Specificity and sensitivity of an indirect fluorescent antibody test to detect antibodies against bovine viral diarrhea virus

Abstract

Since being reported in 1979 and 2006, indirect fluorescent antibody (IFA) tests have not been reported to detect bovine viral diarrhea virus (BVDV) antibodies to our knowledge. Thus, we re-evaluated the efficacy and usefulness of IFA tests for BVDV serology. We tested 4 combinations of 2 antibody conjugates (fluorescein isothiocyanate [FITC]-conjugated rabbit IgG anti-bovine IgG; rabbit IgG F(ab')₂ fragment anti-bovine IgG [F(ab')₂ FITC-IgG]) and 2 washing solutions (PBS; carbonate-bicarbonate–buffered saline [CBBS]) to evaluate the specificity of an IFA test for BVDV. We compared the sensitivity of the optimal combination with virus neutralization (VN) tests and an ELISA, and compared IFA with VN titers against different genotype (subgenotype) strains. For the F(ab')₂ FITC-IgG/CBBS combination, only 1 of the 156 (0.6%) 4-fold diluted cattle sera resulted in a nonspecific reaction; other combinations led to a much higher incidence (22.9–37.2%). For the F(ab')₂ FITC-IgG/CBBS combination, IFA detection rates were identical (36 of 59) for BVDV1 and BVDV2 genotypes, and IFA titers against them were strongly correlated (r = 0.99). The antibody-detection rates of the IFA tests were almost identical to those of VN tests and the ELISA (κ: 0.96 and 0.89, respectively). The IFA titers against 4 strains (BVDV1a, BVDV1j, BVDV2a, and an unidentified strain) were similar, 1,024 to ≥4,096, although the VN titers were different. Thus, our IFA tests were specific and sensitive, and more useful than VN tests given that the IFA tests could evaluate the immune status of cattle using a representative strain, regardless of genotype (subgenotype).

Bovine viral diarrhea virus (BVDV; Flaviviridae) infection causes a notable viral disease that is responsible for serious economic losses in the cattle industry in Japan and other countries. BVDV is classified into 2 genotypes (BVDV1, Pestivirus bovis; BVDV2, Pestivirus tauri) based on the nucleotide sequence of the 5ʹ untranslated region.^20,24 BVDV1 is classified into at least 22 subgenotypes (a–v) and BVDV2 into 4 subgenotypes (a–d).^4,19 The antigenic diversity of intergenotypes and even intersubgenotypes has been well documented by virus neutralization (VN) tests.^1,14,25 BVDV strains are also divided into 2 distinct biotypes, noncytopathic (ncp) or cytopathic (cp), based on their growth in cell culture.^2,5

In calves persistently infected (PI) with ncp BVDV, lifelong excretion of a large amount of virus occurs because of immunotolerance to BVDV, and this excretion is the major reservoir for viral transmission within and between herds.^6,11,22 Therefore, the early detection and removal of PI cattle from herds are critical preventive measures against BVDV infection. Although PI cattle can be identified within a herd by virus isolation and the detection of viral antigens and genes, ¹⁸ it is not feasible to examine a large number of herds using these methods. Therefore, a practical approach to herd screening is important for identifying herds with PI cattle and then removing the infected cattle. In Japan, a reverse-transcription PCR assay to detect BVDV RNA has been used to screen bulk tank milk and identify dairy herds with PI cattle. ¹⁰ However, bulk milk examination is not possible for dry cows, beef cattle, heifers, or calves.

Serologic tests, such as VN tests and ELISAs, can evaluate the immune status of all types of cattle irrespective of breed or age; these test types are currently used to detect BVDV antibodies. ¹⁸ The VN test is widely considered the gold standard for BVDV detection because of its high specificity and sensitivity; however, the setup is laborious, and the test is time-consuming. Although indirect fluorescent antibody (IFA) tests are widely applied for the detection of antibodies to other viruses, unexpectedly the use of IFA tests is not commonly recommended to detect BVDV antibodies. ¹⁸ We retrieved the reports of the use of IFA tests to detect BVDV antibodies in a search of Google Scholar, PubMed, and Scopus, using search terms “bovine viral diarrhea virus and indirect fluorescent antibody test or indirect immunofluorescence assay.” We found 6 available reports published between 1979 and 2000 and one PhD thesis in 2006 on the use of IFA tests for BVDV serology in cattle.^{3,8,12,13,15,16,23} The use of IFA tests has not been reported since 2006.

