Pilot Study of Immunoblots with Recombinant Borrelia burgdorferi Antigens for Laboratory Diagnosis of Lyme Disease

Accurate laboratory diagnosis of Lyme disease (Lyme borreliosis), caused by the spirochete Borrelia burgdorferi (BB), is difficult and yet important to prevent serious disease. The US Centers for Disease Control and Prevention (CDC) presently recommends a screening test for serum antibodies followed by confirmation with a more specific Western blot (WB) test to detect IgG and IgM antibodies against antigens in whole cell lysates of BB. Borrelia species related to BB cause tick-borne relapsing fever (TBRF). TBRF is increasingly recognized as a health problem in the US and occurs in areas where Lyme disease is prevalent. The two groups of Borrelia share related antigens. We have developed a modified WB procedure termed the Lyme immunoblots (IBs) using recombinant antigens from common strains and species of the BB sensu lato complex for serological diagnosis of Lyme disease. A reference collection of 178 sera from 26 patients with and 152 patients without Lyme disease were assessed by WB and IB in a blinded manner using either criteria for positive antibody reactions recommended by the CDC or criteria developed in-house. The sensitivity, specificity, positive and negative predictive values obtained with the reference sera suggest that the Lyme IB is superior to the Lyme WB for detection of specific antibodies in Lyme disease. The Lyme IB showed no significant reaction with rabbit antisera produced against two Borrelia species causing TBRF in the US, suggesting that the Lyme IB may be also useful for excluding TBRF.


Introduction
Lyme disease (Lyme borreliosis) is the most common tick-borne disease in North America and Europe and one where accurate diagnosis is difficult and yet important for initiating treatment to prevent progression to more severe disease [1][2][3]. Lyme disease is caused by spirochete bacteria of the Borrelia burgdorferi sensu lato (BBsl) complex. Globally, an increasing number of species are being characterized within the BBsl complex. Borrelia burgdorferi sensu stricto (BBss) is principally responsible for human infections in the US [3,4], with known contributions from B. bissettii [5] and B. mayonii [6]. Borrelia afzelii, B. garinii, and BBss are important causes of human infections in Europe and Asia [7]. Although 26,203 confirmed and 10,226 probable cases of Lyme disease were reported in the US in 2016, it is estimated that the actual number of people diagnosed with Lyme disease is more likely over 300,000, making it the most common vector-borne disease in the country according to the Centers for Disease Control and Prevention (CDC) [8].
Additionally, when whole cell lysate proteins are separated only by size during SDS gel electrophoresis, multiple proteins of similar size may appear to migrate together causing reduced specificity [34].
Purified recombinant antigens are increasingly being used in many laboratories in a variety of test formats, including EIA and WB, to detect IgG and IgM antibodies to BBsl and BBss for diagnosis [37,38]. To overcome some of the drawbacks associated with WB-based diagnosis, we recently developed IgM and IgG WBs that use a panel of purified recombinant protein antigens from several species of the BBsl complex for identifying the presence of antibodies to the different antigens. These WBs that use purified recombinant antigens as targets are termed Lyme Immunoblots (IBs) here to differentiate them from Lyme WBs that use whole cell lysates as target antigens. The recombinant antigens we used in the Lyme IBs represent all the antigens used for characterizing positive WB reactions in the current in-house [34] and CDC guidelines [10,17]. We investigated the comparative diagnostic performance of Lyme IBs and WBs with a set of well-characterized reference patient sera and also the reaction of antisera produced against two common TBRF Borrelia species in Lyme IBs.

Reference Sera
A total of 178 reference serum samples, including 26 sera from patients with confirmed Lyme disease, were tested. Details of the reference serum samples and test results provided by the suppliers are shown in Table 1. Out of 46 samples with antibodies to viruses provided by New York Biologics, 11 had antibodies to cytomegalovirus; 24 to Epstein-Barr virus; seven to herpes simplex virus; and four to hepatitis C virus. Once received, the sera were stored at 4 • C up to one week and at −20 • C for longer storage. Testing of reference sera was performed by laboratory personnel without prior knowledge of the expected results in the same manner as clinical samples from Lyme disease patients. The CDC provided EIA results, details of the bands that reacted in WBs for CDC Set 1 and 2 reference samples, WB interpretation by CDC criteria and two-tiered testing results. These samples were interpreted by in-house criteria too, for positivity as described in Section 2.6. EIAs and WBs were therefore not performed in-house on these 42 CDC samples, and only IBs were performed. On the remaining 136 samples, EIAs, WBs and IBs were performed.

