A high throughput serum bactericidal assay for antibodies to Haemophilus influenzae type b

Four quality control (QC) sera with very high (QCVH), high (QCH), medium (QCM), or low (QCL) titer sera prepared by mixing sera from 2 to 3 individuals (age range = 26 to 42 years) and were previously described [11, 12]. Their reference ranges of anti-Hib antibody titer were assigned after performing anti-Hib-antibody ELISA assay for more than 50 times [11]. Their reference ranges (mean ± standard deviation [SD]) were 43.00 ± 6.54 μg/mL, 4.38 ± 0.50 μg/mL, 1.52 ± 0.18 μg/mL, and 0.27 ± 0.07 μg/mL for QCVH, QCH, QCM, and QCL, respectively [11]. These sera were stored in 200-μL aliquots at −70 °C.

Ten pre-immune sera and 80 post-immune sera were selected based on their serum availability from a cohort of infants participating in an immunogenicity study of the Hib vaccine in Korean infants [12]. Anti-Hib IgG levels were previously determined for these residual sera [12] and 0.15 μg/mL was used as the lower limit of assay [11]. They were vaccinated with a single Hib vaccine (PRP-T or PRP-OMP).

SBA was performed as described [13] with modifications described below. All serum samples were heated at 56 °C for 30 min before testing was performed in duplicate. The heat inactivated sera were serially (three fold) diluted in a dilution buffer (Hanks’ buffer with Ca²⁺ and Mg²⁺ [Life Technologies, Grand Island, NY, USA] and 0.1 % gelatin). A frozen aliquot of Hib Eagan strain [14] was diluted in the dilution buffer to yield 750 colony forming unit (CFU) in 10-μL. Twenty μl of diluted serum was mixed with 10 μL of Hib solution and 10 μL of baby rabbit complement (Pel-Freez Biologicals, Brown Deer, WI, USA) in a microwell. The mixture was incubated for 2 min at 25 °C with shaking at 700 rpm and incubated for 30 min at 37 °C in a CO₂ incubator without shaking. After stopping the reaction by placing the plates on ice for 15 min, 10 μL of the reaction mixture was plated on an approximately 1 cm by 3 cm area of a brain-heart-infusion (BHI) agar plate with 2 % Fildes enrichment (Becton Dickinson and Company, Sparks, MD, USA). After the fluid was absorbed into the agar, molten BHI agar (0.75 %) with Fildes enrichment (2 %) and 25 mg/L 2, 3, 5-triphenyl tetrazolium chloride (TTC; Sigma, St. Louis, MO, USA) was poured on top of the bottom agar layer.

The plates were incubated at 37 °C in a 5 % CO₂ incubator for 16 h. Surviving bacterial colonies on the plates were counted with NICE (NIST [National Institute of Standards and Technology]’s Integrated Colony Enumerator); a free software [15] available from Dr. J. Whang in the US NIST) (http://​www.​nist.​gov/​pml/​electromagnetics​/​grp05/​nice.​cfm). The colony counts were used to determine the serum bactericidal index (SBI). The SBI of a serum was defined as the dilution of the serum that results in half as many colonies as are seen with complement controls. If an undiluted serum sample killed 50 %, then the SBI is 4 in our system. To reduce the effect of variable activation of the alternative pathway on SBI result, each assay run included a complement control containing bacteria and baby rabbit complement with no serum. This complement control was used as 0 % killing in the SBI calculation. A control serum was included in each assay to monitor assay reproducibility. A detailed SBA method is posted on the following website: http://​www.​vaccine.​uab.​edu/​.

The reproducibility of the Hib SBA was evaluated using the four QC sera, QCVH, QCH, QCM, and QCL. To determine intra-assay variability, sera were tested 5 times in one assay run. To determine short-term inter-assay variability, 11 independent assays using the same lot of reagents were performed but on different days.

To determine the effect of varying assay components, the assay was performed once with two different batches of bacterial stock and once with two different batches of complement. The overall mean ± SD and the coefficient of variation (CV) of SBIs were calculated.

