|Home | About | Journals | Submit | Contact Us | Français|
In felids, three hemotropic mycoplasma species (hemoplasmas) have been described: Mycoplasma haemofelis, “Candidatus Mycoplasma haemominutum,” and “Candidatus Mycoplasma turicensis.” In particular, M. haemofelis may cause severe, potentially life-threatening hemolytic anemia. No routine serological assays for feline hemoplasma infections are available. Thus, the goal of our project was to identify and characterize an M. haemofelis antigen (DnaK) that subsequently could be applied as a recombinant antigen in a serological assay. The gene sequence of this protein was determined using consensus primers and blood samples from two naturally M. haemofelis-infected Swiss pet cats, an experimentally M. haemofelis-infected specific-pathogen-free cat, and a naturally M. haemofelis-infected Iberian lynx (Lynx pardinus). The M. haemofelis DnaK gene sequence showed the highest identity to an analogous protein of a porcine hemoplasma (72%). M. haemofelis DnaK was expressed recombinantly in an Escherichia coli DnaK knockout strain and purified using Ni affinity, size-exclusion, and anion-exchange chromatography. It then was biochemically and functionally characterized and showed characteristics typical for DnaKs (secondary structure profile, thermal denaturation, ATPase activity, and DnaK complementation). Moreover, its immunogenicity was assessed using serum samples from experimentally hemoplasma-infected cats. In Western blotting or enzyme-linked immunosorbent assays, it was recognized by sera from cats infected with M. haemofelis, “Ca. Mycoplasma haemominutum,” and “Ca. Mycoplasma turicensis,” respectively, but not from uninfected cats. This is the first description of a full-length purified recombinant feline hemoplasma antigen that can readily be applied in future pathogenesis studies and may have potential for application in a diagnostic serological test.
Hemotropic mycoplasmas (hemoplasmas) are small (0.3 to 0.8 μm) epierythrocytic bacteria, which previously have been known as Haemobartonella and Eperythrozoon species. In felids, Mycoplasma haemofelis, “Candidatus Mycoplasma haemominutum,” and “Candidatus Mycoplasma turicensis” have been described (5, 6, 19, 36). They vary in their pathogenicity, responsiveness to antimicrobial drugs, and probably in their ability to form a carrier state (5, 26, 36). M. haemofelis in particular may cause severe, potentially life-threatening hemolytic anemia (5).
Real-time PCR assays are the tools of choice for diagnosing and differentiating feline hemoplasma infections (29, 36). However, they may not detect all hemoplasma infections, e.g., due to fluctuating M. haemofelis bacteremia (29), reduced bacterial blood loads after antibiotic treatment (5), or chronic carrier status of infected animals with undetectable numbers of circulating hemoplasmas (28). To overcome the resulting diagnostic gap and to further characterize the course and pathogenesis of feline hemoplasma infections, a diagnostic assay based on serum antibody detection would be desirable.
To date, no routine serological assays for the diagnosis of feline hemoplasma infections are available. The development of such assays has been significantly hampered by the fact that hemoplasmas cannot be cultured in vitro, and therefore antigens have had to be produced by the experimental infection of cats with hemoplasmas. Experimental serological assays have been described using hemoplasma antigen either on blood smears (5) or purified from large volumes of blood (1) from infected cats. Western blot analyses of Haemobartonella felis antigen preparations resulted in the identification of five antigens recognized by sera from experimentally H. felis-infected cats (1). A recent study identified M. haemofelis antigens in crude antigen preparations from erythrocytes collected from an experimentally infected cat (21). Those antigens reacted with plasma antibodies of cats collected at different time points after experimental infection when applied in Western blot analyses. The first recombinant hemoplasma antigen, Mycoplasma suis HspA1, was developed during a study of experimentally M. suis-infected pigs for application in Western blotting and enzyme-linked immunosorbent assays (ELISAs) (10). This antigen belongs to the heat shock protein 70 (HSP70) family. It was found to be DnaK-like and present on the surface of M. suis (10). DnaKs are molecular chaperones consisting of an N-terminal nucleotide-binding domain (ATPase activity) that generates the energy necessary to refold misfolded proteins in cell stress situations (9). Misfolded proteins bind to the C-terminal substrate-binding domain of DnaKs. Most recently, we developed a recombinant feline hemoplasma antigen to demonstrate the seroconversion of experimentally “Ca. Mycoplasma turicensis”-infected cats in preliminary Western blot analyses (18). The antigen described was a truncated M. haemofelis DnaK form recombinantly expressed in E. coli that was only partially purified, leading to large interbatch variations with regard to quality and purity. The described assay did not allow for the quantification of antibody levels.
The aim of this study was to identify the complete DnaK gene of M. haemofelis, to recombinantly produce, highly purify, and characterize the antigen, and to apply it in an ELISA as a serological tool for the detection and quantification of the humoral immune response during experimental feline hemoplasma infection.
