Enzyme immunoassays (EIA) for serodiagnosis of human leptospirosis: specific IgG3/IgG1 isotyping may further inform diagnosis of acute disease

The laborious microscopic agglutination test (MAT) is the gold standard serologic test for laboratory diagnosis of leptospirosis. We developed EIA based serologic assays using recombinant proteins (rLigA, rLigB, rLipL32) and whole-cell extracts from eight Leptospira serovars as antigen and assessed the diagnostic performance of the new assay within each class, against MAT positive (MAT+) human sera panels from Portugal/PT (n = 143) and Angola/AO (n = 100). We found that a combination of recombinant proteins rLigA, rLigB and rLipL32 correctly identified antigen-specific IgG from patients with clinical and laboratory confirmed leptospirosis (MAT+) with 92% sensitivity and ~ 97% specificity (AUC 0.974) in serum from the provinces of Luanda (LDA) and Huambo (HBO) in Angola. A combination of whole cell extracts of L. interrogans sv Copenhageni (LiC), L. kirschneri Mozdok (LkM), L. borgpetersenii Arborea (LbA) and L. biflexa Patoc (LbP) accurately identified patients with clinical and laboratory confirmed leptospirosis (MAT+) with 100% sensitivity and ~ 98% specificity for all provinces of Angola and Portugal (AUC: 0.997 for AO/LDA/HBO, 1.000 for AO/HLA, 0.999 for PT/AZ and 1.000 for PT/LIS). Interestingly, we found that MAT+ IgG+ serum from Angola had a significantly higher presence of IgD and that IgG3/IgG1 isotypes were significantly increased in the MAT+ IgG+ serum from Portugal. Given that IgM/IgD class and IgG3/IgG1 specific isotypes are produced in the earliest course of infection, immunoglobulin G isotyping may be used to inform diagnosis of acute leptospirosis. The speed, ease of use and accuracy of EIA tests make them excellent alternatives to the laborious and expensive MAT for screening acute infection in areas where circulating serovars of pathogenic Leptospira are well defined.


Introduction
Leptospirosis is a neglected emerging zoonotic disease with worldwide distribution that affects essentially all vertebrates, mostly in resource-poor and developing countries [1]. It affects vulnerable populations such as rural subsistence farmers and urban slum dwellers. Urban epidemics are reported mostly in cities of developing countries and will likely increase as the world's slum population doubles to 2 billion by 2030 [1]. In developed countries, unexpected deadly outbreaks have been reported in New York City in 2017 [2] and 2021 [3]. However, outbreaks of leptospirosis are expected in grain-growing rainy regions of Australia when reservoir host populations of house mice skyrocket [4]. Sub-Saharan African countries lack notification surveillance systems for leptospirosis and in most cases the laboratory diagnosis is not done [5]. The estimated prevalence from countries that report the disease is high (75-102 per 100,000 population in Tanzania) compared with occidental countries [6,7]. Laboratory diagnosis of leptospirosis is not established in many Sub-Saharan countries despite the need for differential diagnosis with malaria, dengue, yellow fever and other common febrile illnesses [8,9], as is the case in Angola [10].
Human leptospirosis ranges in severity from a mild, self-limited febrile illness to a fulminant life-threatening disease [11]. A review of published cases estimated that leptospirosis causes~1 million cases a year [1], resulting in~5-10% death rate [1,12]. A number of organs is involved, reflecting the systemic nature of the infection. As a result, the symptoms of leptospirosis are frequently mistaken for other causes of acute febrile syndrome such as dengue, hepatitis [11], and malaria, depending on the overlap of endemic geographic areas. Unlike other spirochetal diseases that are characterized by signs and symptoms that aid in clinical diagnosis (ex. the bull's eye erythema migrans in Lyme disease, and the chancre in primary syphilis), diagnosis of leptospirosis is tentatively based on evaluation of fever and myalgia in patients presenting at the hospital in areas of endemicity, and it is rarely confirmed in most parts of the world due to lack of affordable diagnostic assays.
