengin-Lecreulx D, Chopra I (2009) Functional and bio- 5. Hatch TP (1996) Disulfide cross-linked envelope proteins: The functional equivalent of chemical analysis of the Chlamydia trachomatis ligase MurE. JBacteriol191(24):7430–7435. peptidoglycan in chlamydiae? J Bacteriol 178(1):1–5. 25. Schleifer KH, Kandler O (1972) Peptidoglycan types of bacterial cell walls and their 6. Chopra I, Storey C, Falla TJ, Pearce JH (1998) Antibiotics, peptidoglycan synthesis and taxonomic implications. Bacteriol Rev 36(4):407–477. genomics: The chlamydial anomaly revisited. Microbiology 144(Pt 10):2673–2678. 26. Sabbah A, et al. (2009) Activation of innate immune antiviral responses by Nod2. Nat 7. Barbour AG, Amano K, Hackstadt T, Perry L, Caldwell HD (1982) Chlamydia tracho- – matis has penicillin-binding proteins but not detectable muramic acid. J Bacteriol Immunol 10(10):1073 1080. 151(1):420–428. 27. McCoy AJ, et al. (2006) L,L-diaminopimelate aminotransferase, a trans-kingdom en- 8. Moulder JW (1993) Why is Chlamydia sensitive to penicillin in the absence of pepti- zyme shared by Chlamydia and plants for synthesis of diaminopimelate/lysine. Proc doglycan? Infect Agents Dis 2(2):87–99. Natl Acad Sci USA 103(47):17909–17914. 9. Ghuysen JM, Goffin C (1999) Lack of cell wall peptidoglycan versus penicillin sensi- 28. Inohara N, et al. (2003) Host recognition of bacterial muramyl dipeptide mediated tivity: New insights into the chlamydial anomaly. Antimicrob Agents Chemother through NOD2. Implications for Crohn’s disease. J Biol Chem 278(8):5509–5512. 43(10):2339–2344. 29. Mahapatra S, Crick DC, Brennan PJ (2000) Comparison of the UDP-N-acetylmuramate:L- 10. Patin D, Bostock J, Chopra I, Mengin-Lecreulx D, Blanot D (2012) Biochemical char- alanine ligase enzymes from Mycobacterium tuberculosis and Mycobacterium leprae. acterisation of the chlamydial MurF ligase, and possible sequence of the chlamydial J Bacteriol 182(23):6827–6830. peptidoglycan pentapeptide stem. Arch Microbiol 194(6):505–512. 30. Emanuele JJ, Jr, et al. (1996) Kinetic and crystallographic studies of Escherichia coli 11. Shimada K, et al. (2009) The NOD/RIP2 pathway is essential for host defenses against UDP-N-acetylmuramate:L-alanine ligase. Protein Sci 5(12):2566–2574. Chlamydophila pneumoniae lung infection. PLoS Pathog 5(4):e1000379. 31. Mizuno Y, Yaegashi M, Ito E (1973) Purification and properties of uridine diphosphate 12. Welter-Stahl L, et al. (2006) Stimulation of the cytosolic receptor for peptidoglycan, N-acetylmuramate: L-alanine ligase. J Biochem 74(3):525–538. Nod1, by infection with Chlamydia trachomatis or Chlamydia muridarum. Cell 32. Raymond JB, Mahapatra S, Crick DC, Pavelka MS, Jr (2005) Identification of the namH – Microbiol 8(6):1047 1057. gene, encoding the hydroxylase responsible for the N-glycolylation of the mycobac- 13. Opitz B, et al. (2005) Nod1-mediated endothelial cell activation by Chlamydophila terial peptidoglycan. J Biol Chem 280(1):326–333. pneumoniae. Circ Res 96(3):319–326. 33. Coulombe F, et al. (2009) Increased NOD2-mediated recognition of N-glycolyl 14. Buchholz KR, Stephens RS (2008) The cytosolic pattern recognition receptor NOD1 muramyl dipeptide. JExpMed206(8):1709–1716. induces inflammatory interleukin-8 during Chlamydia trachomatis infection. Infect 34. Dicker LW, Mosure DJ, Berman SM, Levine WC (2003) Gonorrhea prevalence and coinfection Immun 76(7):3150–3155. with chlamydia in women in the United States, 2000. Sex Transm Dis 30(5):472–476. 15. Girardin SE, et al. (2003) Peptidoglycan molecular requirements allowing detection by 35. Miller WC, et al. (2004) Prevalence of chlamydial and gonococcal infections among Nod1 and Nod2. J Biol Chem 278(43):41702–41708. – 16. Pilhofer M, et al. (2013) Discovery of chlamydial peptidoglycan reveals bacteria with young adults in the United States. JAMA 291(18):2229 2236. murein sacculi but without FtsZ. Nat Commun 4:2856. 36. Nsuami M, Cammarata CL, Brooks BN, Taylor SN, Martin DH (2004) Chlamydia and – 17. Horn M (2008) Chlamydiae as symbionts in eukaryotes. Annu Rev Microbiol 62: gonorrhea co-occurrence in a high school population. Sex Transm Dis 31(7):424 427. 113–131. 37. Binet R, Maurelli AT (2005) Frequency of spontaneous mutations that confer antibiotic 18. Liechti GW, et al. (2014) A new metabolic cell-wall labelling method reveals pepti- resistance in Chlamydia spp. Antimicrob Agents Chemother 49(7):2865–2873. doglycan in Chlamydia trachomatis. Nature 506(7489):507–510. 38. Iyer JK, Coggeshall KM (2011) Cutting edge: Primary innate immune cells respond effi- 19. Garrett AJ, Harrison MJ, Manire GP (1974) A search for the bacterial mucopeptide ciently to polymeric peptidoglycan, but not to peptidoglycan monomers. J Immunol component, muramic acid, in Chlamydia. J Gen Microbiol 80(1):315–318. 186(7):3841–3845. MICROBIOLOGY

Packiam et al. PNAS | September 15, 2015 | vol. 112 | no. 37 | 11665 Downloaded by guest on September 23, 2021