Abstract
The nucleoprotein of Lassa virus, strain Josiah, was expressed in Escherichia coli as an N-terminally truncated, histidine-tagged recombinant protein. Following affinity purification the protein was completely denatured and spotted onto nitrocellulose membrane. A total of 1 μg of protein was applied for detection of Lassa virus antibodies (LVA) in a simple immunoblot assay. Specific anti-Lassa immunoglobulin M (IgM) antibodies could be detected by increasing the amount of protein to 5 μg. A panel of 913 serum specimens from regions in which Lassa virus was endemic and from regions in which Lassa virus was not endemic was used for evaluating the sensitivity and specificity of the LVA immunoblot in comparison to those of an indirect immunofluorescence (IIF) assay. The sera originated from field studies conducted in the Republic of Guinea (570 serum samples) and Liberia (99 serum samples), from inpatients of the clinical department of the Bernhard-Nocht-Institute, Hamburg, Germany (94 serum samples), and from healthy German blood donors (150 serum samples). In comparison to the IIF assay the LVA immunoblot assay had a specificity of 90.0 to 99.3%, depending on the origin of the specimens. The sensitivity was found to be highest for the Guinean samples (90.7%) and was lower for the Liberian samples (75%). Acute Lassa fever was diagnosed by PCR in 12 of 59 (20.3%) patients with fever of unknown origin (FUO) from the Republic of Guinea. On admission to the hospital, nine Lassa fever patients (75%) were reactive by the IgM immunoblot assay. One of the patients was infected with a new Lassa variant, which showed 10.4% variation on the amino acid level in comparison to the prototype strain of Lassa virus, Josiah. Seven PCR-negative patients were reactive by immunoblotting. The positive and negative predictive values of a single IgM immunoblot result for acute, PCR-confirmed Lassa fever were therefore 53.6 and 93.0%, respectively. Because of its high negative predictive value, a single IgM immunoblot result will be valuable for excluding acute Lassa fever for cases of FUO in areas where Lassa fever is endemic.
Lassa fever continues to be a public health problem of major significance in certain West African countries, causing up to 16% of all adult medical admissions and approximately 30% of adult deaths in some hospitals in Sierra Leone (13, 14). More than 800 cases of Lassa fever with over 150 fatalities have been reported from this country to the World Health Organization during an ongoing epidemic since 1996 (22). The definitive diagnosis of Lassa virus infection to date depends on virus isolation or serological and molecular techniques. These tests have not been adapted to field use and are normally carried out in laboratories of biosafety level 4, which do not exist in West Africa. Because of a general lack of infrastructure, PCR technology for detection of Lassa virus is not very likely to be introduced into regions of endemicity in the near future. Recombinant protein technology could meet the demand for a simple and reliable Lassa fever test system, and recombinant Lassa virus proteins have been shown to be more or less useful for cross-sectional serological surveys of antibodies (10). A few attempts to diagnose acute Lassa fever cases by demonstrating a rise in immunoglobulin G (IgG) titer have been published (9), but there are no reports on the detection of specific IgM antibodies with these test systems. Lassa virus serology in regions where Lassa fever is endemic is complicated by the high background level of specific IgG antibodies, approaching a prevalence of up to 35% in selected villages (21). Furthermore, epitopes on the structural proteins of Lassa virus variants isolated from different geographic areas have been mapped by monoclonal antibodies and were shown to exhibit a distinct pattern of serological cross-reactivity (18). In consequence, any recombinant assay must be evaluated for its potential to detect variant-specific or cross-reactive serological responses in humans from areas of endemicity. To this end, we have recently expressed an N-terminally truncated recombinant nucleoprotein (NP) of Lassa virus (strain Josiah) in Escherichia coli and found it to react with immunofluorescence-positive sera in an enzyme-linked immunosorbent assay (ELISA) only after a prolonged renaturation process and with a disappointingly low sensitivity of 30% (19). In the present report we describe the evaluation of a simple immunoblot assay utilizing the same antigen but in a completely denatured form and in a quantity 5 to 25 times higher than that used for the ELISA.
MATERIALS AND METHODS
Subjects.
A total of 570 serum samples had been sampled in 1993 during a population-based serosurvey in the Republic of Guinea investigating risk factors for Lassa virus transmission in a high-prevalence area (Gueckedou) and a low-prevalence area (Pita). The prevalence of Lassa virus antibodies was found to be 14% in Gueckedou and 2.6% in Pita, as measured by an indirect immunofluorescence (IIF) assay with a titer of 1:20 as the cutoff (21). A total of 99 serum samples had been collected during a field study conducted in 1985 in Bong County, Liberia, where clinical Lassa fever cases occurred. All specimens had been inactivated for 30 min at 56°C and were stored at −20°C until further use.
An incidence study of Lassa fever among cases of fever of unknown origin (FUO) was carried out in the regional 130-bed hospital of Gueckedou, which serves a population of approximately 250,000 people. The study was conducted throughout 1995 and 1996, but because of intermittent logistic problems the results may not be truly representative of this period of time.
