OUP user menu

Specific recognition of the major capsid protein of Acanthamoeba polyphaga mimivirus by sera of patients infected by Francisella tularensis

Nicolas Pelletier, Didier Raoult, Bernard La Scola
DOI: http://dx.doi.org/10.1111/j.1574-6968.2009.01675.x 117-123 First published online: 1 August 2009


Francisella tularensis, a Gram-negative cocobacillus responsible for tularemia, especially severe pneumonia, is a facultative intracellular bacterium classified as a biological agent of category A. Acanthamoeba polyphaga mimivirus (APM) is a recently discovered giant virus suspected to be an agent of both community- and hospital-acquired pneumonia. During specificity testing of antibody to APM detection, it was observed that nearly all patients infected by F. tularensis had elevated antibody titers to APM. In the present study, we investigated this cross-reactivity by immunoproteomics. Apart from the detection of antibodies reactive to new immunoreactive proteins in patients infected by F. tularensis, we showed that the sera of those patients recognize specifically two proteins of APM: the capsid protein and another protein of unknown function. No common protein motif can be detected in silico based on genome analysis of the involved protein. Furthermore, this cross-reactivity was confirmed with the recombinant capsid protein expressed in Escherichia coli. This emphasizes the pitfalls of a serological diagnosis of pneumonia.

  • tularemia
  • cross-reaction
  • Mimivirus
  • Western blot
  • protein of capsid


Francisella tularensis is a Gram-negative coccobacillus facultative intracellular bacterium that causes a serious disease known as tularemia in humans exposed to infected animals, arthropods or insects, and can be fatal. Because of its extremely low infectious dose and easy aerosolization, the Centers for Disease Control and Prevention have classified F. tularensis ssp. tularensis as a biological agent of category A. Inhalation of F. tularensis, the most likely route of infection for a biological weapon, can cause oropharyngeal and pneumonic tularemia (Ellis et al., 2002). Currently, the diagnosis of tularemia can be made either by culture and PCR (Splettstoesser et al., 2005) or by serology (Porsch-Ozcurumez et al., 2004). Several serological methods (indirect immunofluorescence, seroagglutination, microagglutination, hemagglutination and enzyme-linked immunofluorescent assay) have been used for detection of antibodies to F. tularensis, but the results obtained are frequently ambiguous (Porsch-Ozcurumez et al., 2004; Brouqui & Raoult, 2006). The most reliable and commonly used method is indirect immunofluorescence (Peacock et al., 1983). However, cross-reactions, mainly due to immunoglobulin M (IgM) antibodies, have been described with Brucella spp. (Behan & Klein, 1982), Proteus OX19 and Yersinia spp. (Larson et al., 1951; Saslaw & Carlisle, 1961).

In a recent work, we demonstrated that there was an association between pneumonia and the presence of antibodies to the giant Acanthamoeba polyphaga mimivirus (APM) in patients with pneumonia using a microimmunofluorescence assay (Raoult et al., 2004; Berger et al., 2006). When the specificity of our microimmunofluorescence was tested with sera from patients with diverse arthropod-borne diseases received in our laboratory, no cross-reactivity was observed, except for most patients having tularemia who presented elevated antibody titers to APM. This virus, which is the largest known to date, has a 1.2-Mb genome encoding 911 proteins, among which only 298 predicted functions and 114 proteins have been identified (Renesto et al., 2006). Analysis of its genome did not reveal any similarity to the F. tularensis genome. In recent years, several studies have reported an association between cases of tularemia and exposition to water, suggesting that protozoa may be the site of multiplication or starvation of F. tularensis (Abd et al., 2003; Holowecky et al., 2009). Possible coinfection of amebae by APM and F. tularensis appears as to be likely event, and thus we expected to find a common protein responsible for cross-reactivity as a consequence of lateral gene transfer in amebae. Another hypothesis was that it could not be a cross-reaction, but coexposition to microorganisms living in the same aquatic biotope. In the present study, we investigated the nature the cross-reaction between Mimivirus and F. tularensis ssp. holarctica strain URFTT1 (La Scola et al., 2008) using two-dimensional (2D) immunoblot and proteome analysis.

