Decoding bee cleptoparasitism through comparative transcriptomics of Coelioxoides waltheriae and its host Tetrapedia diversipes

Abstract

Cleptoparasitism, also known as brood parasitism, is a widespread strategy among bee species in which the parasite lays eggs into the nests of the host species. Even though this behavior has significant ecological implications for the dynamics of several species, little is known about the molecular pathways associated with cleptoparasitism. To shed some light on this issue, we used gene expression data to perform a comparative analysis between two solitary neotropical bees: Coelioxoides waltheriae, an obligate parasite, and their specific host Tetrapedia diversipes. We found that ortholog genes involved in signal transduction, sensory perception, learning, and memory formation were differentially expressed between the cleptoparasite and the host. We hypothesize that these genes and their associated molecular pathways are engaged in cleptoparasitism-related processes and, hence, are appealing subjects for further investigation into functional and evolutionary aspects of cleptoparasitism in bees.

Introduction

Parasitism is an interaction between different species, in which one of them, the parasite, benefits at the expense of another, the host^1,2. Parasitic species may play an important role in the dynamics of natural populations of host species, for instance, they may affect the susceptibility of their hosts to predation, modify their reproductive patterns, and influence the abundance of endemic and introduced species^3,4. Cleptoparasitism, alternatively known as brood parasitism, refers to a parasitic behavior where the parasite lays eggs into the nests of the host species. Then during its larval stage, the parasite offspring thrives by consuming the food resources that have been provided by the adult host, ultimately leading to the demise of the host larva or egg, as it is either killed or eaten by the parasite larva. Finally, an adult parasite emerges from the host nest⁵.

Cleptoparasitism is widespread in bees (Hymenoptera: Antophila). It is estimated that approximately 13% of the 20,500 bee species in the world⁶ are obligate cleptoparasites⁷. Currently, it has been inferred that this behavior has arisen independently 18 times in four out of nine bee families: three distinct times in Apidae^8–10; probably five times in Megachilidae^11–13; at least once in Colletidae¹⁴; and possibly nine in Halictidae^5,15. In spite of these multiple and independent origins, most cleptoparasitic bees show important convergent adaptations such as the reduction or complete loss of pollen-collecting (e.g., pollen-manipulating brushes and scopa) and nest-building structures (e.g., basitibial and pygidial plates for ground-nesting species)⁵. Compared to non-parasitic species, many cleptoparasites also have a heavily sclerotized cuticle in addition to spines, ridges, crests, carinae or lamellae protecting them from the jaws or sting of host females^5,7. Moreover, convergent anatomical and physiological changes in the reproductive system of some cleptoparasitic species have been described, such as a greater number of mature oocytes in the ovaries or more ovarioles per ovary¹⁶. These adaptations allow parasitic females to lay several eggs in a short period of time^5,7.

In addition to their evolutionary relevance, cleptoparasitic species also have an ecological value. Cleptoparasitic bees are considered by far the largest protagonist of solitary bee brood mortality among all natural enemies¹⁷. In this context, Sheffield et al.¹⁸ suggest that cleptoparasites, especially generalist ones, perform a stabilizing role in bee communities by attacking dominant and abundant host taxa, which may reduce competition among non-parasitic bee species. These authors also argue that cleptoparasites can serve as indicator taxa for assessing the status of the entire bee community, as their diversity and abundance are closely tied to their host species¹⁸. However, there is still a deep lack of knowledge regarding the general biology, behavior⁷, and particularly the molecular mechanisms underlying the evolution and maintenance of cleptoparasitism. This knowledge gap might be driven by the rarity to find cleptoparasitic bees in nature^7,19.

The abundance of cleptoparasitic bees often exhibits a positive correlation with the density of available host nests^18,20, implying that hosts displaying high levels of gregarious behavior are more susceptible to parasitic attacks. In this context, the species Coelioxoides waltheriae Ducke, 1908 and Tetrapedia diversipes Klug, 1810 comprise a compelling parasite-host species pair to investigate cleptoparasitism. T. diversipes, the host species, has a gregarious behavior, building its nests in naturally pre-existing cavities such as holes in wood²¹, and trap-nests. Indeed, T. diversipes easy aggregatory behavior in trap-nests allowed the description of the cleptoparasitic behavior of C. waltheriae²¹, the first recorded for the Coelioxoides genus. Coelioxoides and Tetrapedia are endemic to the Neotropical region²². Both genera have been formerly grouped within the same tribe, Tetrapediini (Apidae), as they share a number of similar morphological traits²³. However, recent molecular-based phylogenies have placed Coelioxoides within a large cleptoparasitic clade (Nomadinae: Apidae) not sister to Tetrapedia^8–10,24,25. C. waltheriae is considered to be the main natural enemy of T. diversipes²⁶ even though it has been also reported to parasitize nests of other Tetrapedia species^27,28.

Recent studies have successfully employed comparative transcriptomic analysis to unveil molecular features related to complex behaviors. For instance, transcriptomics studies in bees have helped to better understand the molecular pathways related to the honeybee (waggle) dance²⁹, Varroa sensitive hygiene behaviour³⁰, olfactory³¹ and visual³² learning, diapause^33,34, and evolution of task division^35–37. In this context, herein we use gene expression data to perform a comparative analysis between the correspondent reproductive life stages of C. waltheriae and its host, T. diversipes. We aimed at identifying diverging gene expression patterns between the solitary pollen-collecting host and the cleptoparasitic bee species. We explored the function of enriched diverging pathways and discussed whether these could be related at some level to their ecological interactions. Instead of assuming any of the differences here observed are causative, we focused on relating our findings to molecular processes for which ecological and/or behavioural relevance has been previously described in the literature to shed some light on the molecular mechanisms potentially related to cleptoparasitic behavior.