FUO was defined as any febrile condition (>38.5°C) not responsive to antimalarial or antibiotic treatment for more than 3 days. Furthermore, all patients with fever and hemorrhage were enrolled in the study. Patients were clinically investigated, and blood, urine, and, occasionally, cerebrospinal fluid samples were drawn for analysis. A total of 55 patients meeting these criteria were included in the study as were an additional 4 patients with cases of suspected hemorrhagic fever. These four patients were from a Liberian refugee camp in the Prefecture of Gueckedou and were seen by doctors of Médecins Sans Frontiers. A total of 94 patients admitted to the clinical department of the Bernhard-Nocht-Institute, Hamburg, Germany, for various diseases (malaria, typhoid fever, hepatitis, dengue fever, human immunodeficiency virus infection) and 150 healthy German blood donors, i.e., individuals from areas where Lassa fever is not endemic, served as negative controls.
Serological methods. (i) IIF assay.
In our BSL4 facility in Hamburg, Germany, Lassa virus (Josiah strain) was grown in U937 (ATCC CRL 1593) or Vero cells (ATCC CCL 81) which were propagated in RPMI 1640 or minimal essential medium, respectively, and supplemented with 5% fetal calf serum. After approximately 1 week, the cells were harvested, spread on immunofluorescence slides, air-dried, and fixed for 1 h at room temperature. Successful infection (30 to 60% of cells infected) was shown by immunofluorescence with monoclonal antibodies raised against the NP of Lassa virus (8). The slides were stored at −20°C until further use. All sera were tested at a dilution of 1:20 in phosphate-buffered saline (PBS) by IIF assay with a labelled secondary antibody (goat anti-human IgG–fluorescein isothiocyanate [FITC]; Dianova, Hamburg, Germany) at a dilution of 1:50. Positive sera were then diluted in PBS in twofold steps for endpoint titer determination.
For detection of Lassa virus IgM antibodies in FUO cases, sera with low IgG titers by IIF assay (<1:40) were directly diluted 1:20 with PBS. One serum sample with a high IgG titer by IIF assay (>1:1,280) was first preabsorbed with goat anti-human IgG (Dianova) for 2 h at room temperature. The slides were incubated overnight at 4°C and then washed twice for 20 min each time with PBS, which resulted in a low background of nonspecific fluorescence. FITC-conjugated goat anti-human IgM (Sifin, Berlin, Germany) was used as a second antibody in a dilution of 1:70. Positive sera were further serially endpoint diluted in PBS.
RNA extraction of samples, reverse transcription (RT), and nested PCR.
Nucleic acids were extracted by the guanidiniumthiocyanate (GIT)/glassmilk method (3) from serum and urine samples collected from patients with FUO. Briefly, 100 μl of specimen was mixed with 900 μl of GIT lysis buffer (4.5 M GIT, 50 mM Tris-HCl [pH 6.5], 20 mM EDTA [pH 8.0], 1.2% Triton X-100) and 15 μl of HCl-treated silica beads were added. After incubation at room temperature for 15 min the beads were spun down and were washed twice in GIT washing buffer (5 M GIT, 50 mM Tris-HCl [pH 6.5]), twice in 70% ethanol, and once in acetone. The beads were dried at 56°C for 30 min, and the bound nucleic acids were liberated with 50 μl of diethylpyrocarbonate-treated H2O. Samples were stored at −70°C until further use. To avoid contamination problems, preparation and storage of samples and performance of the PCR were routinely performed in different laboratories.
A nested PCR was employed for amplification of a 196-bp fragment of the NP. The NP primer sequences were chosen in a region of the NP which had previously been found to be a suitable target for RT-PCR (11). The detection limit of this nested RT-PCR was determined as 10 50% tissue culture infective doses (data not shown). The numbering of the nucleotide (nt) positions is according to the viral sense sequence of the small RNA (S) segment of the Josiah strain of Lassa virus (EMBL Nucleic Acid Database accession no. JO 4324). The primers used were as follows: N1+, 5′-AAG TGC AGG TGT CTA TAT GGG (nt 2943 to 2923); N4, 5′-CAA CCT AAG CTC ACA GCA ACT TGA C (nt 2922 to 2898); N2+, 5′-TGT ACT GCA TCA TTC AAG TCA AC (nt 2704 to 2726); N2, CTG CCC CTG TTT TGT CAG ACA TGC C (nt 2727 to 2751). Primers N1+ and N4 are reverse complementary to the viral RNA, and N1+ was used for the RT.
RT was started by annealing 50 pmol (2 μl) of primer N1+ to 6 μl of extracted nucleic acid for 10 min at 70°C and then for 3 min at 0°C. A total of 12 μl of preset RT reaction mixture, containing 4 μl of 5× reaction buffer (GIBCO BRL, Eggenstein, Germany), 2 μl of deoxynucleoside triphosphates (dNTPs) (2 mM [each] dATP, dTTP, dCTP, and dGTP), 2 μl of 1 M dithiothreitol (DTT), 3.5 μl of DEPC · H2O, and 100 U (0.5 μl) of Superscript Reverse Transcriptase (GIBCO), was added. The reaction mixture was overlaid with paraffin oil and was incubated for 60 min at 50°C and then for 10 min at 95°C.