Materials and methods

Antigens, sera and microimmunofluorescence assay

Mimivirus was grown in A. polyphaga strain Linc AP-1 as described previously (La Scola et al., 2005). After most amebae were lyzed by APM, intact amebae were removed by low-speed centrifugation at 100 g for 15 min. Mimivirus particles present in the supernatant were centrifuged at 4000 g for 30 min and washed three times in phosphate-buffered saline (PBS). The pellet obtained after the last washing was then resuspended in PBS at a 2 mg mL−1 concentration of protein. Francisella tularensis URFT1 was cultivated on chocolate blood agar (Biomerieux, Marcy-l'Etoile, France) at 37 °C. After 5 days, bacteria were harvested, pelleted, resuspended in PBS and used as antigens at a concentration of 2 mg mL−1.

Antigens were applied with a dip-pen nib to each well of 30-well microscope slides (Dynatech Laboratories Ltd, Billingshurst, UK), air dried and fixed in acetone for 10 min. Sera were diluted at 1/25, 1/50 and 1/100 in PBS with 3% nonfat dry milk and applied to the antigens on the slides, which were incubated in a moist chamber for 30 min at 37 °C, followed by three 10-min washes in PBS. After air drying, bound antibody was detected using a fluorescein isothiocyanate-conjugated goat anti-human total immunoglobulin (Fluoline H, Biomerieux) diluted at 1 : 300 in PBS. Incubation, washing and drying were performed as described above. The slides were mounted in buffered glycerol (Fluoprep, Biomerieux) and read with a Zeiss epifluorescent microscope at × 400 magnification. After this screening for total immunoglobulins, serial twofold dilutions from 1 : 25 to 1 : 3200 were made of positive sera and the titers of reactive IgG and IgM were determined as above using goat anti-human IgG (Fluoline G, Biomerieux) or IgM (Fluoline M, Biomerieux). To remove IgG, rheumatoid factor adsorbent (RF-absorbent, Behringwerke AG, Marburg, Germany) was used according to the manufacturer's instructions before IgM determination. On each test slide, positive control serum (positive patient) and negative control (PBS with 3% nonfat dry milk instead of diluted serum) were added.

For this study, we used the sera of 38 patients with tularemia, 35 from France and three from Sweden. All had clinical presentation compatible with tularemia and were tested positive by microimmunofluorescence. Of these, 10 were considered definitive as they were confirmed by molecular detection of F. tularensis and/or culture. These 10 sera (Table 1) were later used for Western blot studies. We also used three sera with antibodies to APM: one serum was from a patient with accidental laboratory contamination (Raoult et al., 2006) and sera from rabbit and mouse, both immunized with APM.

View this table:

Antibody titers of patients to Francisella tularensis and APM

Francisella tularensisAPM
Patients with F. tularensis infection
11 : 8001 : 200<1 : 251 : 200
21 : 4001 : 251 : 50<1 : 25
31 : 2001 : 200<1 : 25<1 : 25
41 : 16001 : 50<1 : 251 : 200
51 : 8001 : 25NANA
61 : 8001 : 8001 : 2001 : 800
71 : 16001 : 4001 : 50<1 : 25
81 : 16001 : 400<1 : 25<1 : 25
91 : 4001 : 2001 : 1001 : 400
101 : 4001 : 200<1 : 25<1 : 25
Subjects with APM infection
Human<1 : 25<1 : 251 : 800<1 : 25
Mouse<1 : 25<1 : 251 : 1600NA
Rabbit<1 : 25<1 : 251 : 1600NA
  • NA, not available.

1D electrophoresis of protein of capsid D13L

We used the recombinant protein of capsid (D13L) of Mimivirus (S. Azza et al., unpublished data) for testing Mimivirus-infected patients, F. tularensis-infected patients and sera of a control group (healthy blood donors). After protein expression in E. coli and separation on gel sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (7.5%), followed by Coomassie blue staining, the capsid protein was extracted from the polyacrylamide gel by the electroelution method using the ElutaTube Protein Extraction Kit (Euromedex, Mundolsheim, France) according to the manufacturer's protocol. Recombinant protein was mixed 1 : 1 with 2 × concentrated SDS sample buffer and heated at 95 °C for 5 min. SDS-PAGE was performed in a separating 11.25% polyacrylamide gel according to the Laemmli method (Cleveland et al., 1977). Proteins, 1 mg per gel, were separated electrophoretically using a Mini-Protean II cell (Bio-Rad Laboratories, Richmond, CA) as described above.