The serologic categorization of a small number of young unvaccinated stock based on neutralizing antibody titers has been used to identify herds with PI cattle.^7,21,27 However, the selection of the challenge virus used in the VN test is critical because of the antigenic variation between BVDV1 and BVDV2 or among different subgenotypes.^21,28 Therefore, if the antigenicity of the challenge virus used in the VN test is fundamentally different from the virus circulating in the herd under test, false-negative results may be obtained.^21,28

We re-evaluated whether the IFA test, which we modified from earlier studies,^{8,12,13,15,23} is a suitable serologic test to detect BVDV antibodies, and then compared our test with the VN test and an ELISA.

Materials and methods

Cattle sera

We studied 156 serum samples from cattle of different ages. Sixty samples were collected from cattle aged 6 d to 5 y that were brought to Obihiro University of Agriculture and Veterinary Medicine for pathology diagnosis and were kindly provided by the Section of Large Animal Clinical Sciences, Division of Clinical Veterinary Medicine, Department of Veterinary Medicine. The remaining 96 serum samples, collected from healthy adult cattle, were kindly provided by the Hokkaido Branch of the National Institute of Animal Health, Japan. All sera were inactivated at 56°C for 30 min.

Virus

BVDV strains KS86-1CP (BVDV1j)^14,29 and Nose (BVDV1a)^9,14 were kindly provided by the National Institute of Animal Health, Japan; TC89-NCP (BVDV2) by Tochigi Prefectural Central Livestock Hygiene Service Center, Japan; and KZ91-CP (BVDV2a) ¹⁷ by Ishikawa Nanbu Livestock Hygiene Service Center, Japan. Virus titers were quantified by the TCID₅₀ value using the Behrens–Kärber method.

Cell culture

The MDBK-SY cell line ²⁶ was kindly provided by Tochigi Prefectural Central Livestock Hygiene Service Center, Japan. The cells were cultured in Dulbecco modified Eagle medium supplemented with 2 mM L-glutamine and 5% fetal bovine serum (FBS) free of anti-BVDV antibodies. Both the ncp and cp BVDV strains are reported to induce a cytopathic effect (CPE) in MDBK-SY cells. ²⁶

Fluorescent antigen preparation

MDBK-SY cells that were inoculated with BVDV at a multiplicity of infection of 0.01 were incubated for 48 h to obtain 100% infected cells and harvested using a trypsin–EDTA solution. IFA tests using BVDV IFA positive control serum (VMRD) confirmed that almost all of the inoculated cells were infected with BVDV (data not shown). Uninfected cells were cultured and harvested using the same procedure.

The infected and uninfected cells were washed with PBS (pH 7.4) and mixed in a 1:1 ratio. Then, the concentration of the mixture was adjusted to 1 × 10⁷ cells/mL, and the mixed cells were smeared on the wells of multiwell glass slides (10 wells, 5-mm diameter; Immuno-Cell) using 1–2 μL of the mixture per well. After drying at ambient temperatures, the cell mixtures were fixed with cold acetone for 10 min (BVDV-positive cells); ~50% of these cells were confirmed positive for BVDV using BVDV IFA positive control serum as described below. Wells containing uninfected cells (BVDV-negative cells) were analyzed in the same manner. All slides were stored at −80°C until use.

IFA test

Briefly, the wells on glass slides containing BVDV-positive or -negative cells were blocked with PBS (pH 7.4) containing 1% normal rabbit serum (PBS-NRS) at 37°C for 10 min, and then incubated with test sera diluted 4-fold with PBS-NRS at 37°C for 30 min. To determine the IFA titers of the test sera, we performed 2-fold serial dilutions starting at 1:4. After incubation, the slides were washed in washing solution (PBS or carbonate-bicarbonate–buffered saline [CBBS; pH 9.0] comprised of 26.9 mM Na₂CO₃, 100 mM NaHCO₃, and 36.4 mM NaCl) for 10 min, incubated with fluorescein isothiocyanate (FITC)-conjugated secondary antibody (conjugate) diluted with PBS-NRS at 37°C for 30 min as described below, and washed again. Fluorescence signals were observed (BZ-9000 fluorescence microscope; Keyence).

The reaction was considered nonspecific if fluorescence was observed in the cytoplasm of almost all BVDV-positive cells or BVDV-negative cells; the reaction was considered specific (antibody-positive) if fluorescence was observed in the cytoplasm of ~50% of BVDV-positive cells.