PCR mixtures were prepared with 20 pmol (2 μl) each of primers N1+ and N2+, 2 μl of 10× PCR buffer (MBI Fermentas, Vilnius, Lithuania), 2 μl of dNTPs (2 mM each), 1 μl of MgCl2 (25 mM), 5.5 μl of H2O and 1 U (0.5 μl) of Taq polymerase (MBI Fermentas). A total of 5 μl of cDNA of the RT reaction was added, and the mixture was overlaid with paraffin oil. Temperatures and cycles for the PCR were 5 min at 95°C; 25 cycles of 1 min at 95°C, 1 min at 50°C, and 1.5 min at 72°C; and 10 min at 72°C. PCR was performed with a Perkin-Elmer Cycler. Two negative controls were run for each PCR.
PCR mixtures for the second (nested) PCR contained primers N2 and N4 and 9.5 μl of H2O. One microliter of PCR fragments from the first PCR was added, and temperatures for the nested PCRs were set as described above except that 40 instead of 25 cycles were run. PCR products were visualized on ethidium bromide-stained agarose gels.
Cloning and sequencing of RT-PCR products.
All PCR fragments of the expected size were cloned into either the TA cloning vector (Invitrogen, NV, Leek, The Netherlands) or pUC57/T (MBI Fermentas), both utilizing T overhangs. Briefly, 3 μl of crude nested-PCR product was ligated by using T4 ligase (MBI Fermentas) with 2 μl (100 ng) of vector overnight at 12°C. The ligated vector was transformed into E. coli DH5α, and colonies were screened with nested-PCR primers N2 and N4 for the expected insert before preparing the DNA for sequencing. Dideoxy chain termination sequencing reactions were carried out with 10 μl of DNA miniprep by using a sequencing kit (version 2.0; U.S. Biochemicals, Amersham Buchler, Braunschweig, Germany) and [35S]dATP.
Cloning, expression, and purification of the N-terminally truncated Lassa virus NP.
The cloning and expression of a truncated recombinant nucleoprotein has been described elsewhere (19). Briefly, fragments of the gene encoding the NP of the Lassa virus, strain Josiah, were amplified by RT-PCR with restriction sites for BamHI and HindIII incorporated into the 5′ and 3′ ends of the PCR primers, respectively. Fragments of different lengths were then cloned into the T7 polymerase-driven expression vector pJC40 (5), which adds an N-terminal tag of 10 histidine residues to the recombinant protein. Expression was performed in E. coli BL21 (DE3). Neither the whole NP nor the N terminus (amino acids [aa] 1 to 139) could be expressed (data not shown), but a truncated protein (aa 141 to 569) was abundantly overexpressed, extracted from insoluble inclusion bodies with 8 M urea, and purified by nickel-chelate chromatography (17). After purification (>99% as estimated from Coomassie-stained sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis gels), the protein was dialyzed against PBS at 4°C overnight. The precipitated recombinant protein was then redissolved in 2% SDS–100 mM DTT–50 mM Tris-HCl (pH 6.8) and stored at −20°C until further use.
LVA and IgM immunoblot assays.
The concentration of the truncated recombinant Lassa NP was determined photometrically and adjusted to 5 μg/3 μl, 1 μg/3 μl, 100 ng/3 μl, and 10 ng/3 μl. A solution of chicken lysozyme (LYS; SIGMA-Aldrich, Deisenhofen, Germany) was prepared in 2% SDS–100 mM DTT–50 mM Tris-HCl (pH 6.8) and adjusted to 1 μg/3 μl and 5 μg/3 μl. The protein solutions were incubated at 95°C for 10 min and then kept at 0°C. For the Lassa virus antibody (LVA) immunoblot assay, 1 μg, 100 ng, and 10 ng of NP and 1 μg of LYS were spotted onto strips of a nitrocellulose membrane (Protran Nitrocellulose BA 79; Schleicher & Schüll, Dassel, Germany) and allowed to dry. For the IgM immunoblot assay, 5 and 1 μg of NP and 5 μg of LYS were dotted onto the strips. The strips were stored at room temperature for immediate use or were sealed under vacuum in aluminum foil bags for prolonged storage.
The strips were blocked with TBST buffer (10 mM Tris-HCl [pH 7.8], 150 mM NaCl, 0.1% Tween 20) containing 10% nonfat milk (NFM) for 30 min at room temperature. To determine the optimal reaction conditions, human sera were diluted 1:100 to 1:400 for detection of IgG antibodies and 1:25 to 1:100 for detection of IgM antibodies in TBST-NFM and incubated for 1.5 h with gentle agitation. After the strips were washed three times with TBST, either peroxidase-labelled rabbit anti-human IgG or IgM or anti-human IgM (Dianova) was added in a dilution of 1:500 to 1:2,000 in TBST-NFM. The strips were incubated for 1 h and washed twice in TBST and once in NT buffer, and a trace of chloro-1-naphthol was added as substrate in 15% (vol/vol) methanol–75% (vol/vol) NT buffer (100 mM NaCl–10 mM Tris-HCl [pH 7.8])–0.2% H2O2. After further incubation for 15 min at room temperature the color reaction was scored as 3+, 2+, 1+, or 0, corresponding to a very strong, strong, weak, or no signal, respectively. Only homogenously colored spots were scored as positive, and faint rings that were sometimes visible were scored as negative. With each new lot of recombinant protein and conjugate antibody a criss-cross titration usually had to be performed to optimize the reaction conditions. A negative reaction on the LYS protein with the positive and negative control sera defined the optimal reaction conditions free of nonspecific binding.