Preparation of Antigen Strips for Lyme Western Blots
Nitrocellulose strips for WB were prepared from whole cell lysates of a mixture of the two BBss strains, B31 and 297, and used for WBs as previously described [34]. Briefly, proteins present in sonicated whole cell lysate were separated by acrylamide gel electrophoresis. The separated proteins were transferred onto nitrocellulose membrane (Amersham Protran, GE Healthcare Life Science). The membranes with bound proteins were washed in deionized water, blocked in 5% non-fat dry milk for one hour, dried and sliced into 3-mm strips.

Preparation of Antigen Strips for Lyme Immunoblots
Recombinant proteins derived from several US and European species of BBsl were used to prepare antigen strips for Lyme IBs. The recombinant proteins selected included all the proteins used in scoring WBs by the CDC and in-house criteria described in Section 2.6 below. P23 (OspC) and P31 (OspA) proteins from several different BBsl species were used as target antigens in the Lyme IB. Separate P39 (BmpA) antigens derived from European and US BBsl species were included in the panel of test antigens. Additionally, the variable surface antigen of BBss (VslE), and a hybrid protein containing the immunodominant region of VslE from different BBsl species termed C6 [39] were also used in the Lyme IBs as target antigens. Recombinant antigens were prepared by cloning the hybrid gene constructs or portions of the selected genes into pET vectors, and then expressing the proteins in Escherichia coli (GenScript, Piscataway, NJ, USA). The E. coli-produced recombinant BBsl proteins were then purified using metal affinity chromatography followed by gel filtration. All the recombinant proteins were >90% pure by Coomassie blue staining after SDS PAGE.
Purified proteins and two control proteins, diluted to yield 7-19 ng of protein as a line in each 3 mm strip of membrane were sprayed in straight lines onto nitrocellulose membrane (Amersham Protran, GE Healthcare Life Science) using a BioDot liquid dispenser (BioDot, Irvine, CA, USA). The two control proteins were Protein L (Sigma, St. Louis, MO, USA) for detecting the addition of human serum (termed serum control), and a mixture of human IgM and IgG (Sigma, St. Louis, MO, USA) for detecting the addition of alkaline phosphatase conjugated anti-human antibodies (termed conjugate control). The membranes were then blocked with 5% non-fat dry milk and sliced into 3 mm wide strips.

Procedure for Detection of Borrelia Specific Antibodies on Lyme Immunoblots and Western Blots with Test Sera
Prior to use, each strip was labeled and then soaked in 1 mL of diluent (100 mM Tris, 0.9% NaCl, 0.1% Tween-20 and 1% non-fat dry milk) for 5 min in a trough. A 10 µL aliquot of the test or control serum was added to a corresponding IB or WB strip in the trough. The strips were then incubated at room temperature for one hour with serum, followed by three washes with wash buffer (KPL, Gaithersburg, MD, USA) at room temperature. After aspirating the final wash solution, strips for detecting IgG and IgM were incubated with alkaline phosphatase-conjugated goat anti-human IgG at 1:10,000 dilution and IgM at 1:6000 dilution respectively (KPL, Gaithersburg, MD, USA) for one hour. After three washes, bands were visualized by reaction with 5-bromo-4-chloro-3-indolylphosphatenitro-blue tetrazolium (BCIP/NBT, KPL, Gaithersburg, MD, USA). The reactions were terminated by washing with distilled water when a calibration control produced a visible band at 39 kDa. Alkaline phosphatase-conjugated rabbit antibody to the 39/93 kDa BBsl antigens (Strategic Biosciences, Stow, MA, USA) diluted in human serum was used as the calibration control as previously described [34]. Bands with lower intensity than the calibration control were reported as negative. The Lyme WB and IB strips were also reacted with a mixture of human sera from patients with confirmed Lyme disease as a positive control and sera from uninfected persons as a negative control.

Scoring of Positive Serological Reactions
The following antigen bands in kDa were scored on the Lyme IB and WB strips: for IgG, 18 [10,17]. However, the CDC guidelines recommend that a positive IgM test is valid only for the first four weeks after infection, i.e., early Lyme disease, because false positive IgM reactions may develop in advanced Lyme disease [10,17]. By the in-house WB criteria, IgG and IgM WBs were considered positive if two from the following six antigens bands were present: 23

Lyme Immunoblots with Rabbit Antisera against Lyme and TBRF Borrelia Species
Rabbit antisera were raised against whole cell lysates of the following BBsl species grown in culture: B. burgdorferi B31, B. burgdorferi 297, B. afzelii, B. garinii, B. californiensis, B. spielmanii and B. valaisiana. Rabbit antisera were also produced against whole cell lysates of the TBRF Borrelia species B. hermsii and B. coriaceae. These different immune rabbit antisera were used in Lyme IBs to investigate the cross-reactions with BBsl antigens. Antigen strips for Lyme IBs were prepared as described in Section 2.4 and reacted with different rabbit antisera essentially as described in Section 2.5 except for the use of individual Borrelia species-or strain-specific rabbit antisera in place of human sera as the primary antibody, followed by alkaline phosphatase-labelled goat anti-rabbit IgG (KPL, Gaithersburg, MD, USA) as the secondary antibody. Two additional strips were reacted with positive and negative control human sera as described in Section 2.5.