Results

Transcriptome assembly and annotation

We sampled reproductive females of C. waltheriae and T. diversipes from the same trap-nest location and at correspondent life stages. In total, we used five sample replicates per species, consisting of a pool of whole-body extractions from individuals (see “Materials and methods”).

After transcriptome assemblies and data treatment (see “Materials and methods”) we obtained a set of 18,208 transcripts for C. waltheriae and 11,998 for T. diversipes. The main quality parameters associated with each assembly are summarized in Table 1.

Table 1.

Main quality parameters of C. waltheriae and T. diversipes assemblies.

Metric	C. waltheriae	T. diversipes
Total transcripts assembled	87,358	17,866a
		40,796b
Number of clustered SuperTranscriptsc	18,208	11,998
SuperTranscripts ≥ 1000 nt	10,195	8051
p fragments mappedd	0.93176	0.86625
p bases uncoverede	0.0328	0.05047
Transrate optimal scoref	0.5904	0.50711
Complete BUSCOs (%)	93.4	88.5
Missing BUSCOs (%)	4.1	8.2

^aNumber of transcripts obtained after the reference-based assembly.

^bNumber of transcripts obtained after the reference guided de novo assembly.

^cAfter removing contaminating sequences.

^dProportion of reads pairs that mapped successfully.

^eProportion of bases that were not covered by any read.

^fThis discrete score represents the accuracy and completeness of the assembly based on the assessment of alignments between contigs and reads. The minimum score is 0, while the maximum is 1.0. For a description of the components used in the calculation of this metric, see Smith-Unna et al.³⁸.

A total of 10,042 (51.2%) and 7447 (61.9%) transcripts in C. waltheriae and T. diversipes, respectively, blasted (e-value < 1e−5) against protein-coding genes in the UniRef90 database. Moreover, 6253 (31.8%) and 2919 (24.2%) transcripts were estimated to be potential non-coding sequences in C. waltheriae and T. diversipes. It is worth mentioning that 1397 (7.1%) transcripts were identified as potential contaminants in C. waltheriae (Supplementary Material 1). Most were attributed to fungi of the family Tubulinosematidae (Microsporidia): Tubulinosema ratisbonensis (1113 transcripts), Anncaliia algerae (15 transcripts), and Tubulinosematidae itself (10 transcripts). In addition, 95 (0.48%) of C. waltheriae transcripts were identified as very similar to plant protein sequences. Overall, the Fabaceae family was the most representative plants among contaminants (52 transcripts), followed by Asteraceae (11 transcripts) and Solanaceae (6 transcripts). Contaminant transcripts of T. diversipes have already been described previously³⁹. Herein, we identified only 31 (0.25%) transcripts as potential contaminants (Supplementary Material 1), far less than that observed in C. waltheriae, which was expected, since the assembly approaches for T. diversipes used a reference genome.

Differential orthologs expression

Using Orthofinder^40,41 we identified 4859 orthogroups shared between C. waltheriae and T. diversipes, of which 3011 were reported as single-copy orthologues. Using blastp^42,43 searches to identify sequences with the highest homology within common orthogroups we additionally selected 1848 one-to-one orthologs, totaling 4154 one-to-one ortholog pairs between the two species. Of these, around half (2096 or 50.45%) were annotated with at least one functional GO term. After correcting for sampling batch effects accounting for species, the number of samples and sampling year (Supplementary Material 2), the significantly differentially expressed orthologs between the species were obtained by overlapping the results from two strategies (1. TMM normalized counts⁴⁴ and edgeR 3.34.0⁴⁵; 2. Scale Based Normalization [SCBN] method⁴⁶ and NOISeq^47,48). We identified a total of 646 orthologs as differentially expressed (DE) between the two species (Supplementary Material 3), 335 of which were highly expressed in C. waltheriae, and 311 were highly expressed in T. diversipes. Among the annotated DE orthologs, two groups caught our attention prior to GO Enrichment Analysis due to their frequency and function. First, a group of eleven transcripts homologous to sequences of PiggyBac transposable element-derived protein genes (Pgbd4 = 10 transcripts and Pgbd2 = 1 transcript), most of which (7) were highly expressed in C. waltheriae (Fig. 1). Moreover, considering the entire set of transcripts assembled, we found that C. waltheriae has more transcripts (n = 68; 0.34% of the transcriptome) of Pgdb-like genes than T. diversipes (n = 24; 0.19%). Additionally, we found in T. diversipes two sequences identified as homologous to Major Royal Jelly Protein/yellow-like (MRJP/y) that were highly expressed (Fig. 1).

Figure 1.

Relative expression of orthologs differentially expressed between the cleptoparasite Coelioxoides waltheriae and their host Tetrapedia diversipes that were identified as homologs to PiggyBac Transposable Element-derived (1: PGBD2-like; 2: PGBD4-like) and Major Royal Jelly Protein (3) sequences. The respective orthogroups (OGs) are identified according to their IDs in Supplementary Material 3.

Through the Gene Ontology term enrichment analysis for all categories (BP, CC, and MF), we identified 54 GO terms enriched among the DE orthologs (Supplementary Material 4). The redundant terms of each GO category were summarized in a two-level hierarchical list for visualization (Fig. 2). Top enriched Cellular Component (CC) terms were related to ion channel complex, transporter complex, and plasma membrane protein complex. Consistently, top enriched Molecular Function (MF) terms were related to channel activity, transmembrane signaling receptor activity, and potassium ion transmembrane activity. The orthologs annotated with these CC and MF terms are shown in Table 2.

Figure 2.

Gene Ontology (GO) terms enriched among the Differentially Expressed (DE) Orthologs between Coelioxoides waltheriae and Tetrapedia diversipes. Each pie chart represents one of the three main GO categories: (A) Cellular Component, (B) Molecular Function, and (C) Biological Process. Legends next to each chart indicate the representative color, name, and proportion of DE that received an annotation related to the parental GO term. Child terms from the second hierarchical level are represented in the outermost part of the pie chart using shades of the parental representative term color. Parent–child GO term relationships are listed in Supplementary Material 4.