RESULTS
Sensitivity and specificity of the LVA immunoblot assay in comparison to the IIF assay.
Table 1 shows the results of testing sera from regions where Lassa virus is endemic (Guinea and Liberia) and from regions where it is not (Germany) with IIF and immunoblot assays. Compared to the IIF assay the sensitivity of the blot analysis was 90.7% (68 of 75) for sera from Guinea and 75.0% (9 of 12) for sera from Liberia (P = 0.13, Fisher’s exact test). The specificity was 63.6% (189 of 297) for sera from Gueckedou, 94.1% (12 of 190) for sera from Pita, 96.3% (79 of 82) for sera from Liberia, 98.9% (93 of 94) for sera from inpatients from Hamburg, and 99.3% (149 of 150) for sera from healthy blood donors from Hamburg. The differences in specificity observed between that with sera from Gueckedou and those with sera from Pita, Liberia, and the two collectives from Germany are statistically significant, with P = 0.001 (test for heterogeneity in 2 × N contingency tables using a Monte Carlo approach [20]). Antibody prevalences by the IIF assay were 14.0% for sera from Gueckedou and 2.6% for sera from Pita, as had been previously found in a population-based serosurvey (21). The antibody prevalence by IIF assay was 12.1% in the series from Liberia, 0% in inpatients from Hamburg, and 0% in healthy blood donors from Hamburg.
TABLE 1.
Comparison of performance of IIF and LVA immunoblot assaysa
| LVA immunoblot assay result | No. of serum samples from indicated location that were:
|
||||||
|---|---|---|---|---|---|---|---|
| Immunofluorescence positive
|
Immunofluorescence negative
|
||||||
| Guinea | Liberia | Gueckedou | Pita | Liberia | Germany | ||
| Positive | 68 | 9 | 108 | 12 | 8 | 0 | 1 |
| Negative | 7 | 3 | 189 | 190 | 79 | 94 | 149 |
The conjugates used for the IIF and immunoblot assays were anti-IgG–FITC and anti-IgG-IgM–POD, respectively. The subjects from Guinea and Liberia were healthy persons who were evaluated during field studies. The subjects from Germany included 94 inpatients, hospitalized for malaria, hepatitis, human immunodeficiency virus, typhoid fever, and dengue fever, and 150 healthy blood donors. In a population-based serosurvey which used an IIF assay, Lassa antibody prevalence was found to be 14% in Gueckedou and 2% in Pita.
In the Prefecture of Pita (antibody prevalence by IIF assay, 2.6%), 12 of 202 (5.9%) IIF-negative serum samples reacted in the blot (1+ to 3+ reactivity) compared with 108 of 297 (36.4%) serum samples from the Prefecture of Gueckedou with an antibody prevalence by IIF assay of 14.0% (1+ to 3+ reactivity).
A total of 8 of 87 (9.2%) IIF-negative serum samples from Liberia reacted in the LVA immunoblot assay (1+ to 3+). A weak reaction (faint 1+) was observed for 1 of 150 (0.7%) healthy German blood donors.
Performance of the IgM immunoblot assay for diagnosis of acute PCR-confirmed Lassa fever cases.
A total of 9 of 55 (16.4%) Guinean FUO cases admitted to the hospital in Gueckedou were diagnosed as acute Lassa fever by PCR and sequencing, and 3 of 4 Liberian patients referred by physicians from Médecins Sans Frontiers as hemorrhagic fever cases were diagnosed as having Lassa fever. Patients were admitted between days 1 and 14 (mean, 4.3) after self-reported onset of fever and blood samples were drawn. A total of 7 of 12 (58.3%) of the patients presented with or developed hemorrhage (mostly epistaxis and/or bloody diarrhea) and 4 of 12 (33.3%) died. Table 2 shows the results from testing serum samples obtained from 12 cases of PCR- and sequence-confirmed acute Lassa fever by IIF (IgG and IgM) and IgM immunoblot assays. Sera from 9 of 12 (75%) patients reacted in the IgM blot compared to 11 of 12 (91.7%) serum samples being reactive in the IgM IIF assay. A total of 7 of 47 (14.9%) PCR-negative patients also tested positive by the immunoblot assay, and IgM antibodies were detected by the IIF assay for 6 of 7 of these patients. The positive predictive value of a single IgM immunoblot result for correctly identifying PCR-confirmed Lassa fever among cases of FUO was therefore 56.3% (9 of 16) and the negative predictive value was 93.0% (40 of 43).
TABLE 2.
