Background

The ribosomal RNAs constitute the core part of the ribosomes, which are essential for protein synthesis in all cells. The nuclear ribosomal units of eukaryotes contain the genes for 18S rRNA, 5.8S rRNA, and 28S rRNA, separated by the internal transcribed spacers ITS1 and ITS2. The 5S rRNA genes are in different genomic regions. The 18S rRNA is part of the small ribosomal subunit (SSU), and 28S rRNA, 5.8S rRNA, and 5S rRNA are part of the large ribosomal subunit (LSU). The ribosomal units of most eukaryotes are arranged in clusters of tandem repeats on one or several chromosomes, whereby each cluster contains multiple copies of ribosomal units. Due to functional constraints, the nuclear ribosomal RNAs are among the most conserved genes in eukaryotes. Ribosomal units are assumed to evolve according to a model of concerted evolution that leads to homogenization [1, 2]. Mechanisms of concerted evolution likely involve unequal crossing over during recombination, gene duplication, and inter-chromosomal gene conversion [3]. However, the nuclear ribosomal genes of Plasmodium parasites are exceptional because their ribosomal units are assumed to evolve according to a birth-and-death model with new variants originating by duplication and others becoming deleted [4]. Studies on the 18S rRNA genes of human and rodent Plasmodium species found that the sequences of individual units can vary substantially [5], and distinct variants are expressed differentially in the vertebrate and mosquito hosts [6]. The 18S sequences expressed in the vertebrate hosts and mosquito vectors were named A-type and S-type variants, respectively [6, 7]. This pattern was found in rodent, simian, and human Plasmodium species (except for Plasmodium malariae), with A-type and S-type differing by 10% to 17% [8].

The first comprehensive study on nuclear ribosomal genes of avian haemosporidian parasites was published by Harl et al. [8], who sequenced the 18S genes of seven Plasmodium, nine Haemoproteus, and 16 Leucocytozoon lineages. Most avian Plasmodium lineages studied also feature clusters of distinct 18S variants, differing by up to 14.9%. A similar pattern was found in the Leucocytozoon toddi group, but the 18S sequences of other Leucocytozoon and Haemoproteus parasites were less variable and lacked highly diverged clusters of variants, except for Haemoproteus tartakovskyi [8].

The genus Haemoproteus currently comprises about 180 morphologically described species, but molecular genetic data is available from less than half of the species [9]. The MalAvi database (http://130.235.244.92/Malavi/; accessed in December 2022) contains more than 1500 unique Haemoproteus lineages covering the entire or almost entire 478 bp CytB barcode region. The vast majority has not been linked to morphospecies yet, therefore, the known species only constitute a fraction of the species diversity. The genus includes the two subgenera Haemoproteus and Parahaemoproteus. Parasites of the subgenus Haemoproteus are mainly found in birds of the orders Columbiformes, Suliformes, and Charadriiformes and are transmitted by louse flies (Hippoboscidae). Parasites of the subgenus Parahaemoproteus are extremely diverse in passeriform birds globally, particularly in the northern hemisphere, and are transmitted by biting midges (Ceratopogonidae).

For the present study, the 18S rRNA genes of 19 Haemoproteus lineages were sequenced. About half of the investigated lineages are currently linked to the Haemoproteus majoris (hCCF5, hCWT4, hEMSPO03, hPARUS1, hPHSIB1, and hWW2) and Haemoproteus belopolskyi (hACDUM2, hARW1, hMW3, and hSW1) groups, which include some of the most common avian haemosporidian lineages in Palearctic passeriform birds. Moreover, the 18S sequences were sequenced of Haemoproteus balmorali hCOLL3, Haemoproteus fringillae hCCF3, Haemoproteus sp. hROFI1, Haemoproteus nucleocondensus hGRW01, Haemoproteus parabelopolskyi hSYAT02, Haemoproteus payevskyi hRW1, Haemoproteus cf. magnus hCCF6, and the yet unlinked lineages hCCF2 and hCWT7. The sequences of eight Haemoproteus lineages previously published by Harl et al. [8] were included in the analyses.

The main question was whether avian Haemoproteus parasites feature clusters of distinct 18S variants like many Plasmodium species, which would indicate that their ribosomal genes could also be differentially expressed in the bird hosts and dipteran vectors. Moreover, the ribosomal genes constitute a large part of the RNA molecules in cells and therefore are suitable targets for molecular genetic approaches such as in situ hybridization assays. The 18S sequences were the basis for designing species/lineage-specific oligonucleotide probes, which can be used to target parasites in histological sections and differentiate parasites in co-infections.

Methods

Sample collection and preparation

For the present study, the 18S rRNA genes of 19 Haemoproteus lineages found in passeriform birds in the Palearctic were sequenced. The samples were part of the collections of the Nature Research Centre in Vilnius (Lithuania) and the Institute of Pathology at the University of Veterinary Medicine Vienna (Austria). Wild birds were collected at the Ornithological Station in Ventė Cape (Lithuania) using stationary traps (large ‘Rybachy’ type, zigzag and funnel traps) and mist nets between 2018 and 2021, and blood samples were taken with heparinised microcapillaries after puncturing the brachial vein. Drops of fresh blood were used to prepare blood spots on filter paper for DNA analysis and several blood films on glass slides for microscopic examination. The blood films were fixed in absolute methanol for one minute and then stained with 10% Giemsa [10]. The blood films were analysed by microscopic examination and 60 birds with high parasitaemia were euthanised by decapitation according to permits (see Ethical statement). Blood samples were taken from two birds (AH1663 and AH1664) during routine bird ringing at the Biological Station Neusiedler See (Illmitz, Burgenland) in 2018 as described above. Tissue samples (liver) were taken from one dead bird (AH2023) submitted to the Institute of Pathology (Vetmeduni Vienna) for a citizen science study in 2020 [11] and stored at minus 80 °C for DNA analysis. Organ samples of the dead birds were fixed in formalin and embedded in paraffin (FFPE) and deposited in the tissue collection of the Institute of Pathology (Vetmeduni Vienna). Voucher blood films of the Lithuanian samples were deposited at the Nature Research Centre. DNA was isolated either from blood spots on filter papers or frozen liver tissue using the DNeasy Blood & Tissue Kit (QIAGEN, Venlo, Netherlands). The manufacturer’s protocol was followed for isolation of DNA from tissue but two eluates of 100 µl each were made from the same column, the first at 8000 rpm, and the second at 13,000 rpm. The second eluate was used for the PCRs. Information on the samples analysed for the present study is provided in Table 1. The table also includes information on the samples studied by [8], who previously published 18S sequences of eight Parahaemoproteus lineages.

