FEMS Pathogens and Disease, 74, 2016, ftv102 doi: 10.1093/femspd/ftv102 Advance Access Publication Date: 1 November 2015 Research Article

RESEARCH ARTICLE

Selective distribution of Pseudomonas aeruginosa O-antigen among strains producing group I pilin Tara M. Allison and Peter Castric∗ ∗

Corresponding author: 600 Forbes Av., Pittsburgh, PA 15216, USA. Tel: + 412-396-6319; Fax: +412-396-4907; E-mail: [email protected] One sentence summary: The authors examine if there is an association between the expression of group I pilin subunits and specific LPS glycoforms in Pseudomonas aeruginosa; interestingly, they find that there may be a bias in the distribution of specific LPS types in the group I pilin expressing strains and that this in turn could influence and even enhance group I pilus function. Editor: Ake Forsberg

ABSTRACT Strains of Pseudomonas aeruginosa that produce type IVa pili categorized as group I have the potential to covalently attach an O-antigen repeating unit to the pilin C-terminal residue. PCR, employing primers targeting a conserved region of a group-I-specific gene, was used to provide evidence that 110 of 206 clinical isolates studied had the capability of producing this type of pilus. The potential of P. aeruginosa to produce a particular O-antigen type is determined by the presence of a specific biosynthetic gene cluster. The distribution of these gene clusters among the isolates studied was determined using a second PCR procedure. The results of these studies showed that the O-antigen repeating unit types associated with group I pilin producers were significantly different from those found in the non-group I pilin strains. In addition, the predicted ability to express O-antigen repeating units composed of four sugars, and the ability of the glycan to express a negative charge were associated with group I pilin producing strains. The results presented suggest that these properties specifically enhance group I pilus function and that the commonality of pilus and O-antigen types may be useful as targets in disease intervention. Keywords: Pseudomonas aeruginosa; Pili; Type IVa pili; O-antigen repeating unit; Pilin glycosylation; virulence factors; pathogenicity

INTRODUCTION Pseudomonas aeruginosa is a Gram negative, rod-shaped bacterium that is a frequent etiological agent of hospital-acquired infections. The type IVa pili (referred to as T4P in this communication) of this organism are surface-associated protein filaments which are composed of a monomeric subunit called pilin (Mattick 2002; Craig, Pique and Tainer 2004). These fibers may be placed into five groups based on subunit primary structure as well as on the presence of pilus-specific accessory genes (Castric 1995; Kus et al. 2004). Approximately half of P. aeruginosa isolates tested in a single study were found to produce pilin belonging to group I (Kus et al. 2004). This protein has the added characteristic of being post-translationally modified with the

covalent attachment of a single lipopolysaccharide (LPS) Oantigen repeating unit (Castric, Cassels and Carlson 2001). All subunits of P. aeruginosa strain 1244 are glycosylated suggesting that the group I pilus (referred to as T4PI in this communication) is uniformly coated with this glycan (Castric, Cassels and Carlson 2001; Smedley et al. 2005). Work from our laboratory has indicated that the T4PI glycan influences pilus solubility and fiber function at physiological ionic strengths (Allison, Conrad and Castric 2015). Additionally, our previous finding that the absence of this saccharide reduces viability in a mouse respiratory disease model (Smedley et al. 2005) further indicates that this structure is of functional importance. Since the O-antigen repeating unit is highly variable in size, charge and structure (Knirel et al.