Comparison of performance of IIF and IgM immunoblot assays for PCR-confirmed Lassa fever casesa
| Patient | Assay resultb
|
Clinical picture | ||
|---|---|---|---|---|
| IgG IIF | IgM IIF | IgM immunoblot | ||
| Guinea | ||||
| LG2 | Neg. | 1/160 | 3+ | Hc |
| LG32 | 1/1,280 | 1/320 | 1+ | Death |
| LG33 | 1/20 | 1/20 | 0 | |
| LG35 | 1/20 | 1/80 | 2+ | H |
| LG36 | Neg. | 1/80 | 2+ | H, death |
| LG37 | Neg. | 1/40 | 3+ | |
| LG38 | 1/20 | 1/20 | 0 | H, death |
| LG40 | Neg. | 1/20 | 3+ | |
| LG52 | Neg. | 1/20 | 1+ | |
| Liberia | ||||
| MSF3 | 1/40 | 1/40 | 2+ | H |
| MSF5 | 1/40 | Neg. | 0 | H |
| MSF7 | Neg. | 1/20 | 2+ | H, death |
The conjugates used were anti-IgG–FITC and anti-IgM–FITC for the IIF assay and anti-IgM–POD for the IgM immunoblot assay. The Lassa virus variants isolated from patient LG32 from Guinea and patient MSF5 from Liberia showed 11.6 and 8.2% variations on the nucleic acid level and 10.4 and 2.1% variations on the amino acid level, respectively, from the prototype Josiah strain.
Results for IIF assays are in titers. Neg., negative. See text for explanation of scale for IgM immunoblot assay.
H, hemorrhage (mostly epistaxis and/or gastrointestinal bleeding).
Patient LG32, who was infected with a Lassa virus variant, exhibited a high titer of both IgG and IgM antibodies by the IIF assay (1:1,280 and 1:320, respectively), and serum from this patient also reacted in the IgM immunoblot assay. In contrast, serum from patient MSF5, who was also infected with a Lassa virus variant but who had a low antibody titer by the IIF assay, showed no reactivity in the blot. Figure 1 gives a representative example of the IgM immunoblot assay performed with sera from patients with Lassa fever.
FIG. 1.
Lassa IgM antibody immunoblot assay of recombinant NP of Lassa virus, strain Josiah, and LYS. Sera of PCR-confirmed Lassa fever cases (LG2, LG32, LG37, and LG52) and of a German negative control (NEG.) were diluted 1:25 in TBST-NFM. Goat anti-human IgM POD-conjugated antibody was diluted 1:500 in TBST-NFM. For further technical details see Materials and Methods.
Sequencing of PCR products.
In 8 of 9 Lassa fever patients from Gueckedou, Republic of Guinea, the sequence of the fragment generated by RT-PCR was 100% identical to the prototype strain Josiah from Sierra Leone. Patient LG32 showed mutations in 17 of 146 (11.6%) nt, resulting in 5 differences (10.4%) at the amino acid level from the sequence of the prototype strain. Patient MSF5 from Liberia had mutations in 12 of 146 (8.2%) nt, leading to only 1 change (2.1%) at the amino acid level (Fig. 2).
FIG. 2.
Alignment of nucleotide and amino acid sequences. Sequences marked LG32 and MSF5 were separately generated from the serum of two patients with Lassa fever by RT-PCR with primers specific for the NP of Lassa virus, strain Josiah. The published sequence of the corresponding region of the Nigeria strain of Lassa virus (EMBL Nucleic Acid Database accession no. X52400) was included in the comparison. Nucleotides and amino acids are noted only when they differ from the reference strain Josiah; when identical, they are indicated by dots. The numbering of the nucleotides and amino acids is according to the mRNA of the NP, i.e., reverse complementary to the numbering of the published viral RNA sequence. The deduced amino acid sequences (bottom) are given in single-letter code.
DISCUSSION
Immunoblot assay for detection of low-level LVA in healthy subjects.