Table 1 Samples analysed in the present study
Full size table

CytB PCR primers

The nested PCR assay described by [12] was used to obtain a 478 bp fragment of the Cytochrome B (CytB) gene, the common DNA barcode sequence for avian haemosporidian parasites. PCRs were performed using the primers HaemNFI (5′-CAT ATA TTA AGA GAA NTA TGG AG-3′) and HaemNR3 (5′-ATA GAA AGA TAA GAA ATA CCA TTC-3′) in the first PCR, and HaemF (5′-ATG GTG CTT TCG ATA TAT GCA TG-3′)/HaemR2 (5′-GCA TTA TCT GGA TGT GAT AAT GGT-3′) and HaemFL (5′-ATG GTG TTT TAG ATA CTT ACA TT-3′)/HaemR2L (5′-CAT TAT CTG GAT GAG ATA ATG GNG C-3′) in the nested PCRs, respectively [12]. The primers CytB_HPL_intF1 (5′-GAG AAT TAT GGA GTG GAT GGT G-3′) and CytB_HPL_intR1 (5′-ATG TTT GCT TGG GAG CTG TAA TC-3′) [8] were used to obtain 885 bp sections of the CytB, which were used to calculate the phylogenetic trees.

18S PCR primers

To amplify the almost entire 18S rRNA genes of the Haemoproteus lineages, the primers 18S_H_1F (5′-TGG TTG ATC TTG CCA GTA ATA TAT GT-3′) and 18S_H_1R (5′-CGG AAA CCT TGT TAC GAC TTTTG-3′) were used, which are located at the 5′-end and 22 bp from the 3′-end of the 18S, respectively [8]. Since the 18S sequence reads of some Haemoproteus lineages did not overlap, the Haemoproteus-specific primers 18S_H_int_F (5′-AGA TCA AGT TGA AGT GCC AGC ATT-3′) and 18S_H_int_R (5′-CGT TAA ACA CGC GAC GTC-3′) were designed, which are located approximately 550 bp and 1700 bp from the 5′-end of the 18S, to sequence a 1100 bp section covering the middle part. The latter primers only target the 18S rRNA genes of the subgenus Parahaemoproteus, but not those of the genetically highly diverged subgenus Haemoproteus [8].

PCRs and molecular cloning

The PCRs targeting the 478 bp CytB barcode section followed the protocol of [12] and were performed using the GoTaq® G2 Flexi DNA Polymerase (Promega, Wisconsin, Madison, USA). The PCRs started with an initial denaturation for 2 min at 94 °C, followed by 35 cycles with 30 s at 94 °C, 30 s at 50 °C, 1 min at 72 °C, and a final extension for 10 min at 72 °C. Each 1 µl of the first PCR-product was used as template in the two nested PCRs. The PCRs targeting the 885 bp CytB fragment were performed with the GoTaq® Long PCR Master Mix (Promega, Wisconsin, Madison, USA). The PCRs started with an initial denaturation for 2 min at 94 °C, followed by 35 cycles with 30 s at 94 °C, 30 s at 55 °C, 2 min at 68 °C, and a final extension for 10 min at 72 °C. The PCRs targeting the 18S sequences were performed with the GoTaq® Long PCR Master Mix under the same conditions but with 2 min extension time. Each two PCRs were done to obtain sufficient PCR product for direct sequencing and molecular cloning. The PCR products were visualised on 1% LB agarose gels and sent to Microsynth Austria GmbH (Vienna, Austria) for purification and sequencing in both directions using the PCR primers.

The 18S PCR products were then further processed and subjected to molecular cloning as described in [8]. Each 20 µl of PCR-product were run on 1% LB agarose gels and the bands were excised with flamed spatulas. The gel bands were purified using the QIAquick Gel-Extraction Kit (QIAGEN) following the standard protocol and eluted with 20 µl distilled water. Cloning was performed with the TOPO™ TA Cloning™ Kit (Invitrogen, Carlsbad, California, USA) using the pCR™4-TOPO® vector and One Shot® TOP10 competent cells. After ligation and transformation, the Escherichia coli cells were recovered in SOC medium for 1 h at 37 °C, plated on LB agar plates, and grown for 20 h at 37 °C. From each cloning assay, 15 to 20 individual clones were picked with sterilised (flamed) tooth sticks and transferred to fresh LB agar plates. The same tooth sticks with remaining E. coli were twisted in PCR-tubes with 25 µl master mix for the colony-PCRs. Colony-PCRs were performed with the GoTaq® Long PCR Master Mix (Promega) under the same conditions as the 18S PCRs (see above) but using the primers M13nF (5′-TGT AAA ACG ACG GCC AGT GA-3′) and M13nR (5′-GAC CAT GAT TAC GCC AAG CTC-3′) [8]. The PCR-products of up to 15 clones carrying inserts of the expected size were sent to Microsynth Austria GmbH (Vienna, Austria) for purification and sequencing using the colony-PCR primers.