Table 2.

Differentially expressed (DE) orthologs between Coelioxoides waltheriae and Tetrapedia diversipes grouped according to their association to the top enriched GO terms for Cell Component and Molecular Function categories.

DEO IDs	Expression patterns	Homologs (Swiss-Prot organism)	GO hierarchical groups
Calcium channels
OG0000284	C. waltheriae ↓ T. diversipes ↑	CAC (Apis mellifera)	Cell component  GO1 ion channel complex  GO2 transporter complex  GO3 plasma membrane protein complex Molecular function  GO1 channel activity  GO7 passive transmembrane transporter activity  GO10 calcium ion binding
OG0000168	C. waltheriae ↓ T. diversipes ↑	Trp (Drosophila melanogaster)	Cell Component  GO1 ion channel complex  GO2 transporter complex  GO3 plasma membrane protein complex  GO4 intrinsic component of plasma membrane  GO5 integral component of plasma membrane Molecular Function  GO1 channel activity  GO7 passive transmembrane transporter activity
Sodium channels
OG0003129	C. waltheriae ↑ T. diversipes ↓	NaCP60E (Drosophila melanogaster)	Cell component  GO1 ion channel complex  GO2 transporter complex  GO3 plasma membrane protein complex  GO4 intrinsic component of plasma membrane  GO5 integral component of plasma membrane Molecular function  GO1 channel activity  GO7 passive transmembrane transporter activity
OG0002153	C. waltheriae ↑ T. diversipes ↓	unc80 (Drosophila melanogaster)	Cell component  GO1 ion channel complex  GO2 transporter complex Molecular Function  GO1 channel activity  GO7 passive transmembrane transporter activity
Potassium channels
OG0004503	C. waltheriae ↓ T. diversipes ↑	KCNQ1 (Felis catus, Xenopus laevis)	Cell component  GO1 ion channel complex  GO2 transporter complex  GO3 plasma membrane protein complex  GO4 intrinsic component of plasma membrane  GO5 integral component of plasma membrane Molecular function  GO1 channel activity  GO3 potassium ion transmembrane transporter activity  GO7 passive transmembrane transporter activity
OG0000260	C. waltheriae ↓ T. diversipes ↑	Sh (Drosophila melanogaster)	Cell component  GO1 ion channel complex  GO2 transporter complex  GO3 plasma membrane protein complex  GO4 intrinsic component of plasma membrane  GO5 integral component of plasma membrane Molecular function  GO1 channel activity  GO3 potassium ion transmembrane transporter activity  GO7 passive transmembrane transporter activity
OG0004153	C. waltheriae ↑ T. diversipes ↓	KCNT1 (Homo sapiens)	Molecular function  GO1 channel activity  GO3 potassium ion transmembrane transporter activity  GO7 passive transmembrane transporter activity
OG0002734	C. waltheriae ↑ T. diversipes ↓	KCNK9 (Homo sapiens, Rattus norvegicus)	Molecular function  GO1 channel activity  GO3 potassium ion transmembrane transporter activity  GO7 passive transmembrane transporter activity
Chloride channel
OG0000326	C. waltheriae ↓ T. diversipes ↑	Grd (Drosophila melanogaster)	Molecular function  GO1 channel activity  GO2 transmembrane signaling receptor activity  GO6 GABA receptor activity  GO7 passive transmembrane transporter activity  GO8 molecular transducer activity
Ligand-gated ion channels
OG0000029	C. waltheriae ↓ T. diversipes ↑	Grik2 (Mus musculus, Xenopus laevis)	Cell component  GO1 ion channel complex  GO2 transporter complex  GO3 plasma membrane protein complex  GO4 intrinsic component of plasma membrane  GO5 integral component of plasma membrane Molecular function  GO1 channel activity  GO2 transmembrane signaling receptor activity  GO3 potassium ion transmembrane transporter activity  GO5 postsynaptic neurotransmitter receptor activity  GO7 passive transmembrane transporter activity  GO8 molecular transducer activity
OG0000962	C. waltheriae ↓ T. diversipes ↑	nAChRβ1 (Drosophila melanogaster)	Cell component  GO1 ion channel complex  GO2 transporter complex  GO3 plasma membrane protein complex  GO4 intrinsic component of plasma membrane  GO5 integral component of plasma membrane Molecular function  GO1 channel activity  GO2 transmembrane signaling receptor activity  GO5 postsynaptic neurotransmitter receptor activity  GO7 passive transmembrane transporter activity  GO8 molecular transducer activity
G-protein coupled receptors
OG0002068	C. waltheriae ↓ T. diversipes ↑	GABBR1 (Homo sapiens, Mus musculus)	Cell component  GO3 plasma membrane protein complex  GO4 intrinsic component of plasma membrane  GO5 integral component of plasma membrane Molecular function  GO2 transmembrane signaling receptor activity  GO5 postsynaptic neurotransmitter receptor activity  GO6 GABA receptor activity  GO8 molecular transducer activity
OG0001556	C. waltheriae ↑ T. diversipes ↓	Gabbr2 (Mus musculus)	Cell component  GO3 plasma membrane protein complex  GO4 intrinsic component of plasma membrane  GO5 integral component of plasma membrane Molecular function  GO2 transmembrane signaling receptor activity  GO6 GABA receptor activity  GO8 molecular transducer activity
OG0000063	C. waltheriae ↑ T. diversipes ↓	Octβ2R (Drosophila melanogaster)	Molecular function  GO2 transmembrane signaling receptor activity  GO8 molecular transducer activity
OG0001943	C. waltheriae ↑ T. diversipes ↓	HTR2C (Mus musculus, Bos taurus)	Molecular function  GO2 transmembrane signaling receptor activity  GO8 molecular transducer activity
OG0000814	C. waltheriae ↓ T. diversipes ↑	CG31760 (Drosophila melanogaster)	Molecular function  GO2 transmembrane signaling receptor activity  GO8 molecular transducer activity
OG0001490	C. waltheriae ↓ T. diversipes ↑	Gr64f./Gr5a (Drosophila melanogaster)	Molecular function  GO2 transmembrane signaling receptor activity  GO8 molecular transducer activity
G protein
OG0004442	C. waltheriae ↓ T. diversipes ↑	G-sα60A (Anopheles gambiae)	Cell component  GO3 plasma membrane protein complex
Translocases
OG0001155	C. waltheriae ↑ T. diversipes ↓	Atpα (Drosophila melanogaster)	Cell component  GO2 transporter complex  GO3 plasma membrane protein complex  GO4 intrinsic component of plasma membrane  GO5 integral component of plasma membrane Molecular function  GO1 channel activity  GO3 potassium ion transmembrane transporter activity
OG0001472	C. waltheriae ↑ T. diversipes ↓	nrv2/nrv1 (Drosophila melanogaster)	Cell component  GO2 transporter complex  GO3 plasma membrane protein complex  GO4 intrinsic component of plasma membrane  GO5 integral component of plasma membrane
OG0000864	C. waltheriae ↓ T. diversipes ↑	UQCRFS1 (Lagothrix lagotricha, Colobus polykomos)	Molecular function  GO1 channel activity
DNA-binding protein
OG0003803	C. waltheriae ↑ T. diversipes ↓	Hr38 (Drosophila melanogaster)	Molecular function  GO2 transmembrane signaling receptor activity  GO8 molecular transducer activity  GO11 sequence-specific DNA binding  GO12 DNA binding
Low-density lipoprotein-receptor-related
OG0004128	C. waltheriae ↑ T. diversipes ↓	LRP6 (Homo sapiens, Mus musculus)	Cell component  GO3 plasma membrane protein complex Molecular function  GO2 transmembrane signaling receptor activity  GO8 molecular transducer activity
Cell adhesion molecules
OG0000727	C. waltheriae ↑ T. diversipes ↓	NRXN1 (Homo sapiens, Mus musculus)	Cell component  GO4 intrinsic component of plasma membrane  GO5 integral component of plasma membrane Molecular function  GO2 transmembrane signaling receptor activity  GO8 molecular transducer activity  GO10 calcium ion binding
OG0000165	C. waltheriae ↑ T. diversipes ↓	Stan (Drosophila melanogaster)	Molecular function  GO2 transmembrane signaling receptor activity  GO8 molecular transducer activity  GO10 calcium ion binding
Protein phosphatase
OG0000836	C. waltheriae ↓ T. diversipes ↑	Ptp69D (Drosophila melanogaster)	Molecular function  GO2 transmembrane signaling receptor activity  GO8 molecular transducer activity
Tyrosine-protein kinase
OG0002698	C. waltheriae ↓ T. diversipes ↑	Abl (Drosophila melanogaster)	Molecular function  GO2 transmembrane signaling receptor activity  GO8 molecular transducer activity