All animals from which samples have been used during this study are listed in Table Table1.1. For sequencing purposes, samples from the following six hemoplasma-infected felids were used: the experimentally M. haemofelis-infected specific-pathogen-free (SPF) cat QLA5 (Liberty Research, Waverly, NY), the naturally M. haemofelis-infected Swiss domestic pet cats 1008 and 7415 (35), the free-living Iberian lynxes (Lynx pardinus) Dalia and Cicuta, which were naturally infected with M. haemofelis and “Ca. Mycoplasma haemominutum,” respectively (16), and the experimentally “Ca. Mycoplasma turicensis”-infected SPF cat Y (18). For the ELISA, pre- and postinfection samples from a total of 20 SPF cats were used, including the M. haemofelis-infected cat QLA5, 8 “Ca. Mycoplasma haemominutum”-infected cats, such as cat 09NFR2, and 11 “Ca. Mycoplasma turicensis”-infected cats, such as cat Y (Table (Table11).
The experimental infections of the “Ca. Mycoplasma haemominutum”- and “Ca. Mycoplasma turicensis”-infected cats have been described earlier (7, 18). For experimental M. haemofelis infection, the SPF cat QLA5 was inoculated intraperitoneally at the age of 2.7 years with 2 ml of dimethylsulfoxide (DMSO)-preserved (20%, vol/vol) M. haemofelis-positive blood from the experimentally infected cat HF3 (31). The inoculum contained 109 M. haemofelis copies/ml, as determined by TaqMan real-time PCR (35), and had been stored at −80°C until use. All SPF cats were kept in groups (QLA5 was kept together with a female neutered SPF companion cat) and examined clinically prior to the study, and their SPF status was verified as described previously (18).
After hemoplasma inoculation, EDTA-anticoagulated whole-blood samples were collected regularly, hemograms were generated using a Cell-Dyn 3500 (Abbott; Baar/Switzerland), and the quantification of hemoplasma blood loads was performed by TaqMan real-time PCR (35). Serum or plasma samples were collected for serological analyses (see below). Anemia was defined as a hematocrit value of less than 33% (equal to a 5% quantile of the reference range determined in our laboratory using identical methods and blood samples from 58 clinically healthy cats). EDTA-anticoagulated blood samples collected from cat QLA5 (4.4 × 108 copies/ml blood), cat 1008 (2.8 × 108 copies/ml blood), cat 7415 (8.0 × 106 copies/ml blood), and from lynx Dalia (6.6 × 104 copies/ml blood) 10 days postinfection (dpi) were used for M. haemofelis DnaK gene amplification and sequencing. DNA from 1 ml of blood of cat QLA5 was extracted manually using the QIAmp DNA blood mini kit (Qiagen, Hombrechtikon, Switzerland). Total nucleic acids from cats 1008, 7415, and lynx Dalia were extracted from 200 μl of blood using the MagNa pure LC total nucleic acid isolation kit I (Roche Diagnostics, Reinach, Switzerland).
Based upon the sequence information of M. haemofelis DnaK gene fragments AY150993 (303 bp) and FJ463263 (899 bp) and of all other mycoplasma DnaK gene sequences available from the GenBank database until June 2009, several consensus primer pairs were designed manually and tested for the amplification of the potential M. haemofelis DnaK gene. Using the primer pairs F1-35Mp/R934-956Mhf and F600-623Var/R1746-1768Ms (Table (Table2),2), the M. haemofelis DnaK gene was amplified as two overlapping fragments of 976 and 1,307 bp, respectively. The following thermal cycling conditions were applied: initial denaturation at 98°C for 180 s, 35 cycles of 98°C for 10 s, 60 (976 bp) or 63°C (1,307 bp) for 30 s, 72°C for 60 s, and then a final elongation at 72°C for 10 min. All DNA amplification steps of this study were performed using Phusion high-fidelity DNA polymerase (Finnzymes, Espoo, Finland). The resulting PCR products of the expected lengths were extracted from agarose gels using the NucleoSpin extract II kit (Macherey-Nagel, Düren, Germany) according to the manufacturer's instructions. Gene sequences of the PCR products were determined by DNA sequencing (Microsynth, Balgach, Switzerland). Resulting M. haemofelis DnaK gene and deduced protein sequences were compared to nonredundant nucleotide and amino acid database sequences using the BLASTN and BLASTP algorithms (http://blast.ncbi.nlm.nih.gov/Blast.cgi) and were aligned to the mycoplasma DnaK sequences of species shown in Fig. Fig.11 using ClustalW2 (11). The phylogenetic tree of mycoplasma DnaK protein sequences was constructed using MEGA version 4 (27). Bootstrap support (1,000 replicates) was calculated by the neighbor-joining method.