The microscopic agglutination test (MAT) is the current gold standard serologic test for laboratory diagnosis of leptospirosis. However, it cannot be used for immediate case identification because it is insensitive in early infection [11] and it can only be performed in specialized laboratories with highly trained personnel and specific conditions such as reference collections of live Leptospira serovars, darkfield microscopy, among others. Thus, it is rarely performed by clinical diagnostic laboratories [13]. Less cumbersome, accurate, affordable, and accessible

Production of whole-cell extract antigens
Leptospira spp previously maintained in EMJH liquid medium were sub-cultured in EMJH semi-solid medium supplemented with 10% Difco Leptospira enrichment (Thermo Fisher Scientific, Waltham, MA), 1% of 5-Fluorouracil, and 2% inactivated rabbit serum (56˚C, 1h) and grown for 15-30 days at 29˚C shaking (100 rpm) in the dark. The individual cultures successfully recovered were grown in 25mL EMJH liquid for 5-10 days (to 10 8 Cells/mL) and then centrifuged at 9,000 rpm for 15 min at room temperature (RT); the supernatant was discarded leaving an aliquot of 1mL which was centrifuged at 20,000 rpm (3 min) and the supernatant was discarded. The pellet was incubated with 1mL of BugBuster solution (protein extraction reagent containing nuclease and lysozyme) at RT in a shaking incubator (100 rpm) for 20 min, homogenized by vortex, and stocks were saved at -20˚C.

Serologic testing on Enzyme ImmunoAssay (EIA)
EIA was performed according to a comprehensive published method (Protocol 3.3.1. in CPM [15]) with the following modifications (a step-by-step protocol is provided in S1 Text). Briefly, purified recombinant proteins quantified by the Lowry protein assay kit (Thermo Fisher Scientific, Waltham, MA) or whole-cell extract of Leptospira diluted in 1X sodium carbonate coating buffer was used to coat Nunc MaxiSorp flat-bottom EIA plates (eBioscience, San Diego, CA) at 0.5-1 μg/ml (protein) or 10 5 -10 8 cells/well (Leptospira extract) overnight at 4˚C. The following day the plates were washed with 1XPBS, blocked with 1% BSA for 2h, washed again, and incubated with human or murine serum diluted at 1:50 or 1:100. Goat anti-mouse or goat antihuman IgG conjugated to HRP diluted at 1:10000 (Jackson ImmunoResearch, West Grove, PA) was used as the secondary antibody. The OD 450 was read on a SpectraMax Plus EIA reader (Molecular Devices). For determination of Ig class and IgG isotypes the secondary antibody conjugated to HRP was diluted as follows: anti-human -IgG1, -IgG2, -IgG3, -IgG4 (1:1000, Invitrogen), anti-human IgA (1:8000, Southern Biotech), anti-human IgD (1:1200, Southern Biotech) and IgM (1:10,000, Jackson ImmunoResearch Laboratories, Inc.). The EIA cutoff was set at three standard deviations above the average OD 450 of all healthy control samples.

Statistical analysis
For analysis of diagnostic performance, the OD cutoff for each antigen was determined for each leptospirosis panel versus healthy control by constructing a receiver-operating characteristic (ROC) analysis with area under the curve (AUC), and selecting the OD cutoff value which resulted in the maximum sensitivity given a minimum of 97% specificity. Furthermore, Ordinary One-Way ANOVA was used to assess the significance of differences between each of the leptospirosis panels and the healthy control. For the combined algorithm, we chose the recombinant protein or whole cell extract with the highest sensitivity and then included the negative samples that tested positive against any other biomarker candidate (a positive in 1 of the 3 or 4 biomarkers was considered positive). For analysis of differences between two IgG+ panels for Ig class and IgG isotyping the Welch's t test was used.