Received: 12 August 2015; Accepted: 26 October 2015  C FEMS 2015. All rights reserved. For permissions, please e-mail: [email protected]

1

Downloaded from http://femspd.oxfordjournals.org/ by guest on March 18, 2016

Department of Biological Sciences, Duquesne University, Pittsburgh, PA 15282, USA

2

FEMS Pathogens and Disease, 2016, Vol. 74, No. 1

MATERIALS AND METHODS Bacterial strains and culture conditions Luria–Bertani medium (LB; 1% tryptone, 0.5% yeast extract, 1% NaCl, pH7.0) or Casamino Acids–Yeast extract medium (CAYE; 0.75% casamino acids, 0.15% yeast extract) (Silipigni-Fusco 1987) broth cultures were grown at 37◦ C on a rotary shaker at 250 rpm. Solid media were prepared with the addition of either 1.5% (in the case of LB) or 2.0% (CAYE) agar. All clinical P. aeruginosa isolates used in this study (Table S1, Supporting Information) were collected from Pittsburgh Children’s Hospital between 17 September 2010 and 7 December 2012. Infection sites were recorded and repeat samples were not used. Reference strains used in development of the O-antigen typing methodology are listed in Table S2, Supporting Information. Pseudomonas aeruginosa 577B was identified as serotype O4 using polyclonal typing sera specific for the 20 IATS serotypes. Strains IATSO1, IATSO3, IATSO9, IATSO10, IATSO12 and IATSO13 were provided by Dr Joanna Goldberg, Emory University School of Medicine.

Identification of strains producing group I pilin PCR identification was employed using primers (Table S3, Supporting Information) against a conserved region of pilO, a T4PI specific gene (Castric 1995). Genomic DNA was extracted using the Qiagen DNA mini kit as per the manufacturer’s protocol (Qiagen). Thermocycler conditions were as follows : (i) 94◦ C—2.5 min, (ii) 94◦ C—1 min, (iii) 61.5◦ C—-45 s, (iv) 72◦ C—1.5 min, repeat 2–4 for 30 cycles, (v) 72◦ C—5 min and (vi) 4◦ C Hold (Eppendorf, MasterGradient). This protocol produced a product of the expected size using DNA from P. aeruginosa 1244, a strain known to have a pilO gene (Castric 1995) as target. DNA from P. aeruginosa PA103, which lacks this gene (DiGiandomenico et al. 2002), produced no signal.

Identification of strains containing specific O-antigen biosynthesis gene clusters Raymond et al. showed that genetic information for each of the 20 IATS O-antigen serotypes was encoded in one of 11 gene clusters on the P. aeruginosa chromosome (Raymond et al. 2002). The sequences for these gene clusters were obtained from GenBank and aligned using ClustalW multiple sequence alignment, Megalign and Microsoft Word. Genetic loci providing sequences unique to this group were found within the (i) orfA and wzz genes (clusters 1, 2, 3, 4, 6, 7, 8, 9 and 11), (ii) wzz and wbpO genes (cluster 5) and (iii) wzz and wbpK genes (cluster 10). With this information, specific primers for each of the 11 clusters were prepared and tested using standard P. aeruginosa strains (Table S4, Supporting Information) of known IATS serotype. Genomic DNA from these organisms was extracted as described above. Thermocycler conditions were as follows (i) 94◦ C—3 min, (ii) 94◦ C—1 min, (iii) 57◦ C—45 s, (iv) 72◦ C—45 s, repeat 2–4 for 30 cycles, (v) 72◦ C—5 min and (vi) 4◦ C Hold (Eppendorf, MasterGradient). The reactions of these primer sets with the stock strains are shown in Fig. 1. Here, it could be seen that these specific primers recognized DNA from their IATS targets, producing fragments of anticipated size. In addition, these primers produced no reaction with DNA from any of the remaining target strains (Allison and Castric, unpublished observations). PCR amplification using these primers was applied to DNA from the clinical isolates described above. Strains which produced DNA that was recognized by these primers were referred to as cluster types (abbreviated as C-types). Ambiguous results with (IATS O7) and (IATS O11) primer sets required minor redesign and retesting. The relationship between IATS serotype and C-type is shown in Fig. 1. In order to increase the efficiency of C-typing with these primers, sets were combined and tested for specificity against a single type for each DNA sample. This was done to reduce the number of reactions necessary to type all isolates. Primer set combinations are shown in Table S5, Supporting Information. While the majority of cluster types express distinctive and predictable physical characteristics, C3 and C9 are each composed of two subtypes composed of IATS O3 and O15 in the former and IATS O11 and O17 in the latter. Members of these subtype pairs differ from each other in sugar number and structure (Knirel et al. 2006; King et al. 2009; Lam et al. 2011). This indicates that the use of the PCR method described would have to be extended when examining populations where C3 and C9 types are important components.