The recombinant M. haemofelis DnaK (M. haemofelis rDnaK) gene was obtained in two steps: first, the M. haemofelis DnaK gene was amplified as two overlapping fragments from DNA extracted from blood of cat QLA5 (using primer pairs F1-35Mp/R934-956Mhf and F666-691Mhf/R1750-1783Mhf; Table Table2).2). The resulting 976- and 1,119-bp-long PCR products were extracted from agarose gels as described above and assembled in a second step, which combined 10 ng of each of those fragments in an overlap extension PCR (50-μl reaction volume) using the overhang primers FDnaKMhfpET and RDnaKMhfpET (Table (Table2).2). Those primers also inserted NdeI and XhoI cleavage sites at the 5′ and 3′ ends of the gene, respectively. The following cycling conditions were used: 5 cycles with an annealing temperature of 85°C, followed by the addition of primers, and 35 cycles with an annealing temperature of 60°C (remaining cycling conditions were as described above). The resulting PCR product (1,824 bp) was extracted from an agarose gel, digested with restriction enzymes NdeI and XhoI (New England Biolabs, Ipswich, MA) according to the manufacturer's instructions, and ligated to the 4,690-bp XhoI-NdeI fragment of vector pMG211 (24). The vector contained an ampicillin resistance gene, a salicylate-inducible promoter, and the genetic information for a C-terminal 6×His tag followed by a stop codon. The correct M. haemofelis rDnaK gene sequence within pMG211 was confirmed by DNA sequencing before protein production. For the DnaK complementation assay, the gene of E. coli chorismate mutase (EcCM) (12, 25) was NdeI and XhoI digested and ligated to the XhoI-NdeI fragment of plasmid pMG211 as described above.
Plasmid pMG211, containing the M. haemofelis rDnaK gene (naturally without UGA readthroughs), was transformed into the recA-deficient Escherichia coli strain XL1 blue (Stratagene, LaJolla, CA) for plasmid storage and multiplication. Cells were grown on LB agar containing 150 μg/ml ampicillin and in LB medium containing 200 μg/ml ampicillin at 37°C and 250 rpm, respectively. Plasmid DNA was purified using a Jetquick Plasmid Miniprep Spin kit (Genomed, Löhne, Germany). Transformed XL1 blue cells were stored as glycerol stocks at −80°C. For protein production, plasmid pMG211 M. haemofelis rDnaK was transformed into the kanamycin-resistant strain JW0013, an in-frame DnaK knockout mutant of E. coli K-12 (2). Preparative cultures were inoculated from overnight starter cultures and grown at 30°C and 250 rpm in LB medium containing 150 μg/ml ampicillin and 25 μg/ml kanamycin. Gene overexpression was induced with 1 mM salicylate at an optical density at 600 nm (OD600) of 0.6, and the culture was incubated for an additional 20 h at 25°C and 250 rpm. After cell lysis using 1 mg/ml lysozyme and ultrasonication, protein was purified from the soluble fraction by affinity chromatography on Ni2+-nitrilotriacetic acid (NTA) agarose (Qiagen). The purification progress was assessed by SDS-PAGE analysis (see below) after each purification step. Fractions containing monomeric M. haemofelis rDnaK were isolated by size-exclusion chromatography on a calibrated Superdex 200 10/300 GL column (Amersham Pharmacia Biotech, Uppsala, Sweden) in Tris-buffered saline (TBS), pH 7.4. Those fractions then were subjected to anion-exchange chromatography on a Mono Q HR 16/10 column (Amersham Pharmacia Biotech) in TBS, pH 7.4, using a salt gradient from 150 to 500 mM NaCl. Fractions containing protein with a molecular mass of about 66 kDa were combined and concentrated using Amicon Ultra Centrifugal Filter 10 K (Millipore, Carrigtwohill, Cork, Ireland), and their protein concentration was determined by the Bradford assay (Coomassie plus protein assay reagent calibrated with bovine serum albumin [BSA]; Thermo Scientific, Rockford, IL).
The molecular mass of M. haemofelis rDnaK protein was determined at the protein service unit of the Functional Genomics Center Zurich (FGCZ), University of Zurich, Zurich, Switzerland. The purified protein solution was analyzed using electrospray ionization mass spectrometry. The experimentally determined molecular mass then was compared to the mass calculated by the ProtParam tool (www.expasy.ch/tools/protparam.html) based on the deduced protein sequence of M. haemofelis rDnaK.
Circular dichroism (CD) spectroscopy was performed on an Aviv circular dichroism spectrometer model 202 (Aviv Instruments Inc., Lakewood, NJ) in quartz cuvettes of 0.2-cm path length (d). Far-UV spectra were recorded from 260 to 200 nm in 1-nm steps at 25°C and a 1 μM M. haemofelis rDnaK protein concentration (c) in TBS, pH 7.4 (50 mM Tris base, 150 mM NaCl). For stability studies, KCl (100 mM) together with MgCl2 (2.5 mM) and/or ATP (0.1 mM) were added. Data were collected for 5 s at each step. Five scans were averaged, and buffer spectra determined under identical conditions were subtracted. The observed ellipticity (θλ) at wavelength λ was transformed into molar ellipticity per residue (θm,r) using equation 1 (where n is the number of residues), resulting in θm,r = θλ/(10·c·d·n).