FITC-conjugated rabbit IgG anti-bovine IgG antibody (whole FITC-IgG; MP Biomedicals Japan) and FITC-conjugated rabbit IgG F(ab')₂ fragment anti-bovine IgG antibody (F(ab')₂ FITC-IgG; Abcam) were used as conjugates. Four-fold dilutions of the 156 serum samples were analyzed for the occurrence of nonspecific reaction in the following 4 combinations of conjugate and washing solution: whole FITC-IgG and PBS; whole FITC-IgG and CBBS; F(ab')₂ FITC-IgG and PBS; and F(ab')₂ FITC-IgG and CBBS (Table 1).

Table 1.

Incidence of nonspecific reactions using different combinations of conjugate and washing solution in indirect fluorescent antibody tests for bovine viral diarrhea virus (BVDV) antibodies.

BVDV status of cells	Incidence of nonspecific reaction*
	Combination of conjugate and washing solution
	Whole FITC-IgG + PBS	Whole FITC-IgG + CBBS	F(abʹ)2FITC-IgG + PBS	F(abʹ)2 FITC-IgG + CBBS
Negative	43/96‡	32/96	22/96	0/96
Positive†	15/60	NT	NT	1/60
Total	58/156 (37.2%)a 95% CI: 29.9–44.9	32/96 (33.3%)a,b 95% CI: 24.6–43.2	22/96 (22.9%)b 95% CI: 15.5–32.3	1/156 (0.6%)c 95% CI: −0.2–3.9

CBBS = carbonate-bicarbonate–buffered saline; FITC = fluorescein isothiocyanate; NT = not tested. Data followed by different superscript letters are significantly different between combinations (p < 0.05).

^*Four-fold diluted sera were used. When fluorescence was observed in the cytoplasm of almost all BVDV-negative and/or -positive cells, the reaction was considered nonspecific.

^†BVDV-positive cells were prepared using strain KS86-1CP.

^‡Number of nonspecific reactions/total number of sera tested; 96 sera were chosen from the serum stock preserved in our laboratory, and 60 sera were collected from cattle introduced to our university (Obihiro University of Agriculture and Veterinary Medicine, Inada, Obihiro, Hokkaido, Japan) for pathology diagnosis.

VN test

VN tests were performed using microtitration. Briefly, 4-fold diluted serum was mixed with an equal volume of the virus (200 TCID₅₀/0.1 mL) and then incubated at 37°C for 1 h. MDBK-SY cells cultured in a 96-well flat-bottomed microplate were inoculated with 0.1 mL/well of the mixture and incubated for 1 h. Next, the culture medium was replaced with a maintenance medium containing 0.5% FBS, and the plates were incubated for 7 d. Test sera that neutralized 50% of viral growth were considered antibody-positive.

For antibody titration, we performed 2-fold serial dilutions of the test sera starting at 1:4. As preliminary tests revealed that the CPE of strain TC89-NCP was somewhat unclear, the mixtures were inoculated into 8-chamber slides (Thermo Fisher) instead of 96-well microplates, and we determined the VN titers by IFA tests with the positive control serum, as described above. The titer of the test serum was the reciprocal of the highest dilution of serum that neutralized viral growth by 50%.

ELISA

We used a commercial ELISA kit (BVDV ELISA kit; Bio-X Diagnostics) per the manufacturer’s instructions to detect BVDV antibodies.

Statistical analysis

The Fisher exact test was used to analyze the frequency of nonspecific reaction in the IFA test of 4 combinations of conjugates and washing solutions using statistical computation (js-STAR_XR+ v.1.9.6, https://www.kisnet.or.jp/nappa/software/star/freq/2x2.htm). The 95% CI was calculated using statistical computation methods (Keisan, https://keisan.casio.jp/exec/user/1490184062).

We calculated the κ coefficient to evaluate the disagreement between IFA and VN tests or between IFA tests and the ELISA. ³⁰ We calculated the Pearson correlation coefficient to evaluate the correlation in FA titers against BVDV1 and BVDV2 (Excel 2016; Microsoft).

Results

When whole FITC-IgG conjugate was used, we observed a high incidence of nonspecific reactions with both BVDV-negative (33.3%) and/or BVDV-positive (37.2%) cells, regardless of the washing solution used (Fig. 1A, 1B). Although a high incidence of nonspecific reactions was observed, it was significantly lower with the combination of F(ab')₂ FITC-IgG and PBS than with the combination of whole FITC-IgG and PBS (22.9% vs. 37.2%, respectively; p < 0.05).

Figure 1.

