|Home | About | Journals | Submit | Contact Us | Français|
To examine the host association of Tula virus (TULV), a hantavirus present in large parts of Europe, we investigated a total of 791 rodents representing 469 Microtus arvalis and 322 Microtus agrestis animals from northeast, northwest, and southeast Germany, including geographical regions with sympatric occurrence of both vole species, for the presence of TULV infections. Based on serological investigation, reverse transcriptase PCR, and subsequent sequence analysis of partial small (S) and medium (M) segments, we herein show that TULV is carried not only by its commonly known host M. arvalis but also frequently by M. agrestis in different regions of Germany for a prolonged time period. At one trapping site, TULV was exclusively detected in M. agrestis, suggesting an isolated transmission cycle in this rodent reservoir separate from spillover infections of TULV-carrying M. arvalis. Phylogenetic analysis of the S and M segment sequences demonstrated geographical clustering of the TULV sequences irrespective of the host, M. arvalis or M. agrestis. The novel TULV lineages from northeast, northwest, and southeast Germany described here are clearly separated from each other and from other German, European, or Asian lineages, suggesting their stable geographical localization and fast sequence evolution. In conclusion, these results demonstrate that TULV represents a promiscuous hantavirus with a large panel of susceptible hosts. In addition, this may suggest an alternative evolution mode, other than a strict coevolution, for this virus in its Microtus hosts, which should be proven in further large-scale investigations on sympatric Microtus hosts.
Hantaviruses (genus Hantavirus, family Bunyaviridae) are characterized by a tripartite RNA genome of negative polarity. The small (S) genome segment of about 1.7 kb encodes the nucleocapsid (N) protein that is associated as a multimer with the viral RNA genome. The medium (M) segment of about 3.6 kb encodes a glycoprotein precursor that is cotranslationally cleaved at a highly conserved WAASA motif into the G1 and G2 envelope glycoproteins. These proteins form oligomers which mediate the interaction of the virus with the cellular receptor. The large (L) segment of about 6.5 kb encodes the RNA-dependent RNA polymerase that functions as transcriptase and replicase (for a review, see reference 57).
In general, hantaviruses are harbored by persistently infected rodent reservoir hosts which shed the hantaviruses by urine, feces, and saliva. Therefore, the major route of transmission to humans is by inhalation of aerosols originating from virus-contaminated urine or feces (for a review, see reference 58). The high stability of hantaviruses in nature allows indirect transmission and underlines the importance of environmental factors on the frequency of transmission (31). An alternative route of virus transmission to humans is by rodent bites (10). Human-to-human transmission has exclusively been observed for the South American Andes virus (42).
The congruent phylogenetic affinities of hantaviruses and their corresponding reservoir hosts are currently explained by a virus-host coevolution hypothesis (46). According to this theory, each hantavirus species is associated with a single rodent species or a closely related species of the same genus. This close host/pathogen association is also believed to determine viral properties and therefore the pathogenicity in humans (73).
Tula virus (TULV) has primarily been identified in the European common voles Microtus arvalis and Microtus rossiaemeridionalis collected from the Tula area of Russia, located 200 kilometers south of Moscow, and in M. arvalis in western Slovakia, near the town of Malacky (43, 61). Subsequently, TULV has been detected in other related rodent species of the subfamily Arvicolinae, i.e., M. agrestis, Microtus gregalis, Microtus (Pitymys) subterraneus, and Lagurus lagurus (56, 65; A. E. Dekonenko and V. V. Yakimenko, unpublished data) (GenBank accession number AF442620-21). TULV-positive rodents have been detected in a number of European countries, such as Russia, Slovakia, Croatia, the Czech Republic, Austria, Poland, Belgium, Germany, France, Hungary, The Netherlands, and Slovenia (4, 6, 21, 30, 32, 35, 43, 44, 47, 51, 56, 61, 65, 66).
Hantaviruses associated with the genus Microtus (subfamily Arvicolinae) have currently only rarely been associated with human disease. Although the North American viruses Prospect Hill virus (PHV), isolated from the meadow vole Microtus pennsylvanicus (38), Bloodland Lake virus, detected in the prairie vole Microtus ochrogaster (22), and Isla Vista virus (ISLAV), detected in the Californian vole Microtus californicus (64), have not been shown to cause significant disease in humans, experimental infections of nonhuman primates with PHV caused an acute nephropathy (74). Similarly, the Eurasian hantavirus species Khabarovsk virus hosted by the reed vole Microtus fortis (23) and TULV are believed to have no or low pathogenicity for humans. However, a few human TULV infections have been reported, with two case reports about potential TULV-induced disease in humans (32, 59, 70, 71). This low frequency in the detection of human TULV infections might be explained by its low pathogenic potential (36); however, human TULV infections might have been overlooked because their differentiation from infections with the TULV-cross-reactive Puumala virus (PUUV) would require a neutralization assay (70).