Thermal denaturation experiments were performed in TBS, pH 7.4, at a 1 μM protein concentration by monitoring the CD signal at 222 nm from 10 to 95°C and reverse in 0.5 K steps with 60 s of temperature equilibration, 60 s of data collection, and a 1-K-per-min heating/cooling rate between temperature steps. Tm, the melting point of the M. haemofelis rDnaK ATPase domain, was defined as the inflection point of the melting curve and was determined from the first derivative after curve smoothing using the SigmaPlot v11.0 software package (Systat Software Inc., Richmond, CA).
ATPase activity was measured using a spectrophotometric assay (32), which quantified the released amount of inorganic phosphate (Pi) during ATP hydrolysis. The reaction of 2-amino-6-mercapto-7-methylpurine-ribonucleoside (MESG) with Pi was catalyzed by the purine nucleoside phosphorylase (PNP; Sigma-Aldrich, Buchs, Switzerland) and led to a measurable change in absorbance at 360 nm. MESG was synthesized and purified as previously published (32). All measurements were performed at 25°C and 1 U/ml PNP. The change in absorbance at 360 nm was calibrated to the Pi concentration using MESG (190 μM) and Pi (0, 1, 5, 25, and 50 μM; from a 200 μM Na2HPO4 solution). Michaelis-Menten kinetic measurements of M. haemofelis rDnaK then where performed under the following ATPase assay conditions: 50 mM Tris, 100 mM KCl, 2.5 mM MgCl2, 400 nM purified M. haemofelis rDnaK, and 380 μM MESG. The change in absorbance at 360 nm was measured in duplicate for 26 min at ATP concentrations of 10, 50, 100, 250, and 500 μM after an equilibration time of 4 min in a Lambda 35 spectrophotometer (PerkinElmer, Waltham, MA). Blank values measured without M. haemofelis rDnaK were subtracted, and the reaction rates were calculated. The catalytic parameters kcat and Km were determined from curve fitting to the Michaelis-Menten equation using the SigmaPlot v11.0 software package (Systat Software Inc.).
The E. coli DnaK knockout mutant strain JW0013 was transformed with plasmid pMG211 containing either M. haemofelis rDnaK or the EcCM gene (as a negative control). Liquid cultures were prepared from both transformants and incubated at 30°C and 250 rpm overnight. Cell densities of both cultures were adjusted to an OD600 of 3.0, and cultures then were diluted sequentially 10-fold down to 10−6. One 5-μl drop of each culture and dilution was placed on LB agar plates containing 150 μg/ml ampicillin, 25 μg/ml kanamycin, and 1 mM salicylate. Agar plates were incubated without ventilation at 30 or 43°C, respectively. Bacterial growth was assessed after 22.5 h by counting colonies.
SDS-PAGE and protein transfer to nitrocellulose membranes were performed in a PhastSystem high-speed electrophoresis system, a PhastSystem development unit, and a PhastTransfer semidry electrophoretic blotting unit (all Amersham Pharmacia Biotech). Recombinant M. haemofelis DnaK (540 ng/lane) was separated on PhastGel 20% (wt/vol) homogenous SDS polyacrylamide gels (Amersham Pharmacia Biotech), transferred, and Coomassie stained according to the manufacturer's instructions. Western blots were probed with pre- and postinfection serum or plasma samples from the experimentally infected cats QLA5 (M. haemofelis; prior to and 21 dpi), 09NFR2 (“Ca. Mycoplasma haemominutum”; prior to and 56 dpi), and Y (“Ca. Mycoplasma turicensis”; prior to and 109 dpi). Serum and plasma samples were diluted 1:100 in blocking buffer (150 mM NaCl, 10 mM Tris base, 20 g/liter skimmed milk powder). Serum antibodies bound to M. haemofelis rDnaK were visualized using peroxidase-conjugated, affinity-purified goat anti-cat IgG antibodies (diluted 1:2,000 in blocking buffer; Jackson ImmunoResearch Laboratories, West Grove, PA) and 4-chloro-1-napthtol as the chromogenic agent.