The NP and GP1 and GP2 glycoproteins of the Josiah and Nigeria strains of Lassa virus have previously been expressed in E. coli (1), baculovirus (2, 9), and vaccinia virus (6, 15). Only the recombinant NP of the Josiah strain, expressed in baculovirus and purified from cell culture supernatant by high-pressure liquid chromatography, has been evaluated in a large field study in comparison with the IIF assay for the detection of LVA in healthy subjects (10). No reports on the use of recombinant proteins to detect IgM antibodies have been published so far. We chose to express the NP of Lassa virus, strain Josiah, in E. coli and investigate its use in a immunoblot assay for several reasons. In searching for a recombinant antigen useful for LVA detection assays in our field studies in the south of the Republic of Guinea, we assumed that most of the circulating Lassa virus variants would be similar or identical to the Josiah strain from neighboring Sierra Leone, even though to our knowledge only one single Lassa virus isolate of a Guinean patient, admitted to the hospital of Zorzor in Liberia, has to date been serologically characterized (18). The NP seemed more useful than the glycoproteins because it is not glycosylated and human sera readily react with it in Western blot assays, whereas this is not the case for GP1 (4). Furthermore, serological diagnosis of other hemorrhagic fever virus infections has successfully been carried out with the recombinant nucleoproteins of, e.g., Crimean-Congo virus (12) and hantavirus (7) in simple immunoblot assays. We therefore expressed a truncated NP in E. coli by using a pET-derived vector system and adding an N-terminal histidine tag to the recombinant protein to enable its high-grade purification via nickel-chelate chromatography (17). As described before, neither expression of the full-length protein nor expression of the N terminus was possible (1). Codon usage problems in E. coli, with six arginine residues clustering in the N terminus of the NP, coded by AGG and AGA (codon usage is 0.13 and 0.2% for these codons, respectively, in E. coli versus 1.7 and 2.1% in Lassa virus), could account for this result. In initial experiments with an ELISA, the truncated recombinant protein reacted with 97 IIF-positive serum samples only after a prolonged renaturation process and with a disappointingly low sensitivity of approximately 30% (19). However, when the protein was totally denatured in SDS and used in a large amount of 1 μg in an immunoblot assay, the same serum panel now reacted with a sensitivity of 90.7%, indicating that antibodies recognizing linear epitopes of the Josiah NP were present in the majority of the sera. We next tested several hundred IIF-negative serum samples from healthy subjects from a population-based serosurvey in the Republic of Guinea (21). In the Prefecture of Pita, where a low prevalence of LVA by IIF assay had been found, 5.6% of IIF-negative sera reacted in the blot compared with 36.4% of the sera from the Prefecture of Gueckedou, where there is a high prevalence of LVA in sera. We conclude that a substantial number of LVA-positive sera in Gueckedou had fallen below our cutoff titer of 1/20 for the IIF assay. To avoid false-positive results, an IIF titer of ≥1/16 is recommended as the cutoff for detecting antibodies against Lassa virus, even though it has repeatedly been stated that this will probably lead to an underestimation of the true LVA prevalence (14). Recalculating the data from Gueckedou with the results generated by the immunoblot assay, the LVA prevalence would be 45.3% instead of 14%. The low reactivity with IIF-negative sera from Germany and Liberia further corroborates this finding.
False-positive and false-negative LVA immunoblot assay results.
One serum sample from the 244 German controls (area of nonendemicity) reacted weakly (faint 1+ reaction) in the LVA blot. This 0.41% rate of false positivity closely parallels the rate of 1% recently reported for an immunoblot assay for hantavirus (7). Preabsorption of the serum with a lysate of nontransformed E. coli cells resulted in the disappearance of the signal. Despite affinity purification, contamination of the recombinant NP with traces of E. coli proteins therefore results in a small number of false-positive results. Modifications of the purification procedure are currently being tested to eliminate this problem.
Of the IIF-positive sera from Guinea and Liberia, 9.3 and 25%, respectively, did not react in the LVA blot. Lack of reactivity did not depend on the LVA titer measured in the IIF assay, as sera with a low titer of 1/20 and with a high titer of 1/160 tested negative. We offer two explanations for this observation. First, some serological reactivity is probably missed because the recombinant NP lacks the N terminus. Second, Lassa virus variants serologically distinct from the Josiah strain circulate in Liberia (18) and we have identified a new variant in the Republic of Guinea, with 10% variation in the NP on the amino acid level (Fig. 2). Because the LVA immunoblot assay uses a completely denatured protein, it might be rather specific for antibodies against linear epitopes of the NP of Lassa virus strain Josiah. These explanations are also applicable to the IgM immunoblot assay (see below). Expression of the truncated NP of the variant of patient LG32 is under way to increase the sensitivity of the blot for Guinean sera.
IgM immunoblot assay for detecting IgM antibodies in acute Lassa fever cases.
We further evaluated the immunoblot assay for its ability to detect specific anti-Lassa IgM antibodies in acute cases of Lassa fever. Detection of IgM antibodies in Lassa fever cases has been described by researchers using an IIF assay (23) and a μ-capture ELISA, with Lassa virus-infected cell culture supernatant being used as the antigen for the latter assay (16). In the rhesus monkey model of Lassa fever, IgM antibodies were found to appear as early as day 10 after infection, and they peaked between day 13 and 17 (maximum mean titer, 1/420 in an IIF assay) and were detectable through day 126 but not 370 (16). ELISA IgM titers rose later but were detectable throughout the 532-day observation period. We detected IgM antibodies in 75% of our PCR-positive Lassa fever patients using the immunoblot assay but only after applying as much as 5 μg of recombinant protein to the assay. This did not result in increased nonspecific binding, which was controlled by raising the amount of negative control protein to 5 μg. The blot was also reactive in 14.9% of Lassa PCR-negative FUO cases, which by definition were not regarded as acute Lassa fever cases. In 85.7% of these FUO cases Lassa IgM antibodies were also detectable by IIF assay. The IgM blot assay was less sensitive than the IgM IIF assay. Two patients (LG33 and LG38) were positive by the IgM IIF assay but despite being infected with the Josiah strain of Lassa virus their sera did not react in the IgM immunoblot assay. A possible explanation for this finding could be the delayed appearance of IgM antibodies directed against linear epitopes, which are exclusively detected by the blot assay.