To assess long term assay performance, SBA was performed with three QC sera (QCVH, QCH, and QCL) over a 6 year period and their bactericidal indices were plotted on a Levey-Jennings chart. During this period, the assay involved four different lots of baby rabbit complement, two different lots of Hib bacteria, and three assay operators.

Correlations between SBAs were determined by Pearson’s correlation. Significant differences among assays were determined by Student’s t test. Comparisons between paired data were done by chi-square or Fisher’s two-tailed exact test. The significance level was set at a P value of <0.05.

The study protocol was approved by the Institutional Review Board of Ewha Womans University Hospital. The study was conducted in accordance with good clinical practices (national regulations and ICH E6) and the principles of the Helsinki Declaration. Informed written consent was obtained from all participants or their parents or legal guardians following a detailed explanation of the study.

Automation of pneumococcal colony counting was attempted using an overlay of agar containing TTC [16]. However, TTC can be toxic to some bacteria resulting in the possibility that the overlay agar approach would not be feasible. Additionally, the overlay agar approach could have potentially failed because many bacteria may have simply detached from the base agar and floated over the overlay agar. Fortunately, Hib did not detach from the base agar and the overlay agar approach was determined to be feasible. To find a TTC concentration that gave optimal color without toxicity, we next investigated the effects of various concentrations (1–400 μg/mL) of TTC in the overlay agar. We found that the best results were obtained with 0.75 % agar containing 25 μg/mL TTC and 2 % Fildes enrichment [10]. The number of colonies observed with this overlay agar was equal to that of colonies observed without the overlay, indicating that the selected concentration of TTC was not toxic. With TTC, even very small colonies with a diameter less than 1 mm were clearly visible (Fig. 1), and allowed us to have up to 300 distinct colonies in a 1 cm by 3 cm area of an agar plate. In a conventional assay, a 96-well assay plate generates 96 agar plates. However, in our new procedure, a petri dish can have bacterial colonies from 48 reaction wells (i.e., half of the 96 well plate). Consequently, this allowed us to miniaturize the bacterial culture step and reduced the number of agar plates used for the assay by 48-fold.

To efficiently count colonies on the agar plates using the freely available program, NICE, we compared two different methods for acquiring images of the bacterial colonies on the petri-dish. We found that a camera can be used to obtain the images and that the colony counts obtained with digital camera images were highly correlated (r = 0.96) with the true manual counts as previously published [16], however, but the automated counts were slightly lower (by about 10–15 %) than the true counts (Fig. 2a). The deviation was noticeable when the number of colonies per spot exceeded 50. We subsequently found that images of four Petri dishes could be simultaneously obtained using a document scanner (Scanjet G4010, HP [Hewlett Packard Korea Ltd. Seoul, Korea]), which is readily available in many offices [15], and that the counts obtained with the scanner were highly correlated (r = 0.98) with the true counts. The slope of the best fit line was 0.89, suggesting that the colony counts determined with the scanner were slightly lower than the true counts.

As anti-Hib antibody ELISA has been extensively used in studies of Hib vaccines and protective surrogates have been established for Hib antibody ELISA, it was important to compare our high throughput SBA results with the ELISA results. The SBI and the anti-Hib IgG concentrations are highly correlated (Fig. 4). The correlation coefficient (r) was 0.61 for the 80 post-immune sera and became 0.84 when all 90 samples were included. For the statistical analysis, it should be noted that we assigned an SBI of 2 and an anti-Hib antibody concentration of 0.075 μg/mL to the 10 pre-immune samples (Fig. 4). Interestingly, the 80 post-immune serum samples had anti-Hib IgG concentrations of greater than 0.15 μg/mL and, not surprisingly, all 80 samples had detectable SBI. In contrast, the 10 pre-immune samples contained 0.15 μg/mL anti-Hib IgG and all of them resulted in undetectable SBIs. Based on this data, we believe that 0.15 μg/mL and 1 μg/mL IgG anti-Hib antibody, the widely accepted protective levels, may correspond to SBIs of 10 and 64, respectively. Nevertheless, additional studies would be necessary to establish the corresponding cut off values.