2D electrophoresis and matrix-assisted laser desorption/ionization-time-of-flight (MALDI-TOF) of F. tularensis and Mimivirus

APM was purified and the protein fraction was obtained as described by Renesto et al. (2006), and then stored at −80 °C until isoelectric focusing (IEF) was performed. Francisella tularensis ssp. holarctica strain URFT1 (La Scola et al., 2008) was cultivated on chocolate blood agar, harvested after 5 days, suspended in PBS and lysed by sonication. The insoluble fraction was removed by centrifugation (12 000 g, 4 °C, 10 min) and proteins were prepared as for Mimivirus.

Immobiline DryStrips (7 cm, pH 3–10, Amersham Biosciences) were rehydrated overnight with 20 μg of a whole-cell extract (of Mimivirus or F. tularensis) in a rehydration solution supplemented with 0.5% v/v IPG buffer (pH 3–10) (Amersham Biosciences). IEF was carried out according to the manufacturer's instructions (Multiphor II System, Amersham Biosciences). For the next steps, we used the same protocol as Renesto et al. (2006). Following migration, gels were used for Western blot or stained by a method compatible with MS. The molecular weight was determined by running standard protein markers (LMW, Bio-Rad Laboratories). The 2D gels thus stained were digitalized by transmission scanning (ImageScanner, Amersham Biosciences). For identification of F. tularensis proteins by MALDI-TOF, each immunoreactive gel spot was excised from gels and stored at −20 °C. Before MS analysis, the protein spots were subjected to in-gel digestion with trypsin. Digested peptides were extracted from the gels and applied for analysis as described previously (Renesto et al., 2005). For identification of Mimivirus proteins, we used mapping performed in our laboratory in previous studies (Renesto et al., 2006; Raoult et al., 2007).

Western blotting

For Western blot, the proteins of F. tularensis and Mimivirus resolved by two-dimensional electrophoresis (2-DE) were transferred onto a nitrocellulose membrane in a transblot cell (Bio-Rad Laboratories) at 100 V for 1 h 20 min in an ice bath. Polyacrylamide gels were also stained with silver stain as described previously (Nesterenko et al., 1994) to check the quality of migration. Membranes were then blocked in PBS supplemented with 0.2% Tween and 5% nonfat dried milk (blocking buffer) overnight, before incubation with the serum of a patient (dilution 1 : 500) for 2 h. After this step, the membranes have been treated as described previously by Renesto et al. (2006) with anti-human secondary antibodies. A software was used for the analyses of the stained 2D gels and blot images (imagemaster 2D Platinum version 6.0, GE Healthcare). One gel was silver stained and used as a reference map. For 2D gels, analyses included spot detection and matching.


Microimmunofluorescence results (Table 1)

Among the 38 tularemia sera tested, 20 presented cross-reactivity to APM. All human sera of patients with tularemia tested by microimmunofluorescence had titers of ≥1 : 200 (IgG). All these sera also had titers of ≥1 : 25 for IgM. Among the sera from patients with a definitive diagnosis of tularemia, 6/10 were positive for APM, two for IgG and IgM, two for IgG only and two sera for IgM only. The serum of the patient infected by APM was positive for IgG with titers ≥1 : 800 to APM, but this serum tested by microimmunofluorescence on F. tularensis antigen was negative. The IgG titer of both APM-immunized mouse and rabbit had a titer of 1/1600 to APM.