Analysis of raw sequence data

The forward and reverse reads and electropherograms of the CytB and 18S sequences were aligned manually and checked by eye in BioEdit v.7.0.8.0 [13]. Then the sequences were aligned and sorted with MAFFT v.7 [14] using the default option (FFT-NS-2), and primer and cloning vector sequences were cut from the 18S alignments. Since some 18S clones did not overlap, the middle part was re-sequenced from 65 clones with the primers 18S_H_int_F and 18S_H_int_R. The long and the short 18S sequences were combined and realigned with MAFFT v.7. and all aberrant positions were rechecked in the corresponding electropherograms.

CytB haplotype networks

To visualise the geographic and host distribution of the H. majoris and H. belopolskyi lineages, two DNA haplotype networks were calculated. The CytB lineages of both groups cluster in reciprocally monophyletic clades, which currently contain 21 (H. majoris) and 40 (H. belopolskyi) lineages, most of which have not been linked to morphospecies yet. The clades were identified by calculating a Maximum Likelihood (ML) tree based on all complete Haemoproteus lineages listed in the MalAvi database (http://130.235.244.92/Malavi/index.html) and a few other lineages available from NCBI GenBank only. The sequences were aligned with MAFFT v.7. [14] applying the default option (FFT-NS-2) and the first and last two bp were trimmed because they were erroneous or incomplete in some sequences. A ML bootstrap tree (1000 replicates) was calculated with IQ-TREE v.1.6.12. [15] based on the trimmed 474 bp alignment (1325 unique lineages) applying the substitution model GTR+F+I+G4. For each lineage contained in the H. majoris and H. belopolskyi clades, the information on hosts, countries, and references was extracted from the MalAvi “Hosts and Sites” table (http://130.235.244.92/Malavi/) and organised in a Microsoft Excel (Microsoft, Redmond, WA, USA) sheet. Moreover, sequences and related information for some lineages, which were available on NCBI GenBank only, were added. The Median-Joining haplotype networks were calculated with Network 10.2.0.0 (Fluxus Technology Ltd, Suffolk, UK) using the default settings. The networks were graphically arranged and provided with information on hosts species/families and geographic regions according to the United Nations geoscheme with Network Publisher v.2.1.2.5 (Fluxus Technology Ltd). To show the geographic and host distribution of the other lineages, pie charts were created with Microsoft Excel based on the MalAvi “Hosts and Sites” data. All graphics were finalised with Adobe Illustrator CC v.2015 (Adobe Inc., San José, CA, USA).

Phylogenetic trees

ML and Bayesian Inference (BI) trees were calculated for both the CytB and 18S data sets. CytB trees were calculated based on the 885 bp alignment of the 19 MalAvi lineages obtained in the present study and eight lineages published by [8]. A sequence of Plasmodium matutinum pLINN1 (MT912161) was included as outgroup. The best fit substitution model suggested by IQ-TREE v.1.6.12. [15] according to the corrected Akaike Information Criterion (AICc) was GTR+F+I+G4. The ML bootstrap tree was calculated with IQ-TREE v.1.6.12. [15] by performing 10,000 bootstrap replicates. The BI tree was calculated with MrBayes v.3.2. [16]. The analysis was run for 5 million generations (2 runs each with 4 chains, one of which was heated) and every thousandth tree was sampled. The first 25% of trees were discarded as burn-in and a 50% majority rule consensus tree was calculated from the remaining 3750 trees.

The alignment of 18S sequences contained 201 clones of 19 MalAvi lineages obtained in the present study and 71 clones of seven MalAvi lineages published by [8]. Thirteen clones were excluded from the analyses because they originated from other lineages present in co-infections. The 18S sequences of the sample AH0002H (hRB1) from [8] were also excluded because they lacked a 115 bp section at the 5′-end. The sequences were aligned with MAFFT v.7. [14] using the option G-INS-I (globally alignment based on Needleman-Wunsch algorithm). To reduce the number of sequences, subsets of two to five distinct clones per MalAvi lineage (74 sequences in total) were selected. An outgroup was not included because the 18S sequences of other haemosporidian genera differ strongly from those of Parahaemoproteus spp. and the removal of gap position would have led to the loss of information. The subset of sequences was realigned with MAFFT v.7. (G-INS-I option), resulting in a 2526 bp alignment. The first 65 and the last 18 positions were removed because they were not present in the sequences of samples AH1982 (hCOLL3) and AH1895 (hSYAT02) (obtained by direct sequencing), and in the sample AH0608 (hEMCIR01; [8]). After trimming the latter sites, the alignment featured 2443 positions. After removing all sites containing gaps with trimAl v.1.2. [17], the final alignment featured 1605 positions. The best fit substitution model suggested by IQ-TREE v.1.6.12. [15] according to the corrected Akaike Information Criterion (AICc) was TVM+F+I+G4. Since the latter model is not implemented in MrBayes v.3.2. [16], the second-best model GTR+F+I+G4 was used for both the ML and BI analyses. The ML bootstrap and the BI trees were calculated with IQ-TREE v.1.6.12. [15] and MrBayes v.3.2. [16] applying the same parameters as used for inferring the CytB trees.

Sequence comparison and recombination tests

Prior to the analysis of sequences, the 18S clones of each MalAvi lineage were placed in separate files to determine the minimum and maximum lengths of sequences. The mean GC-contents of 18S sequences from each MalAvi lineage were calculated with Microsoft Excel. The sequences of each MalAvi lineage were aligned separately with MAFFT v.7. [14] using the option G-INS-I, and maximum p-distances between the variants were calculated with MEGA X v.10.0.5 [18]. The latter alignments were also used to test whether distinct 18S clones from the same MalAvi lineages showed chimeric features, thus indicating recombination between different 18S variants from the same MalAvi lineages. RDP5 v.5.3. [19]) was used to perform the following recombination tests: RDP [20], Bootscan [21], GENECONV [22], Maxchi [23], Chimaera [24], SiSscan [25], and 3Seq [26].