Here we report on data from the first longitudinal monitoring study of TULV in Europe, with samples collected over a period of 12 years. These investigations showed the sympatric occurrence of TULV in M. arvalis and M. agrestis at several localities and in different years. Phylogenetic analysis of the TULV sequences demonstrates a geographical clustering which is independent from the rodent reservoir species. This provides strong evidence for isolated replication and transmission cycles of TULV in both species, along with frequent multiple spillover infections of TULV between M. arvalis and M. agrestis populations.
Between 1994 and 2005, Microtus rodents were trapped in six districts of the Federal State of Brandenburg, northeast Germany, and in 2005, they were trapped at one site (Sennickerode [SEN]) in the district Göttingen (GÖ) in the Federal State of Lower Saxony, northwest Germany. Additional rodents were trapped between 1994 and 1997 at four different military training areas, including Grafenwöhr, district Neustadt/Waldnaab (NEW), located in the Federal State of Bavaria, southeast Germany (Fig. (Fig.11 and Tables Tables11 and and2).2). Captured rodents were weighed and measured, and their gender and species were identified. For serological analysis, transudate samples were collected from the thoracic cavity of rodents from Brandenburg. From rodents trapped from 1994 to 1997 in Brandenburg or Bavaria, only brain or kidney samples were available. From rodents trapped from 2002 to 2005 in Brandenburg and Lower Saxony, lung, heart, liver, spleen, kidney, and brain samples were collected. Tissue samples were stored at −20°C until processing with reverse transcriptase PCR (RT-PCR). Morphological species determination of TULV RT-PCR-positive animals was confirmed by using a PCR specific for the mitochondrial cytochrome b gene (11).
To detect hantavirus-specific immunoglobulin G (IgG) antibodies, rodent transudates were screened by using enzyme-linked immunosorbent assays (ELISA) and Western blot (WB) tests based on Saccharomyces cerevisiae-expressed His-tagged N proteins of PUUV strain Vranica/Hällnäs (9, 50) and TULV strain Moravia (44; M. Mertens, R. Petraityte, K. Sasnauskas, R. Friedrich, and R. G. Ulrich, unpublished data). Assays were performed as described previously (11). Selected seroreactive samples were confirmed by focus reduction neutralization test analysis, according to protocols described earlier (3).
Potential associations between gender and age, using body mass as a proxy, and seropositivity to hantavirus in voles were analyzed using the chi-square test, Mann-Whitney test, or Fisher exact test included in the software package SPSS (SPSS version 12.0.1, SPSS Inc., Chicago, IL) and by Win Episcope 2.0 (69).
RNA from all animals from Lower Saxony and Bavaria and from TULV-seroreactive animals from Brandenburg was extracted from brain, heart, or lung tissue samples using commercial kits (11). Single-step and nested RT-PCRs using primers targeting the S segment were performed with primer pairs DOBV-M6/DOBV-M8, PUUV 342/cPUUV 1102, SNMa1/MaS4C, and S1/S10PC and nested primers PUUV390/cPUUV721 as described previously (11, 55, 61, 62). The partial M segment was amplified using primers C1, 5′-CCCCCTGATTGTCCTGGTGTAG-3′, and C2, 5′-CCAACTCCTGAACCCCATGC-3′ (corresponding to nucleotides [nt] 2369 to 2390 and nt 3011 to 3031 of PUUV strain Sotkamo, GenBank accession number X61034). Direct sequencing of the purified PCR products was done using the S and M segment-specific primers described above. GenBank accession numbers of the novel TULV sequences are shown (see Table Table44).
Nucleotide sequences were aligned in BioEdit 5.0.9 (17) and revised manually. Phylogenetic relationships among nucleotide sequences were reconstructed with neighbor-joining (NJ) (54) and maximum likelihood (ML) algorithms implemented in PAUP* 4.0b (67) using two PUUV strains as the outgroup and 5,000 bootstrap replicates. Hierarchical likelihood ratio tests and the Akaike information criterion (1) implemented in Modeltest 3.06 (48) were used to estimate the most suitable model of nucleotide substitution. The best substitution model for the S segment was the general time-reversible (GTR) model with gamma distribution (68) with the following parameters: substitution rate matrix, A ↔ C, 4.1689; A ↔ G, 14.7655; A ↔ T, 3.1028; C ↔ G, 0.6093; C ↔ T, 22.7757; and G ↔ T, 1.0000; gamma distribution shape parameter, 0.1931. The base frequencies were estimated as follows: A, 0.3318; C, 0.2114; G, 0.1966; and T, 0.2602. For the M segment, the best substitution model was the GTR model with invariable sites and gamma distribution (52). The following parameters for the model were estimated: substitution rate matrix, A ↔ C, 4.8675; A ↔ G, 16.5929; A ↔ T, 1.5562; C ↔ G, 4.9664; C ↔ T, 36.3308; and G ↔ T, 1.0000; gamma distribution shape parameter, 0.5635; proportion of invariable sites, 0.4103; base frequencies, A, 0.3568; C, 0.1519; G, 0.1657; and T, 0.3256.
