Development of an interferon-γ release assay (IGRA) for detection of Brucella abortus and clinical diagnosis of brucellosis

Introduction: Brucellosis, caused by Brucella abortus (B. abortus), is an important zoonosis posing a great risk to both livestock and humans. Currently, most assays for clinical diagnosis of brucellosis have been developed based on serological principles; however, these assays have a number of limitations and disadvantages. Methodology: To address this concern, the aim of this study was to develop a gamma interferon (IFN-γ) release assay (IGRA) for the diagnosis of brucellosis. Towards this end, the stimulating effect induced by different somatic antigens of B. abortus on the secretion of IFN-γ was evaluated. Results: The best antigen candidate, B. abortus strain 2308, able to induce high levels of IFN-γ expression in peripheral blood (PB) cells from cattle, was used for the development of the IGRA. The optimal concentration for stimulation was determined as 1.0×10 CFU/mL. This study demonstrated that IFN-γ was detectable on day 5 post infection (p.i.) and peaked on day 14 p.i.. Finally, the IGRA developed was used for detection of B. abortus in clinical samples, and a higher level of IFN-γ was detected in Brucella-infected samples compared to vaccination samples and negative controls. Conclusions: The optimal somatic antigen for B. abortus was identified and used to establish a robust IGRA. The IGRA developed is suitable for clinical diagnosis of brucellosis, especially in the early stages of infection.


Introduction
Brucellosis is a critical zoonotic disease caused by B. abortus, potentially leading to chronic multiple organ dysfunction and posing a high risk to livestock and public health [1]. Brucellosis is widely prevalent around the world, and nearly 170 countries or districts have so far reported an epidemic of this disease [2]. In the past decade, brucellosis re-emerged in China, as evidenced by increasing reports on epidemics of this disease. Although the rate of infection of B. abortus in animals and humans has been decreasing in recent years, it is still a great threat to animal husbandry and public health [3]. It is thought that prevention and control of brucellosis relies largely on reliable diagnosis. The existing methods for the diagnosis of brucellosis, however, have been developed based mainly on serology, such as the Rose Bengal test (RBT), standard tube agglutination test (SAT), complement fixation test (CFT), fluorescence polarization test (FPA) and enzyme-linked immunosorbent assay (ELISA) [4]. Non-specific reactions are common in these approaches since specific epitopes of Brucella and parasitic bacterium are localized in the cell [5].
Brucella is an intracellular bacterium, and this property elicits the host cellular immune response and accordingly, detection of the cellular immune response could be used to develop methods for the diagnosis of brucellosis. The most common method used in the clinic based on host cellular immune response is the Brucella skin delayed-type hypersensitivity (SDTH) test, which was developed to detect the allergic response in the host after subcutaneous injection of brucellins. It was reported that the rate of detection was 88% and 95% for the SAT and CFT assays [6]. It was demonstrated that the specificity of SDTH was higher than RBT and CFT when used in animals without vaccination with brucellosis vaccine [6]. IFN-γ is a significant cytokine that plays a crucial role in the host immune response. Previous studies demonstrated that the number of CD4 + cells and the expression of IFN-γ increased in the early stages of B. abortus infection [7][8]. Interestingly, the level of CD4, CD8 and CD21 in the blood of cattle increased when animals received the Brucella S19 vaccine and were stimulated by the Brucella 2308 antigen. Meanwhile, the IFN-γ expression mediated by CD4 was dramatically increased [7,8].
Similarly, Mycobacterium bovis is an intracellular bacterial pathogen inducing high levels of IFN-γ in PB cells when animals are stimulated by tuberculin [9]. Therefore, the secretion of IFN-γ is an indicator for Mycobacterium infection, and was used to establish an IGRA, which was subsequently applied clinically for the detection of bovine tuberculosis [10]. At the present time, a commercial kit based on the detection of IFN-γ expression for diagnosis of Mycobacterium bovis is available.
IFN-γ is an indictor cytokine in the infection process of Brucella and Mycobacterium. The practical value has been confirmed for detection of Mycobacterium. The applicability of the IGRA test for detection of brucellosis will be investigated.
This aim of this study was to screen B. abortus antigens capable of stimulating PB cells to secret IFNγ. The final goal was to establish an IGRA based on alterations in IFN-γ expression for the detection of B. abortus. To do this, four somatic antigens, including B. abortus 2308, B. melitensis 28, B. suis 1330 and B. suis 2 (vaccine strain), were prepared and used to stimulate PB cells collected from healthy and brucellosis-positive and -negative cattle. The stimulation of IFN-γ by the four antigens was compared and the optimum antigen was selected and used to establish the IGRA. Finally, clinical samples were screened using the IGRA to evaluate the developed method.