Design of probes for in situ hybridization

Based on the alignments of all Haemoproteus 18S sequences, oligonucleotide probes were designed, which specifically target the investigated lineages. The alignment was inspected by eye to identify suitable regions for probe binding. Most probes were placed in variable sequence regions specific to each one MalAvi lineage. The quality of the probes was checked with AmplifX v.2.0.7 (Nicolas Jullien, Aix-Marseille Univ., CNRS, INP, Marseille, France; https://inp.univ-amu.fr/en/amplifx-manage-test-and-design-your-primers-for-pcr). All probes were blasted against genomes of apicomplexan parasites and birds in NCBI GenBank to exclude unintentional binding.

Results

The 18S sequences of 19 Parahaemoproteus lineages, half of which belong to the H. majoris and H. belopolskyi groups, were analysed. The 18S sequences published for seven Haemoproteus lineages by [8] were included in the statistical analyses and phylogenetic trees.

CytB haplotype networks

To visualise the geographic and host distribution of CytB lineages belonging to the H. majoris and H. belopolskyi groups, DNA haplotype networks were calculated based on 474 bp alignments containing data of all lineages clustering in two clades.

The H. majoris clade included 21 unique lineages differing by up to nine bp in the 474 bp CytB sequence (Fig. 1). Most lineages were predominantly found in songbirds in Europe and Western Asia, and occasionally in Northern Africa and Southern Asia, except for hPOEATR01, hTUMIG08, hTUMIG21, and hPHYBOR04, mainly from Northern American thrushes. According to the MalAvi database (http://130.235.244.92/Malavi/), the following six lineages were attributed to H. majoris by [27, 28]: hCCF5, hCWT4, hPARUS1, hPHSIB1, hPHYBOR04, and hWW2. The 18S sequences were obtained from all the latter lineages except for hPHYBOR04, because the only sample available did not contain sufficient DNA. The three lineages hPARUS1, hPHSIB1, and hWW2 are among the most common haemosporidian parasite lineages in Eurasian songbirds. hPARUS1 (876 records) was mainly found in Northern Europe (510), Western Europe (135), Eastern Europe (104), Western Asia (65), and Southern Europe (48); the main hosts are Cyanistes caeruleus (463) and Parus major (244) of the family Paridae. hPHSIB1 (868) was mostly found in Northern Europe (791) and Eastern Europe (44); the Muscicapidae species Ficedula albicollis (499) and Ficedula hypoleuca (172) are the most common hosts. hWW2 (505) was mostly found in Northern Europe (445) and has a wider host spectrum with about half of the records from Sylviidae (247), followed by Phylloscopidae (138), Paridae (55), and Muscicapidae (31); the most common host species are Phylloscopus trochilus (108), Sylvia borin (90), Sylvia atricapilla (79), and Sylvia communis (64). hCCF5 (36) was almost exclusively found in Fringilla coelebs (35) in Northern Europe (32) and Northern Africa (4). hCWT4 (22) was mainly found in Northern Europe (8), Eastern Europe (7), and Western Asia (5), with most records from Sylviidae (15), particularly S. communis (8). hEMSPO03 (8) has not been linked to a morphospecies yet; it was found in Western Asia (5), Western Europe (2), and Eastern Europe (1) in Pyrrhula pyrrhula (3) and Phylloscopus nitidus (2) and others. Records of hEMSPO03 from Phylloscopus humei (1) and Phylloscopus trochiloides (3) in India [29] were not included because the sequences covered only 448 bp of the CytB barcode region.

Fig. 1
figure 1

Median‑Joining DNA haplotype network of partial (474 bp) CytB sequences belonging to the Haemoproteus majoris group. The two figures show the distribution in A bird families and B geographic areas according to the United Nations geoscheme. Each circle represents a unique haplotype/lineage. The frequency is indicated for all haplotypes with more than one record and roughly corresponds to the size of circles. Bars on branches indicate the number of substitutions between two haplotypes. Small white circles represent median vectors, which are hypothetical (often ancestral or unsampled) sequences required to connect existing haplotypes with maximum parsimony. The lineages analysed in the present study are marked with asterisks