Estimates of evolutionary raw divergence and standard error estimates (500 bootstrap replicates) over sequence pairs between groups were obtained by pairwise analysis, as supplied by MEGA4 (37). Codon positions included were first plus second plus third plus noncoding. All positions containing gaps and missing data were eliminated from the data set (complete deletion option).
Morphological species designations of TULV-positive rodents were verified by performing BLAST searches of the novel cytochrome b sequences with sequences available in GenBank (http://www.ncbi.nlm.nih.gov). The cytochrome b sequences of TULV-positive M. agrestis and M. arvalis were further compared to those from larger surveys of genetic diversity in these rodents across Germany and neighboring countries (14, 19, 25). Representative sequences from all evolutionary lineages present in Europe in M. arvalis and M. agrestis were obtained from GenBank. Technical details on phylogenetic reconstructions from cytochrome b are given elsewhere (15, 16, 19).
Between 1994 and 1997, a total of 427 Microtus rodents, with 377 M. arvalis rodents, 115 male and 262 female, and 50 M. agrestis rodents, 24 male and 26 female, were trapped at 10 sites in two different districts (Ostprignitz-Ruppin [OPR] and Prignitz [PR]) in the Federal State of Brandenburg, northeast Germany (Fig. (Fig.11 and Tables Tables11 and and2).2). In addition, a total of 336 Microtus rodents, including 58 male and 8 female M. arvalis rodents and 122 male and 148 female M. agrestis rodents, were trapped at five sites in four other districts (Barnim [BAR], Oder-Spree [LOS], Potsdam-Mittelmark [PM], and Oberspreewald-Lausitz [OSL]) in the Federal State of Brandenburg between 2002 and 2005 (Fig. (Fig.11 and Tables Tables11 and and2).2). Altogether, M. arvalis and M. agrestis were trapped at nine sites in Brandenburg, with a sympatric occurrence of M. arvalis and M. agrestis in the same year of trapping (Tables (Tables11 and and2).2). In 2005, 16 M. arvalis rodents, 5 female and 11 male, and 2 male M. agrestis rodents were trapped at a single site in Lower Saxony (SEN, GÖ) (Fig. (Fig.1).1). From 1994 to 1997, 10 M. arvalis rodents were trapped at one site in the military training area Grafenwöhr (NEW) in Bavaria (Fig. (Fig.11).
Sequencing of cytochrome b genes of selected M. arvalis and M. agrestis rodents from Brandenburg and Lower Saxony confirmed the morphological species determination (see Table Table4).4). Phylogenetic reconstructions of rodent host relationships based on the mitochondrial cytochrome b gene showed a clear differentiation between M. arvalis and M. agrestis and an additional substructure among voles from different European regions (Fig. (Fig.2).2). All M. arvalis rodents from the German study sites belonged to the Central evolutionary lineage of this species, and all M. agrestis rodents from these sites belonged to the Western lineage. This is in agreement with the general distribution of these lineages, as determined in larger surveys of the two species (14, 19, 25).
In total, 791 Microtus rodents were trapped at 17 sites located in eight different districts in northeast, southeast, and northwest Germany.
An initial IgG ELISA screening using recombinant PUUV N protein as the antigen revealed a total of 35 seroreactive animals out of 763 animals (4.6%) from Brandenburg (data not shown). For 28 out of 35 samples (80%), the immunoreactivity was confirmed by a corresponding PUUV WB test (data not shown). In parallel, all transudate samples were tested by an IgG ELISA using recombinant TULV N antigen (Tables (Tables11 and and2).2). Using this test format, TULV-reactive antibodies were detected in 71 of 443 M. arvalis transudates (16%) and 17 of 320 M. agrestis transudates (5.3%), demonstrating a much higher level of sensitivity for the TULV ELISA than that for the PUUV test. For 25 out of 30 TULV- and PUUV-IgG ELISA-reactive samples (83%), the endpoint titer was higher in the TULV assay (Table (Table3).3). For the majority (75 of 88; 88%) of transudates, the TULV ELISA reactivity was confirmed in a TULV WB test (data not shown). The M. arvalis-seroreactive samples originated from both male (31 of 173; 17.3%) and female (40 of 270; 14.8%) animals. In contrast, seroreactivity in M. agrestis transudates demonstrated a strong gender bias, with 15 out of 146 male (10.3%) but only 2 out of 174 female (1.1%) animals being seroreactive, which proved to be statistically significant (chi-square test, P of <0.001; odds ratio, 8.9; Fisher's exact test, P of <0.001). In contrast, the differences in the TULV seroprevalence between the genders in M. arvalis were not statistically significant (chi-square test, P of 0.416; odds ratio, 1.210; Fisher's exact test, P of 0.513). The body mass of M. agrestis, which was used as a proxy for the age of the animals, was found to be positively associated with the seropositivity (Mann-Whitney test, P of 0.0058), whereas this was not statistically significant for M. arvalis (Mann-Whitney test, P of 0.0703).