Western blotting of whole cell lysates One 2% CAYE agar plate was inoculated with an 8-h CAYE broth starter culture and incubated 14 h at 37◦ C. Cells were resuspended with phosphate buffered saline (PBS) and normalized to an equal OD650 using the same buffer. All samples were lysed in Laemli buffer and subjected to polyacrylamide gel electrophoresis using a 16% T separation gel as previously described (Castric, Sidberry and Sadoff 1989). Proteins were electroblotted onto 0.45 μm nitrocellulose membranes at 100V for 20 min as previously described (Horzempa et al. 2006a). The membranes were blocked in protein solution (Sidberry et al. 1985) for 20 min and incubated overnight with monoclonal antibody 5.44 (Castric, Cassels and Carlson 2001), which is specific for group I pilin. The membranes were treated with an alkaline-phosphatase-labeled secondary antibody and developed as previously described (Horzempa et al. 2006a).

Downloaded from http://femspd.oxfordjournals.org/ by guest on March 18, 2016

2006; King et al. 2009; Lam et al. 2011), it is possible that specific types of O-antigen repeating unit might have a direct effect on pathogenicity through an influence on T4PI function. If this is the case, natural selection could produce dominance of particular glycan types found among clinical isolates while suppressing the presence of others. This would suggest a linkage between T4PI and LPS of P. aeruginosa in terms of function as well as distribution within natural populations. The International Antigenic Typing Schema (IATS) allows LPS from this organism to be placed into one of 20 serotypes (Liu et al. 1983; Liu and Wang 1990). Work by Raymond et al. has shown that the variety of O-antigen types produced by P. aeruginosa are coded for by only 11 different gene clusters (Raymond et al. 2002). Nine of eleven of these clusters produces structurally predictable O-antigen oligosaccharides (Knirel et al. 2006; King et al. 2009; Lam et al. 2011) which can be related to the IATS as well as to previous serotyping systems (Fisher, Devlin and Gnabasik 1969; Lanyi and Bergan 1978). The present paper describes development of a typing system that is based on unique sequences present in each of these O-antigen biosynthetic gene clusters rather than on antigenic type. Application of this system to a clinical isolate population indicated that while one particular category predominated in strains producing T4PI pilin, others were rare. The evidence for an association between O-antigen and pilin types among strains of known disease origin suggests a functional relatedness between these virulence factors as well as a possible targeted therapeutic approach to P. aeruginosa infections.

Allison and Castric

3

Dot blot determination of whole cells Overnight broth cultures of all clinical isolates typed as C5, P. aeruginosa strain PAK (positive control), and P. aeruginosa strain 1244 (negative control) were diluted to the same OD650 using CAYE broth. Five microliters of each culture were then spotted onto 0.45 μm nitrocellulose membranes. The membranes were blocked in protein solution (Sidberry et al. 1985) for 20 min and incubated for 4 h in the presence monoclonal antibodies (128.1.1, 2-34-3-7 and 7-72-1) (Sadoff et al. 1985) specific for IATS serotype O6. Membranes were then washed three times in PBS and incubated with a goat anti-mouse Alexa Fluor 532 labeled secondary antibody for 1 h with constant motion in darkness. They were again washed three times in PBS for 10 min each time in darkness and allowed to dry. Fluorescent protein spots were visualized using a FluorImager 595 (Molecular Dynamics) set at 514 nm emission and using a 570 DF30 filter.

Twitching assays Subsurface twitch assays (McMichael 1992) were conducted as previously described (Castric, Cassels and Carlson 2001). Fisherbrand 100 × 15 mm polystyrene plates were used for the standard subsurface assay.