The orthologs are represented by the ID of their respective orthogroups. Expression patterns of orthologs comparatively between C. waltheriae and T. diversipes are indicated by arrows: ↑ (upregulated) and ↓ (downregulated). GO hierarchical groups refer to CirGo terms as shown in Fig. 2.

For the Biological Process (BP) category, the most significant terms in the enrichment analysis were related to camera-type eye morphogenesis, angiogenesis, and regulation of osteoblast differentiation. The main DE orthologs annotated within these terms were identified as homologs of Fibrillin-2-like, Low-density lipoprotein receptor-related protein 6, and serine/threonine-protein kinases. Moreover, the highest number of DE orthologs were associated with cell communication, signaling, and signal transduction BP terms (Supplementary Material 5).

Discussion

We employed comparative transcriptomics to investigate molecular distinctions during the reproductive stage of Coelioxoides waltheriae (a cleptoparasite) and its host species, Tetrapedia diversipes. The present study represents an initial effort to obtain some insights about the molecular mechanisms involved in bee cleptoparasitism by comparing orthologs expression data between a cleptoparasitic species and their host, both inhabiting the same location and at equivalent developmental stages. Considering the constraints of cross-species transcriptomic comparisons, our findings shed light on broad molecular variations between the cleptoparasite and its host, suggesting possible ecological implications.

We overlapped two statistical approaches to perform a cross-species comparison and to reduce the number of false positives in our analyses. The first of these methods was the SCBN, a recently proposed method for count normalization optimized to deal with cross-species comparisons⁴⁶. Secondly, we used a traditional differential expression workflow with the edgeR Bioconductor package⁴⁵. By overlapping the results from these two approaches we consistently retrieved 646 differentially expressed orthologs between the two species, with PiggyBac Transposable Element-derived like (Pgbd-like) genes being the most frequent (Fig. 1). Overall, we found many orthologs annotated as Pgbd-like, and several of them were found to be highly expressed in C. waltheriae. The PiggyBac transposon superfamily is widespread among eukaryotes. The first PiggyBac element was described from a cell culture of Trichoplusia ni (Lepidoptera: Noctuidae), the cabbage looper moth^49,50. Since then, several PiggyBac-like sequences have been described in a variety of organisms. Some of these transposable elements lost transposase activity and are called domesticated⁵¹. Molecular domestication is a process that occurs due to a transposition inactivating mutation, resulting in the loss of mobility of the transposable element⁵². These insertions may lead to the emergence of new cellular activities, either by altering the coding and/or regulatory regions in which these elements are inserted in the genome or by the evolution of the former TE genes into new genes^52,53. Thus, these findings lead us to hypothesize whether the domestication of PiggyBac-like transcripts and its regulatory consequences could be one of the molecular mechanisms involved in the evolution and adaptation of C. waltheriae, justifying their differential expression between the studied host-parasite species. Indeed, in a recent study, it was observed through comparative genomic analyses that retroviral or transposable elements have undergone a recent or ongoing spread in the genome of a Nomadinae cleptoparasite⁵⁴. This indicates a possible involvement of TEs in the evolution of cleptoparasitism. Future research should further explore this hypothesis, investigating the evolution of transposable and transposable-derived elements in host-parasites genomes and their expression pattern in other cleptoparasitic species of Apidae.