To prove the validity of the ELISA and WB tests, five M. arvalis and four M. agrestis transudates were tested by focus reduction neutralization assays using TULV, PUUV, Dobrava virus, and Saaremaa virus. For all nine transudates, the highest endpoint titer was observed for TULV, confirming TULV infections in both rodent species (Table (Table33).
The serological investigations demonstrated the presence of TULV-reactive antibodies in both M. arvalis and M. agrestis, with differences depending on the species, the trapping site, the gender, and the age of the rodents.
For 19 out of 20 seroreactive animals from Brandenburg whose brain, heart, or lung tissue samples were available, the investigations with S segment-specific RT-PCR or nested RT-PCR revealed specific amplification products (Table (Table4).4). The positive samples originated from four male and three female M. arvalis rodents and 11 male and one female M. agrestis rodents. One sample (Mu/04/151) from a seropositive female M. agrestis rodent tested negative in all RT-PCRs used. Sequencing of the amplification products resulted in the identification of TULV sequences in samples from both M. arvalis and M. agrestis (Table (Table4).4). As expected for RT-PCR investigations of lung samples, sequence information for all tested animals was obtained, with the majority of sequences being about 700 nt in size. Interestingly, the nested RT-PCR approach was successful for all tested brain samples from seven animals when no other tissue samples were available. In addition, for four out of five investigated heart samples, partial S segment-specific sequences were obtained (Table (Table44).
Screening of lung samples from all 16 M. arvalis and 2 M. agrestis rodents from Lower Saxony (district GÖ, site SEN) by S segment-specific RT-PCR revealed six positive M. arvalis results from one female and five male animals. In addition, one male M. agrestis animal was PCR positive (Table (Table4).4). Larger PCR fragments (from 1,347 nt up to 1,708 nt in length) (Table (Table4)4) could be derived from all positive samples.
Out of the 10 M. arvalis animals from Bavaria (Grafenwöhr, NEW), two kidney samples allowed the amplification of TULV S segment sequences of 334 and 1,832 nt in length (Table (Table44).
In addition to the S segment sequences, M segment sequences were obtained by RT-PCR amplification of the region spanning nt 2369 to 3031 (numbering according to PUUV strain Sotkamo, GenBank accession number X61034) for four M. arvalis animals, three from Lower Saxony and one from Brandenburg, and five M. agrestis animals from Brandenburg.
In summary, TULV S and M segment sequences were detected in both M. arvalis and M. agrestis.
In contrast to the rodent data, phylogenetic analyses of both the S and M segments revealed very strong geographical but no host-specific affinities of TULV sequences. Phylogenetic trees reconstructed from 80 sequences of a 333-nt-long S segment (nt positions 355 to 686 in TULV strain Lodz-2, GenBank accession number AF063897) demonstrated that the novel TULV sequences from northwest and southwest Brandenburg and Lower Saxony (lineages Germany I and II) were clearly separated from other German lineages originating from Bavaria, southeast Germany (Germany IV), and southeast Brandenburg (Germany III) (Fig. (Fig.3A).3A). A more detailed evaluation of clusters Germany I and II showed a geographical clustering of these sequences irrespective of the rodent host species (Fig. (Fig.3B).3B). The sublineage IA from the trapping sites Granzow and Bendelin (district PR) and Nackel (district OPR) was predominantly detected in M. arvalis but also in one M. agrestis animal. A similar pattern was observed for TULV infections in Microtus from SEN (district GÖ; lineage Germany II). In Eberswalde (EBE) (district BAR; sublineage IB), TULV infections were found mostly in M. agrestis but also in one M. arvalis animal. At the trapping site Marzehns (MRZ) (district PM) in southwest Brandenburg, exclusively TULV-infected M. agrestis animals were found between 2002 and 2005, but no TULV-infected M. arvalis animal was found (Tables (Tables1,1, ,2,2, and and44).
The analysis of M segment nucleotide sequences (nt positions 2390 to 3010) (Fig. (Fig.4)4) and corresponding amino acid sequences (amino acid [aa] positions 780 to 985 of GPC in the G2 part) supported the patterns detected in the S segment (data not shown). The novel sequences from Germany were clearly distinct from all M segment sequences published so far. Again, the M. arvalis-associated sequence from Mar137_04_EBE (district BAR) grouped together with four M. agrestis-associated TULV sequences in the cluster Germany I. Similarly, the Germany II clade consisted of TULV M sequences associated with M. arvalis and M. agrestis.
