A New Microarray System to Detect Streptococcus pneumoniae Serotypes

Streptococcus pneumoniae, one of the most common gram-positive pathogens to colonize the human upper respiratory tract, is responsible for many severe infections, including meningitis and bacteremia. A 23-valent pneumococcal vaccine is available to protect against the 23 S. pneumoniae serotypes responsible for 90% of reported bacteremic infections. Unfortunately, current S. pneumoniae serotype testing requires a large panel of expensive antisera, assay results may be subjective, and serotype cross-reactions are common. For this study, we designed an oligonucleotide-based DNA microarray to identify glycosyltransferase gene sequences specific to each vaccine-related serotype. Out of 56 isolates representing different serotypes, only one isolate, representing serotype 23A, was not detected correctly as it could not be distinguished from serotype 23F. Our data suggest that the microarray provides a more cost-effective and reliable way of monitoring pneumococcal capsular types.


Introduction
Streptococcus pneumoniae is an important cause of bacteremia, community-acquired bacterial pneumonia, and meningitis, especially among young children and older adults [1][2][3]. Capsular polysaccharide is the primary S. pneumoniae virulence factor and encapsulated pneumococci are responsible for more diseases than unencapsulated strains [4]. After comparing the differences in capsular polysaccharides composition, S. pneumoniae can be divided into more than 90 serotypes [5] and the 23 serotypes responsible for 90% of disease cases [6] are represented in a 23-valent pneumococcal vaccine. Pneumococcal serogroup and serotype identification is currently performed by using large panels of expensive antisera by various methods, including the capsular swelling (Quellung) reaction, latex agglutination, and coagglutination. Cross-reactions between serotypes and discrepancies between methods can occur and some strains are nonserotypable. On the other hand, 2.3. Chromosomal DNA Labeling. 500 ng of genomic DNA was suspended in 21 µL dH 2 O and 20 µL of 2.5 × Random Primer Solution (Invitrogen, Carlsbad, CA), heated to 95 • C for 5 min, and chilled on ice for 3 min. The DNA was labelled in a reaction including 5 µL of 10X dCTP Nucleotide Mix (Invitrogen, Carlsbad, CA), 5 µL Cy3 or Cy5-dCTP (GE Healthcare, Buckinghamshire, UK), and 1 µL of Exo-Klenow Fragment (Invitrogen, Carlsbad, CA ). After a 2hour incubation at 37 • C, 5 µL of sodium acetate, 125 µL of ethanol and 1 µL of glycogen was added to 25 µL of Cy3 and Cy5 labeled DNA, which was purified previously by QIAprep Spin Miniprep Kit (250) (Qiagen, Tokyo, Japan). Following a 30-minute incubation at −80 • C in the dark, the probe mixture was centrifuged for 30 min at 14,000 rpm at 4 • C. The supernatant was removed and the probe was air-dried for 5 min in the dark. The probe mixture was diluted in 70 µL of the hybridization buffer (25% formamide, 0.1% SDS, 6 × SSPE), incubated for 30 min at room temperature in the dark, heated for 8 min at 75 • C, and incubated for 30 min at 42 • C.

Probe Hybridization and Microarray Signal Detection.
Prewarmed probe mixture was applied to the prepared microarray slide, placed in a hybridization chamber and incubated for 20 h at 42 • C. After hybridization, the plastic seal was removed and the slide was washed with 1 × SSC/0.1% SDS solution for 3 min, 0.05 × SSC for 3 min, and 95% ethanol for 90 s at room temperature. The washed microarray slide was dried by centrifugation and scanned using the DNA Microarray Scanner (Agilent, Santa Clara, CA).

Data Analysis.
The signal and background intensities of each spot were quantified using GenePix Pro 6.0 software and the average was calculated with Microsoft Excel software.

Evaluation of the Microarray.
A total of 274 oligonucleotide probes were used in this microarray, including positive and negative controls and GT gene-specific probes. The microarray probes were tested using 36 pneumococcal isolates from 23 vaccine-associated serotypes and 19 additional pneumococcal isolates belonging to other serotypes (Table 1). Figure 1(b) shows the examples of scanned pictures of 6 strains representing different serotypes. Examples of the same serotype were tested repeatedly and shown to have an identical signal pattern, for example, 5 times for serotype 3 (data not shown). Of 23 strains representing 23-valent vaccine serotype, 18 strains hybridized to all the specific set of probes, and four strains hybridized to almost all the specific set of probes (Table 4). The strain representing serotype 22F may actually belong to serotype group 22F/22A, since this sample failed to hybridize specifically to wchF and wcwA probes but hybridized to the rest of group 22F/22A specific probes. Of the 13 strains representing the 23 vaccinerelated serotypes, only 1 isolate (serotype 46), failed to hybridize to a specific probe while the other 12 strains hybridized perfectly. Of the 20 nonvaccine serotypes, 19 strains either hybridized partially to GT-specific probes or did not hybridize to any probes. One strain, representing serotype 23A, hybridized to most of the 23F-specific probe; thus, 23A may be indistinguishable from 23F using GT gene sequences.

Discussion
In order to develop a more effective S. pneumoniae vaccine, simple detection methods are required to serotype large numbers of clinical isolates. Conventional serotyping methods using large panels of antisera are labourious and require technical expertise. Our microarray method can determine serotype of a strain at one time and needs no expertise.
In addition, the microarray method described here has the potential to be automated. To our knowledge, our report describes the first microarray to utilize GT genes to predict serotype of any bacteria. Several molecular typing methods have been developed based on serotype-specific sequences [12][13][14][15][16][17][18][19][20][21]. Wang et al. [21] described microarray method using wzy and capA genes. Our approach is different in that GT genes were selected as serotype-specific genes. Since GTs catalyze the transfer of the sugar moiety to an acceptor and generate a serotype-specific capsular polysaccharide, detecting GT genes can directly reflect polysaccharide structure. We discovered considerable variability within S. pneumoniae GT genes, which provides groundwork for future investigations into new S. pneumoniae capsular types. Our method using GT genes can not only discriminate serotypes but can give information of the capsular polysaccharide structure.

Conclusion
We developed a S. pneumoniae DNA microarray that identifies GT gene polymorphisms to distinguish capsular types. We believe that our microarray system is more reliable and cost-effective and will help to survey the emergence of new S. pneumoniae serotype.