We also identified two MRJP/Yellow-like sequences as highly expressed in T. diversipes while none were upregulated in C. waltheriae. The MRJP/Yellow gene family encodes multifunctional yellow-like proteins identified in arthropods and in several bacteria^55,56. Nonetheless, the MRJP-like genes, as part of this family, seem to be restricted to Hymenoptera⁵⁶. Even though these genes have been associated with olfactory learning—particularly regarding mrjp1 expression in the mushroom body (Kenyon cells)^31,57,58—and functional studies have shown that MRJPs may have immunoregulatory and antibacterial effects⁵⁶, in bees, MRJPs are mostly associated with larvae feeding, development and with the regulation of phenotypic plasticity and age-polyethism in workers⁵⁹. In honeybees, MRJPs are known to be essential components of the larvae diet, with MRJP1–3 and 5 being the most abundant proteins of larval food⁵⁶. These proteins are synthesized by honeybee nurse workers in the specialized hypopharyngeal glands and offered to the immature offspring through a special food jelly (royal jelly)⁶⁰. In contrast to honeybees, bumblebees do not produce royal jelly. However, their hypopharyngeal glands express one MRJP ortholog⁶¹. The production of MRJP is not dependent on brood-feeding activity in bumblebees, and protein signals were identified in the abdominal parts of the digestive tract in queens and workers. This suggests that bumblebee MRJP does not have a nutritional function but is instead involved in food digestion and/or modification, consistent with the putatively ancient function of bee hypopharyngeal glands in food digestion⁶¹. In solitary bees the role of MRJP-like genes is currently poorly understood. One hypothesis for the high expression of MRJPs in T. diversipes could be the association of these proteins with larval food. As a solitary bee species, founder females of T. diversipes are also responsible for larvae food provisioning and an increased expression of transcripts related to MRJP-like genes could suggest that these proteins may be component of larval food. Yet, the role of MRJPs-like may not be necessarily nutritional/developmental, as they could alternatively play an antibacterial role and/or aid in the processing of bee products (e.g., formation of pollen-pellets and pollen-bread)^62–64.

Several Biological Process (BP) GO terms enriched among DE orthologs were found to be typical or exclusively related to vertebrate species, such as camera-type eye morphogenesis, angiogenesis, and osteoblast differentiation. It is important to note that molecular data banks, including UniRef, predominantly rely on scientific evidence derived from a limited number of model species⁶⁵. As a result, the nature and extent of GO annotations reflect the aspects investigated in these organisms being only extrapolated to other non-model species under the assumption that putative orthologs are functionally equivalent⁶⁶. However, when dealing with species that are evolutionarily distant, orthologs may have completely different biological functions⁶⁷. Therefore, extrapolating the biological significance of these top enriched BP terms to the invertebrate species investigated in this study is challenging. On the other hand, the enriched Molecular Functions (MF) and Cellular Components (CC) GO terms results were more relatable. The top enriched GO terms for the CC category were related to “ion channel complex”, “transporter complex”, and “plasma membrane protein complex”. Consistently, top enriched MF terms were related to “channel activity”, “transmembrane signaling receptor activity”, and “potassium ion transmembrane activity”. Most of the DE orthologs annotated along with these terms (Table 2) express proteins with a variety of functional roles that are putatively relevant for cleptoparasitic adaptations. In summary, in C. waltheriae upregulated orthologs annotated to these terms were related to olfaction, learning, neuronal excitability, synaptic organization, and other neuronal processes. Adaptations in these genetic pathways could be related to the successful exploitation of host resources and the development of specialized behaviors in cleptoparasitic bees. Meanwhile, orthologs highly expressed in T. diversipes were involved in chemosensory and sensory perception, particularly gustatory receptor genes involved in sweet taste response, motor, and mitochondrial activity, as well as neurotransmission, genetic mechanisms that could be associated with foraging and nest provisioning performed by T. diversipes foundresses. In boxes 1 and 2, we detailed the differentially expressed orthologs associated with the top MF and CC GO terms enriched by addressing their putative functions.

Cleptoparasitic bees are thought to rely on olfactory signals to locate host nests^68,69. Indeed, morphological differences in antennae sensillae composition have been observed between cleptoparasites and their host species, with cleptoparasites having a greater prevalence of olfactory structures⁷⁰. Previous studies have also shown that chemosensory-related genes (CRGs) are highly expressed in hymenopteran parasitoids^71–74. Chemosensory cues, along with other external stimuli, are processed in sensory systems, triggering the formation of both long-term and short-term memories⁷⁵. These adaptations may contribute to the localization and memorization of host nest aggregations⁷⁶. In addition to the sensory-related genes described in box 1, we identified other potential chemosensory-related genes (CRGs) upregulated in the cleptoparasite, including Ionotropic Receptors (IRs), a gene family involved in olfaction, gustation, and other sensory perceptions^77,78. Unlike many other chemosensory gene families, IRs exhibit a remarkable degree of conservation across species⁷⁹. The IR25a homolog was found highly expressed in C. waltheriae. This gene, along with IR8a, functions as a co-receptor (IRco) for the formation of functional IR complexes in conjunction with a ligand-binding receptor protein (IR X)⁸⁰. Furthermore, we detected an upregulated putative odorant receptor (OR) homologous to the Polistes dominula Or85c-like sequence in C. waltheriae. Although this specific OR's role in odour perception remains unknown, computational analysis in P. dominula suggests its involvement in parasitoid sensory perception. It is important to note that our current approach may have missed lineage-specific CRGs, and further investigation into these lineage-specific CRGs in cleptoparasitic lineages is crucial to gain insights into the underlying processes of parasitism in bees.