M. haemofelis rDnaK was heated in coating buffer (100 mM Na2CO3, 0.1% [wt/vol] SDS, pH 9.6) to 100°C for 1 min and then diluted 1:20 in coating buffer without SDS. Flat-bottomed 96-well microtiter plates with medium binding capacity (Greiner Bio-One, Frickenhausen, Germany) then were coated with 100 μl of this M. haemofelis rDnaK solution per well for 3 h at 37°C and overnight at 4°C. Plates subsequently were washed three times with 200 μl/well washing buffer (150 mM NaCl, 0.05% [vol/vol] Tween 20) and incubated for 1 h at 37°C with 100 μl/well blocking buffer (150 mM NaCl, 50 mM Tris base, 1 mM Titriplex III, 0.1% [wt/vol] BSA, 0.1% [vol/vol] Tween 20, pH 7.4). After being washed as described above, 100 μl of serum samples diluted in serum buffer (blocking buffer without BSA) was added per well and incubated for 1 h at 37°C. The plates were washed, and each well was filled with 100 μl of peroxidase-conjugated, affinity-purified goat anti-cat IgG antibodies (diluted 1:3,000 in serum buffer; Jackson ImmunoResearch Laboratories) and incubated for 1 h at 37°C. After being washed, 100 μl/well substrate solution [150 mM citric acid pH 4.0, 1% (vol/vol) H2O2 2%, 1% (vol/vol) 2,2′-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) diammonium salt] was added, and plates were incubated at room temperature for 10 min. Absorbance then was measured at a wavelength of 415 nm (OD415) using a Spectramax Plus 348 microplate spectrophotometer (Molecular Devices, Sunnyvale, CA).
Optimal serum dilutions and recombinant antigen concentrations were established by a checkerboard titration using pre- and postinfection sera from the three experimentally infected cats QLA5 (M. haemofelis; prior to and 21 dpi), 09NFR2C (“Ca. Mycoplasma haemominutum”; prior to and 56 dpi), and Y (“Ca. Mycoplasma turicensis”; prior to and 109 dpi) at serum dilutions of 1:50, 1:100, and 1:200. Antigen amounts of 200, 100, 50, and 10 ng/well were tested in duplicate. Wells containing only antigen without serum served as blanks, and wells containing preinfection serum samples served as negative controls.
Serum samples from 20 hemoplasma-free SPF cats before and after hemoplasma infection (Table (Table1)1) were tested using optimized conditions. For cat QLA5, samples taken during the whole course of experimental M. haemofelis infection (0 to 800 dpi) were assayed additionally. The signal-to-noise ratio was determined by dividing the postinfection ELISA OD415 values by the corresponding preinfection OD415 values.
Statistical analyses were performed using the SigmaPlot v11.0 software package (Systat Software Inc.). Correlations between ELISA antibody levels and M. haemofelis blood loads in the experimentally M. haemofelis-infected cat QLA5 were assessed using the Spearman rank-order correlation test.
The nucleotide and deduced amino acid sequences of M. haemofelis DnaK have been deposited in the GenBank database (HM594280, HM594281, and HM594283). Additionally, a partial “Ca. Mycoplasma haemominutum” DnaK sequence, acquired using identical methods, was submitted (HM594282).
The obtained sequences of the potential M. haemofelis DnaK gene included, according to BLASTN search results, an open reading frame (ORF) most similar to mycoplasma DnaK gene sequences (data not shown). Compared to mycoplasma DnaK complete gene sequences from the GenBank database, the highest identities with this ORF were found for M. suis DnaK (HspA1; 72%) and Mycoplasma penetrans DnaK (69%). Identities with the two partial M. haemofelis DnaK gene sequences FJ463263 (899 bp) and AY150993 (303 bp) were 99 and 97%, respectively. Accordingly, the ORF was named the M. haemofelis DnaK gene. Further comparisons of M. haemofelis DnaK gene sequences derived from four different M. haemofelis-infected animals resulted in three distinct M. haemofelis DnaK gene sequences: from cats 7415 and QLA5 an identical sequence was obtained (HM594280). Compared to the 7415/QLA5 sequence, sequences obtained from cat 1008 (HM594283) and lynx Dalia (HM594281) shared the same 21 silent point mutations as well as one point mutation causing a serine-to-glycine exchange at position 580. The sequence from lynx Dalia (HM594281) additionally contained a point mutation causing a proline-to-serine exchange at position 577. When the deduced M. haemofelis DnaK protein sequence (cats 7415 and QLA5; HM594280) was compared to mycoplasma DnaK sequences from the GenBank database using the BLASTP search algorithm, again M. suis (70%) and M. penetrans (65%) shared highest identities. The phylogenetic analyses of the M. haemofelis DnaK protein sequence revealed that it clustered within the haemofelis group of the hemoplasmas, which is distinct from other mycoplasma groups (Fig. (Fig.11).
Ni-affinity chromatography of crude extracts of salicylate-induced JW0013 pMG211 M. haemofelis rDnaK cells yielded one predominant protein band corresponding to a molecular mass of about 66 kDa (data not shown). After size-exclusion chromatography, anion-exchange chromatography, and protein concentration, M. haemofelis rDnaK was judged to be pure by SDS-PAGE (Fig. (Fig.2A2A).
Mass spectrometry analysis determined the molecular mass of M. haemofelis DnaK to be 66,406 Da, while the calculated molecular mass based on its deduced protein sequence was 66,537 Da. This difference in mass of 131 Da corresponds to the N-terminal loss of methionine during mass spectrometry analysis.
