Single‐Cell Transcriptomics and Development of Gametocyte‐Specific Molecular Markers for Avian Malaria Parasites

ABSTRACT

Avian malaria, caused by Plasmodium parasites, poses a significant threat to bird populations worldwide, particularly in vulnerable island ecosystems. Yet, progress in understanding avian malaria transmission dynamics has been hampered by the lack of molecular tools to quantify and sex the transmissible stages of the parasite: the male and female gametocytes. This challenge is compounded by the nucleated erythrocytes of avian hosts and the absence of an in vitro culture system, which have historically hindered the advancement of molecular approaches. Here, we develop and validate the first molecular markers to discriminate between asexual, male and female gametocytes in the widespread avian malaria parasite Plasmodium relictum. Using single‐cell RNA sequencing and orthology‐guided stage mapping, we identified conserved, stage‐specific transcripts and leveraged this information to develop molecular markers capable of quantifying and distinguishing male and female gametocytes in two parasite cyt‐b lineages, pSGS1 and pDELURB4. These markers outperformed microscopy in sensitivity, detected gametocytes earlier and for longer and revealed consistently higher male‐to‐female gametocyte ratios than estimated by blood smears. The high degree of conservation of these markers across Plasmodium species suggests broad applicability of these markers to other avian lineages.

1. Introduction

Although malaria parasites (Plasmodium sp.) are largely known for infecting humans, these protozoans can also be found infecting hundreds of other terrestrial vertebrate species, including non‐human primates, ungulates, rodents, bats, lizards and birds. Avian parasites are the most widespread, prevalent and diverse of all malaria parasites known to date (Rivero and Gandon 2018). Transmitted by mosquitoes, avian Plasmodium parasites pose a major threat to bird biodiversity and have caused population declines and extinctions, most notably among native Hawaiian species (Atkinson et al. 2014; Garamszegi 2011). The ongoing impacts of global warming are exacerbating the invasive potential of these parasites, expanding the geographic and ecological range of their vectors and increasing the vulnerability of endemic bird populations. For instance, Plasmodium infections are imperilling endemic bird species in the Galapagos Islands and New Zealand, underscoring the urgent need for effective tools to study and mitigate avian malaria (Miranda Paez et al. 2022).

The erythrocytic life cycle of avian Plasmodium parasites is similar to that of its human and rodent counterparts (Howick et al. 2019). Within red blood cells, the haploid parasite undergoes multiple rounds of asexual multiplication before irreversibly differentiating into sexual stages, called gametocytes. Male and female gametocytes are the only stages capable of being transmitted to a mosquito vector. As in human and rodent malaria parasites, once ingested by the mosquito vector, avian male gametocytes undergo mitosis to produce eight flagellated male gametes, while female gametocytes mature into single female gametes. Male and female gametes fuse in the mosquito midgut to produce a zygote, which continues the parasite's cycle inside the mosquito, eventually producing sporozoites that migrate to the salivary glands and can then be transmitted to a new bird (Valkiūnas 2005).

The timing and quantity of gametocyte production, along with the sex ratio of gametocytes, are critical epidemiological parameters as they directly influence the success of transmission to the mosquito vector. The ability of the asexual stage to modulate its conversion rate into gametocytes is widely interpreted as a form of adaptive phenotypic plasticity, allowing the parasite to make dynamic life‐history adjustments and optimise its transmission strategy in response to environmental cues. Research on human and rodent malaria parasites has indeed shown that Plasmodium can adjust both the number and sex ratio of gametocytes in response to a hostile host environment, such as when the host is anaemic (Cameron et al. 2013; Reece et al. 2005) or exposed to antimalarial drugs (Buckling et al. 1999; Schneider et al. 2018; Talman et al. 2004). Other work has shown that gametocyte sex ratios become more male‐biased in mixed‐genotype infections (Reece et al. 2008) or to ensure female fertilisation in the mosquito when gametocytaemias are low (Mitri et al. 2009).

Being able to quantify and sex gametocytes is therefore essential for elucidating both the epidemiology and evolutionary biology of avian Plasmodium. Blood smear microscopy is a standard method for detecting malaria parasites, but it has several inherent limitations. Its detection threshold is influenced by factors such as smear quality, the expertise of the examiner, counting method and the time dedicated to the examination. While examining a larger number of cells and extending the observation period can improve the detection of gametocytes, routine microscopy often underestimates gametocyte prevalence and provides inaccurate counts. Male and female gametocytes can exhibit subtle morphological differences, which are not always distinct from each other or from asexual stages transitioning to gametocytes (Valkiūnas 2005). Misidentification of parasite stages or host cell artefacts can therefore lead to inaccurate estimates of the gametocyte sex ratio. These issues underscore the need for molecular tools to quantify and sex gametocytes in avian Plasmodium.