Statistical analysis Frequency data were analyzed using the 2 × 2 Contingency Table (GraphPad software: http://graphpad.com/quickcalcs/ contingency1.cfm).

RESULTS Identification of P. aeruginosa T4PI pilin producers A total of 206 P. aeruginosa isolates from defined clinical sources (chronic respiratory, acute respiratory, and non-respiratory) were screened for the potential to produce T4PI . This was done

Table 1. Distribution of isolates potentially able to produce group I pilin. PCR reaction Isolate source Chronic respiratory Acute respiratory Non-respiratory Total a b

Western blot

Positives/totala

(%)

Positives/PCR positivesb

(%)

38/82 28/48 44/76 110/206

46.5 58.3 57.4 53.4

30/38 27/28 42/44 99/110

79.0 96.4 95.5 90.0

Group I positives/total source number. Group I western blot positives/group I PCR positives.

by PCR using primers specific to a conserved region of pilO (tfpO), a gene found in strains producing such pili (Castric 1995). These results (Table 1) show that strains potentially capable of producing T4PI were well represented in these clinical isolates with a frequency of 110/206 (53.4% of all isolates). Western blotting using monoclonal 5.44, an antibody recognizing a conserved T4PI pilin epitope (Castric and Deal 1994), was carried out in order to confirm the PCR response and to determine whether the identified strains actually produced pilin (Table 1). While all strains were subjected to this analysis, it was found that a positive western blot reaction was limited to strains producing a pilO-specific PCR signal. Further, this work showed that out of all of the PCR-positive strains tested, only one acute respiratory isolate (out of 28) and two non-respiratory strains (out of 44) were not recognized by monoclonal 5.44. The respiratory isolate failed to show twitching motility (Table S1, Supporting Information) that suggested that this strain may be unable to produce pilin. Both of the non-respiratory isolates were able to display twitching motility suggesting that pilin from these strains may express a variant epitope which the antibody was unable to recognize. This low rate of variation can be compared to that seen with the chronic respiratory isolates where 8 of 38 PCR-positive strains did not produce a western blot signal. Since five of these strains were unable to twitch (Table S1, Supporting Information), it is possible that these are pilin-negative

Downloaded from http://femspd.oxfordjournals.org/ by guest on March 18, 2016

Figure 1. Validation of C-type PCR primer specificity using DNA from strains of known O-antigen serotype. (A) Relationship between C-type and IATS serotype. (B) Positive control PCR results. Primer sets recognize only one C-type and produce the expected product size. The first and last lanes contain DNA size standard.

4

FEMS Pathogens and Disease, 2016, Vol. 74, No. 1

Table 2. Immunoblot reaction frequency of R5 clinical isolates with O6-specific monoclonal antibodies. Antisera tested Sample source Chronic Respiratory Acute Respiratory Non-respiratory Strain PAK Strain 1244 a

mutants. These results may again be due to mutant selection since selection for mutants of various phenotypes (such as Oantigen production) is a characteristic of chronic P. aeruginosa infections of cystic fibrosis patients (Smith et al. 2006; Cullen and McClean 2015).

C-typing of isolates Each of the 11 known P. aeruginosa O-antigen biosynthetic gene clusters (Raymond et al. 2002) can be distinguished from one another by PCR. Since these clusters each produce structurally distinct repeating units (Knirel et al. 2006; King et al. 2009; Lam et al. 2011), this procedure allows for a novel form of O-antigen typing as well as the prediction of the glycan type available for pilin glycosylation. The results of analysis of the 206 clinical isolates using this methodology are presented in Fig. 2. While the entire collection of T4PI strains could be unambiguously typed, 8 of the 96 non-group I pilin producing strains failed to react with any of the primer sets and were listed as non-typeable. Repeated attempts to amplify DNA from these strains using all primer sets gave negative results. These non-reacting strains may contain O-antigen gene clusters that have previously not been described or are variants that do not contain complete primer recognition sequences. An immunoblot assay was carried out to determine if there is a correlation between C-typing and serotyping. Since the C5 gene cluster produces the IATS O6 serotype, all of the T4PI /C5 isolates were tested by immunoblot using a panel of three monoclonal antibodies specific for IATS O6. Table 2 shows that while the pattern of reaction varied, at least 56 of the 65 strains tested reacted with at least one of the sera employed. Although lack of antibody recognition could be due to epitope variation, five of the nine non-reactors were from patients with cystic fibrosis suggesting the possibility of loss of this trait during chronic infection (Ernst et al. 1999; Smith et al. 2006; Cullen and McClean 2015).