Representative fluorescence images of nonspecific and specific reactions of cattle serum in the indirect fluorescent antibody test. A. Nonspecific reaction of 4-fold diluted serum with bovine viral diarrhea virus (BVDV)-negative cells using the combination of whole FITC-IgG conjugate with PBS as the washing solution. All uninfected cells had cytoplasmic fluorescence. B. Nonspecific reaction of serum with virus neutralization antibodies in BVDV-positive cells using the combination of whole FITC-IgG conjugate with PBS washing solution. All cells had cytoplasmic fluorescence. C. Specific reaction of serum with BVDV-negative cells using the combination of F(ab′)₂ FITC-IgG conjugate with carbonate-bicarbonate–buffered saline (CBBS) as the washing solution. None of the uninfected cells fluoresced. D. Specific reaction of the same antibody-positive serum with BVDV-positive cells using the combination of F(ab′)₂ FITC-IgG conjugate with CBBS as the washing solution. Almost half the cells possessed specific cytoplasmic antigens because the BVDV-positive cells were prepared by mixing infected cells with uninfected cells in a 1:1 ratio.

In contrast, when F(ab')₂ FITC-IgG was used in combination with CBBS as the washing solution, nonspecific reactions were not observed in either BVDV-negative cells (Fig. 1C) or BVDV-positive cells (with the exception of one sample with cytoplasmic fluorescence in almost all cells). When sera containing BVDV antibodies were tested using BVDV-positive cells, the reaction was considered specific because ~50% of the cells had cytoplasmic fluorescence (Fig. 1D). The combination of the F(ab')₂ FITC-IgG conjugate with CBBS had a significantly lower frequency of nonspecific reactions than the other combinations (p < 0.01). For subsequent experiments, we therefore used F(ab')₂ FITC-IgG as the conjugate and CBBS as the washing solution.

Of the 60 serum samples collected from cattle that were introduced to our university for pathology diagnosis, only 1 had a nonspecific reaction with the combination of F(abʹ)₂ FITC-IgG with CBBS (Table 1). The IFA detection rates agreed between BVDV1 (KS86-1CP)-positive and BVDV2 (TC89-NCP)-positive cells (36 of 59; 61%). Additionally, IFA titers against both BVDV1 and BVDV2 were strongly positively correlated (r = 0.99; Fig. 2).

Figure 2.

Comparison of fluorescent antibody (FA) titers against bovine viral diarrhea virus (BVDV)1-positive and BVDV2-positive cells. The BVDV1 strain KS86-1CP and the BVDV2 strain TC89-NCP were used to prepare BVDV-positive cells for the indirect FA tests (n = 15).

The antibody-detection rates between IFA and VN tests using the KS86-1CP strain and between IFA tests and the ELISA in 59 sera were in almost-perfect agreement (κ: 0.96 and 0.89, respectively; Tables 2, 3).

Table 2.

Agreement of bovine viral diarrhea virus (BVDV) antibody-detection rates between indirect fluorescent antibody (IFA) and virus neutralization (VN) tests.

IFA test*	VN test		Total	κ
	+	−
+	35	1	36
−	0	23	23	0.96
Total	35	24	59

⁺= antibody-positive; − = antibody-negative. Four-fold dilutions of 59 of the 60 sera. One sample had a nonspecific reaction and was excluded (see Table 1).

^*BVDV-positive cells were prepared using strain KS86-1CP. The same viral strain was used as the challenge virus in the VN test.

Table 3.

Agreement of bovine viral diarrhea virus (BVDV) antibody detection rates between the indirect fluorescent antibody (IFA) test and ELISA.

IFA test*	ELISA		Total	κ
	+	−
+	34	2	36
−	1	22	23	0.89
Total	35	24	59

⁺= antibody-positive; − = antibody-negative. Four-fold dilutions of 59 sera were used in IFA tests, and 100-fold dilutions were used in ELISAs per the manufacturer’s instructions.

^*BVDV-positive cells were prepared using strain KS86-1CP, as detailed in the Materials and methods section.

We compared the IFA and VN titers of 8 sera selected from the 60 samples with antibodies against BVDV1 (Nose, 1a; KS86-1CP, 1j) and BVDV2 (KZ91-CP, 2a; TC89-NCP; Table 4). The IFA titers of the 4 BVDV strains were similar, irrespective of genotype or subgenotype, although there were differences in the VN titers between different genotypes (subgenotypes).

Table 4.

Comparison of antibody titer to bovine viral diarrhea virus (BVDV) strains in virus neutralization (VN) and indirect fluorescent antibody (IFA) tests.