Unlike the chromosomal or genetic sex determination systems found in many other organisms, Plasmodium relies on a distinct molecular mechanism to regulate sexual differentiation. In human and rodent Plasmodium species sexual differentiation begins when intraerythrocytic parasites commit to a sexual fate, a process regulated by the transcription factor AP2‐G (Kafsack et al. 2014; Sinha et al. 2014). Following this initial commitment, the sexually destined progenitor parasites make a secondary decision to differentiate into either male or female gametocytes (Gomes et al. 2022). Whilst there have been molecular tools available for quantifying and sexing gametocytes in human in rodent malaria for almost two decades (Babiker and Schneider 2008) the development of comparable molecular tools for avian Plasmodium has been hindered by several significant challenges. Firstly, the nucleated nature of avian erythrocytes results in an overwhelming abundance of host nucleic acids, which obscures and complicates the detection and quantification of parasite‐specific genetic material. Secondly, the asynchronous development of the parasite means there is no specific time point when only gametocytes are present in the blood. Finally, no in vitro culture systems exist for avian Plasmodium, making it difficult to isolate and study gametocyte‐specific stages under controlled conditions.

In this study, we present the first single‐cell (SC) transcriptome analysis of an avian malaria parasite and leverage it to develop and test molecular markers for quantifying and sexing gametocytes in Plasmodium relictum, one of the most widespread and prevalent species within the avian Plasmodium genus (Rivero and Gandon 2018). The single‐cell transcriptome provided a global view of gene expression, allowing us to apply a ‘guilt‐by‐association’ approach by comparing expression patterns with those documented in the Malaria Cell Atlas (Dogga et al. 2024; Howick et al. 2019). We selected two stage‐specific markers for female gametocytes, male gametocytes and asexual stages and validated these markers in an experiment using two different P. relictum cyt‐b lineages (pSGS1 and pDELURB4). These markers represent a significant step forward in our ability to study avian malaria at a molecular level, providing a foundation for future research into the epidemiology and evolutionary dynamics of host–parasite interactions under changing environmental conditions.

2. Materials and Methods

2.1. Single Cell Transcriptomics and Stage‐Specific Marker Identification

We infected five canaries ( Serinus canaria ) with Plasmodium relictum (lineage pSGS1) through intraperitoneal infection using 100 μL of infected blood from our infected canary stock. The birds were sampled at the peak of the acute infection (10 days post infection) and their parasitaemias and gametocytaemias? were quantified via the microscopic examination of blood smears (Table S1). For each bird, 100 μL of infected blood was subjected to enrichment with magnetic cell separation (MACS, Miltenyi Biotec) columns based on the paramagnetic properties of hemozoin. Each enriched fraction was then counted and pooled with a sample of blood prior to enrichment (to capture the non‐magnetic stages) at a ratio of 8:2. The five preparations for each bird were then pooled 1:1:1:1:1 and diluted to 300 cells per μL. The pooled blood sample was processed on a microfluidic chip and single cells were partitioned into individual wells for long‐read cDNA library creation (ArgenTag Single‐Cell RNA Library kit) using the manufacturers recommendation. The PCR cycles were increased to a total of 22 based on experience in prior Plasmodium species (PMID: 29580379). The quality of the amplified cDNA was monitored with the high‐sensitivity DNA chip on an Agilent 2100 Bioanalyser. The library was prepared with the LSK114 kit. The library was sequenced on one FLO‐PRO114M flow cell of a PromethION P2 Solo (Oxford Nanopore) to produce long‐read single‐cell transcriptomes and the high accuracy base caller was used concurrently with sequencing.

Sequencing data was converted from bam to fastq format using samtools (v1.19.2). Barcode demultiplexing and UMI tagging of reads was subsequently performed using taggy_demux (commit bb9ad27, https://github.com/argentagsw/taggy_demux) resulting in tagged fastq files. Demultiplexed reads were aligned to the concatenated pSGS1 (NCBI RefSeq GCF_900005765.1, Böhme et al. 2018) and Serinus canaria (NCBI RefSeq GCF_022539315.1) genomes using minimap2 ‐‐max‐intron‐len 50000. Gene annotation files were downloaded from Ensembl (July 2025). For each read, only the best scoring alignment was kept and multiple‐mapping was not allowed. Reads with low mapping quality (MAPQ < 20) and length < 200 bp were discarded. Read counts were assigned to genes using the FLAMES pipeline (gene_quantification step, v2.4.1) (Wang et al. 2025) and transcript isoform counts were summed to the gene count level. The single‐cell dataset was processed using Seurat (v5.3.1) (Hao et al. 2024). Host and parasite count matrices were split based on the presence of parasite genes and cells having more than 150 parasite genes while also having more than 250 and less than 15,000 parasite UMI were retained for subsequent analysis (Figure S1). After filtering, the P. relictum dataset retained 2027 single‐cells, detecting 5075 unique genes. Gene expression values were normalised by log‐transformation and subsequently centred and scaled for each gene.