Lyme IgM and IgG Immunoblots
Typical Lyme IgM and IgG IBs are illustrated with results from 10 sera in Figure 1. Sample 1 is positive only by the in-house criteria, samples 2-5 are positive by both CDC and in-house criteria and samples 6-10 are negative by both sets of criteria. The positive sera recognize P23 and/or P31 from different BBsl species in the IgG IBs. Three of the five positive sera recognize P39 from both US and European BBsl species in the IgG IBs. Four of the five positive sera also recognized VlsE and C6.

Comparative Reactions of Four Sera in Lyme IgM Western Blots and Lyme IgM Immunoblots
The results of Lyme IgM WB and IgM IB performed on four reference sera are shown in Figure 2. They show that the Lyme IgM IB is more sensitive than Lyme IgM WB with these four samples. Serum samples 1 and 3 are negative by either the CDC or in-house criteria in both Lyme IgM WB and IB. Serum samples 2 and 4 are negative in the Lyme IgM WB but positive in the Lyme IgM IB by both criteria. The IB-positive sera 2 and 4 also reacted with VlsE and C6 in the IBs.

Lyme IgM and IgG Immunoblot and Western Blot Results with Known Lyme Disease-Positive Sera
The results obtained with the 26 known Lyme disease-positive sera among the 178 reference sera tested by Lyme IgM and IgG IBs as well as Lyme IgM and IgG WBs are presented in Table 2. They show that the both IgG and IgM Lyme IBs identified the same number or more positive sera overall than the corresponding IgG and IgM WBs by the in-house and CDC criteria.

Lyme IgM and IgG Immunoblots and Western Blots Detecting Different Stages of Lyme Disease
There were 17 sera from patients with confirmed Lyme disease among the 42 reference CDC samples (Table 1). Of these, the CDC had identified 10 patients with early Lyme disease, four with Lyme arthritis and three with Lyme neuroborreliosis ( Table 3). The WB and IB results suggest that IgG and IgM IBs tend to be as good as or better than the corresponding IgG and IgM WBs for detecting the three stages of Lyme disease by either the in-house or CDC criteria for positivity. A high proportion of early Lyme disease and all three Lyme neuroborreliosis sera had detectable IgM antibodies in both IBs and WBs. When the two-tiered results were compared with Lyme IB read by in-house criteria, the specificity of the Lyme IB was comparable with the two-tiered testing (p = 1.0) and the sensitivity was improved significantly (p = 0.044).

False Positive Reactions on Reference Sera in Lyme IgM and IgG Immunoblots and Western Blots
The results obtained with the 152 reference sera that were known to be negative for Lyme disease (Table 1) when tested by Lyme IBs and WBs are presented in Table 4. The Lyme WBs with in-house criteria detected more false positive reactions with control sera from persons with syphilis and viral infections than the other tests performed, with the majority of these detecting IgM antibodies.

Clinical Parameters of the Lyme Immunoblot and Western Blot Tests with Reference Sera
The results obtained with the 178 reference sera presented in Tables 2 and 4 were used to calculate the sensitivity, specificity, PPV and NPV of serological detection regardless of disease stage with either IgG or IgM antibodies, with IgG antibodies and with IgM antibodies. The parameters obtained are shown in Table 5. The results suggest that because the 95% confidence intervals do not overlap, the specificity of detection with either IgG or IgM antibodies, IgG antibodies or IgM antibodies and PPV with either IgG or IgM antibodies is better in WBs on application of the CDC than the in-house criteria. The findings also show that the IBs tended to perform better than WBs with either in-house or CDC criteria for the detection of Lyme disease. This is particularly the case for specificity and PPV of either IgG or IgM antibodies or IgM antibodies by the in-house criteria where the 95% confidence intervals do not overlap between WBs and IBs.   Neuroborreliosis  3  3  3  3  3  3  3  3  3  3  3  3  3  3  3   Total Positive  17  13  9  12  13  8  8  13  12  14  15  17  12  8  14 CDC provided the samples and the EIA, Western Blot bands, Western Blot interpretation by CDC criteria and two-tiered results.