Therefore, phylogenetic analysis revealed a geographical, but not host-specific, clustering of TULV S and M segment sequences from two different Microtus species. In addition, these investigations showed for the first time that TULV is (probably) able to establish an isolated transmission cycle in M. agrestis.
A pairwise comparison of TULV S and M segment nucleotide sequences from animals within the clusters of TULV revealed a high level of divergence. The levels of the average nucleotide (amino acid) partial S segment sequence divergence in groups Germany I and Germany II were up to 7.2% (0.2%) and 5.4% (0.2%), respectively (Table (Table5).5). Similarly, sequences of clusters Russia I, Russia II, and Russia III showed divergence levels of 5.3 to 6.8% (0 to 2.3%). In contrast, the other clusters, including Germany III, Slovakia III/Czech Republic, Austria I and II, Poland, and Slovakia I and II, displayed lower average intercluster divergence levels of 0 to 1.9% (0 to 0.3%).
The divergence level of the S segment nucleotide sequences between the clusters from Germany was surprisingly high (about 12 to 20%) (Table (Table6).6). The divergence level of about 18% between sequences from clusters I and II and those from cluster III is particularly important, as all sequences in clusters I (districts OPR, PR, and BAR) and III (district Spree-Neisse [SPN]) and about one-half of the sequences in cluster II (districts PM and OPR) originated from closely related geographical sites in Brandenburg (Fig. (Fig.1).1). The nucleotide sequence divergence level of TULV sequences from other European regions also ranged from 17 to 20%. Interestingly, the amino acid sequence divergence level between sequences from clusters Germany I and II and those from cluster Germany IV was higher than the divergence levels for all other sequences from Europe.
The full-length open reading frame (ORF) of the N protein made up of 430 codons was able to be amplified from four RT-PCR-positive Microtus sequences from SEN (district GÖ; SEN 174, 175, 204, and 205) and one from Grafenwöhr (district NEW; AF164093). A putative second ORF encoding a hypothetical nonstructural protein, NSs, made up of 90 aa (corresponding to nt 84 to 356) was determined for these four sequences of Microtus from SEN and the one from Grafenwöhr. The entire N-encoding sequence in the samples from Bavaria was 84.2 to 84.6% identical at the nucleotide level and 95.3 to 95.8% identical at the amino acid level to the corresponding sequences from SEN (data not shown). For the N ORF, the nucleotide and amino acid sequences of Mar204_05_SEN, Mar205_05_SEN, and Mar174_05_SEN differed from 0.1 to 0.5% and 0 to 0.5%, respectively. The sequence divergence of these sequences from M. arvalis at the nucleotide and amino acid levels was slightly higher than that of Mag175_05_SEN from M. agrestis trapped at the same site and time, with 0.8 to 1.2% and 0.5 to 1.0%, respectively (Tables (Tables55 and and66).
For the partial M segment, within-group calculations resulted in cluster Germany I having an average nucleotide divergence level of up to 3.5% (amino acid divergence level, 0.4%) and in cluster Germany II having an average nucleotide divergence of 4.9% (amino acid divergence, 0%), whereas in clusters Czech Republic and Poland, the average nucleotide divergences were 1.3% (amino acid divergence, 0%) and 0%, respectively (Table (Table5).5). The divergence pattern observed for the M segment sequences seems to be comparable to that found for the S segment sequences (Table (Table7).7). However, this conclusion does not take into account that different numbers of sequences were available for the various clusters.
As expected, the nucleotide and amino acid sequence divergences of S and M segment sequences from other Microtus-associated viruses, i.e., Yakeshi virus, Fusong virus, Vladivostok virus, PHV, ISLAV, and Khabarovsk virus, were found to be much higher.
These investigations revealed a surprisingly high level of nucleotide sequence divergence between the cluster of the novel German TULV sequences and TULV sequences from other European countries.
A comparison of the partial N protein spanning aa 105 to 215 or aa 121 to 215 (according to numbering in the N protein of TULV, strain Moravia, GenBank accession number Z69991) of the novel sequences revealed only three amino acid exchanges, comprising residues with identical (M188L and R213K in Mar1384_95) or similar (A208T in Mag235_03; data not shown) properties.
The observed higher level of amino acid sequence divergence between sequences from clusters Germany I and II and those from cluster Germany IV is reflected also in the amino acid sequence alignment of the N protein between aa residues 231 and 332 (numbering according to TULV reference strain NC_005227), showing the sequences from Germany IV to be much more similar to sequences from Slovakia, Czech Republic, and Croatia (Fig. (Fig.5).5). However, the other TULV sequences from Germany had a more unique amino acid sequence pattern than those from clusters Germany I and II. Interestingly, TULV sequences from trapping sites from the districts BAR (site EBE) and PR/OPR (sites Granzow, Bendelin, and Nackel) had unique amino acid residues at positions 248 and 258.