Dimentionality reduction was performed within Seurat (v5.3.1) (Hao et al. 2024). We determined the number of principal components (PCs) based on the difference between variance of consecutive PCs and the UMAP (Uniform Manifold Approximation and Projection) was performed with the following parameters: dims = 1:13, n.neighbors = 30, min.dist = 0.7. We performed clustering (resolution = 0.4) and marker identification (using default parameters except from min.pct = 0.25) with the same package. Stage mapping was performed by transforming the data by orthology to P. falciparum using the OrthoMCL data (Chen et al. 2006) (September 2025) and selecting all orthologs available on PlasmoDB (Amos et al. 2022). The transformed data was then mapped onto a reference P. falciparum dataset (Dogga et al. 2024) using scmap (v1.32.0) (Kiselev et al. 2018) with a similarity threshold of 0.4. Downstream analyses were greatly facilitated by using PlasmoDB (Amos et al. 2022).

We selected and designed primers to amplify fragments of two female gametocyte, two male gametocyte and two asexual stage genes based on cluster marker metrics (adjusted p‐value, fold change and percentage of cells in the cluster expressing the marker) (Table 1). Primers were initially designed using the pSGS1 genome with Primer3 software (version 4.1.0) and whenever possible, primer pairs were placed on either side of an intron to ensure that the amplified fragments were derived from cDNA rather than contaminating genomic DNA. The P. relictum lineages pSGS1 and pDELURB4 are phylogenetically very close (Bensch et al. 2009). To verify cross‐lineage applicability, we confirmed that the primers designed for pSGS1 are fully conserved in pDELURB4 using a draft genome recently generated in our laboratory, which is currently under assembly (Bardil et al. unpublished).

TABLE 1.

Stage‐specific genes Plasmodium relictum (male, female, asexual), the molecular marker for conversion (AP2‐G) and the corresponding primers.

P. relictum (stage)	P. falciparum orthologue	Gene function	qPCR primers	DNA/cDNA size (bp)
M0509 (male) PRELSG_ 0509700	PF3D7_0414500	Male development protein MD5, putative	qPrM0509700dir 5′‐GAAGTTAGAAAAGCTGAACCCA‐3′ qPrM0509700rev 5′‐GTTATAGGGAAATGGGGCCA‐3′	209/114
M1409 (male) PRELSG_1409000	PF3D7_1311100	Meiosis‐specific nuclear structural protein 1, putative	qPrM1409000dir 5′‐AGAGAGAAAGCTTTGCAACA‐3′ qPrM1409000rev 5′‐TGCTTCTTTAACTTGGTTTTCTC‐3′	114/114
F0512 (female) PRELSG_0512300	PF3D7_0417000	Conserved protein, unknown function	qPrF0512300dir 5′‐TCATTTACCTTTCAAAATCCCT‐3′ qPrF0512300rev 5′‐AGGGCATGATGTTTCTCTACA‐3′	267/127
F1229 (female) PRELSG_1229400	PF3D7_0109600	LCCL domain‐containing protein, putative	qPrF1229400dir 5′‐ATAGAACATGACTGCGGTGT‐3′ qPrF1229400rev 5′‐CGGGAGAAGAGATATAGGCC‐3′	117/117
A0415 (asexual) PRELSG_0415200	PF3D7_020700	Merozoite surface protein 5, putative	qPrLS0415200dir 5′‐AGTGGACATGAACAAGAACT‐3′ qPrLS0415200rev 5′‐TCCTAATTTTCTTCCTTGAGT‐3′	291/114
A1229 (asexual) PRELSG_1229000	PF3D7_1452000	Rhoptry neck protein 2, putative	qPrLS1229000dir 5′‐CCCACTAAACAAACCGCATT‐3′ qPrLS1229000rev 5′‐AACTGCACCAAACCATAACG‐3′	114/114
AP2‐G PRELSG_1438200.1	PF3D7_1462800	AP2 domain transcription factor, putative	qPr_AP2Gdir2 5′‐TGGGAAAACGGGAGAAGATT‐3′ qPr_AP2Gdir2 5′‐TGCAATTGTTCTTTTGGAGC‐3′	119/119

Note: Primers were designed based on the pSGS1 genome and confirmed to be fully conserved in pDELURB4.

