Signs and symptoms suggestive of clinical sepsis were: Unstable temperature, lethargy, irritability, gastrointestinal dysfunction with milk intolerance, vomiting, abdominal distension or bloody stool, respiratory dysfunction, sudden increase in respiratory rate or persistent tachypnoea, and tachycardia or bradycardia. These signs and symptoms were described in detail in a previous study (9).

From January to May 2012, 60 neonates who were suspected of neonatal septicemia according to the above diagnostic criteria in the Department of Neonatology, Children's Hospital of Chongqing Medical University were enrolled in the present study. Paired blood samples were collected after careful skin disinfection and sent for blood culture and molecular analysis. Blood samples from 10 neonates diagnosed with jaundice (caused by ABO hemolytic disease, without any evidence of infection or antibiotic treatment) were collected and served as negative controls. Venous blood (3 ml) was collected, of which 1 ml was inoculated in the corresponding culture bottles for aerobic blood culture in the BACTEC 9120 system (BD Diagnostics, Bergen, NJ, USA), and another 1 ml was used for anaerobic blood culture in the BacT/Alert system (BioMérieux, Marcy-l'Etoile, France). The remaining 1 ml venous blood was collected in a sterile blood collection tube containing EDTA (Shanghai Kehua Bio-engineering, Shanghai, China) and stored at −70°C until molecular processing.

DNA extraction was performed according to the manufacturer's instructions using the QIAamp DNA Blood Mini kit (Qiagen, Hilden, Germany) with sterile water as a negative control. The blood samples were frozen, thawed for 3 cycles, and incubated at 37°C for 1 h with mixing every 20 min followed by the addition of 180 μl (40 mg/ml) lysozyme (Sigma-Aldrich, St. Louis, MO, USA). All DNA extraction reagents except the lysozyme solution were filtered through a 0.22-μm filter prior to bacterial DNA extraction. Whole genomic DNA (100 ml), including bacterial genomic DNA, was dissolved in buffer AE (10 mM Tris·Cl, 0.5 mM EDTA, pH 9.0) and stored at −20°C. PCR amplification was performed using a PCR amplifier (Eppendorf, Humburg, Germany), with previously described thermocycling conditions (8).

Primers for amplification of the variable region of 16S rDNA for reaction 1 were as follows: B5, 5′-TCAGATTGAACGCTGGCGGC-3′ and B4, 5′-TATTACCGCGGCTGCTGGCA-3′ (8). The amplified 493-bp products from reaction 1 were then amplified in reaction 2 with primers P2 (5′-CCTACGGGAGGCAGCAG-3′) and P3 (5′-ATTACCGCGGCTGCTGG-3′) starting from nucleotide 341 and 534 of the 16S rDNA, respectively (8). A 40-bp GC clamp was attached to the 5′ end of the P2 primer to prevent complete separation of PCR amplicons during DGGE analysis. The sterile water was filtered through a 0.22-μm filter to avoid possible contamination. The PCR mixture of the first amplification was adjusted to a final volume of 25 μl with sterile water after 12.5 μl of Premix Taq (Takara Bio Inc., Otsu, Japan), 0.5 μl each primer (10 μM), and 5 μl DNA template were added. The PCR mixture of the second amplification was adjusted to a final volume of 50 μl with sterile water after 1 μl of PCR product from the first amplification, 25 μl Premix Taq (Takara Bio Inc.), and 1 μl each primer (10 μM) were added, with sterile water used as a negative control.

DGGE analysis was performed using the Dcode Universal Mutation system (Bio-Rad, Hercules, CA, USA) following the manufacturer's instructions. The products from PCR amplification were first electrophoresed on a 2% (wt/vol) agarose gel and stained with 4S nucleic acid (Sangon Biotech Co. Ltd., Shanghai, China). Only samples with visible target bands were adopted for further DGGE analysis. Electrophoresis was performed on 8% polyacrylamide gels with a denaturing gradient ranging from 35 to 65% in 1X TAE buffer (0.04 M Tris-acetate, 0.00l M EDTA; pH 8.0) at 85 V and 60°C for 16 h. Then the gels were incubated in 1X TAE buffer containing SYBR Green I (Biotech, Beijing, China) for 30 min and scanned using a Benchtop 3UV Transilluminator (UVP, Inc., Upland, CA, USA). Each visible band was excised from the DGGE polyacrylamide gels, placed in 30 μl of sterile water, and incubated at 4°C overnight. PCR-DGGE analysis of each sample was repeated twice.