A comparison of the G2 part covering aa positions 780 to 985 of the GPC (numbering in GPC of TULV, strain Moravia, according to GenBank accession number Z66538) confirmed the geographical but not host-specific clustering of TULV sequences from Germany (data not shown). Conservative amino acid exchanges were observed at position 834 with neutral V and I amino acid residues (in both M. agrestis and M. arvalis from trapping site EBE) and L residues (in both M. agrestis and M. arvalis from trapping sites MRZ and SEN) and at position 885 with neutral L amino acid residues (in both M. arvalis and M. agrestis from trapping site EBE as well as in the other five TULV sequences from Europe) and I residues (in both M. agrestis and M. arvalis from trapping sites MRZ and SEN). Interestingly, at position 834, the other five TULV sequences from Poland, Czech Republic, and Serbia had a conserved V residue. At position 976, all other European TULV strains showed a V residue, whereas the German strains had an I residue.
The amino acid sequence comparison of TULV sequences from certain sites resulted in the identification of trapping site-specific amino acid signatures.
In this study, we demonstrated for the first time that TULV occurs in Germany simultaneously in two different Microtus species, M. arvalis and M. agrestis, which is apparently uncommon for hantaviruses. However, TULV seems to be a very special hantavirus, as it has been found in a large number of different species, including M. rossiaemeridionalis, M. agrestis, M. gregalis, M. subterraneus, and Lagurus lagurus (43, 56, 65; Dekonenko and Yakimenko, unpublished), after its initial description in M. arvalis (43, 61). Similarly, other Old World hantaviruses, like Seoul virus, and various North and South American hantaviruses have been found in multiple rodent reservoir hosts (for reviews, see references 8a and 40a). These observations raise the question whether the occurrence of hantaviruses in different related rodent hosts might be a general phenomenon which has frequently been overlooked so far due to the lack of large-scale screenings of sympatrically occurring animals.
This paper also describes the first comprehensive study of the presence of TULV in three different regions of Germany. The initial serological detection of TULV-specific antibodies in M. arvalis and M. agrestis by ELISA using a novel homologous N antigen was confirmed by RT-PCR investigations targeting the S and M segments. RT-PCR using S segment-specific primers resulted in the detection of TULV RNA not only in lung tissue samples but also in heart and/or kidney tissue samples, which is in line with our previous observations in PUUV-infected bank voles (11). Interestingly, we were able to detect TULV-specific RNA in brain samples of seven animals from which no other tissues were available, even after storage at −20°C for 7 to 10 years. Previously, Black Creek Canal virus, a North American, Sigmodon hispidus-transmitted hantavirus causing hantavirus cardiopulmonary syndrome (53), was detected in rodent brain samples (24). The observed presence of TULV and Black Creek Canal virus in the brain of their natural hosts might be explained by crossing of the blood-brain barrier due to infection of newborn animals lacking an intact blood-brain barrier or of animals with pathological changes in the central nervous system microenvironment, resulting in a blood-brain barrier dysfunction. Alternatively, the presence of the virus in the brain might be mediated by infected migrating Trojan horse-like cells, such as monocytes. If crossing of the brain barrier is an outstanding property of TULV, as it is also in terms of host specificity and pathogenicity, the finding of TULV RNA in the brain should be proven in a comparative study of different tissue samples from a larger number of voles.
Phylogenetic analyses of the novel S and M segment sequences indicated that these Microtus-borne sequences belong to TULV and are clearly separated from sequences originating from other Microtus-borne viruses such as PHV, ISLAV, Yakeshi virus, Khabarovsk virus, VLA virus, Vladivostok virus, and Fusong virus.
The high nucleotide divergence level of 12 to 20% among the three novel TULV clades from northeast, northwest, and southeast Germany, the already known clade from central east Germany (32), and clades from other European countries, as well as the Omsk region in the Asian part of Russia, is remarkable. This high level of sequence divergence was even observed among sequences from trapping sites that are only about 200 km or less apart from each other, i.e., in the districts SPN (clade Germany III) (32) and PM (clade Germany II). In contrast to the high level of nucleotide sequence divergence, the level of amino acid sequence divergence was much lower, e.g., only about 1% between clades Germany II and III. This is consistent with strong purifying selection. Taken together, these results showed the presence of at least four clades of TULV sequences in Germany, suggesting a quickly evolving virus species with a strong genetic substructure.