2.2. Experimental Validation of Stage‐Specific Markers

To validate the stage‐specific markers, we designed and conducted an experiment aimed at testing their specificity and accuracy using bulk RNA transcriptomics. Birds were infected by intraperitoneal injection of 100 μL blood infected with either pSGS1 or pDELURB4 (3 birds each). The blood of these experimental birds was sampled on days 0, 3, 5, 7, 10, 12, 14, 17, 19, 21, 24 and 31 post infection. At each sampling event, we carried out a blood smear and collected a further 20 μL of blood for molecular analyses. Of these, 10 μL of blood was mixed with 100 μL of PBS and stored at −20°C while the other 10 μL was mixed with 500 μL of Trizol Reagent (Invitrogen) and stored at −80°C for subsequent DNA and RNA extraction, respectively. Parasitaemia and gametocytaemia were quantified by microscopic examination of Giemsa‐stained blood smears, by counting 10,000 erythrocytes per sample (Table S1). DNA was used for molecular quantification of the total parasite load, while RNA was used to quantify the different stages. DNA and RNA were extracted with the Blood and Tissue kit (Qiagen) or with Trizol (Invitrogen), respectively, following the manufacturer's instructions. DNA was eluted in 80 μL of Nuclease‐free water (Qiagen) and the RNA pellet was resuspended in 40 μL of DEPC‐treated water (Invitrogen). Both of DNA and RNA concentrations were measured using a Spark spectrophotometer (Tecan).

Reverse transcription was conducted on the RNA samples using SuperScript II RT (Invitrogen) following the manufacturer's instructions with a few minor modifications. To this end, 100 ng of RNA was mixed with 0.5 pg. of oligo(dT) (Invitrogen), 1 μL of random decamers (Invitrogen), 1 μL of 10 mM each dNTP mix and completed to 12 μL with nuclease‐free water (Qiagen). The mixture was heated at 65°C for 5 min, then quickly chilled on ice. Next, 4 μL of 5X First‐Strand Buffer, 2 μL of 0.1 M DTT and 40 U of RNase OUT (Invitrogen) were added to each well. The plate was incubated at 42°C for 2 min, followed by 25°C for 2 min. Finally, 200 U of SuperScript II RT was gently added to each well and the plate was incubated first at 25°C for 10 min, then at 42°C for 50 min. The reaction was inactivated by heating at 70°C for 15 min.

For each bird, stage‐specific quantification was performed using qPCR on a LightCycler 480 (ROCHE) in a single run. Reactions were conducted in a final volume of 6 μL, containing 1 μL of cDNA, 0.6 μM of the corresponding primer pair (Table S2) and 3 μL of SensiFAST SYBR No‐ROX kit mix (OZYME). Amplifications were conducted in triplicate, using the following cycle: 8 min at 95°C, followed by 45 cycles (4 s at 95°C, 20 s at 62°C, 20 s at 72°C) and ending with the melting curve program (70°C–98°C). When the primer pairs were designed on either side of an intron, we checked the specificity of the amplicon by monitoring the melting curve of the amplified fragment. Calibration curves were established using a synthetic plasmid (Eurofins) containing a single copy of each target gene. For the quantification of male and asexual stages, the plasmid contained one copy each of the two gene fragments of interest (Table 1) and one GAPDH copy (housekeeping gene). For the quantification of female stages, due to the complexity of the gene fragment sequences it was not possible to synthesise a single plasmid. Consequently, two plasmids each containing a single copy of one of the markers plus a copy of GAPDH were synthesised.

Molecular quantification of the total parasitaemia was carried out using 2 qPCRs in a single run. The first one amplified a fragment of the Plasmodium cyt b gene with the primers L9 (5′‐AAACAATTCCTAACAAAACAGC‐3′) and NewR (5′‐ACATCCAATCCATAATAAAGCA‐3′) from Knowles et al. 2009 and the second one amplified the bird's 18 s rDNA gene with the primers 18sAv7 (5′‐GAAACTCGCAATGGCTCATTAAATC‐3′) and 18sAv8 (5′‐TATTAGCTCTAGAATTACCACAGTTATCCA‐3′) (Cellier‐Holzem et al. 2010). Each qPCR reaction was performed in triplicate for each sample, in a final volume of 6 μL containing 5 ng of DNA, 0.6 μM of each primer, 3 μL of SensiFAST SYBR No‐ROX kit mix (Ozyme) on a LightCycler 480 (Roche). The amplification cycle followed the same conditions as previously described.