Screening recombinant protein candidates for specificity and potential sensitivity
To evaluate EIA specificity of the 10 candidate biomarkers we tested purified recombinant proteins using L. interrogans whole cell extract (WCE) as control, against serum (1:50) from healthy individuals from an area (FL/US) that is not endemic for leptospirosis (Fig 1). We found that seven recombinant proteins (rFlaB, rFlgC, rLigA 7-13 , rLigB 1-6 , rLemA, rLipL32 and rTolC) were more specific (less cross-reactive) to the healthy human serum than the WCE control. Of these seven, FlaB, LigA, LigB and LipL32 were shown to be reactive with serum from leptospirosis patients [16], and three (LigA, LigB, LipL32) had 95-99% homology between all pathogenic Leptospira serovars [17]. To confirm the potential sensitivity of the three leads in comparison with other candidates, we tested purified rLigA, rLigB, rLipL32, rTolC and rFliE using serum from C3H-HeJ mice infected with L. interrogans sv Copenhageni FioCruz. We confirmed that unlike TolC and FliE, the three recombinant proteins LigA, LigB and LipL32 were highly immunogenic in mice (S2

Diagnostic performance of recombinant proteins and Leptospira whole-cell extracts against human leptospirosis serum panels from two distinct geographic regions
Four MAT+ leptospirosis serum panels from two countries, representing Northern and Southern Hemispheres geographic regions, were used to evaluate the diagnostic potential of each antigen (recombinant protein and whole-cell extract) by measuring their sensitivity and specificity. We tested the three lead recombinant proteins (rLigA, rLigB and rLipL32) selected after our initial specificity analysis against four distinct serum panels by EIA and found that if keeping a specificity >96%, each recombinant protein detected each serum panel with different sensitivities (Fig 2, Table 1 and S1 Data). For Angola (AO), rLigA and rLipL32 performed better than rLigB: rLigA detected MAT+ serum with 88% sensitivity for LDA/HBO (AUC 0.958)
We did a side-by-side analysis of each serum sample against each of the three recombinant antigens to assess if an improvement in sensitivity could be measured when screening the MAT+ samples against one, two or the three proteins. A sample that was positive in 1 of the 3 biomarkers was considered positive. We found that this combination dramatically increased the diagnostic performance of the assay (Fig 3, Table 1 and S1 Data). While each recombinant protein detected the Angola AO (LDA/HBO) and AO (HLA) serum panels with sensitivities ranging from 35% to 88% and 32% to 76%, respectively, inclusion of positive samples to one of  the three biomarkers (LigA or LigB or LipL32) detected the same panels with 92% (AUC 0.974) and 76% sensitivity (AUC 0.883). The same trend was observed for the Portugal (PT) panels: single recombinant protein detection ranged from 44% to 72% (PT AZ) and 67% to 87% (PT LIS), whereas inclusion of positive samples to one of the three biomarkers (LigA or LigB or LipL32) resulted in 74% (AUC 0.849) and 88% (AUC 0.938) sensitivities for the same panels. Next, we tested whole cells extracts from eight Leptospira serovars against the same serum panels by EIA and found that keeping a specificity >97%, each extract detected the four panels with different sensitivities (Fig 4, Tables 2 and S1 and S1 Data). For Angola, two whole-cell extracts performed well (sensitivity >90%) only against the sera panel from HLA: L. interrogans sv Copenhageni and L. borgpetersenii sv Arborea detected MAT+ serum with 97.67% and 93% sensitivity (AUC 0.995 and 0.976, respectively). For Portugal, L. interrogans sv Copenhageni was the best whole cell extract detecting MAT+ sera from AZ and LIS with 97.67% and 93% sensitivity (AUC 0.997 and 0.993, respectively).
We did a side-by-side analysis of each serum sample against each of the four Leptospira extracts to assess if an improvement in sensitivity could be measured when screening the MAT + samples against one, two, three or four whole cell extracts. A sample that was positive to 1 of the 4 biomarkers (LiC or LbA or LkM or LbP) was considered positive. We found that this combination improved the diagnostic performance of the assay for all 4 sera panels tested (Fig 5, Table 2 and S1 Data) with the largest gain in the AO (LDA/HBO) sensitivity which increased from 62.69% to 100% (AUC 0.997).
Lastly, we analyzed IgM, IgD and IgA class, and IgG isotypes between two MAT+, LiC IgG+ serum panels, one from Portugal (PT-AZ) and one from Angola (AO-HLA) (Fig 6 and  S1 Data). We found that MAT+ IgG+ serum from Angola had a significantly higher presence of IgD than MAT+ IgG+ serum from Portugal; and that IgG3/IgG1 isotypes were significantly increased in the MAT+ IgG+ serum from Portugal in contrast MAT+IgG+ serum from Angola. As expected, there were no significant differences in IgM or IgA between the two panels which were included based on IgG positivity to LiC.