Immunoreactivity pattern of APM sera and sera from patients and rabbit with tularemia to APM proteins revealed by 2DE immunoblots (Table 2, Fig. 1)

View this table:

Summary of Mimivirus proteins immunoblot with sera from patients with tularemia, Mimivirus infection and immunized animals

ProteinsL425 (capsid protein D13L)L725 (ORFan)L442 (ORFan)R135 (choline dehydrogenase)
Mimivirus sera
Human sera
Mouse sera
Rabbit sera
Tularemia sera
  • We have noted (•) when the protein is detected by Western blot (and known with previous studying) and (▪) when the protein is not found previously.


2D immunoblotting patterns of the immunoreactive proteins of Mimivirus. Silver-stained 2D reference map for control of a whole viral particle lysate (a) probed with sera from a Mimivirus-infected patient (b), mouse immunized with Mimivirus (c), rabbit immunized with Mimivirus (d), patient 8 with tularemia (e) and with a laboratory Francisella tularensis-immunized rabbit (f).

MIMI_L425 (the capsid protein), MIMI_R135 (choline dehydrogenase) and MIMI_L725 (unknown function) were recognized by human and animal sera. MIMI_L442 was also recognized by the human sera and by the sera of mice experimentally infected with APM, but not by rabbit serum. The 10 sera from patients with tularemia recognized different proteins of Mimivirus whole-cell lysate. All 10 reacted with the capsid protein D13L (MIMI_L425) even if only 6/10 were observed to be cross-reactive by microimmunofluorescence, nine reacted with the protein encoded by MIMI_L725 (unknown function), three reacted with the protein encoded by MIMI_L442 (unknown function) and one reacted with choline dehydrogenase encoded by MIMI_R135. An F. tularensis-immunized rabbit also recognized proteins of APM (Fig. 1f). No reaction was detected with the sera obtained from healthy blood donors.

Immunoreactivity pattern of patients with tularemia to recombinant protein of capsid D13L

SDS-PAGE performed with recombinant protein D13L showed a band at 80 kDa, which corresponded to the recombinant protein D13L (Fig. 2), the level corresponding to the capsid protein that has a theoretical mass of about 70 kDa (when using Bio-rad Laboratories Original Prestained SDS-PAGE standards). Western blots performed with the sera from the patient infected with APM and rabbit serum showed reactivity to 80-kDa protein. The Western blots obtained for Francisella-infected patients (except patient 5, for whom we did not have enough serum to perform the test) showed reactivity to this 80-kDa band for all tested sera. No reaction was detected with sera obtained from healthy blood donors. No reaction was also observed with mouse serum immunized with Mimivirus.


Recombinant capsid D13L on 11.25% SDS-PAGE blue-stained 1D (RCP), representative 1D immunoblot with rabbit Mimivirus serum (R), mouse Mimivirus serum (M) and tularemia sera from patients 1 to 10 (patient 5 not tested).

Immunoreactivity pattern of patients with tularemia to F. tularensis proteins revealed by 2-DE immunoblots

The sera of the 10 patients with tularemia tested on F. tularensis as a control showed that 20 different proteins of the whole-cell bacterial lysate were immunoreactive (Table 3 and Fig. 3). No reaction was detected with sera obtained from healthy blood donors.

View this table:

Summary of Francisella tularensis proteins immunoblot with sera of patients with tularemia

(10) Chaperone Hsp90, heat shock protein HtpG (FTL_0267)
(11) Chaperone protein dnaK (FTL_1191)
(15) Peroxidase/catalase (FTL_1504)
(16) Peroxidase/catalase (FTL_1504)
(25) groEL chaperone (FTL_1714)
(30) 3OS ribosomal protein S1 (FTL_1912)
(32) Succinate dehydrogenase, catalytic and NAD/flavoprotein subunit (FTL_1786)
(36) ATP synthase alpha chain (FTL_1797)
(55) UDP-glucose/GDP-mannose dehydrogenase (FTL_0596)
(62) 3-oxoacyl-[acyl-carrier-protein] synthase II (FTL_1137)
(66) NAD(P)-specific glutamate dehydrogenase (FTL_0269)
(75) Cell division protein FtsZ (FTL_1907)
(86) Citrate synthase (FTL_1789)
(118) Glyceraldehyde-3-phosphate dehydrogenase (FTL_1146)
(130) Universal stress protein (FTL_0166)
(148) Purine nucleoside phophorylase (FTL_1461)
(175) Intracellular growth locus, subunit C (FTL_1159)
(185) Acetyl-CoA carboxylase, biotin carboxyl carrier protein subunit (FTL_1592)
(197) 50S ribosomal protein L7/L12 (FTL_1745)
(204) Hypothetical protein FTL_0617 (FTL_0617)
(231) ATP synthase β subunit (FTL_1795)
  • For definition of symbols, see Table 2.