Expression of the AP2‐G transcription factor, which regulates differentiation from asexual to sexual stages, was quantified by qPCR. qPCR primers (Table 1) were designed based on the homology of the orthologous P. falciparum gene (PF3D7_1222600, PlasmoDB). Reactions were performed in triplicate for each sample in a final volume of 6 μL, consisting of 1 μL of cDNA, 0.6 μM of each primer and 3 μL of SensiFAST SYBR No‐ROX kit mix (Ozyme), using a LightCycler 480 (Roche).

2.3. Statistical Analyses

To analyse the experimental validation of stage‐specific markers we used the R statistical package (v4.1.1). To account for repeated measures within each bird, we constructed mixed models with bird identity as a random effect (all models are summarised in Table S3). Model selection followed a stepwise approach: starting with a maximal model that included all fixed effects and their higher‐order interactions, we sequentially removed non‐significant terms to obtain a minimal model. The significance of the explanatory variables was assessed using likelihood ratio tests (LRT) which follow an approximate chi‐square distribution (Bolker 2008), with a significance threshold of p = 0.05. Proportion data (sex ratio) was analysed using a binomial distribution. Least square means were estimated using the emmeans function. When appropriate, we calculated the marginal R ² (the proportion of the variance explained by the fixed effects) following the method of Nakagawa and Schielzeth (2013).

2.4. Ethical Statement

Bird manipulations were carried out in strict accordance with the ‘National Charter on the Ethics of Animal Experimentation’ of the French Government. Experiments were approved by the Ethical Committee for Animal Experimentation established by the authors' institution (CNRS) under the auspices of the French Ministry of Education and Research (permit number CEEA‐LR‐221 1051).

3. Results

3.1. Single Cell Transcriptomics and Stage‐Specific Marker Identification

Blood samples enriched in infected erythrocytes originating from five infected canaries were pooled, partitioned by single cells and sequenced to produce long‐read single‐cell transcriptomes. After separating host and parasite transcripts in each cell, 2027 P. relictum single‐cell transcriptomes were obtained. Clustering analysis grouped these parasite transcriptomes into six distinct clusters, representing different putative life stages (Table S4, Figure S1). To formally assign developmental stages to the clusters, we transformed the data by orthology to P. falciparum and performed cell mapping onto a reference P. falciparum dataset (Dogga et al. 2024) (Figure 1A). The mapping allowed us to discriminate between sexual and asexual stages, with male gametocytes corresponding to cluster 4 and female gametocytes corresponding to cluster 5, while asexual stages corresponded to clusters 0, 1, 2 and 3. Concordantly, clusters 4 and 5, respectively, expressed orthologs of markers corresponding to established female (P47) and male (HAP2) gametocytes in rodent and human malaria parasites (Howick et al. 2019), whereas cluster 3 was characterised by the expression of the MSP1 ortholog, indicative of maturing asexual stages (Figure 1B). Despite the phylogenetic divergence between avian and mammal malaria parasites (Böhme et al. 2018), the stage mapping produced a clear distinction between sexual and asexual stages (Figure 1C). Only a minor proportion of cells remained unassigned, with few ambiguous mappings restricted to the schizont‐ring transition stages, thereby allowing the assignment of each P. relictum cluster to the developmental stages commonly observed in blood smears (Figure 1D).

FIGURE 1.

(A) UMAP of P. relictum transcriptomes wih their native clustering (top) or their stage mapped by orthology to a P. falciparum reference (Dogga et al. 2024) (bottom). P. relictum cells that have met the mapping similarity threshold of 0.4 are coloured according to the P. falciparum life stage in the bottom panel (unassigned cells are presented in grey). (B) Proportion of cells in each P. relictum cluster, annotated using the P. falciparum reference dataset. Cell proportions for main annotations are displayed on the left of each P. relictum cluster, while the right axis represents cell counts. Developing gametocyte stages are marked with a star and represent in total ~4% of P. relictum cells. (C) Microscopy images of P. relictum life stages corresponding to the cell clusters that were annotated. (D) Expression of P. relictum clusters, annotated using the ortholog genes of established mammalian Plasmodium markers for asexual stages (MSP1), male gametocytes (HAP2) and female gametocytes (P47). Gene expression values were normalised by log transformation and subsequently centred and scaled for each gene.