CD spectrum analysis of M. haemofelis rDnaK showed two distinct minima at 208 and 222 nm (Fig. (Fig.3A).3A). The CD spectrum of M. haemofelis rDnaK recorded without ATP or K+ and Mg2+ ions was not markedly different from those recorded with additives. The temperature-dependent CD signals at 222 nm showed the thermally induced unfolding of M. haemofelis rDnaK with well-defined (Tm1) and less-well-defined (Tm2) temperature transitions (Fig. (Fig.3B).3B). The melting temperature of the nucleotide-binding domain of M. haemofelis rDnaK without additives (Tm1a) was determined to be 42°C, the addition of the nucleotide ATP increased it to a Tm1c of 46°C, and ATP together with K+ and Mg2+ ions further increased it to a Tm1d of 50°C. The addition of K+ and Mg2+ ions alone caused only a minimal increase to a Tm1b of 43°C.
Enzymatic activity of the ATPase domain of M. haemofelis rDnaK was determined in a spectrophotometric assay. Fitting reaction rates at various substrate concentrations to the Michaelis-Menten equation yielded catalytic parameters for ATP hydrolysis (Fig. (Fig.3D3D).
During the DnaK complementation assay, the in vivo biological function of DnaKs, to allow bacterial growth at cell-stressing temperatures by repairing damaged enzymes, was tested for M. haemofelis rDnaK. No difference in the extent of bacterial growth between M. haemofelis rDnaK and the EcCM transformant could be seen for any of the tested culture dilutions at 30°C (Fig. (Fig.3D).3D). At 43°C, however, cells overexpressing EcCM and lacking DnaK protein grew only until a dilution of 10−1 (20 colonies), while for rDnaK-expressing cells growth until a 10−3 dilution (1 colony) was observed (Fig. (Fig.3D3D).
Western blot analyses of M. haemofelis rDnaK showed that the protein was recognized by serum antibodies from cats experimentally infected with M. haemofelis (cat QLA5), “Ca. Mycoplasma haemominutum” (cat 09NFR2), and “Ca. Mycoplasma turicensis” (cat Y). Preinfection serum or plasma samples from the same cats did not result in a positive Western blot signal (Fig. (Fig.2B2B).
For the M. haemofelis rDnaK ELISA, 50 ng M. haemofelis rDnaK/well and a dilution of 1:200 for sera from M. haemofelis-infected cats and 1:100 for sera from “Ca. Mycoplasma haemominutum”- and “Ca. Mycoplasma turicensis”-infected cats were found to be the optimal conditions. OD415 values for serum samples from 20 SPF cats prior to hemoplasma infection ranged from 0.12 to 0.33 under these conditions. After experimental infection, for the M. haemofelis-infected SPF cat QLA5 the OD415 peaked at 1.4. For the eight “Ca. Mycoplasma haemominutum”-infected SPF cats (Table (Table1),1), the OD415 peak values ranged from 0.5 to 1.2, and for the 11 “Ca. Mycoplasma turicensis”-infected SPF cats (Table (Table1),1), they were between 0.7 and 1.5. The signal-to-noise ratio of the M. haemofelis-, “Ca. Mycoplasma haemominutum”- and “Ca. Mycoplasma turicensis”-infected SPF cats reached a maximum of 10.4, 10.4, and 10.6, respectively. A signal-to-noise ratio of at least 1.5 was considered serologically positive for anti-M. haemofelis rDnaK antibodies.
The experimentally M. haemofelis-infected cat QLA5 turned M. haemofelis TaqMan real-time PCR positive within 4 dpi and became anemic within 10 dpi (Fig. (Fig.4A).4A). On the day of infection (0 dpi) the cat was mildly anemic (hematocrit of 28%), probably due to a baseline blood collection of 26 ml 11 days prior to M. haemofelis infection. However, the cat recovered to hematocrit values within the reference range within a few days (7 to 9 dpi) before a decrease in the hematocrit was observed starting at 10 dpi. The minimum hematocrit value of 15% was measured 36 dpi (Fig. (Fig.4A).4A). However, no severe clinical signs were observed that necessitated blood transfusion or antibiotic treatment during the course of infection, and the cat subsequently recovered from anemia. From 148 dpi (5.3 months postinfection [mpi]) onwards, the hematocrit values stayed within the reference range until the end of the experiment, 28.6 mpi (Fig. (Fig.4A4A and data not shown).
The peak M. haemofelis load in blood (2.2 × 108 copies/ml blood) was recorded at 29 dpi. Between 4 and 42 dpi the first marked M. haemofelis copy number fluctuations were observed; they ranged from 103 to 108 M. haemofelis copies/ml blood within a minimum of 2 days (Fig. (Fig.4A).4A). From 3.8 to 8.3 mpi a second episode of copy number cycling was observed; the loads ranged from 102 to 105 copies/ml blood within a minimum of 8 days. Five distinct M. haemofelis load peaks were observed in 1- to 2-month intervals during this second cycling period. QLA5 stayed PCR negative from 260 dpi (9.3 mpi) until the end of the observation period at 28.6 mpi (Fig. 4A and B).