Comparison of Performance of Lyme IBs with Two-Tiered Testing (Whole-Cell EIA Followed by Confirmation with WBs)
The data (Table 5) demonstrates that the specificity of the Lyme IBs when read by in-house criteria is equivalent to two-tiered testing with WBs (p = 1.0) and the sensitivity is improved with Lyme IBs (p = 0.05). Interestingly with the CDC samples covering the full spectrum of disease, the sensitivity of the Lyme IBs was significantly improved (p = 0.044).

Antisera to TBRF Borrelia species B. hermsii and B. coriaceae Tested with Bbsl Antigens in Lyme IgG Immunoblots
The results of Lyme IgG IBs performed with rabbit antisera raised against different BBsl species and TBRF Borrelia species are shown in Figure 3. All the rabbit anti-BBsl group sera were positive on the Lyme IgG IB whereas the rabbit antisera to the two TBRF Borrelia species, B. hermsii and B. coriaceae, only showed a weak cross-reaction with BBsl P41 (flagellin) in the Lyme IgG IB. Antisera against the different BBsl species showed variable strength of reaction with different BBsl antigens used in the IB. For example, the reaction with P23 (OspC) was more variable than that with P31 (OspA). None of the rabbit anti-BBsl sera reacted with VlsE or C6.

Discussion
The two-tiered serological approach for laboratory diagnosis of Lyme disease [10,17] has continued to be widely used with EIA as the preferred first test followed by WB as the common second test. This two-step procedure was initiated because first-generation EIAs for the detection of anti-Borrelia antibodies lacked specificity. The inclusion of a second, more specific, serological method such as WB made it possible to exclude false-positive EIA samples. The shortcomings and advantages of this approach have been extensively examined [38,41,42]. The main drawbacks identified in such studies were the (i) relatively poor sensitivity in early disease [38,42]; (ii) persistence of antibodies even after the infection is cleared by treatment; (iii) less than optimal specificity of the WB used as a confirmatory test [41]; (iv) variability in sensitivity and specificity of EIAs and WBs employing recombinant target antigens for detection of anti-Borrelia antibodies [40] leading to unsatisfactory concordance in results between different laboratories using similar test formats and (v) possible subjectivity in scoring WB results. Nevertheless, the studies [38,41,42] highlight the continuing usefulness of the two-tiered diagnostic approach, whole-cell lysate EIA, followed by confirmation