Full size image

The H. belopolskyi clade included 40 unique lineages differing by up to 38 bp in the 474 bp CytB sequence (Fig. 2). Hence, the distances between some lineages are much higher than in the H. majoris clade. They were almost exclusively found in various species of the family Acrocephalidae (marsh- and tree-warblers) in Europe, Asia, and Africa. According to the MalAvi database (http://130.235.244.92/Malavi/), the following nine lineages belong to H. belopolskyi: hARW1, hHIICT1, hHIICT3, hHIICT4, hHIICT5, hMW1, hRW3, hSW1, and hSW3. However, only HIICT1 and hMW1 were studied morphologically [30], and hHIICT5 (1 bp difference from hHIICT3) does not cover the entire barcode region. The clade also contains the lineage hGRW01, which was linked to H. nucleocondensus by [31], but differs only in 2 bp from H. belopolskyi hMW1. Further taxonomic studies might reveal that the lineages in the H. belopolskyi clade belong to multiple cryptic species. 18S sequences were obtained from the following five lineages: hACDUM2, hARW1, hGRW01, hMW3, and hSW1. hACDUM2 (52 records) was found mainly in Acrocephalus agricola in Eastern Europe (22) and Acrocephalus dumetorum in Southern Asia (22). hARW1 (25) was found in Acrocephalus scirpaceus (12), Acrocephalus palustris (5), and a few other bird species in Northern, Western, and Eastern Europe, Western Asia, and Northern and Eastern Africa. hGRW01 (225) was almost exclusively detected in Acrocephalus arundinaceus (216) throughout its distribution range in Europe (195) and Africa (18), and Western Asia (3). hMW3 (6) was so far only found in A. palustris in Turkey (3) and in the present study in Lithuania (2). hSW1 (171) was mainly detected in Acrocephalus schoenobaenus (158) in Eastern Europe (152), Northern Europe (2), Western Europe (1), Western Asia (2), and Western Africa (1), and in A. scirpaceus (8) in Eastern Europe (3), Southern Europe (2), Western Asia (3), and Western Africa (1). The distributions of the other lineages (hCCF2, hCCF3, hCCF6, hCOLL3, hCWT7, hEMCIR01, hCULKIB01, hLK03, hROBIN1, hROFI1, hRW1, hSISKIN1, hSTAL2, hTURDUS2, and hSYAT02) in bird hosts and geographic areas (UN geoscheme) are shown as pie charts (Fig. 3). Haemoproteus sp. hCCF2 (95 records) was mainly found in Fringilla coelebs (81) in Northern Africa (52), Northern Europe (18), and Southern Europe (16). Haemoproteus fringillae hCCF3 (58) was detected mostly in F. coelebs (31), Carduelis chloris (14), and Pyrrhula pyrrhula (7) in Northern Europe (36), Western Asia (12), Eastern Europe (5), Northern Africa (4), and Southern Europe (1). Haemoproteus cf. magnus hCCF6 (87) was almost exclusively found in F. coelebs (85) in Northern Africa (53), Europe (27), Western Asia (5), and Southern Asia (2). Haemoproteus balmorali hCOLL3 (108) was detected in Ficedula albicollis (76), F. hypoleuca (24), and other Muscicapidae (9) in Northern Europe (56), Eastern Europe (41), Southern Europe (7), Western Asia (2), and Northern Africa (1). Haemoproteus sp. hCWT7 (29) was mostly found in S. communis (26) in Western Asia (24). Haemoproteus syrnii hCULKIB01 (22) was found in the owl species Strix aluco (10), Strix uralensis (9), Bubo bubo (1), and Strix nebulosa (1) in Europe. Haemoproteus dumbbellus hEMCIR01 (47) was almost exclusively found in Emberiza citrinella (40) in Europe. Haemoproteus brachiatus hLK03 (19) was mainly found in Falco tinnunculus (13) and other falcons (2) in Eastern Asia (8), Western Asia (1), and Europe (6). Haemoproteus sp. hROFI1 (61) was mainly found in F. coelebs (22), Carduelis chloris (20), and other Fringillidae species (12) in Northern Europe (38), Northern Africa (16), Eastern Europe (4), and Western Asia (3). Haemoproteus attenuatus hROBIN1 (103) was mostly found in the Muscicapidae species Erithacus rubecula (40), Luscinia luscinia (34), Luscinia megarhynchos (7), and Saxicola rubetra (7) in Europe and Western Asia. Haemoproteus sp. hRW1 (115) was detected in A. scirpaceus (62) and other Acrocephalidae (4), in Cinclus cinclus (24), Lanius meridionalis (16), Luscinia svecica (8), and Cisticola nigriloris (1) in Southern Europe (87), Eastern Europe (12), Northern Europe (11), and other regions (5). Haemoproteus tartakovskyi hSISKIN1 (87) was mostly found in Fringillidae (83) in Central America (37), Northern America (22), and Europe (26). The most common hosts were Haemorhous mexicanus (42), Acanthis flammea (7), Loxia curvirostra (6), Loxia leucoptera (5) in the Americas and Spinus spinus (16) in Europe. Haemoproteus parabelopolskyi hSYAT02 (250) was almost exclusively found in S. atricapilla (246) in Western Europe (103), Southern Europe (69), Eastern Europe (38), Northern Europe (30), Western Asia (8), and Africa (2). Haemoproteus syrnii hSTAL2 (37) was detected in owls in Europe (36) and Northern Africa (1), mainly in Strix aluco (19) and Strix uralensis (15). Haemoproteus minutus hTURDUS2 (170) was found mainly in Turdidae (140) in Europe (92), Western Asia (37), and the Americas (14). Most records originate from Turdus merula (121).

Fig. 2
figure 2

Median‑Joining DNA haplotype network of partial (474 bp) CytB sequences belonging to the Haemoproteus belopolskyi group. The two figures show the distribution in A bird families and B geographic areas according to the United Nations geoscheme. Each circle represents a unique haplotype/lineage. The frequency is indicated for all haplotypes with more than one record and roughly corresponds to the size of circles. Bars on branches and numbers in squares indicate the number of substitutions between two haplotypes. Small white circles represent median vectors, which are hypothetical (often ancestral or unsampled) sequences required to connect existing haplotypes with maximum parsimony. The lineages analysed in the present study are marked with asterisks

Full size image
Fig. 3
figure 3

Pie charts showing the distribution of parasite lineages in geographic areas according to the United Nations geoscheme (left) and bird hosts (right)

Full size image

The information shown in the DNA haplotype networks and pie charts is summarised in Additional file 1: Table S1.