Preparation and inactivation of somatic antigens
Four B. abortus strains, including B. abortus 2308, B. melitensis M28，B. suis S1330 and B. suis S2, were purchased from the China Veterinary Culture Collection Center (CVCC) and were grown in Tryptic Soy Broth media (Difco, Sparks, USA) at 37⁰C with agitation (150 rpm) for three to four days. The bacteria were harvested and the cell pellet was re-suspended in phenolic-normal saline (0.9% NaCl buffer, 0.5% phenolic), and then inactivated at 80⁰C for 30 minutes. Incubation of the inactivated B. abortus for three days on agar plates confirmed that the inactivation was successful.

Sample collection
The bovine peripheral bloods were provided by the Dairy Cow Center, Shandong Academy of Agricultural Science, China. The Brucella-positive bloods, verified by RBT, were collected from cattle without vaccination. Brucella-negative bloods were collected from healthy cattle at the Dairy Cow Center, Shandong Academy of Agricultural Science, China. Cattle were infected with 1×10 8 CFU of B. abortus, and bloods were collected on day 1, 3, 7, 10, 14, 21, 35 and 42 p.i.

Determination of the optimal concentration of antigen for IGRA
A serial dilution of the antigen was prepared and the concentration was determined, as previously described in the methodology. Subsequently, the PB cells collected from cattle were added to a 24-well plate with a volume of 1mL blood per well. The blood cells were stimulated by different concentrations of the aforementioned four somatic antigens, ranging from 1×10 5 CFU/100μl to 1×10 8 CFU/100μl of the final concentration. The blood cells were incubated for 18-24 h at 37⁰C, and the culture supernatant was collected and tested for the expression of IFN-γ using the bovine IFN-γ indirect ELISA test kit (Bovigam 2G, Prionics, 63330, Schlieren-Zurich, Switzerland), according to the supplier's instructions. Briefly, the culture supernatant was diluted in sample dilution buffer in a ratio of 1:1 (supernatant:buffer), and was added into the antigencoated plate in a volume of 100 µl per well. The plates were incubated at room temperature (25⁰C) for 60 minutes and then washed 4 times, and 100 µl of conjugate reagent was added to each well, and then incubated for 60 minutes at room temperature. Plates were washed 4 times, and 100 µl of substrate solution was added to each well. After incubation for 30 minutes in a dark room at room temperature, 50 µl of stop solution was added. Reaction index (RI) values were measured using an automatic microplate reader. An RI higher than 0.8 was considered positive and an RI less than 0.2 was considered negative.

Evaluation of IFN-γ expression and Brucella antibodies
The level of IFN-γ expression and antibodies specific for Brucella were measured. The detection of IFN-γ is described above. The antibodies were detected by indirect ELISA based on B. abortus lipopolysaccharides (LPS). Briefly, a flat-bottomed 96well polystyrene micro-titer plate (Nunc, DK4000, Roskilde, Denmark) was coated with 100 µl of LPS (10 µg/mL), from B. suis biotype 2, suspended in 0.05 mM sodium bicarbonate buffer (pH 9.6), and each plate was incubated at 4°C for 18 h. Plates were washed three times with phosphate buffered saline (PBS) containing 0.05% Tween 20 (PBST [pH 7.4]) and then blocked with 5% gelatin from cold water fish skin (Sigma, St.Louis, MO 63103, USA) at 4°C for 16-18 h. After washing three times with PBST, serum samples were diluted in PBST in a ratio of 1:50 and added to the LPScoated wells. The plates were incubated for 30 minutes at 37⁰C. For detection of IgG, 100 µl of a 1:10,000 dilution of rabbit anti-bovine IgG peroxidaseconjugated antibodies (A5295; Sigma, St. Louis, USA) was added to each well. Each plate was incubated for 30 minutes at 37°C and then washed twice with PBST. Substrate solution (100µl) containing 3, 3', 5, 5'tetramethylbenzidine was added to each well and incubated for 15 minutes in a dark room at room temperature. The reaction was stopped by adding of 100 µl 1M HCl to each well. The optical density (OD)450 values were measured using an automatic microplate reader (SPECTRA MAX190, Molecular Devices LLC., Sunnyvale, USA). The percentage of inhibition values were calculated using the following formula: P%=OD450nm sample value/OD450nm positive control×100%. The recommended cutoff value was 20% for IgG. Samples with a P% equivalent to or higher than 20% were considered to have specific antibodies against Brucella LPS.