Together, these data provide reference transcriptomes for asexual, male and female gametocytes of P. relictum . We further identified stage‐specific marker genes for each cluster and selected two robust candidate biomarkers of asexual, female and male parasites (Figure 2). The orthologs of the selected gametocyte markers were also markers of the same stage in P. berghei (Russell et al. 2023) and P. falciparum (Gomes et al. 2022).

FIGURE 2.

(A) Intensity of expression and proportion of cells expressing P. relictum stage‐specific markers across the annotated P. relictum life stages. (B) Expression profiles of the P. relictum stage‐specific markers across 2027 P. relictum cells. Gene expression values were normalised by log‐transformation and subsequently centred and scaled for each gene.

3.2. Experimental Validation of Stage‐Specific Markers

We assessed the performance of male and female gametocyte markers in two different ways. First, by regressing the number of male and female gametocytes as estimated by the markers against the number of male and female gametocytes observed in the blood smears by microscopy (Figure 3A). Statistical analyses revealed a very significant correlation between blood counts and the number of marker copies, although the estimated slope of the relationship was steeper for pDELURB4 than for pSGS1 (Table S3, model #1, interaction blood counts: Plasmodium lineage χ² ₁ = 29.91, p < 0.0001, estimated marginal means and standard errors: pDELURB4 = 0.811 ± 0.049, pSGS1: 0.563 ± 0.049, marginal R ² = 0.61). This difference between the lineages was independent of the type of marker (model #1, 3‐way interaction χ² ₃ = 4.10, p = 0.2506). Using the newly developed markers, we were able to detect gametocytes in 32% of the pSGS1 samples and 21% of the pDELURB4 samples that were negative for gametocytes through microscopic examination of the blood. Second, by regressing the number of male and female gametocytes as estimated by the markers against the number of copies of the AP2‐G gene, a conserved gene responsible for the sexual commitment of Plasmodium parasites in blood (Figure 3B). Statistical analyses revealed a very significant correlation between AP2‐G counts and the number of marker copies (model #2, AP2‐G effect χ² ₁ = 158.01, p < 0.0001, marginal R ² = 0.60), which was independent of the lineage (model #2, interaction AP2‐G: lineage, χ² ₃ = 3.519, p = 0.318). The markers were less effective at detecting gametocytes than AP2‐G. Specifically 31% of pSGS1 and 30% of pDELURB4 samples that tested positive for gametocytes based on AP2‐G expression were not detected as gametocyte‐positive using the new markers. Male and female molecular markers outperformed the microscopic estimation of male and female gametocytes from blood smears (Figure 4). These markers detected gametocytes as early as, or earlier than, blood counts and remained detectable for a longer duration. Two of the markers, the male marker M1409 and the female marker F0512, outperformed their counterparts in both pSGS1 and pDELURB4 by detecting gametocytes for a longer duration following the peak of infection.

FIGURE 3.

Regression of marker copy number (log‐transformed) on blood smear gametocyte counts (A) and AP2‐G copy number (B). Male markers (M1409 and M0509) in dark blue, female markers (F0512, F1229) in light blue.

FIGURE 4.

Gametocyte detection window for the different gametocyte quantification techniques: Blood smear counts (males: blood_m, females: blood_f), sex‐specific molecular markers (males: M1409, M0509, females: F0512, F1229) and gametocyte conversion molecular markers (ap2g). The figure shows the results in birds infected with pSGS1 (S1–S3) and pDELURB4 (D1–D3).

The temporal dynamics of asexual, male and gametocyte stages in each bird is shown in Figure 5. With the exception of one bird infected with DELURB4 (Figure 5F), which did not exhibit any particular dynamic pattern, all other birds showed male and female gametocyte dynamics that mirrored those of the asexual stages (Figure 5A–E). In pSGS1 the first gametocytes were detected between days 5–7 post infection and peaked on days 10–12, whereas in pDELURB4 gametocytes appeared and peaked later (days 10–14 and days 14–19, respectively).

FIGURE 5.

Temporal dynamics of marker copy number (log‐transformed) across birds infected with either pSGS1 (A–C) or pDELURB4 (D‐F). Asexual markers (A0415, A1229) in yellow, male markers (M1409 and M0509) in dark blue, female markers (F0512, F1229) in light blue.

Overall, molecular markers estimate a higher proportion of male gametocytes compared to blood smears, suggesting a difference in detection efficiency of male and female gametocytes between the methods (Figure 6). Consequently, the sex ratio (% males) estimated from molecular markers was significantly higher than that estimated from blood smears (lsmeans = 0.53 and 0.33, respectively, model #3, χ² ₁ = 5.18, p = 0.022, no difference between the strains χ² ₁ = 0.002, p = 0.9619).