The detection of a large number of novel sequences forming clusters in the phylogenetic tree makes an analysis of the intercluster differences possible. The observed levels of intercluster differences of 5.4 and 7.2% for the novel S segment sequences and 3.5 and 4.9% for the novel M segment sequences of clades Germany I and Germany II, respectively, are in a similar range as those observed for TULV clades Russia I, II, and III. A previous study revealed a range of diversity of 1.5 to 4.9% for the S segment and 0.2 to 1.2% for the M segment between TULV strains circulating within a location of 20 km (43). It is interesting to note that the intercluster differences for 22 partial S segment PUUV sequences derived from bank voles from the city of Cologne, Germany, and for 10 partial S segment PUUV sequences from Lower Bavaria, southeast Germany, were only 1.2% and 3.1%, respectively (11, 12). Compared to these studies, the intercluster sequence differences of TULV clusters Germany I and II were doubled.
In contrast to previous investigations (45), no indication of a quasispecies population in the investigated Microtus animals has been obtained. The novel nucleotide sequence from Bavaria (Germany IV) clustered with nucleotide sequences from Austria, Slovakia, Czech Republic, and Croatia, and the previously described sequences from cluster Germany III grouped with sequences from Poland. Although the novel sequences represented by clusters Germany I and Germany II did not cluster with sequences from other parts of Europe in the phylogenetic tree, a multiple amino acid sequence alignment reflected some similarities in the amino acid sequences between the novel strains (clades Germany I and II) and those from clades Germany III and Poland (Fig. (Fig.55).
Phylogenetic reconstructions of rodent host relationships based on the mitochondrial cytochrome b gene showed a clear differentiation between M. arvalis and M. agrestis. This is in accord with the current taxonomy of Microtus (16). In both species, separation of populations during glacial cycles has led to intraspecific genetic divergence into several evolutionary lineages, which nowadays occupy large regions in Central Europe which do not overlap (19). These evolutionary lineages of M. arvalis and M. agrestis can be identified by phylogenetic analysis of the cytochrome b gene. The new sequences of M. arvalis presented here cluster clearly within the Central lineage, an old evolutionary lineage which currently inhabits most of Central Europe (Germany, Denmark, The Netherlands, and Switzerland) (19). Variation in cytochrome b was not high enough for a further geographical resolution among rodents at the regional level. All new sequences are distinct from the Eastern lineage of M. arvalis. The western border of this lineage is in Poland and the Czech Republic (19). All new cytochrome b sequences of M. agrestis clustered with the Western lineage of this species. The Western lineage in M. agrestis has a very large distribution range, encompassing most of Western, Central, and Eastern Europe and Scandinavia (25, 26). The absence of a substructure within the Western lineage is consistent with detailed analyses of cytochrome b in M. agrestis (25). Dedicated analyses of more variable genetic markers are needed to resolve the finer genetic structures that exist among local populations within evolutionary lineages of voles (8, 13, 19, 60). However, movements of individual Microtus animals are unlikely to exceed a few kilometers (60). The clustering of TULV strains according to locality within the clades Germany I and II is in accord with relatively low levels of migration among regional Microtus populations.
The detection of naturally TULV-infected M. agrestis contrasts with a previous study, where attempts to infect this rodent species with TULV failed (33). The failure in this experimental setting might have been due to the use of a cell culture-adapted TULV (33), as previously observed for PUUV (39). In addition, the TULV strain Moravia used for the experimental infections contains a stop codon in the putative NSs ORF (71), which may reduce its potential activity as an interferon antagonist (28, 29). Interestingly, in our field study, TULV infections in M. arvalis were found at a similar frequency in male and female animals, whereas M. agrestis TULV-positive males were significantly more frequent than M. agrestis TULV-positive females. This might indicate gender-dependent limitations for the transmission of TULV by spillover infections from M. arvalis or by horizontal transmission between M. agrestis individuals mediated by gender-specific differences in the TULV-specific immune response, as previously demonstrated for PUUV patients (34). Further studies are required to prove whether differences in the immune response are responsible for the clearance or the establishment of a persistent infection in spillover-infected M. agrestis. Additionally, species-, age-, and gender-specific differences in behavior, such as aggression, territoriality, or social status, may contribute to the detected differences in the prevalences. Males tend to be more mobile and aggressive than females in many Microtus species (see reference 60 and references therein), which may increase the risk of infection for males overall. However, comparative analyses of space use and social behavior of M. arvalis and M. agrestis living in sympatry are needed to clarify whether these factors could cause gender-specific differences in infection rates or prevalences between species. In line with previous studies for other hantaviruses (for a review, see reference 39a), our investigations demonstrated a positive association of the age and seropositivity for M. agrestis.