The seroconversion of cat QLA5, defined as a signal-to-noise ratio of at least 1.5, occurred between 8 (signal ratio, 1.3) and 14 dpi (signal ratio, 6.9) (Fig. (Fig.4B).4B). QLA5 stayed serologically positive until the end of the observation period at 28.6 MPI. Twelve and 18 mpi the signal-to-noise ratio dropped to a minimum of 2.8, followed by a signal ratio increase to 6.0 without detectable amounts of M. haemofelis DNA in the cat's blood (Fig. (Fig.4B).4B). The reinfection of QLA5 by its SPF companion cat was excluded as a cause for this ratio increase by negative PCR and serology results for the companion cat before, during, and at the end of the observed infection period (data not shown).
There was a significant positive correlation between the signal-to-noise ratio and M. haemofelis load when all pairs of variants throughout the whole course of infection (n = 32) were included (Spearmen correlation coefficient, r = 0.619; P = 0.0002) as well as when only pairs of variants from the PCR-positive period (n = 21; r = 0.502; P = 0.0202) were included.
This is the first study to identify, characterize, and recombinantly produce a full-length, highly pure antigen of feline hemotropic mycoplasmas. The protein is an HSP70 protein and belongs to the DnaK protein family. It is most closely related to HspA1, the DnaK of M. suis (CAK22359); the latter protein was demonstrated to be expressed on the surface of M. suis cells and to have immunogenic potential (10). In analogy to this, we found that hemoplasma-infected cats readily produced antibodies to M. haemofelis rDnaK.
The protein cross-reacted with sera from cats experimentally infected with M. haemofelis, “Ca. Mycoplasma haemominutum” and “Ca. Mycoplasma turicensis,” but not with serum samples from SPF cats. However, the optimization of the ELISA resulted in higher sample dilutions for M. haemofelis samples than for “Ca. Mycoplasma haemominutum” and “Ca. Mycoplasma turicensis” samples, which indicates that the immunogenicity of M. haemofelis rDnaK is caused by conserved as well as species-specific epitopes of this antigen. This would be in agreement with the high identity (71%) that we found between the M. haemofelis DnaK and the partial “Ca. Mycoplasma haemominutum” DnaK gene sequence (1,304 bp; HM594282) but also could explain why we were unable to amplify the 3′ end of the “Ca. Mycoplasma haemominutum” DnaK gene sequence using consensus primers despite several attempts (data not shown). The observed cross-reactivity is also in agreement with a previous study using whole feline hemoplasma antigen preparations (5). In the latter study, antigen derived from H. felis large form (today known as M. haemofelis) was tested with sera from cats infected with M. haemofelis and “Ca. Mycoplasma haemominutum” (formerly known as H. felis small form). M. haemofelis-derived whole hemoplasma antigen cross-reacted with sera from M. haemofelis and “Ca. Mycoplasma haemominutum”-infected cats, while “Ca. Mycoplasma haemominutum”-derived antigen was recognized only by sera from “Ca. Mycoplasma haemominutum”-infected cats.
The antigen M. haemofelis rDnaK was purified to homogeneity from potentially antigenic proteins originating from the production process to improve the signal quality of the serological assays and to minimize interbatch variations in antigen quality. Indeed, M. haemofelis rDnaK protein expression and purification was repeated with identical results. The identity of the protein was proven by the comparison of the observed and calculated molecular masses of M. haemofelis rDnaK. CD spectrum analysis of M. haemofelis rDnaK revealed two minima (at 208 and 222 nm), suggesting that it consisted mostly of α-helices (8), which is in good agreement with known DnaK structures, e.g., of E. coli (Protein Database identiy [PDB ID]: 2KH0) and G. kaustophilus (PDB ID: 2V7Y). The structure profile of M. haemofelis rDnaK was insensitive to a change in the presence of nucleotide, as has been shown before for Bacillus licheniformis DnaK (13), and also in the presence of K+ and Mg2+ ions. The thermal denaturation of M. haemofelis rDnaK was characterized by two temperature transitions. This corresponded well to an earlier study (17), where deletion mutants of E. coli DnaK were used. The authors demonstrated that the first transition (Tm1) was related to the unfolding of the DnaK N-terminal nucleotide-binding domain, while the second transition was related to the unfolding of the C-terminal substrate-binding domain. A raising of Tm1 in the presence of nucleotide, as also observed for M. haemofelis rDnaK, was reported for E. coli DnaK to be caused by a stabilizing effect occurring due to the ligand binding to the nucleotide-binding domain of the protein (20). As found for M. haemofelis rDnaK, this stability was supposed to be further enhanced in the presence of nucleotide together with K+ and Mg2+ ions, which mediate contacts between DnaK and nucleotide (15, 33). The kinetic constants for ATP hydrolysis by pure M. haemofelis rDnaK (kcat = 0.015/min; Km = 23 μM; kcat/Km = 650/M/min) (Fig. (Fig.3C)3C) were comparable to those published for E. coli DnaK, which showed kcat values ranging from 0.02 to 0.2/min (3) and Km values ranging from 20 nM to 20 μM (4, 14). This indicated that M. haemofelis rDnaK possesses a typically low ATPase activity when evaluated without its cochaperones DnaJ and grPE. The DnaK complementation assay confirmed the molecular chaperone activity of M. haemofelis rDnaK in an E. coli DnaK knockout mutant at heat shock temperatures. This heat shock protein activity serves as another piece of true evidence for the identity of M. haemofelis DnaK.