DNA recovered from excised DGGE bands was amplified by PCR as described previously (10), and the PCR products were electrophoresed on a 2% agar gel and purified using the Gel DNA Extraction kit (Takara Bio Inc.). The purified PCR products (2 ml) were used to perform TA cloning using the pMD 18-T Vector kit (Takara Bio Inc.) according to the manufacturer's instructions. Then the pMD 18-T plasmids containing the PCR amplicons were transformed to Escherichia coli DH5α-competent cells (TIANGEN Biotech (Beijing), Co., Ltd., Beijing, China) and cultured on ampi-cillin-resistant Luria-Bertani (LB) broth at 37°C overnight.

Clones on ampicillin-resistant LB media were collected and blended in 10 μl sterile water. Then colony PCR was performed to confirm the cloned DNAs using primers M13-RV (5′-CAGGAAACAGCTATGAC-3′) and M13-M3 (5′-GTAAAACGACGGCCAGT-3′). The PCR mixture containing 1 μl template (the monoclone blended in sterile water), 1 μl of each primer (10 μM), and 12.5 μl Premix Taq (Takara Bio Inc.) was adjusted to a final volume of 25 μl with sterile water. The PCR mixture was first incubated at 94°C for 7 min followed by a total of 30 cycles of 30 sec at 94°C, 30 sec at 60°C, and 30 sec at 72°C, with a final step at 72°C for 5 min. The length of each amplicon was confirmed by agarose gel electrophoresis.

The sequencing of the PCR products was performed on an ABI 3730×l DNA Analyzer with M13+(−47) primers and BigDye terminator v3.1 (Applied Biosystems, USA) at Sangon Biotech (Shanghai, China), and the sequences were analyzed and blasted on NCBI (http://blast.ncbi.nlm.nih.gov/Blast.cgi?CMD=Web&PAGE_TYPE=BlastHome).

E. coli and Staphylococcus epidermidis cultures at 1×10⁵ CFU/ml (10 ml) each were blended in blood samples (200 μl) from neonates diagnosed with jaundice to prepare spiked, 'infected' blood samples, in order to imitate the bacterial load in septicemia (4). The bacterial DNA was then extracted following the procedures described above and analyzed by the 16S rDNA PCR-DGGE and sequencing method.

Pathogen-specific PCR was conducted to detect the DNA of the bacteria confirmed by blood culture but not by the 16S rDNA PCR-DGGE and sequencing method. Two patients who were diagnosed with Klebsiella pneumoniae septicemia were selected. The primers for amplification of K. pneumoniae were as follows: Forward: 5′-GCGTGGCGGTAGATCTAAGTCATA-3′ and reverse: 5′-TTCAGCTCCGCCACAAAGGTA-3′. The PCR conditions and method used to confirm the length of the PCR products were the same as described above.

Enterococcus faecalis, S. epidermidis and Pseudomonas aeruginosa, often observed in septicemia, were used to imitate the situation present in infected blood by a pyrosequencing method. Genomic DNA of each bacterium was extracted as described above. The concentration of the DNA was quantified using NanoDrop 1000 (Nanodrop, Wilmignton, DE, USA) and diluted to 65 ng/μl. The DNA of E. faecalis was 5-fold serially diluted (served as less dominant bacteria) and blended with a constant concentration of DNA from the other two bacteria (both served as dominant bacteria). The set of the mixed DNA served as the template and was subjected to a series of 16S rDNA PCR amplifications as described previously (10). In addition, the set of diluted DNA of E. faecalis alone served as the template and was subjected to 16S rDNA PCR amplification.