Paleozoological and molecular investigations suggest a last common ancestor of M. arvalis and M. agrestis more than 0.5 million years ago (16, 27). Following the coevolution hypothesis, this ancient separation of the Microtus species should have resulted in very distinct genetic lineages of hantaviruses associated with M. arvalis and M. agrestis. Indeed, the coexistence of two different virus lineages has been described for Dobrava-Belgrade virus in natural foci with sympatric populations of Apodemus agrarius and Apodemus flavicollis in Slovenia and Slovakia (5, 63). Similar observations were made for different New World hantaviruses (for a review, see reference 46). However, there is increasing evidence that besides a general coevolution of the reservoir host and the associated hantavirus species, host switch events might have been played an important role in hantavirus evolution. Such a host switch event has been postulated for the Arvicolinae-associated Khabarovsk virus (72). We found a clustering of TULV sequences depending on the geographical origin but not on the sympatrically occurring M. arvalis and M. agrestis hosts. In addition, our data may indicate an ongoing process of establishing M. agrestis as a novel reservoir host for TULV. These data thus indicate that spillover infections are less rare than believed so far. Moreover, our finding of TULV infections in M. agrestis at different time points without any presence of M. arvalis might even suggest an already established isolated replication and transmission cycle of TULV in M. agrestis. Taken together, these findings and the recent detection of novel hantaviruses in shrews and moles in different parts of the world suggest a richer evolutionary history and more complex transmission dynamics of hantaviruses (2, 20).
The potential for spillover infections between host species depends for instance on the frequency of cooccurrence determined by the specific ecological preferences. M. arvalis and M. agrestis both have very large distribution ranges in Europe, with regions of overlap extending from Spain into Russia (40). In many regions, M. arvalis prefers somewhat drier and more open habitats than M. agrestis, but cooccurrence is relatively common (18, 40, 41). Our trapping results confirmed the sympatric occurrence of M. arvalis and M. agrestis in exactly the same habitats at different places in northeast and northwest Germany. Information on the frequency and nature of interspecific interactions in the field is lacking for these species and other small rodents, but territoriality and the establishment of kin associations suggest aggressive interactions, with biting and scratching as the most likely route of transmission between species (7, 60).
In conclusion, this paper demonstrates that TULV is a promiscuous virus able to infect different Microtus species, including M. arvalis, M. agrestis, M. rossiaemeridionalis, M. gregalis, M. subterraneus, and other related species such as Lagurus lagurus. Moreover, initial evidence at one trapping site in Brandenburg suggests that TULV not only causes multiple spillover infections of M. agrestis but also seems to establish an isolated replication and transmission cycle in this putative novel reservoir host. Although we cannot rule out coevolutionary mechanisms, the observations described here may be interpreted against the background of the following two alternative evolution mechanisms for hantaviruses. (i) After initial multiple spillover infections of a hantavirus, e.g., TULV, from the established host to a sympatrically occurring potential novel host in different geographical regions, host adaptation of the virus in the novel host at different geographical localizations may result in a convergent evolution. This would then lead to a change from a geographical clustering to a host-specific clustering of the hantavirus sequences. Therefore, the finally observed host-specific clustering of hantavirus sequences might be, under certain circumstances, misinterpreted as a coevolution mechanism. In line with this assumption, recent studies have postulated that similarities between the phylogenies of hantaviruses and their hosts also seem to result from preferential host switching and local host-specific adaptation (49). (ii) The geographical clustering of hantavirus, e.g., TULV, sequences might have been caused by an isolation-by-distance mechanism. If this hypothesis is true, one would postulate that the viruses are less adapted to their rodent host, e.g., representatives of the genus Microtus, allowing frequent spillover or host switch events in overlapping rodent populations. This might be supported by a strong similarity of host receptor molecules that may have evolved slowly since the separation of the different species in a rodent genus, e.g., Microtus.
Future investigations on different sympatrically occurring Microtus species should address the frequency of spillover and host switch events for TULV and may thus allow for the definition of the host range of TULV and its viral and host determinants. Similar investigations of other hantaviruses and their putative rodent or insectivore hosts should greatly improve our current knowledge on the molecular evolution and host adaptation of hantaviruses.
We kindly acknowledge Martina Steffen, Heike Kubitza, Andreas Micklich, Claudia Dettmer, Roswitha Mattis, Lieselotte Minke, and Ulrich Löschner (Wusterhausen, Germany) for their support. We are very grateful to forest officers Wolfgang Michelson (AfF Belzig, Obf. Treuenbrietzen, Revier MRZ), Steffen Pauly (AfF EBE, Obf. Gross Schönebeck, Revier Trämmersee), Regina Thanisch (AfF Doberlug-Kirchhain, Obf. Altdöbern, Revier Lug), Joachim Schmelz (AfF Wünsdorf, Obf. Schwenow, Revier Schwenow), and their collaborators. We thank K. Weiss and A. Lorber for their support during rodent trapping.
J. Schmidt-Chanasit acknowledges support from the Förderverein of the Friedrich-Loeffler-Institut.
Published ahead of print on 4 November 2009.