Detection of IFN-γ in clinical samples
The clinical bovine PB from Brucella-infected, vaccinated or negative control cattle were stimulated respectively with the optimal antigen concentration determined above; a total of 24 samples were collected from three groups and used subsequently to verify the reliability of the IGRA method. The IGRA approach was assessed by comparison of the level of IFN-γ expression. Samples were incubated at 37°C for 16-24 hours, the supernatant collected and the level of IFN-γ release was detected as described above.

Statistical analysis
The level of IFN-γ expression in cattle was analyzed for statistical significance using a repeated measures t test (GraphPad 5.0). A p-value of < 0.05 was considered significant.

Ethics statement
All experiments were performed according to the Regulations for the Administration of Affairs Concerning Experimental Animals published by the Ministry of Science and Technology, China in 2004 and approved by the Animal Care and Use Committee from the China Institute of Veterinary Drug Control, Beijing, China.

Comparison of IFN-γ secretion induced by different B. abortus antigens
In an effort to establish an IGRA for the diagnosis of brucellosis, four Brucella strains, including B.
abortus 2308, B. melitensis M28，B. suis S1330 and B. suis S2 were selected. These strains were derived from B. abortus, B. melitensis and B. suis which are common across the world. The B. suis S2 is widely used to control and prevent brucellosis in China. The optimal concentration of these four somatic antigens was first determined. Our results demonstrate that the concentration of Brucella antigens was 2.5×10 9 CFU/mL when the optical density value of OD600 was 1.0. The somatic antigens were diluted to a concentration of 1×10 9 CFU/mL before use. Subsequently, the antigens were used to stimulate PB cells collected from brucellosis-negative and -positive cattle for secretion of IFN-γ. The stimulatory ability of antigens was represented by their RI, which was indicative of IFN-γ production. As shown in Figure 1, the four somatic antigens were able to induce production of IFN-γ, albeit the level of IFN-γ induced varied across the four antigen groups. As expected, the results indicated that, overall, the level of IFN-γ secretion was dependent on the concentration of antigen, as higher RI values were obtained when higher concentrations of antigens were used (Figure 1). For blood cells collected from Brucella-positive samples (PS), the RIs in A2308-or S2-stimulation groups were approximately 2-fold higher than that of S1330-or M28-stimulation groups when the same concentration of antigen was used (Figure 1, compare 1A and1B to 2A and 2B), indicating that Brucella strains A2308 and S2 were able to induce higher level of IFN-γ, compared to S1330 and M28. Significantly, when blood cells from Brucella-negative samples (NS) were stimulated by the four antigens, all RIs detected in the four groups were significantly lower than those detected in the Brucellapositive samples (PS) (Figure 1, compare PS and NS in each panel). Furthermore, for NS samples, the RIs detected in S1330 and M28 groups were slightly higher than those detected in A2308 and S2 groups (Figure 1, compare C and D panels with A and B panels). This result is contrasts with the results obtained in the PS groups where RI values in A2308 and S2 groups were higher than those of S1330 and M28 groups. As shown in Figure 1, when 1×10 8 CFU/mL of antigen was used, no significant difference in IFN-γ expression was detected among the four groups ( Figure 1). In order to avoid false positives and reduce antigen loading, a concentration of 1×10 7 CFU/mL antigen was selected for stimulation.

Detection of IFN-γ expression induced by B. abortus 2308
Currently, the majority of methods for brucellosis detection are based on serological reactions. In order to evaluate the level of IFN-γ expression and verify the effect induced by stimulation, four cattle were infected with B. abortus 2308. The results revealed that IFN-γ was detectable on day 5 p.i. (RI>0.2, Figure 2). The level of IFN-γ expression increased rapidly over time, and peaked on day 14 p.i (up to 1.7) before decreasing slowly over time. However, IFN-γ was still detectable on day 42 p.i. (Figure 2). This result suggested that IFNγ was secreted at an early time from PB cells of cattle after B. abortus infection, and the secretion of IFN-γ lasted as long as 42 days.

Detection of antibody produced after B. abortus 2308 infection
Contrary to IFN-γ, the antibody against Brucella was detectable (value higher than the cutoff) at a later time after infection with the same B. abortus strain 2308 IS represents infected sample. The horizontal dashed line represents the cut-off value (RI > 0.2), above the threshold, IFN-γ was considered present in samples.
( Figure 3). The results showed that the earliest time the antibody could be detected was 21 days p.i.. However, it increased over time and peaked on day 42 p.i. (Figure  3).

Evaluation and application of IGRA for clinical samples
The effect of IGRA for clinical samples, which were infected, vaccinated and non-infected with B. abortus was evaluated. A large amount of IFN-γ was produced in the PB cells from infected samples, as illustrated by the high RI value (averaged to 1.0), compared to that of the vaccinated group (with averaged RI value 0.8) (p < 0.01, as shown in Figure 4). As expected, IFN-γ was minimal in the negative control group, where PB cells were collected from healthy and non-vaccinated cattle (Figure 4).