Serum sample	Antibody titer to BVDV1 strain				Antibody titer to BVDV2 strain
	KS86-1CP (1j)		Nose (1a)		KZ91-CP (2a)		TC89-NCP
	VN	IFA	VN	IFA	VN	IFA	VN	IFA
1	256	2,048	1,448	2,048	64	2,048	32	2,048
2	256	2,048	1,448	2,048	64	1,024	23	1,024
						2,048 (Q)
3	512	≥4,096	724	≥4,096	≥5,792	≥4,096	1,448	≥4,096
4	362	≥4,096	512	≥4,096	≥5,792	≥4,096	≥5,792	≥4,096
5	362	≥4,096	≥5,792	≥4,096	362	2,048	362	≥4,096
						4,096 (Q)
6	2,048	≥4,096	≥5,792	≥4,096	1,024	≥4,096	256	≥4,096
7	2,048	2,048	≥5,792	≥4,096	2,048	2,048	1,448	≥4,096
		4,096 (Q)				4,096 (Q)
8	512	≥4,096	2,896	≥4,096	64	≥4,096	91	≥4,096

Q = questionable reaction.

Discussion

Between 1979 and 2006, there were at least 7 reports of IFA tests for BVDV antibody detection. In these reports, whole FITC-IgG conjugate with PBS as the washing solution was used (2 studies did not describe IFA procedures).^{3,8,12,13,15,16,23} However, background fluorescence of negative and/or positive sera was observed in BVDV-negative or -positive cells.^13,15,23 Similarly, when we used whole FITC-IgG conjugate and PBS as the washing solution for our IFA tests with 4-fold diluted sera, the frequency of nonspecific reactions was also high. In VN tests, serial 2-fold dilutions of the test sera from a starting dilution of 1:4 are recommended to examine the BVDV antibody level. ¹⁸ Therefore, we used the same serum dilution in IFA tests when comparing specificity and sensitivity with VN tests. In contrast, when we used F(ab')₂ FITC-IgG conjugate and CBBS as the washing solution, a nonspecific reaction occurred in only one sample. Thus, the result may indicate that our IFA tests are specific and able to detect low antibody levels in cattle sera similarly to VN tests. It is unclear why the combination of F(ab')₂ FITC-IgG conjugate and CBBS significantly reduced the frequency of nonspecific reactions compared with the combination of whole FITC-IgG conjugate with PBS, which is generally used in IFA tests.

The antigenic diversity between BVDV1 and BVDV2, as well as among subgenotypes, has been well documented by VN tests.^1,14,25 However, the high correlation between the IFA titers that we noted suggests that antigenic diversity among BVDV1 and BVDV2 strains was not apparent from results of the IFA test. Thus, antigenic diversity among strains exists, and the IFA test was not affected by it. This result was confirmed because the IFA titers against 4 strains derived from different genotypes or subgenotypes were similar, although the VN titers were not. We found almost perfect agreement between IFA and VN detection rates and IFA and ELISA detection rates.

The IFA test was highly specific and sensitive for BVDV detection. Moreover, the IFA test is more useful than the VN test because it is simple and rapid, usually with same-day turnaround, which would enable a fast evaluation of the immune status of cattle herds.

Two of the most effective BVDV control measures are the detection and exclusion of PI cattle from herds. A study ⁷ demonstrated that herds with PI cattle could be differentiated from herds without PI cattle by categorizing only 5 randomly selected unvaccinated 9–18-mo-old cattle as having VN titers of ≥128 or ≤64, respectively. The usefulness of this screening test has been confirmed by other research groups, although the number and age of the cattle examined or the cutoff antibody titers were slightly different.^21,27 However, accurate VN titers may not be obtained owing to the antigenic variation among BVDV strains if the genotype or subgenotype of the challenge viruses used in the VN test differ from the BVDVs that have recently invaded or are circulating in the target cattle herds; in this situation, false-negative results may occur.^21,28 Therefore, correctly selecting the challenge viruses used in the VN tests is critical. In one study, ²⁸ using several genotypes was recommended, including those mainly circulating within the country where the test is being performed, as challenge viruses in the VN test to avoid herd misclassification resulting from antibody titer levels. It is usually not easy to define the genotypes and subgenotypes to which cattle herds were exposed before examination. As mentioned above, the IFA test that we used measures antibody titers using only one representative strain, regardless of BVDV genotype (subgenotype); thus, the IFA test appears to be useful as a screening test to detect herds with PI cattle.

According to the manufacturer, the ELISA kit that we used detects antibodies to the common BVDV viral antigen NS3. Given that our results indicated almost perfect agreement in the antibody-detection rates between the IFA test and the ELISA, the IFA test most likely also detects antibodies to common antigens, such as conserved nonstructural proteins NS2 and NS3, as well as envelope antigens. This may be one of the reasons why we found similar IFA titers against different strains of BVDV1 and BVDV2.