FIGURE 6.

Male and female gametocyte abundance (log‐transformed) estimated using molecular markers (blue dots, average of two markers) or blood smear counts (red dots). Data are shown for birds infected with pSGS1 (A) or pDELURB4 (B). The diagonal line (intercept = 0, slope = 1) represents a balanced sex ratio.

4. Discussion

Molecular markers to quantify and sex gametocytes have long existed for rodent and human malaria parasites (Babiker et al. 2008), but until now, equivalent tools were lacking for avian systems. In this study, we report the first development and validation of molecular markers for quantifying and sexing gametocytes of Plasmodium relictum, a key model in avian malaria research. Our approach, combining single‐cell transcriptomics with orthology‐guided stage mapping, allowed us to overcome major technical challenges posed by the nucleated red blood cells of avian hosts and the absence of an in vitro culture system, both of which have traditionally hindered the development of molecular markers for key parasite stages.

Clustering of the single‐cell transcriptomes revealed distinct transcriptional profiles corresponding to asexual stages as well as male and female gametocytes. These assignments were robustly supported by orthology to known stage‐specific markers in P. falciparum (Dogga et al. 2024; Howick et al. 2019). The identification of reliable and conserved molecular markers for each stage enabled us to design a set of qPCR assays capable of separately quantifying male and female gametocytes. Two of the markers in particular, the male marker M1409 and the female marker F0512, outperformed the others in terms of sensitivity in the two lineages tested: pSGS1 and pDELURB4. These markers consistently detected gametocytes earlier and remained detectable for a longer duration following the peak of infection, making them especially valuable for tracking transmission‐relevant stages over time. pSGS1 and pDELURB4 are phylogenetically close and represent but a small fraction of the over 600 cyt‐b lineages described thus far for avian Plasmodium parasites (Bensch et al. 2009). Further validation is needed to determine whether these markers reliably capture gametocyte dynamics in lineages with greater genetic divergence. However, given the high degree of conservation of stage‐specific gene expression patterns across Plasmodium species (Howick et al. 2019), we suggest that these markers are likely to be broadly applicable across other avian Plasmodium lineages.

Our experimental validation confirms that these molecular markers significantly outperform traditional microscopic examination in terms of sensitivity and temporal resolution. Importantly, the markers were able to detect gametocytes in over 30% of the pSGS1 and 20% of the pDELURB4 samples scored as negative by microscopy, particularly in the early and late phases of infection. These results are consistent with previous findings in human malaria that suggest microscopy underestimates gametocyte prevalence (Babiker et al. 2008). The molecular estimates also showed a strong correlation with AP2‐G expression, the master regulator of sexual commitment in P. falciparum and P. berghei (Kafsack et al. 2014; Sinha et al. 2014). The fact that a proportion of AP2‐G‐positive samples lacked detectable gametocyte markers suggests the potential presence of early committed stages not yet expressing mature gametocyte markers. Future work should explore additional markers that span the entire differentiation process, from early commitment to fully mature gametocytes.

Interestingly, we observed differences in the temporal dynamics of gametocyte development between P. relictum lineages, with pSGS1 producing gametocytes earlier than pDELURB4. While these differences could reflect intrinsic differences in the developmental pace of the lineages, they might also stem from variation in parasite load or host response. Canaries, originally from the Macaronesian Islands (Azores, Madeira and the Canary Islands), have been bred in captivity in Europe since the 17th century. Today, birds housed in outdoor aviaries frequently become natural hosts for Plasmodium relictum, including these two lineages. It remains to be determined whether these developmental differences also manifest in wild birds with a longer co‐evolutionary history with the parasite. More systematic investigations will be needed to disentangle these effects and assess their implications for transmission.

In Plasmodium, sex ratios are generally expected to be female‐biased, as each male gametocyte produces up to eight gametes in the mosquito vector, whereas each female gametocyte yields only a single gamete (Sinden 2015). This reproductive asymmetry favours an overproduction of female gametocytes to maximise fertilisation success in the mosquito. However, the degree of female bias observed in peripheral blood varies and is influenced by multiple factors, including the parasite's inbreeding (Reece et al. 2008; Schneider et al. 2019; West et al. 2002) and the total density of circulating gametocytes (West et al. 2002).