The PCR mixture containing 25 μl Premix Taq (Takara Bio Inc.), 1 μl of each primer (10 μM), 1 μl diluted DNA of E. faecalis, or 3 μl (mixed DNA with 1 μl of each bacterial DNA) of bacterial DNA was adjusted to a final volume of 50 μl with sterile water. The PCR products were then analyzed by the DGGE and sequencing method as described above.

From January to May 2012, 60 neonates who were suspected of neonatal septicemia were enrolled in the present study. Positive blood culture results occurring >72 h after birth were reported for 14 of these 60 neonates (Table I). Of these 14 neonates with positive blood culture, 12 neonates were diagnosed with mono-bacterial infection and 2 neonates were confirmed to be infected by two species of bacteria.

The sequencing results showed diverse bacterial species in the blood samples, the majority of which were not detected by blood culture (Table I). Only in five cases did the sequencing results match partly or wholly with the blood culture results. One of these five patients was diagnosed with K. pneumoniae bloodstream infection, which was confirmed by molecular methods and blood culture. In the other four of these five cases, the molecular method detected a more complex bacterial spectrum, which contained the blood culture proven bacteria (Table I). In the other 9 cases, blood culture failed to detect bacteria, such as Neisseria sp., Moraxella sp., that were detected by the 16S rDNA PCR-DGGE and sequencing method. In addition, the 16S rDNA PCR-DGGE and sequencing method also failed to detect the blood culture-proven bacteria, such as Klebsiella pneumonia and Enterococcus faecalis.

The three steps investigating the possible reason for failure of the molecular method in detecting culture-proven bacteria were as follows. Firstly, efficiency of the DNA extraction method was tested. Then pathogen-specific PCR was applied to confirm the existence of culture-proven bacterial DNA in the sample DNA solutions. Thirdly, spiked samples were made to imitate the infection revealed by 16S rDNA PCR-DGGE and sequencing (presence of 'culture-proven pathogen' bacteria with other bacteria detected by molecular method). The hypothesis that the molecular method-proven bacterial DNA could interfere with the detection of blood culture-proven bacteria was tested in these spiked blood samples by the 16S rDNA PCR-DGGE and sequencing method as described in the materials and methods.

The results of the PCR-DGGE and sequencing method for the spiked 'infected' blood samples demonstrated that the DNA of the corresponding bacteria in blood was successfully isolated. The bands on the polyacrylamide gel of the DGGE were identical to those of E. coli and S. epidermidis, which were confirmed by subsequent sequencing (Fig. 1).

No visible band was found on the gels of two culture-proven cases after PCR-agarose gel electrophoresis. Hence, a second PCR amplification of the products from the first amplification was performed. After the second amplification, the target gene of K. pneumoniae was observed on the agarose gel (Fig. 2).

The detection limits of 16S rDNA PCR-DGGE for a single certain species of bacterial DNA versus mixed species of bacterial DNA (mixture of other bacterial DNA with this certain species of bacterial DNA) were completely different with a higher sensitivity in detecting this certain species of bacterial DNA in the spiked samples which were free of other bacterial DNA (Fig. 3A and B). When the E. faecalis DNA was amplified alone even at a 1×5⁷-fold dilution, the faint band amplified from E. faecalis DNA was still observed (Fig. 3A). When the DNAs of the other two species of bacteria were introduced, no band was observed for E. faecalis DNA even at dilution by 1×5³-fold (Fig. 3B). The detection limit of E. faecalis was different under the above two situations where the amount of E. faecalis DNA and 16S rDNA were identical throughout, indicating a possible influence of the introduced bacterial DNA on detecting the E. faecalis DNA. As the amplifying condition of 16S rDNA PCR was consistent, the introduced bacterial DNA caused reduced capability in amplifying E. faecalis DNA. There was lower sensitivity of 16S rDNA PCR in detecting the same copies of E. faecalis DNA when other bacterial DNA was introduced, compared with detecting the E. faecalis DNA in the samples containing E. faecalis DNA alone. This suggests a competitive inhibitory effect of other bacterial DNA on the E. faecalis DNA during the 16S rDNA PCR amplification.