Discussion
The two-tiered serological approach for laboratory diagnosis of Lyme disease [10,17] has continued to be widely used with EIA as the preferred first test followed by WB as the common second test. This two-step procedure was initiated because first-generation EIAs for the detection of anti-Borrelia antibodies lacked specificity. The inclusion of a second, more specific, serological method such as WB made it possible to exclude false-positive EIA samples. The shortcomings and advantages of this approach have been extensively examined [38,[41][42][43]. The main drawbacks identified in such studies were the (i) relatively poor sensitivity in early disease [38,42,43]; (ii) persistence of antibodies even after the infection is cleared by treatment; (iii) less than optimal specificity of the WB used as a confirmatory test [41]; (iv) variability in sensitivity and specificity of EIAs and WBs employing recombinant target antigens for detection of anti-Borrelia antibodies [41] leading to unsatisfactory concordance in results between different laboratories using similar test formats and (v) possible subjectivity in scoring WB results. Nevertheless, the studies [38,[41][42][43] highlight the continuing usefulness of the two-tiered diagnostic approach, whole-cell lysate EIA, followed by confirmation with WB (prepared from cell lysate); and using the CDC criteria for positive WBs because of the high specificity compared with other methods including those employing recombinant target antigens in EIA and WB. Our present findings also show that the CDC criteria for WBs yield better PPVs and similar NPVs to the in-house criteria for WBs. However our results, based on procedures to minimize or eliminate subjective assessment, suggest that the sensitivity, specificity, PPV and NPV values tend to be better with the newly developed IBs than WBs by either the CDC or in-house criteria for positivity. In addition, the results demonstrate that the specificity of the Lyme IBs is equivalent to two-tiered testing (using whole cell lysate EIA and WBs) with improved sensitivity ( Table 5).
Seventeen of the 26 reference samples obtained from the CDC were from patients with Lyme disease and of these 10 were from patients with early Lyme disease as classified by the CDC. A high proportion of the 10 early Lyme sera had IgM antibodies that were detectable in WBs and IBs by either the CDC or the in-house criteria for positivity. This is consistent with the early formation of IgM antibodies during infection [10,17]. The nine Lyme positive proficiency test samples from the College of American Pathologists and the New York State Department of Health were not classified according to the stage of the Lyme disease. However all nine samples had IgM antibodies in WBs and IBs using either the CDC or the in-house criteria for positivity.
Because the precise time after infection when many positive reference sera were collected from patients with Lyme disease was not available, the presence of either IgM or IgG antibodies as well as IgM and IgG antibodies alone were separately analyzed for the clinical diagnostic parameters of sensitivity, specificity, PPV and NPV. The results obtained from the 178 reference serum samples show that IBs tend to have in all cases superior clinical diagnostic parameters to WBs for detecting BBsl-specific antibodies by either the current CDC or the in-house developed criteria for positivity. The decrease in sensitivity when only IgG antibodies are considered, compared with either IgG or IgM, may partly be due to some cases of early Lyme disease in the samples where IgG production lags behind IgM formation and partly due to the continued formation of specific IgM anti-BBsl antibodies in late disease as reported by others [44]. The tendency for a higher specificity in WBs by the in-house criteria when only IgG antibody responses are considered in comparison with either IgG or IgM antibodies may partly be due the prominent presence of cross-reacting IgM antibodies in viral infections and syphilis in the 74 reference samples from New York Biologics. The PPV of 83.9% with detection of either IgG or IgM antibodies in IBs read by the in-house criteria was obtained with 178 reference sera of which only 26 (or 14.6% prevalence) were from patients with Lyme disease. The PPV can be expected to improve when sera from patients who are suspected to have Lyme disease on clinical criteria are tested because of the expected higher disease prevalence in this population. Further studies on patient sera are therefore needed to fully evaluate the clinical diagnostic parameters of IBs described here.
A disadvantage with WBs is the presence of non-specific proteins migrating at the same positions as the specific antigens used for scoring WBs. An example of this is a binding of a non-specific protein of 31kDa at the same position as Osp A on WB strips [34]. Removing patients who showed a positive 31kDa band on WB but tested negative for antibodies to recombinant OspA antigen from the analysis improved the specificity to >97% for both IgM and IgG Lyme WBs [34]. The inclusion of purified recombinant OspA protein as antigen in Lyme IBs therefore maintains sensitivity but improves specificity of Lyme IB compared to Lyme WB. Thus with Lyme IB no further confirmatory testing for OspA antigen reactivity is required. Other advantages of IBs over WBs are that IBs permit selection of proteins of diagnostic importance, including those from different Bbsl species, and avoid variations due to genetic factors and culture conditions [12,[34][35][36] that can affect the antigenicity of scored proteins in WBs.
The IBs showed that different Lyme-positive sera differentially recognized P23 and P31 from different BBsl species and P39 from US and European BBsl. This suggests that the use of recombinant proteins from several species of BBsl in the Lyme IBs contributes to their better diagnostic performance compared to Lyme WBs. However, other proteins used for scoring IBs show sequence homology among different BBsl species and are therefore expected to show varying degrees of antigenic cross-reactivity. Further studies with patient sera in conjunction with PCR-based tests for identifying species are required to determine whether IBs can help identify the BBsl species causing the infection.
The findings with rabbit antibodies reported here also demonstrate the potential of IBs for differentiating sera from patients with Lyme disease and TBRF. Whole cell lysates of Borrelia species causing Lyme disease and TBRF contain antigens that are shared between the two groups of Borrelia [15,16]. Based on the present findings that rabbit antisera to the two TBRF Borrelia species B. hermsii and B. coriaceae do not cross-react with the recombinant antigens used in Lyme IBs with the exception of a weak reaction with flagellin, it may be expected that sera from patients with B. hermsii and B. coriaceae infections will be similarly unreactive in Lyme IBs. However these predictions need to be confirmed in further studies. Our additional unpublished data suggest that it is possible to identify patients with mixed infections of BBsl species and TBRF Borrelia species and also clearly differentiate Lyme disease from TBRF in patients with either disease through the use of Lyme IBs and similar IBs developed for TBRF using TBRF Borrelia-specific recombinant antigens.
A limitation of our study is that it is based on a collection of reference serum samples obtained from 26 patients with Lyme disease, but only 17 were characterized by stage of disease. Studies on a larger number of samples from patients with more clearly defined stages of this infection would be useful to better demonstrate the utility of Lyme IBs in laboratory evaluation of Lyme disease. Since 6.6% of Lyme disease patients can only be detected by the LM-PCR assay on urine and blood and not by the currently recommended WB serological assay [25], additional testing by the LM-PCR assay would also be useful in further evaluating Lyme IB tests.
Funding: This research received no external funding.