We demonstrated for the first time that an experimentally M. haemofelis-infected cat mounted antibodies to M. haemofelis rDnaK within 8 to 14 days after experimental infection and shortly after the cat's blood was M. haemofelis PCR positive. Moreover, we found a correlation between the M. haemofelis blood loads and antibody levels. This indicates that M. haemofelis DnaK is immunogenic and that the recombinant antigen is suited for use in quantitative serological assays and to demonstrate seroconversion in infected animals.
The experimentally M. haemofelis-infected cat stayed serologically positive for more than 2 years postinfection and for more than 1.5 years after turning PCR negative for M. haemofelis in the blood. So far, we have data from only one M. haemofelis-infected cat. However, earlier results from “Ca. Mycoplasma turicensis” infection (18) and preliminary follow-up data from these cats (M. Novacco, G. Wolf-Jäckel, H. Lutz, and R. Hofmann-Lehmann, unpublished data) confirm the persistence of anti-M. haemofelis rDnaK antibodies in the absence of PCR positivity in blood. This indicates that there is active antigen stimulation in the chronic phase of hemoplasma infection, possibly by antigen sequestered in the tissue. We have postulated that the decline of Western blot signal in two cats after the antibiotic treatment of experimental “Ca. Mycoplasma turicensis” infection could have been due to therapy-induced “Ca. Mycoplasma turicensis” clearance from blood and tissues (18), whereas “Ca. Mycoplasma turicensis” antigen sequestered in the tissues of 10 untreated cats could have resulted in the continuous low-level stimulation of the humoral immune system (18). The latter would be in agreement with the findings from two experimentally “Ca. Mycoplasma turicensis”-infected cats: in nearly all tissues collected from those two cats at 20 and 94 dpi, more “Ca. Mycoplasma turicensis” copies were found than expected to result from blood contamination alone, which suggested that “Ca. Mycoplasma turicensis” can become concentrated in organs of infected cats (S. Tasker, personal communication). However, for M. haemofelis no evidence for significant tissue sequestration has been found so far (30).
In conclusion, this study provides evidence that M. haemofelis rDnaK, the antigen we have identified, recombinantly produced, and characterized herein, has biochemical and molecular chaperone properties of the HSP70/DnaK family. It has been applied successfully to quantify anti-feline hemoplasma antibodies in samples from experimentally infected cats, as well as to monitor seroconversion after experimental infection when used in Western blot assays and ELISA. The antigen and the assays are prerequisites to gain more insight into the course of hemoplasma infections, e.g., by investigating hemoplasma pathogenesis in experimental infection setups. In addition, the described antigen may have the potential to be used in a rapid test for clinicians, supporting quick diagnosis and faster choice of adequate therapy of feline hemolytic anemia. However, further studies will be necessary to fully evaluate the cross-reactivity of this antigen with antibodies directed against other bacterial pathogens and its potential benefit for testing samples from naturally hemoplasma-infected cats.
We are very grateful to Donald Hilvert from the Laboratory of Organic Chemistry at the ETH Zurich for his generosity in providing laboratory infrastructure and material for the protein production and characterization. We thank Hajo Kries for his indispensable support in MESG synthesis, Valentino Cattori for scientific support and fruitful discussions, Marilisa Novacco for her effort in conducting parts of the M. haemofelis infection project, Uwe Sauer from the Institute of Molecular Systems Biology at ETH Zurich for kindly providing the E. coli strain JW0013, and the Environmental Council of the government of Andalusia in Southern Spain for providing the Iberian lynx samples. We thank Marychelo Rios and Daniela Brasser for their loving care and commitment in attending the laboratory cats, Barbara Riond for managing the cattery, and the whole staff of our routine diagnostic laboratory for the hematological analyses.
The M. haemofelis inoculum used to infect cat QLA5 was derived from a study funded by the Wellcome Trust (grant no. 077718). Laboratory work was performed partly using the logistics of the Center for Clinical Studies at the Vetsuisse Faculty of the University of Zurich. R.H.L. is the recipient of a professorship from the Swiss National Science Foundation (PP00P3-119136).
Published ahead of print on 28 September 2010.
‡The authors have paid a fee to allow immediate free access to this article.