A particularly notable result is the discrepancy between gametocyte sex ratios estimated by microscopy and those obtained with molecular markers. Molecular data consistently indicated a higher proportion of male gametocytes (average sex ratio 0.53) compared to microscopy (0.33). This may reflect a differential detectability of male and female gametocytes under the microscope, or a bias in staining or morphological identification. Alternatively, it may point to differences in the expression levels or stability of the male and female markers used. Variability in transcript abundance between male‐ and female‐specific genes could lead to an overestimation of male gametocytes if the selected marker is more abundantly expressed or has a longer transcript half‐life. Nevertheless, transcript‐based methods remain a valid and powerful tool for comparing sex ratios of a given lineage between individual hosts and for monitoring changes within individuals over time. Importantly, the sex ratio estimated by our molecular markers (0.53) is consistent with theoretical expectations: classical sex allocation theory predicts near‐equal investment in male and female gametocytes when parasite populations exhibit high clonal diversity and potential for outcrossing (West et al. 2002), as may be the case here given that our parasite isolate was recently obtained from the field. Thus, the estimated sex ratio falls well within the range predicted by theory and does not necessarily indicate an experimental artefact.

Because P. relictum lacks validated gametocyte‐stage annotations, we relied on orthology‐based mapping to P. falciparum, whose gametocyte markers are well characterised. This approach allowed us to focus on genes likely conserved across Plasmodium species, but it necessarily biases the analysis towards shared loci and may miss P. relictum ‐specific gametocyte markers. Identifying truly species‐specific markers remains a goal for future work and would offer important insights into the evolution and regulation of sex determination in Plasmodium parasites. In addition, the calibration plasmids for male and female markers were synthesised differently: the male plasmid contained both markers on a single construct, whereas female markers were split across two separate plasmids. While this could introduce minor variability in quantification efficiency, all PCR reactions were carefully calibrated using standard curves and GAPDH normalisation, making it unlikely that this affected the main conclusions regarding gametocyte sex ratios. Our results open new opportunities for avian malaria research, as the ability to accurately quantify and sex P. relictum gametocytes provides a powerful tool for assessing transmission potential in wild bird populations. By enabling more precise investigations of host–parasite interactions and disease dynamics, these markers also lay the groundwork for studying how environmental factors, including climate change and habitat disruption, shape parasite transmission strategies. Together, these tools offer a framework for integrating molecular, ecological and evolutionary perspectives in the study of avian malaria.

Funding

This work was supported by Agence Nationale de la Recherche, ANR‐23‐CE02‐0013, ANR‐19‐CE15‐0007, ANR‐17‐CE35‐0012.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: Total parasitaemias (% of red blood cells infected with any stage) and gametocytaemias (% of red blood cells infected with gametocytes) of the birds used in the single cell transcriptomic analysis.

Table S2: Reverse transcription reagents used.

Table S3: Description of the statistical models used to analyse the data. ‘Maximal model’ represents the complete set of explanatory variables (and their interactions) included in the model. ‘Minimal model’ represents the model containing only the significant variables and their interactions. Round brackets indicate that the variable was fitted as a random factor. Square brackets indicate the error structure used (n: normal errors, b: binomial errors). mk = number of copies of marker, bs = number of parasites in blood smear, ap2g = number of AP2‐G copies, id = marker identity, lin = lineage (pSGS1, pDELURB4), bird = bird identity.

Table S4: (.xlsx file) List of differentially expressed genes identified using the Wilcoxon Rank Sum test across P. relictum clusters and annotated based on the P. falciparum reference dataset. For each gene, the p‐value was adjusted with the Bonferroni correction and pct.1 and pct.2 represents cell proportions expressing the gene in a given cluster compared to all other clusters, respectively.

Figure S1: UMI (Unique Molecular Identifiers, left panel) and gene counts (right panel) per cell across P. relictum clusters. Host‐derived UMI counts are shown in grey in the left panel, indicating that P. relictum transcripts dominate over host transcripts in infected cells.

Berthomieu, A. , Rizos I., Bagnoli J. W., Gandon S., Talman A. M., and Rivero A.. 2026. “Single‐Cell Transcriptomics and Development of Gametocyte‐Specific Molecular Markers for Avian Malaria Parasites.” Molecular Ecology Resources 26, no. 5: e70157. 10.1111/1755-0998.70157.

Data Availability Statement

The single‐cell transcriptomes generated in this study have been deposited in the European Nucleotide Archive (ENA) under the accession number PRJEB107802 (https://www.ebi.ac.uk/ena/browser/view/PRJEB107802). Scripts, count matrices and metadata necessary to run all the analyses included in this work are publicly available on the GitHub page: https://github.com/Parasitedevbio/Avian_PlasmoGam.