2D immunoblotting patterns of the immunoreactive proteins of Francisella tularensis. Silver-stained 2D reference map for control of whole-cell lysate (a) probed with sera of patient 9 (b) and patient 1 (c), both with tularemia.


In this study, we investigated the cross-reactions between sera from patients infected by F. tularensis and APM. Immunoproteomic analysis with antibodies to APM was in agreement with previous studies performed in our laboratory (Raoult et al., 2006; Renesto et al., 2006). From 2D gel Western blots performed with the sera of F. tularensis patients, the capsid protein D13L (MIMI_L425) was recognized in all cases, even by sera 3, 8 and 10, for which no reactions were observed by microimmunofluorescence. The Western blots performed on the recombinant protein D13L confirmed these data and provided evidence that the cross-reactivity was due to a protein antigenic compound and not due to carbohydrates as observed previously with other bacteria such as Proteus and Brucella (Saslaw & Carlisle, 1961; Behan & Klein, 1982), or Shigella (Pozsgay, 1998; 1999). A peculiar finding was the lack of reactivity of mouse serum immunized with Mimivirus with recombinant protein, whereas it was recognized on 2D immunoblot. It was previously shown that the capsid protein D13L was not resolved into a single spot, but rather as a train of spot and that it was glycosylated (Renesto et al., 2006). We believe that the difference in reactivity of the serum of immunized mouse serum between recombinant protein and native protein is due to post-transcriptional modifications such as glycosylation. However, recognition by immunized rabbit, APM-infected human and tularemia patients demonstrates that this protein is responsible for cross-reactivity.

In silico, we could not found any homology between the capsid protein D13L with F. tularensis, especially with immunoreactive proteins identified on the Western blot. If peptides comprised of a short succession of amino acid residues are usually the support of antigenicity to protein antigens, they may not be contiguous, comprising atoms from distant residues, but close in 3D space and on the surface of the protein. Because of this, structural methods, especially X-ray crystallography, are the best approaches to identify epitopes (Van Regenmortel, 1989; Gershoni et al., 2007). Our first hypothesis to explain why patients with tularemia have antibodies to APM was that infection occurred from the same reservoir as F. tularensis is suspected, similar to APM, to infect humans from a water reservoir (Holowecky et al., 2009). However, all patients with tularemia recognize the capsid protein D13L, except one, Choline dehydrogenase (R135), which was recognized by all animals infected by APM as by humans with APM infection. Moreover, we observed a comparable pattern of reactivity to APM as patients in an F. tularensis laboratory-infected rabbit with the recognition of the capsid protein D13L (Fig. 1f). Thus, this is cross-reactivity rather than coinfection.

The efficiency of the immunoproteomic procedure used in this study was confirmed by data on the proteins of this bacterium obtained on the sera of patients infected by F. tularensis (Table 3, Fig. 3). The results were in agreement with those of previous studies (Havlasova et al., 2002; Twine et al., 2006; Janovska et al., 2007). We even identified additional proteins that were not previously recognized as immunoreactive, especially the NAD(P)-specific glutamate dehydrogenase, which is among proteins that are overproduced in the subspecies tularensis compared with subspecies holarctica (Pavkova et al., 2006) and among candidates that could explain differences in the virulence between the two subspecies.

In conclusion, this study emphasizes the pitfalls of a serological diagnosis of respiratory pathogens. If testing of specific epitopes in immunoenzymatic assays is supposed to improve specificity as compared with whole-agent testing such as microimmunofluorescence, the results presented here show that this does not guarantee lack of cross-reactions with other pathogens responsible for the same clinical picture.


  • Editor: Jan-Ingmar Flock


View Abstract