Distribution of O-antigen repeating unit genes among isolates Overall, while the isolates typing as C5 were most numerous, those making C10 and C11 were absent. More importantly, the

2.34.3.7a

7.72.1a

15/20 21/22 19/23 1/1 0/1

10/20 20/22 19/23 1/1 0/1

7/20 13/22 15/23 1/1 0/1

Positive reactions/strain tested.

isolates possessing the pilO gene, and therefore potentially capable of producing T4PI , showed a distinctively different pattern of distribution as compared to the remainder of strains. For example, T4PI producing strains that express genetic information for the C5 cluster were found far more frequently (59.1%) as compared to non-group I typeable isolates (16.7%). Strains expressing this co-frequency made up 32.8% of all typeable strains. On the other hand, C2 and C9 types were rarely found among the T4PI producing strains (0.9 and 1.8%, respectively), although clearly represented (16.7 and 24.0%, respectively) among the non-group I pilin producing isolates. The frequency of differences between strains producing these three C-types and these two pilin groups were noted to be significant (P < 0.0001) in each case. Predominance of the C5 phenotype among strains producing T4PI as compared to non-group I strains was seen with each disease type (Fig. 3). However, the overall correlation between C-type and pilin group was found to be especially strong among acute respiratory isolates where the C5 type was associated with 78.6% of T4PI producers. By comparison, 52.5% and 52.0 % of the chronic respiratory and non-respiratory strains, respectively, shared the T4PI /C5 phenotype. Statistical analysis of the comparison of this trait between acute respiratory isolates and the remainder of typed strains gave a P value of 0.0249. The C2 and C9 types, which are rarely found among the T4PI producing strains, do not show a statistically significant disease preference.

Relationship between oligosaccharide structure and glycan distribution While an important characteristic of the O-antigen repeating units is their high degree of structural variability, they are usually composed of either three or four sugars and are normally neutral or express a single net negative charge (Knirel et al. 2006). Although variation in charge can occur through the addition or deletion of functional groups (Knirel 1990), these characteristics are largely predicted by O-antigen biosynthetic gene cluster (Knirel et al. 2006). The distribution of these features among the isolates tested is presented in Table 3. It can be seen that O-antigen repeating units composed of four sugars are predominant among the T4PI isolates, but are in a distinct minority among the non-group I strains. This asymmetrical distribution is strongly significant (P < 0.0001) as determined using the 2 × 2 Contingency Table. The predominance of oligosaccharides expressing a negative charge is also predicted for strains producing T4PI pilin, although with a somewhat lower degree of significance (P = 0.0056). 74.6% of all T4PI strains tested are predicted to produce pilin containing a negatively charged, four sugar glycan.

Downloaded from http://femspd.oxfordjournals.org/ by guest on March 18, 2016

Figure 2. C-type distribution of group I and non-group I pilin producing clinical isolates. Histogram shows C1–C11 typing response frequency. N and P refer to non-typeable and polytypeable strains, respectively. P value: ∗ < 0.0001.

128.1.1a

Allison and Castric

Table 3. Distribution of predicted O-antigen sugar number and charge frequency. Sugar number Strains tested Group I Non-Group I

4 95 36

Selective distribution of Pseudomonas aeruginosa O-antigen among strains producing group I pilin.

Strains of Pseudomonas aeruginosa that produce type IVa pili categorized as group I have the potential to covalently attach an O-antigen repeating uni...
1KB Sizes 1 Downloads 8 Views