C. waltheriae and T. diversipes lineages, both belonging to the Apidae family, diverged over 77 million years ago¹¹. Thus, compared to several cleptoparasites, that prefer hosts from other bee families (e.g., cleptoparasites of the Nomadinae clade)¹⁰, the phylogenetic proximity between C. waltheriae and T. diversipes could minimize the “phylogenetic noise” in transcriptomic comparisons. Still, cross-species transcriptomic comparisons present challenges, and certain caveats should be considered in the light of the present results. First, our analytical strategy focused on conserved orthologs, thus neglecting non-orthologous or rapidly evolving genes that may play a crucial role in cleptoparasitism evolution. Additionally, due to the divergent life history trajectories of these two species, the differentially expressed orthologs identified could be associated with non-behavioral or random species-specific adaptations and not directly related to cleptoparasitic adaptations. Thus, we argue that applying the methodology used in this study to other bee host-parasite species pairs should allow the differentiation of species-specific adaptations from shared molecular mechanisms. Lastly, we conducted a broad-scale transcriptome analysis, meaning that the RNA-Seq data used were obtained from whole-body extractions. As an exploratory analysis, this approach allowed some initial broad insights, but it is limited in capturing genes with tissue-specific expression or complex expression patterns across the body⁸¹. We also used sample replicates containing a variable number of pooled individuals for C. waltheriae, to account for this sampling strategy we normalized and treated the gene expression counts for batch effects across replicates, and this treatment might have reduced even further our power to detect genes with subtle expression differences. Further research should therefore delve into the obtained results by investigating the expression of these genes in a tissue-specific context, in order to elucidate their functions more accurately. Finally, we analyzed a specific life stage of each species by comparing females performing reproductive activities. We considered that this stage comprises some of the most distinguishing behavioural differences between the host and cleptoparasite species, yet comparative analyses of multiple life stages could provide a more detailed profile of cleptoparasitism across developmental stages.

Box 1: Orthologs upregulated in C. waltheriae and associated to top enriched GO terms

In C. waltheriae, we found upregulated orthologs that may have roles in the olfactory and learning pathways, potentially representing adaptations to cleptoparasitic behavior^68,69.

Among the upregulated sequences in C. waltheriae, we discovered homologs to the invertebrate Octβ2R and vertebrate HTR2C, receptors for biogenic amines octopamine (OA) and serotonin (5-HT), respectively. Biogenic amines play important physiological roles in organisms, modulating neuronal, metabolic, and physiological processes⁸². In hymenopteran species, OA has been linked to locomotor activity, sensory (gustatory, olfactory, and visual) sensitivity, aggressive behavior, and (associative and non-associative) learning^76,83,84. Octβ2R has been associated with rewarding reinforcement signaling in Drosophila, indicating its potential role in behavioral responses⁸⁵. In honeybees, octopaminergic signaling is implicated in appetitive learning⁸⁶. Overexpression of octopamine receptors in honeybees has also been associated with oxidative stress, neuroinflammation, and olfactory dysfunction⁸⁷. Similarly, 5-HT acts as a neurotransmitter, neuromodulator, and hormone in insects⁸⁸, regulating various behaviors such as aggression, mating, feeding, locomotion, and olfaction^89–95. Invertebrate 5-HT receptors are found in the central and peripheral nervous systems (CNS and PNS, respectively), mediating both excitatory and inhibitory actions^96,97. The signaling of 5-HT receptors in Drosophila is linked to aggression, sleep, circadian behavior, feeding, mating, learning, and memory^98–103.

We also observed upregulation of the subunit of a neurotransmitter receptor in C. waltheriae, a homolog of the mammalian Gabbr2 (Gamma-aminobutyric acid type B receptor subunit 2). Meanwhile the Gabbr1 (Gamma-aminobutyric acid type B receptor subunit 1) was upregulated in T. diversipes. Gamma-aminobutyric acid (GABA) is the major neurotransmitter for inhibitory synaptic transmissions in the CNS of both vertebrates and invertebrates^104,105. GABA signaling influences insect behaviors such as learning and memory, locomotor activity, and odor processing^106–110. GABA_B receptors, coupled with ionotropic GABA_A receptors, mediate the action of GABA¹¹¹. These receptors regulate complex behaviors and nervous system functions by inhibiting GABA release and reducing the release of other neurotransmitters¹¹². For instance, in Drosophila and Heliothis virescens, GABA_B receptors are involved in the olfactory pathway^89,113. In Apis mellifera, GABA_B is associated with locomotor behavior¹⁰⁶.

In C. waltheriae, we also observed the upregulation of NaCP60E, a voltage-gated sodium channel. This gene has been mainly associated with olfactory function, but it is also expressed in other tissues^114,115 and in invertebrates without olfaction¹¹⁶. In Drosophila, decreased expression of NaCP60E has been shown to decrease the olfactory response to benzaldehyde¹¹⁷. Also, a dense concentration of NaCP60E proteins in neuron axons of chemosensory organs suggests that this gene is involved in the olfactory system¹¹⁸. Thus, the upregulation of NaCP60E suggests its involvement in olfaction-related processes in the cleptoparasitic bee.