The duration of viremia and viruria after acute Lassa virus infection has not yet been determined by PCR, but infectious virus has been rescued as long as 2 months after infection from a patient’s urine. Given this observation and the persistence of IgM antibodies of up to 532 days, the Lassa virus infection of PCR-negative, IgM immunoblot-positive FUO cases in our sample therefore probably dates back between 2 and 18 months. The high background level of Lassa virus IgM in cases of FUO resulted in a positive predictive value of a single IgM immunoblot result for acute Lassa fever of 56% and a negative predictive value for the same of 93%. This makes a single IgM test under field conditions useful for preliminary exclusion of Lassa fever but not for the diagnosis of acute cases, which will still rely on the demonstration of a rise in IgM and/or IgG titers. Studies to evaluate the LVA blot and IgM blot assays to this end are under way.
Patient LG32 was infected with a Lassa virus variant but reacted in the IgM immunoblot assay in contrast to patient MSF5, who was also infected with a Lassa virus variant, albeit with a lower degree of variation on the amino acid level. This observation is probably best explained by the much higher antibody titer in patient LG32 (1/1,280) than in patient MSF5 (1/40). From these data it can be concluded that the IgM immunoblot assay is most likely type specific for IgM antibodies against the NP of strain Josiah if the antibody titer is low. For sera with high titers of antibodies a cross-reactivity can be expected with Lassa virus variants which have up to 10% mutations at the amino acid level.
Lassa fever in the Republic of Guinea.
The repeated description of a high serological activity of Lassa virus has so far contrasted with a lack of reports of clinical cases from the Republic of Guinea (10, 21). By PCR we detected nine cases of Lassa fever in Gueckedou, which represent to our knowledge the first confirmed cases of severe Lassa fever reported from this country. The high case fatality rate and the substantial number of hemorrhagic complications in our small series of patients clearly demonstrates the existence of severe Lassa fever in the Republic of Guinea. Sequencing of the PCR products revealed that most patients were infected with the Josiah strain of Lassa virus. However, patient LG32 was infected with a new Lassa variant, showing 11.6% mutations on the nucleotide and 10.4% mutations on the amino acid levels. We have thus shown that both our PCR and recombinant tests are able to pick up infections with Guinean Lassa virus variants which are as distantly related to the Josiah strain as is the Nigerian strain of Lassa (10% variation in the NP on the amino acid level).
In conclusion, we have devised a simple and rapid immunoblot assay with a single, easily expressed and purified Lassa virus protein. The test detects IgG and IgM LVA directed against linear epitopes of the NP of Lassa virus, strain Josiah, with a sensitivity and specificity comparable to those of the IIF assay for both acute- and convalescent-phase sera. The LVA immunoblot assay seems useful for serological studies, being capable of detecting low-titer antibodies which are missed by the routinely used cutoff of 1:20 in the IIF assay. Performed in the south of the Republic of Guinea, the IgM immunoblot assay had a positive predictive value of 56% and a negative predictive value of 93% for accurately identifying patients with acute Lassa fever among cases of FUO. Inclusion of the NPs of other Lassa virus variants (e.g., LG32 and the Nigeria strain) should increase the sensitivity of the test and render it useful for field purposes in other parts of West Africa. Given the absence of any simple Lassa fever test for field conditions to date, this test should be of some help to the clinician working in areas where Lassa fever is endemic.
REFERENCES
- 1.Barber G N, Clegg J C S, Chamberlain J. Expression of Lassa virus nucleocapsid protein segments in bacteria: purification of high-level expression products and their application in antibody detection. Gene. 1987;56:137–144. doi: 10.1016/0378-1119(87)90166-1. [DOI] [PubMed] [Google Scholar]
- 2.Barber G N, Clegg J C S, Lloyd G. Expression of the Lassa virus nucleocapsid protein in insect cells infected with a recombinant baculovirus: application to diagnostic assays for Lassa virus infection. J Gen Virol. 1990;71:19–28. doi: 10.1099/0022-1317-71-1-19. [DOI] [PubMed] [Google Scholar]
- 3.Boom R, Sol C J A, Salimans M M M, Jansen C L, Wertheim-van Dillen P M E, van der Noordaa J. Rapid and simple method for purification of nucleic acids. J Clin Microbiol. 1990;28:495–503. doi: 10.1128/jcm.28.3.495-503.1990. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Clegg J C S, Lloyd G. Structural and cell-associated proteins of Lassa virus. J Gen Virol. 1983;64:1127–1136. doi: 10.1099/0022-1317-64-5-1127. [DOI] [PubMed] [Google Scholar]