Discussion
The current definitive serologic test for leptospirosis is the microscopic agglutination test (MAT) which has been used for 100 years. Infection can be confirmed if paired acute and convalescent samples from the same patient show a 4X increase in antibody titer [11]. In addition, MAT allows for identification of the serogroup of the infecting serovar. This is important information to acquire in areas where a medley of pathogenic, intermediate and non-pathogenic Leptospira strains circulate. Thus, MAT is the assay of choice to confirm active infection. However, the following major limitations have been noted: MAT cannot be used for immediate case identification because it is insensitive in early infection [11]; it is not suitable for epidemiologic studies [18]; it is unable to distinguish between IgM and IgG [19]; it requires maintenance of large collections of different Leptospira serovars that can only be cultured in expensive specialized reference laboratories managed and operated by highly trained personnel. Furthermore, MAT is subjective to human interpretation and cross-reactivity among several Leptospira serovars and false positive reactions due to auto-agglutination have also been reported [11,12,20]. Given these weaknesses, several molecular techniques are currently being used for diagnosis of acute infection. These techniques can detect the etiologic agent itself in blood or urine by PCR amplification of Leptospira DNA or RNA [21][22][23][24][25][26]. However, there still is a need for versatile and affordable serologic assays that can be performed routinely in clinical laboratories in endemic areas where circulating strains are well known. A recent study showing high incidence of leptospirosis in Vanuatu highlighted the need for improved diagnostic capabilities in developing countries [27]. Interest in redefining the gold standard testing for laboratory diagnosis of leptospirosis has been mounting recently and led to the proposal of several options: combinations of molecular tests and MAT [28,29], as well as molecular tests and EIA [30][31][32] [25,[33][34][35][36][37][38]. Molecular tests have dramatically improved patient care and reduced deaths due to leptospirosis in a developed country [25]. However, molecular tests do not offer information on the immunity status of the individual. This is important as it has been recently shown that infection of mice with L. interrogans (sv Copenhageni, sv Manila, sv Icterohaemorrhagiae) induce an antibody based immune response that protects against homologous re-infection [39]. This opens the field of Leptospirosis diagnostics to applications of easy to deploy serologic assays. We and others [18,29,40] explored the use of a ubiquitous and highly sensitive technique-EIA-to develop an in-house serologic test for laboratory diagnosis of leptospirosis.
We found that in Angola where the predominant circulating Leptospira serovars are L. interrogans Copenhageni and L. borgpetersenii Ballum Arborea a combination of the L. interrogans recombinant proteins LigA, LigB and LipL32 correctly identified antigen-specific IgG from patients with clinical and laboratory confirmed leptospirosis (MAT+) with 92% sensitivity and specificity~97% (AUC 0.974) in the province of Luanda and Huambo (AO LDA/ HBO), whereas a combination of whole cell extracts of L. interrogans sv Copenhageni (LiC), L. kirschneri Mozdok (LkM), L. borgpetersenii Arborea (LbA) and L. biflexa Patoc (LbP) accurately identified patients with clinical and laboratory confirmed leptospirosis (MAT+) with 100% sensitivity and specificity~98% for both provinces (AUC 0.997 for LDA/HBO and 1.000 for HLA). Others have shown that combinations of recombinant proteins containing rLigA, rLigB and rLipL32 on EIA detected antigen-specific IgM and IgG in serum from leptospirosis patients from Peru with 82% sensitivity and 86% specificity compared to MAT [41]. Further, others found that LigA-IgM EIA was more sensitive, but not more specific, than whole-cell based IgM EIA for the early diagnosis of leptospirosis in the Philippines [42]. Another study done recently using a commercial, whole-Leptospira-based IgM EIA, reported low accuracy [43]. However, a weakness of that study was the comparison between different classes of tests using a molecular test as the gold standard to gauge sensitivity and specificity of a serologic test. These two different classes of tests complement each other given that a molecular test is expected to detect the infectious agent early in the course of infection, whereas a serologic test should be positive 7-14 days post a new infection. Furthermore, it was recently found that long term production of IgM to Leptospira interrogans sv Copenhageni and sv Manilae correlate with colonization of the kidney, in contrast to sv Icteroheamorrhagiae which induced a classical temporary IgM response without kidney colonization [44]. These data further support development of EIA based serologic assays for research purposes.