C. waltheriae also showed increased expression of potassium channel orthologs, specifically homologs of human KCNT1 and KCNK9 genes. KCNT1 encodes a sodium-gated potassium channel that is activated by neuromodulators¹¹⁹. Therefore, it is suggested that it regulates neuronal excitability and may play a role in several behaviors¹²⁰. KCNK9 encodes a two-pore domain potassium channel that is involved in resting membrane potentials¹²¹. It can also be a target for modulatory molecules, such as volatile anesthetics and neurotransmitters¹²². Additionally, in Drosophila, KCNK9 orthologs (Task6 and Task7) are preferentially expressed in the antennal lobes and may play a role in olfactory memory formation¹²³.

Furthermore, we identified in C. waltheriae the upregulation of genes associated with neuronal excitability, including circadian rhythm, memory formation, and development. These include Unc80, Hr38, and Lrp6. Unc80 is a regulatory component of the sodium leak channel NALCN complex¹²⁴, which play a role in modulating resting membrane potential, neuronal excitability, firing rates, and pacemaker activity^125,126. In Drosophila, for instance, Unc80 is necessary for circadian rhythmicity¹²⁴. It is thought that Unc80 may also be involved in bee learning and memory³¹. The gene Hr38, a nuclear receptor, is involved in the transcriptional control of the dopamine synthesis pathway¹²⁷, the cuticle gene expression¹²⁸, and the ecdysteroid signaling pathways¹²⁹. In addition, it has been suggested that Hr38 may play an important role in high neuronal functions such as memory formation, courtship behavior, and circadian rhythm^130–132. Lrp6 belongs to the low-density lipoprotein receptor family of cell surface receptors and is an essential component of the Wnt signaling pathway, which controls several biological processes throughout the development and adult life of metazoans¹³³. Expression of the Drosophila ortholog of Lrp6 (arr) has been associated with CNS morphogenesis and organization¹³⁴, as well as long-term memory formation¹³⁵.

We also found upregulated orthologs of cell adhesion molecules (CAMs) genes, including NRXN1 and Fmi/Stan. In Drosophila, NRXN1 gene is required for proper organization and growth at neuromuscular junctions¹³⁶, and it also regulates sleep and synaptic plasticity¹³⁷. In honeybees, the expression of NRXN1 is increased under sensory stimulation, suggesting a link between sensory processing and associative learning¹³⁸. Drosophila Fmi/Stan encodes a cadherin that regulates planar cell polarity¹³⁹, axon guidance in photoreceptor neurons¹⁴⁰, dendrite morphogenesis of sensory neurons¹⁴¹, and neuronal morphogenesis of the mushroom body¹⁴². The upregulation of CAMs genes can be involved with their role in synaptic organization and plasticity, which may be relevant to the cleptoparasitic lifestyle of C. waltheriae.

Lastly, we found upregulated orthologs of ATPα and nrv2/nrv1 genes in C. waltheriae, which encode components of the Na + /K + ATPase^143,144. Na + /K + ATPase is important for maintaining the balance of sodium and potassium ions across the cell membrane¹⁴⁵. Defects in the sodium–potassium pump can lead to neuronal dysfunctions, such as circadian rhythm disturbances, locomotor problems, and auditory mechanosensation^146–148.

Box 2: Orthologs upregulated in T. diversipes and associated to top enriched GO terms

T. diversipes exhibited upregulated sequences related to chemosensory perception, particularly in the context of gustatory receptor (GR) genes. One such sequence was identified in T. diversipes as the Gr5a gene while in C. waltheriae, the corresponding sequence was identified as the Gr64f. gene. Despite the difference in annotation, both genes belong to the same conserved subfamily that is involved in the sweet taste response observed in Drosophila¹⁴⁹. Nevertheless, they are located on different chromosomes in Drosophila (Gr5a: X chromosome; Gr64f.: third chromosome)¹⁵⁰. These receptors are considered the primary basis for sugar reception in Drosophila, and their co-expression in gustatory receptor neurons is necessary for the response to certain sugars^149,151. Since bees primarily rely on flower nectar as an energy source, sugar detection is crucial for their survival¹⁵². Thus, a higher expression of these receptors in T. diversipes may be attributed to their intense foraging activity while provisioning food in their nests, an activity not performed by the parasite. Another upregulated sequence in T. diversipes was also related to GR, the transcript homologous to the G-sα60A gene of Anopheles gambiae. Studies on female antennae of A. gambiae have suggested that G-sα60A isoforms participate in olfactory signal transduction¹⁵³. It has also been proposed that Gsα in Drosophila is responsible for Gr5a-mediated sweet taste perception¹⁵⁴.

Apart from GR genes, several other upregulated genes in T. diversipes were associated with sensory perception but were notably involved in distinct pathways when compared to those upregulated in C. waltheriae. For example, the Drosophila Sh (Shaker) gene encodes a voltage-gated potassium channel (Kv)¹⁵⁵ that is involved in shaping and firing the action potential¹⁵⁶. It is also expressed in the retina and different regions of the CNS, including the mushroom body neuropil^157,158. Furthermore, evidence suggests that Sh regulation affects olfactory learning and memory¹⁵⁹. Another putative Kv upregulated in T. diversipes was a homolog to the vertebrate KCNQ1 gene, which regulates essential physiological processes all over the body¹⁶⁰. In Drosophila, KCNQ1 ortholog (dKCNQ) is expressed in the brain cortical neurons, cardia, and in the nurse cells and oocytes in the ovary^161,162. Apart from its role in the fly's normal heartbeat¹⁶² and early embryonic development¹⁶¹, dKCNQ has been implicated in age-dependent memory impairment and is required in α/β Mushroom Body Neurons for setting short-term memories¹⁶³.

Another upregulated sequence in T. diversipes, homologous to the Drosophila trp (Transient Receptor Potential) gene, may also be related to sensory perception. TRP proteins have been associated with phototransduction and olfactory response to CO₂ in Drosophila^164,165.