18S sequence analyses

Molecular cloning was required in most cases to retrieve the 18S sequences because the variants differed in their lengths. Only the samples AH1895H (hSYAT02) and AH1982H (hCOLL3) were not cloned because their 18S sequences were fully readable following direct sequencing. Almost the complete 18S sequences were cloned from 20 samples featuring 17 Haemoproteus lineages. Two to 15 clones were sequenced per sample (201 in total, 10.1 clones in mean). We intended to sequence up to 15 clones per lineage, but the yield of clones was extremely low in some cases. Of the 201 clones, 13 were excluded from the analyses because they originated from co-infections, which were not detected when sequencing the CytB barcode section with the standard primers by [12]. However, in some cases co-infections were visible either in the 885 bp sequences obtained with the primers CytB_HPL_intF1 and CytB_HPL_intR1 by [8] or the 18S clones. Mostly the 18S clones could be clearly assigned to one of the MalAvi lineages because other samples had single infections with the same lineages. The sample AH2168H featured a mixed infection with the lineages hCCF3 (6 clones), hCCF6 (7 clones), and hCCF5 (2 clones) of which only the hCCF3 clones were included in the analyses; the hCCF6 clones were excluded because samples AH2153H and AH2154H contained single infections with the same lineage, and the two hCCF5 clones were excluded because the mid parts of the sequences were unreadable. The sample AH1973H featured a co-infection with lineages hCCF3 (9 clones), hCCF5 (2 clones), and hROFI1 (1 clone). The hROFI1 clone was excluded from the analyses because the sample AH2171H provided sufficient clones of lineage hROFI1, but the two hCCF5 clones were kept because they were the only complete 18S sequences of this lineage. The sample AH2171H contained a co-infection with hROFI1 (12 clones) and hCCF6 (3 clones); the hCCF6 clones were excluded from the analyses because sufficient hCCF6 clones were available from samples AH2153H and AH2154H. Moreover, sample AH1664H likely contained a co-infection with hSW1 and another H. belopolskyi lineage. The sample featured four clones (3, 5, 6, 14) closely resembling those of H. belopolskyi hARW1 (AH1903H) and hMW3 (AH1899 and AH1902), and seven clones (2, 4, 7, 8, 10, 11, 12) forming a separate branch within the H. belopolskyi clade. The first four clones might belong to lineage hSW3, which was exclusively found in the same host species (A. schoenobaenus) and differs only in one bp from hARW1 (Fig. 2). Double peaks in the electropherograms of the 885 bp CytB fragment matched lineage hSW3 but were too faint to clearly confirm its presence. Apart from AH1973H, AH2168H, AH2171H, and AH1664H, the other samples most likely contained mono-infections with 18S clones belonging to single parasite lineages. The 18S sequences of H. majoris hEMSPO03 (AH2023H) could not be fully retrieved because it featured poly-A and poly-T motives from position 1660 to 1800. The 18S sequences were uploaded to NCBI GenBank under the accession numbers OR337936–OR338136. The 885 bp sections of the mitochondrial CytB were deposited under the accession numbers OR283176–OR283196.

Phylogenetic trees were calculated with the 18S and CytB sequences of the present study and those published by [8]. The 18S tree (Fig. 4), containing a selection of two to five distinct 18S clones per MalAvi lineage (74 sequences in total), was mid-point rooted. The CytB tree (Fig. 5) was calculated with the 885 bp sequences and rooted with a sequence of Plasmodium matutinum pLINN1. The deeper nodes obtained low support values in both trees and the topology differed partially, but the 18S and CytB trees shared some common patterns. The lineages H. brachiatus hLK03, H. minutus hTURDUS2, H. syrnii hCULKIB01, and H. syrnii hSTAL2 take more basal positions in the trees, whereas the other lineages cluster together in one clade with low (18S; bs/bp = 72/0.76) and moderate support (CytB; bs/pp = 82/0.98). The latter clade featured four highly supported subclades in both trees. The first subclade includes H. dumbbellus hEMCIR01, H. fringillae hCCF3, Haemoproteus sp. hROFI1, H. tartakovskyi hSISKIN1, and Haemoproteus sp. hCCF2. The second subclade includes the H. balmorali hCOLL3 and H. attenuatus hROBIN1. The third subclade features the H. majoris lineages hCCF5, hPARUS1, hWW2, hCWT4, hPHSIB1, and hEMSPO03 (not yet linked to a morphospecies). The fourth subclade includes the group of four H. belopolskyi lineages hMW3, hARW1, hSW1, hACDUM2, H. nucleocondensus hGRW01, and the lineages H. parabelopolskyi hSYAT02, H. payevskyi hRW1, H. cf. magnus hCCF6, and H. sp. hCWT7. The 18S clones of most lineages clustered into reciprocally monophyletic clades. However, as mentioned above, the clones of sample AH1664H probably belong to two different Haemoproteus lineages (hSW1 and hSW3).

Fig. 4
figure 4

Bayesian inference tree of Haemoproteus 18S sequences. Posterior probabilities and maximum likelihood bootstrap values are indicated at most nodes. The scale bar indicates the expected mean number of substitutions per site according to the model of sequence evolution applied. The tree was midpoint-rooted, no outgroup was used

Full size image
Fig. 5
figure 5

Bayesian inference tree of Haemoproteus CytB sequences (885 bp). Posterior probabilities and maximum likelihood bootstrap values are indicated at most nodes. The scale bar indicates the expected mean number of substitutions per site according to the model of sequence evolution applied. The tree was rooted with a sequence of Plasmodium matutinum pLINN1

Full size image

The p-distances between the 18S variants of the 27 Parahaemoproteus lineages ranged from 0.46% in H. lanii hRB1 to 20.54% in H. tartakovskyi hSISKIN1 (Table 2). The latter lineage is exceptional because it featured two similar 18S variants and a highly diverged third one. Excluding this sample, the mean maximum p-distance between 18S clones of the other lineages was 1.84%. As mentioned above, some clones of sample AH1664H might belong to hSW3 (not confirmed) in co-infection. When testing the two sequence clusters separately, maximum p-distances between clones of hSW1 and hSW3 were less with 1.26 and 0.73, respectively (Table 2).

Table 2 Sequence features of the Haemoproteus 18S rRNA genes
Full size table

The approximate total lengths of the 18S sequences (including missing parts and primer regions) ranged from 1943 bp in H. minutus hTURDUS2 to 2278 bp in H. majoris hCCF5. The largest differences between the shortest and longest 18S sequences were found in H. tartakovskyi hSISKIN1 with 53 bp and in H. fringillae hCCF3 with 40 bp; the mean difference over all 27 Parahaemoproteus lineages was 6.8 bp. The GC contents ranged from 39.4% in H. tartakovskyi hSISKIN1 to 47.4% in H. syrnii hSTAL2; the overall mean was 45.2% (Table 2). Interestingly, H. tartakovskyi hSISKIN1 featured three 18S clones with 42.7% (short branches in Fig. 4) and seven clones with 37.4% GC-content (long branch in Fig. 4).