For Portugal, the best diagnostic performance was achieved with a combination of whole cell extracts of L. interrogans sv Copenhageni (LiC), L. kirschneri sv Mozdok (LkM), L. borgpetersenii sv Arborea (LbA) and L. biflexa sv Patoc (LbP) which accurately identified patients with clinical and laboratory confirmed leptospirosis (MAT+) with 100% sensitivity and~98% specificity (AUC 1.000). In-house ELISAs done by others using whole cell extracts of the most prevalent serovar in Germany (Leptospira kirschneri sv Grippotyphosa) performed with a clinical sensitivity of 83% and clinical specificity of 98.5% in MAT+ serum [18], which is equivalent to our results in regards to specificity (98%). However, in our study, whole-cell in-house EIAs made with L. interrogans sv Copenhageni produced higher sensitivities in MAT+ serum (93%-98%) than the study in Germany, which could be attributed to the very high prevalence of this serovar in Portugal. Our data also suggests that sensitivity of the EIA can be improved by adding extracts from other prevalent serovars in the region such as the leptospires isolated from the Angola specimens [10]. Our data further supports [40] the development of in-house IgG EIAs using Leptospira serovars known to circulate in the region/country [29] as an alternative method for the diagnosis of Leptospirosis. Overall, the ease of use and accuracy of the EIA make them excellent serologic alternatives to the laborious MAT for diagnostic screening and for epidemiologic studies in resource-limited countries, as we offer an example for Angola.
Interestingly, we found that MAT+ IgG+ serum from Angola had a significantly higher presence of IgD and that IgG3/IgG1 isotypes were significantly increased in the MAT+ IgG + serum from Portugal. As expected, there were no significant differences in IgM or IgA between the two panels given that both sera panels used had equivalent levels of IgG as this was the criteria to do the IgG subtyping analysis. Circulating IgD is found at low levels in serum, it has a short half-life and its function is not clear. Because there is a FcR for IgD in CD3 T cells, it has been proposed that IgD might serve as a bridge for antigen presentation by B cells to T cells [45]. Presence of transient IgD in serum may be indicative of a pre-adaptive immune response ongoing in acute leptospirosis when mature B cells producing IgM+/IgD + reach the spleen. Increased IgD in the Angola panel suggests a very early stage of disease. This is corroborated by the clinical inclusion criteria for patients in the Angola study, which was fever. IgG3 is the first IgG to appear in serum as switching from IgM/D to IgG takes place and it is an early effector, independent of T cell help [46]. IgG1 soon follows after T cell help has been engaged. Presence of IgG3/IgG1 in the Azores panel suggests that these patients also presented at the hospital with acute leptospirosis. Production of IgG2 and IgG4 is associated with long exposure to antigen and switching to IgG4 is associated with induction of tolerance [46,47]. Absence of IgG2 and IgG4, two isotypes commonly associated with later stages of disease, further corroborate the clinical characterization of the patients included as acute leptospirosis. Thus, immunoglobulin G isotyping provides another measure that can aide in clinical characterization and should be further tested to discriminate infection from reinfection.
The increased sensitivity and specificity of this assay may be due to the use of Bugbuster reagent in our protein extraction protocol. This reagent breaks up the Leptospira membrane gently but preserves integrity of the proteins. The Leptospira LPS induces specific immune responses to each serovar leading to low cross-reactivity/specificity between strains, and thus it defines the serovars. Thus, the high specificity of our assay might be associated to the protein extracted by the Bugbuster reagent. This is further corroborated by the detection of IgG3/IgG1 rather than IgG2/IgG4 given that IgG1 and IgG3 are generally induced by protein antigens, whereas IgG2/IgG4 are generally induced by polysaccharide antigens (in this case, Leptospira LPS) [45,46].
One limitation of this study was that we did not have access to serum from healthy individuals from the endemic areas used in this study (Portugal and Angola). Use of serum from healthy individuals from an endemic area may lower the overall specificity of the assay. Another limitation is that serum from other diseases used in differential diagnosis of leptospirosis in co-endemic areas (e.g. malaria) was not available for cross-reactivity testing.
In conclusion, in areas where circulating serovars of pathogenic Leptospira are well defined, accurate assays such as EIA which can be operated at a fraction of the cost of MAT should be explored as effective serodiagnostic tools for leptospirosis applicable to humans and animals. Addition of IgG3/IgG1 isotyping further increases likelihood that the samples tested represent acute human leptospirosis.
Supporting information S1 Fig. Purified recombinant proteins rLigA, rLigB and rLipL32. The SDS-PAGE gel was stained by coomassie blue to confirm the molecular weight of the protein and antigenicity was confirmed by western blot of proteins electrotransferred into PVDF membranes against antigen-specific mouse polyclonal antibodies.