Discussion
The role of IFN-γ in the host immune response is diverse, and the importance of this cytokine in the innate immune defense has been well documented [11]. Suboptimal concentration of IFN-γ does not trigger immune cells, instead, it primes for subsequent response upon stimulation, but, in excess, can eventually cause deleterious consequences. This IFN-γ priming effect has been increasingly implicated in the immune response to several infectious diseases, including viral, bacterial, and parasitic diseases [12][13][14]. For brucellosis, the crucial role of IFN-γ was recognized, as it was shown that IFN-γ knockout mice died due to B. abortus infection and IFN-γ producing CD4+ T-cells from infected donors were able to protect the recipient mice against challenge with B. abortus [15]. IFN-γ was induced by reactive nitrogen intermediates and reactive oxygen intermediates after mouse macrophages were activated [16]. In addition, IFN-γ persists throughout the whole infection process and is the mechanism by which hosts resisted B. abortus infection.
Brucella is an intracellular bacterium, and the cellular immune response is the major mechanism for the host to counteract B. abortus [17]. However, most of the diagnostic methods available were developed based on the humoral immune responses [18]. However, in the later stage of vaccination or the early period of B. abortus infection, antibodies induced by B. abortus infection were undetectable. Therefore, it is unlikely that brucellosis antibodies can be detected using serological methods in the early stages of infection. IFN-γ is a significant cytokine that plays a critical role in the host immune response after infection with B. abortus, hence a diagnostic method for B. abortus detection based on IFN-γ expression has potential.
To develop an IGRA for detecting B. abortus, in this study four somatic B. abortus antigens were evaluated and the antigen 2308 was optimal, and was subsequently used as stimulator for inducing IFN-γ. An IGRA has been established based on detecting IFN-γ expression. It has remained unclear why B. abortus was capable of stimulating host cells to induce a higher level of IFN-γ. This might be related to LPS or some nonstructural proteins. In addition, the concentration of somatic antigens was closely related to IFN-γ expression, nevertheless, we demonstrated that the level  Significant differences between groups are indicated by *p < 0.05, **p < 0.01, ***p < 0.0001 (t test). The vertical bars represent a 95% confidence interval. of IFN-γ expression remained unaltered when the concentration of B. abortus was higher than 1×10 7 CFU/mL. Thus, we chose 1×10 7 CFU/mL as the optimal concentration of B. abortus 2308 for stimulation and the development of an IGRA.
Intriguingly, there was a significant difference in the initiation of the host cellular immune response and humoral response. This study showed that the host cellular immune response initiated earlier than the humoral response (compare Figure 2 and Figure 3). Some researchers have reported that the level of antibodies, IgG1 subtype in particular, peaked 28-42 days after vaccination with B. abortus S19 [19]. However, in this study, for infected individuals, the level of induced antibodies continued to increase, but in the acute infection period, the antibodies were not detected. This indicated that detection methods based on the humoral response have some limitations. Comparatively, IFN-γ secreted in the period of acute B. abortus infection peaked as early as 14 days p.i. Therefore, IGRA can be used to detect B. abortus at an early stage, including acute infection.
In addition, an interesting finding in this study was that the level of IFN-γ expression in the infected group was significantly higher than the vaccinated group ( Figure 4). The explanation for this could be that: 1) the virulent B. abortus wild type strain replicated more efficiently than that of the vaccine strain; 2) The wild type strain persisted for a longer time in the host cell, resulting in a stronger immune response, compared to the vaccine strain. This biological property of B. abortus facilitates the detection of IFN-γ induced in the host cell after infection, leading to the establishment of IGRA, as described in the present study. It should be noted that the host immune response induced by B. abortus is weaker than that induced by Mycobacterium bovis. Therefore, the bovine IFN-γ indirect ELISA test kit, which was developed for Mycobacterium bovis detection, might not be sensitive enough for detection of B. abortus detection. As there was no commercially available kit, we used it as an alternative. There is an urgent need to develop a reliable approach for clinical detection of brucellosis. The IGRA established in this study provides a robust method for diagnosis of brucellosis. Additionally, our IGRA could serve as a potential method to differentiate B. abortus from natural infection and vaccination.

Conclusion
B. abortus 2308 somatic antigen is an ideal stimulator for induction of IFN-γ, and the optimal concentration of 2308 for stimulation is 1.0×10 7 CFU/mL. IFN-γ was secreted on day 5 p.i. and peaked on day 14 p.i.. The level of IFN-γ expression in the Brucella-infected group was highest, followed by the vaccinated group. IFN-γ expression was not detected in the Brucella-negative group.