Tests for recombinant signals were performed separately for the alignments of 18S clones from each Parahaemoproteus lineage using RDP5 [19]. Recombinant signals were detected in ten of the 27 lineages investigated, including H. fringillae hCCF3, Haemoproteus sp. hROFI1, H. nucleocondensus hGRW01, H. syrnii hCULKIB01, H. syrnii hSTAL2, H. tartakovskyi hSISKIN1, and the H. majoris lineages hCWT4, hPHSIB1, hPARUS1, and hWW2 (Table 2). The strongest recombination signals were detected in hPHSIB1, hWW2, and hCCF3 with each five or more of the seven tests providing significant results (Additional file 2: Table S2).

Probes for in situ hybridization

Based on the alignment containing all Haemoproteus 18S sequences available, oligonucleotide probes were designed to specifically target the 18S rRNA of the lineages investigated in this study. The 18S sequences of most Haemoproteus lineages featured unique sequence regions, which could be targeted with oligonucleotide probes by in situ hybridization (Table 3). However, 18S sequences of the H. belopolskyi lineages hARW1, hMW3, and hSW3 were too similar and a probe targeting all three lineages in parallel was designed. The probe designed for lineage hROFI1 only targets one of the two main variants (C02, C08, C09, C11); the probe was already tested, confirming the expression in the bird hosts (unpublished results). The lengths of the probes and the annealing temperatures ranged from 23 to 31 nucleotides and 54.2 °C and 68.1 °C, respectively. Most probes obtained a maximum (100) quality score in AmplifX v.2.0.7.

Table 3 In silico tested probes for in situ hybridization
Full size table

Discussion

For the present study, the 18S rRNA genes of 19 Haemoproteus lineages, all belonging to the subgenus Parahaemoproteus, were sequenced. More than half of these lineages were previously linked to or are closely related to the H. majoris (hCCF5, hCWT4, hEMSPO03, hPARUS1, hPHSIB1, hWW2) and H. belopolskyi (hACDUM2, hARW1, hMW3, hGRW01, hSW1) species groups based on morphological characters of their blood stages and/or similar CytB barcode sequences. The two species groups include some of the most common Haemoproteus lineages found in songbirds of the Palearctic region. Moreover, the 18S rRNA genes of eight other Haemoproteus lineages were sequenced: hCOLL3, hCCF2, hCCF3, hCCF6, hCWT7, hROFI1, hRW1, and hSYAT02. The data on eight Haemoproteus lineages published by [8], hCULKIB01, hEMCIR01, hLK03, hROBIN1, hRB1, hSISKIN1, hSTAL2, and hTURDUS2, were included in the sequence analyses.

Most mammalian Plasmodium species feature two highly diverged sequence clusters, each containing one or two similar 18S variants. The 18S variants of Plasmodium falciparum, Plasmodium vivax, and Plasmodium berghei were shown to be differentially expressed in the vertebrate hosts (A-type) and mosquito vectors (S-type) [32,33,34]. Plasmodium vivax is the only species with an additional O-type variant, which is expressed in the ookinetes and oocysts [33]. Another exception is Plasmodium malariae because its 18S genes are almost identical and do not form separate clusters. The distances between 18S variants were also comparably high in Plasmodium vaughani pSYAT05 (9.3%), Plasmodium matutinum pLINN1 (10.8%), and Plasmodium elongatum pGRW06 (14.9%) [8]. Recombination tests detected chimeric features in the 18S sequences of several Plasmodium species, suggesting that distinct variants do not evolve independently according to a model of birth-and-death evolution under strong purifying selection [4] but rather in a semi-concerted fashion [8, 35].

The Haemoproteus parasites studied also featured distinct 18S variants, but the averaged maximum distances between the variants were considerably lower (mean of 1.84% excluding the aberrant H. tartakovskyi hSISKIN1) than in most Plasmodium species investigated. Unlike in many Plasmodium species, the 18S variants of most lineages clustered into reciprocally monophyletic clades. An exception was sample AH1664H, whose 18S variants differed by 5.64% and fell into separate clades within the H. belopolskyi group, however, the latter sample likely featured a co-infection with H. belopolskyi hSW3 and H. belopolskyi hSW1 (see “Results”). Haemoproteus sp. hROFI1 featured two 18S variants diverged by 7.5%, but they clustered together in a weakly supported clade (Fig. 4); apart from double peaks matching hCCF6 in the long CytB sequences, no other lineage in co-infection was identified. The most exceptional patterns were found in H. tartakovskyi hSISKIN1, which featured three 18S variants, two similar ones separated by 6.5% and a third one separated from the latter two by 18.3% and 20.5%, respectively. The third variant showed a high number of substitutions (mostly changes G/C to A/T) in sequence regions, which were conserved in most other Haemoproteus lineages, therefore its GC-content was considerably lower with 37.4% compared to that of the other variants with 43.2% and 42.4%. This aberrant 18S variant might be non-functional and therefore acquired a much higher number of substitutions compared to other copies of the Haemoproteus 18S rRNA genes. Despite the high sequence divergence, all three variants clustered in a well-supported clade (Fig. 4). The nuclear genome sequences of H. tartakovskyi hSISKIN1 published by Bensch et al. [36] also included the aberrant third variant (PRJNA309868, contig 65) but only one of the other two (PRJNA309868, contig 602). The mean GC-content of all Haemoproteus 18S sequences analysed was 45.2%, which is considerably higher than that of avian Plasmodium spp. with 34.0% and common Leucocytozoon spp. with 37.3%. The GC-content was only higher in parasite lineages belonging to the Leucocytozoon toddi species group with 49.3% [8]. The approximate total lengths of the Haemoproteus 18S sequences ranged from 1943 bp in H. minutus hTURDUS2 to 2278 bp in H. majoris hCCF5 and, therefore, was more variable than in avian Plasmodium spp. (2100 to 2177 bp), Leucocytozoon spp. (2094 to 2138 bp), and the Leucocytozoon toddi species group (2125 to 2308 bp). The recombination tests conducted with the 18S sequences of the Haemoproteus parasite lineages also indicated chimeric features potentially resulting from recombination between 18S variants of the same species or lineages. The results differed between species, e.g., no recombinant 18S variants were detected in the H. belopolskyi group compared to the H. majoris group with four out of five lineages featuring recombinant variants. Thus, the results suggest that the nuclear ribosomal genes of Haemoproteus species rather evolve in a semi-concerted fashion as suggested for several Plasmodium species by Corredor and Enea [35] than according to a model of birth-and-death evolution as proposed by Rooney [4]. Corredor and Enea [35] used the term semi-concerted evolution because they found that some (but not all) 18S rRNA gene copies of Plasmodium spp. evolve in concert, thus requiring some form of sequence interaction (conversion) other than unequal crossing over. In contrast, the ribosomal genes of most eukaryotes evolve in a fully concerted fashion, leading to the homogenization of individual gene copies [3] involving mechanisms such as unequal crossing over during recombination, gene duplication, and inter-chromosomal gene conversion [1, 2, 37]. The model of birth-and-death evolution on the other hand assumes that multigene families involved in the immune system, such as immunoglobulins and the major histocompatibility complex (MHC), do not evolve in a concerted fashion; new copies originate by gene duplication, whereas others become non-functional and deleted over time [38, 39].