- 5.Clos J, Brandau S. pJC20 and pJC40—two high-copy-number vectors for T7 RNA polymerase-dependent expression of recombinant genes in E. coli. Protein Expr Purif. 1994;5:133–137. doi: 10.1006/prep.1994.1020. [DOI] [PubMed] [Google Scholar]
- 6.Fischer-Hoch S P, McCormick J B, Auperin D, Brown B G, Castor M, Perez G, Ruo S, Conaty A, Brammer L, Bauer S. Protection of rhesus monkeys from fatal Lassa fever by vaccination with a recombinant vaccinia virus containing the Lassa virus glycoprotein gene. Proc Natl Acad Sci USA. 1989;86:317–321. doi: 10.1073/pnas.86.1.317. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Hjelle B, Jenison S, Torrez-Martinez N, Herring B, Quan S, Polito A, Pichuantes S, Yamada T, Morris C, Elgh F, Lee H W, Artsob H, Dinello R. Rapid and specific detection of Sin Nombre virus antibodies in patients with hantavirus pulmonary syndrome by a strip immunoblot assay suitable for field diagnosis. J Clin Microbiol. 1997;35:600–608. doi: 10.1128/jcm.35.3.600-608.1997. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Hufert F T, Lüdke W, Schmitz H. Epitope mapping of Lassa virus nucleoprotein using monoclonal anti-nucleocapsid antibodies. Arch Virol. 1989;106:201–212. doi: 10.1007/BF01313953. [DOI] [PubMed] [Google Scholar]
- 9.Hummel K B, Martin M L, Auperin D D. Baculovirus expression of the glycoprotein gene of Lassa virus and characterization of the recombinant protein. Virus Res. 1992;25:79–90. doi: 10.1016/0168-1702(92)90101-e. [DOI] [PubMed] [Google Scholar]
- 10.Lukashevich I S, Clegg J C S, Sidiba K. Lassa virus activity in Guinea: distribution of human antiviral antibody defined using enzyme-linked immunosorbent assay with recombinant antigen. J Med Virol. 1993;40:210–217. doi: 10.1002/jmv.1890400308. [DOI] [PubMed] [Google Scholar]
- 11.Lunkenheimer K, Hufert F T, Schmitz H. Detection of Lassa virus RNA in specimens from patients with Lassa fever by using the polymerase chain reaction. J Clin Microbiol. 1990;28:2689–2692. doi: 10.1128/jcm.28.12.2689-2692.1990. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Marriot A C, Polyzoni T, Antoniadis A, Nuttall P A. Detection of human antibodies to Crimean-Congo haemorrhagic fever virus using expressed viral nucleocapsid protein. J Gen Virol. 1994;75:2157–2161. doi: 10.1099/0022-1317-75-9-2157. [DOI] [PubMed] [Google Scholar]
- 13.McCormick J B, King I J, Webb P A, Johnson K M, O’Sullivan R, Smith E S, Trippel S, Tong T C. A case-control study of the clinical diagnosis and course of Lassa fever. J Infect Dis. 1987;155:445–455. doi: 10.1093/infdis/155.3.445. [DOI] [PubMed] [Google Scholar]
- 14.McCormick J B, Webb P A, Krebs J W, Johnson K M, Smith E S. A prospective study of the epidemiology and ecology of Lassa fever. J Infect Dis. 1987;155:437–444. doi: 10.1093/infdis/155.3.437. [DOI] [PubMed] [Google Scholar]
- 15.Morrison H G, Bauer S P, Lange J V, Esposito J J, McCormick J B, Auperin D D. Protection of guinea pigs from Lassa fever by vaccinia virus recombinants expressing the nucleoprotein or the envelope glycoproteins of Lassa virus. Virology. 1989;171:179–188. doi: 10.1016/0042-6822(89)90525-4. [DOI] [PubMed] [Google Scholar]
- 16.Niklasson B S, Jahrling P B, Peters C J. Detection of Lassa virus antigens and Lassa virus-specific immunoglobulins G and M by enzyme-linked immunosorbent assay. J Clin Microbiol. 1984;20:239–244. doi: 10.1128/jcm.20.2.239-244.1984. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Qiagen. The QIA expressionist. Germany: Hilden; 1992. [Google Scholar]
- 18.Ruo S L, Mitchell S W, Kiley M P, Roumillat L F, Fischer-Hoch S P, McCormick J B. Antigenic relatedness between arenaviruses defined at the epitope level by monoclonal antibodies. J Gen Virol. 1991;72:549–555. doi: 10.1099/0022-1317-72-3-549. [DOI] [PubMed] [Google Scholar]
- 19.Schmitz, H., P. Emmerich, and J. ter Meulen. 1996. Imported tropical virus infections in Germany. Arch. Virol. 11(Suppl.):67–74. [DOI] [PubMed]
- 20.Sham P C, Curtis D. Monte Carlo tests for associations between disease and alleles at highly polymorphic loci. Ann Hum Genet. 1995;59:97–105. doi: 10.1111/j.1469-1809.1995.tb01608.x. [DOI] [PubMed] [Google Scholar]
- 21.ter Meulen J, Lukashevich I, Sidibe K, Inapogui A, Marx M, Dörlemann A, Yansane M L, Koulemou K, Chang-Claude J, Schmitz H. Hunting of peridomestic rodents and consumption of their meat as possible risk factors for rodent-to-human transmission of Lassa virus in the Republic of Guinea. Am J Trop Med Hyg. 1996;55:661–666. doi: 10.4269/ajtmh.1996.55.661. [DOI] [PubMed] [Google Scholar]
- 22.World Health Organization. Lassa fever in Sierra Leone. Weekly Epidem Rec. 1997;72:157–164. [Google Scholar]
- 23.Wulff H, Johnson K M. Immunoglobulin M and G responses measured by immunofluorescence in patients with Lassa or Marburg virus infections. Bull W H O. 1979;57:631–635. [PMC free article] [PubMed] [Google Scholar]