The 18S rRNA gene has been used as the standard reference sequence when screening for apicomplexan parasites and determining their phylogenetic relationships. However, due to the presence of distinct variants and recombination between them, long inserts/deletions, and GC-contents strongly varying between genera/subgenera, the 18S sequences of haemosporidian parasites are less suitable as a phylogenetic marker than in other groups of apicomplexan parasites (e.g., Eucoccidiorida, Piroplasmorida, and Eugregarinorida), which mostly possess identical and more conserved copies of nuclear ribosomal genes. Therefore, PCR screening assays for human Plasmodium species only target one of the main 18S variants [40, 41]. More recent approaches even established quantitative reverse transcription PCR (qRT-PCR) to directly target the 18S rRNA, resulting in even higher sensitivity [42, 43]. These approaches are practical when screening for human Plasmodium infections because they comprise only four species, but less suitable when screening for haemosporidian parasites of wild birds because they include a vastly higher number of species and obtaining their 18S sequences would require molecular cloning or the use of next generation sequencing methods.

However, the 18S and other nuclear ribosomal RNA genes, respectively the corresponding ribosomal RNAs, are extremely useful targets for in situ hybridization assays to label parasites in the host tissue. That opens new perspectives for pathology research by targeting certain parasite species/lineages during avian haemoproteosis, which can markedly damage various bird organs but remains insufficiently investigated [44, 45]. One of the main advantages compared to other target sequences is that each cell contains numerous ribosomes, resulting in a higher sensitivity. Moreover, ribosomal RNAs are considerably more stable than other RNAs, which is particularly important when analyzing pathological samples that were not prepared from fresh material [46]. Genus-specific oligonucleotide probes for in situ hybridization have been established to target and differentiate between avian Plasmodium, Haemoproteus, and Leucocytozoon parasites in bird tissue [47, 48]. A Plasmodium-specific probe was successfully used to detect blood and tissue stages in paraffin-embedded organs of captive penguins and wild passeriform birds, showing that tissue stages of Plasmodium spp. can cause mortality in both groups [47, 49]. Chromogenic in situ hybridization assays have also been performed to characterise tissue meronts in accipitriform raptors [50], strigiform raptors [51], thrushes [52], and other songbirds [11]. The latter studies provided valuable information regarding the development of exo-erythrocytic parasite stages in host tissue. For example, the combination of histological methods and in situ hybridization led to the first report of megalomeronts in Haemoproteus syrnii hSTAL2 and the characterization of a new mode of exo-erythrocytic development in Leucocytozoon sp. lSTAL5 [51]. The new Haemoproteus 18S sequences and oligonucleotide probes could be used to specifically target certain parasite lineages/species in host tissue. This particularly important and remains the only available approach when samples contain co-infections, which predominate in wildlife.

Conclusion

For the present study, the 18S rRNA genes of 19 Haemoproteus lineages belonging to the subgenus Parahaemoproteus, the most common blood parasites of Palearctic birds, were sequenced, thereby focusing on the H. belopolskyi and H. majoris species groups. The 18S sequences of eight additional Haemoproteus species, published previously by Harl et al. [8], were included in the analyses. To show the geographic and host distribution of the lineages investigated, DNA haplotype networks and pie charts were prepared based on the CytB data available in the MalAvi database. Like most Plasmodium parasites, the Haemoproteus lineages also featured two or more 18S variants, but the intraspecific distances between variants of the same lineages were generally lower. Moreover, the 18S sequences of all but one parasite lineage clustered into reciprocally monophyletic clades, which was not the case for most mammalian and avian Plasmodium species. The presence of chimeric features in the 18S variants of more than one third of the Haemoproteus lineages indicates that their ribosomal units evolve in a semi-concerted fashion rather than according to a strict model of birth-and-death evolution. Based on an alignment of the 18S sequences, oligonucleotide probes were designed in silico, which could be used to specifically target parasites species/lineages in the host tissue with chromogenic or fluorescent in situ hybridization methods. This would allow the detection of certain parasite lineages even in samples with co-infections, which are extremely common in songbirds.