An upregulated homolog of the Drosophila abl (Abelson) gene, which encodes a non-receptor tyrosine kinase, was also identified upregulated in T. diversipes. The Drosophila abl gene is expressed in the axons of the CNS and plays a very important role in establishing axonal connections during CNS development¹⁶⁶. Also, it has been suggested that Abl-mediated phosphorylation is an important mechanism for the fly visual development¹⁶⁷.

Motor activity-related genes also exhibited higher expression levels in T. diversipes. One such gene is the cacophony (Cac), which encodes a voltage-gated calcium channel that plays a role in regulating neurotransmitter release at neuromuscular synapses in Drosophila¹⁶⁸. Similarly, a homolog of the Drosophila Ptp69D gene, which encodes a receptor of tyrosine phosphatase that is associated with mechanosensory neuron development¹⁶⁹, was also upregulated gene in T. diversipes. Ptp69D might influence some elements of motor function in adult T. diversipes females.

Additionally, in T. diversipes, there was a higher expression of a homolog to the UQCRFS1 gene. This gene encodes an iron-sulfur protein (Rieske Fe-S protein—RISP)¹⁷⁰ and is involved in electron transfer in bc1 complex¹⁷¹. Furthermore, T. diversipes displayed higher expression of homologs of other genes associated to electron transport activity, including genes with mitochondrial activity, indicating an increased energy demand during the founder life stage. These upregulated genes included, ETFA, MRPS18C, mdh2, Cox6al, and Ndufs2. The ETFA gene encodes an electron transfer flavoprotein subunit alpha, which is involved in mitochondrial fatty acid beta-oxidation and amino acid catabolism¹⁷². MRPS18C encodes a mitochondrial ribosomal protein that plays a role in protein synthesis within mitochondria¹⁷³. The mdh2 gene encodes a malate dehydrogenase, an enzyme involved in the citric acid cycle¹⁷⁴. Cox6al encodes a subunit of cytochrome c oxidase¹⁷⁵, and Ndufs2 encodes a core subunit of NADH dehydrogenase (Complex I)¹⁷⁶, another two of the critical enzymes in the electron transport chain.

Finally, several upregulated genes in T. diversipes were homologous to neurotransmitter receptors, including nAChRβ1, Grik2, and Grd. The nAChRβ1 gene encodes a subunit of the nicotinic acetylcholine receptor (nAChR), which is the primary excitatory neurotransmitter in insects. It is involved in olfactory learning and memory formation, as well as modulation of aggressive behavior^177,178. Grik2 encodes the vertebrate glutamate receptor ionotropic kainate 2, which contribute to rapid synaptic transmission¹⁷⁹. However, the role of kainate receptors in CNS glutamatergic circuits of insects is not well understood¹⁸⁰. The Grd gene encodes a Drosophila chloride-channel homolog of the mammalian GABA receptor delta subunit that may respond to different neurotransmitters¹⁸¹. This gene has been linked to mediating the glycine response¹⁸¹ and forming functional ionotropic GABA receptors¹⁸². Additionally, an upregulated homolog of the Drosophila CG31760 gene was found in T. diversipes. CG31760 is a probable G-protein coupled receptor (GPCR), whose expression was highest in the adult fly brain however its exact function is unknown¹⁸³.

Conclusion

In conclusion, our study provides global insights into the molecular mechanisms of cleptoparasitism in bees. We observed differential gene expression between the cleptoparasite C. waltheriae and its host T. diversipes particularly involving genes related to signal transduction, sensory perception, learning, and memory formation. These findings suggest the importance of sensory adaptations and learning in host-parasite interactions. We also identified a higher abundance of transcripts derived from transposable elements in C. waltheriae transcriptome, indicating these could be involved in gene neofunctionalization for parasite-specific adaptations. Additionally, the host species exhibited highly expressed genes from the Major Royal Jelly Proteins family. In bees, these genes are associated with various functions, including nutritional, immunity, and developmental regulation. We hypothesize that the differential expression of these genes could be related to nest provisioning in T. diversipes, a task not performed by the parasite. Further investigation is required to fully understand the role of all these molecular mechanisms in cleptoparasitism and their potential tissue-specific functions. Moreover, we propose that the methodology employed in the present study should be extended to other bee host-parasite species pairs, allowing for better differentiation of species-specific adaptations and shared molecular mechanisms underlying cleptoparasitism and its convergent traits in bees.

Material and methods

Sample collection and RNA sequencing

T. diversipes samples were collected and sequenced in a previous study³³. Briefly, females were collected during their reproductive period (November to December of 2012, here called G1, and March to July of 2013, here called G2), at Universidade de São Paulo (23° 33′ 53.2″ S 46° 43′ 51.7″ W), while provisioning their nests. G1 and G2 represent the foundresses of different reproductive generations. Individuals were always sampled between 10:00 A.M. and 12:30 P.M., immediately frozen in liquid nitrogen and stored at − 80 °C. RNA extractions were performed individually using the whole body. Each sample replicated contained pooled RNA from three individuals. Here we included in the analyses RNA-Seq data from five replicates: three from G1 and two from G2. For more details see³³.

Adult females of C. waltheriae were collected at the same location as T. diversipes samples while hovering at the entry of T. diversipes trap nesting spots. This behaviour is typical of C. waltheriae females looking for locations to lay their eggs²¹. Two individuals were collected in March 2013 (along with T. diversipes samples), and six more were collected later between November 2015 and March 2016. These individuals were also sampled between 10:00 A.M. and 12:30 P.M., immediately frozen in liquid nitrogen, and stored at − 80 °C. Total RNA extractions were performed for each individual separately using the RNeasy® Mini Kit (Qiagen, Austin, Texas, USA), following the manufacturer’s instructions. RNA was extracted from the whole body of the samples to enable a broad-scale transcriptomic analysis. Quantification and quality assessment of RNA was performed initially using NanoDrop™ Lite (Thermo Fisher Scientific, Wilmington, Delaware, USA) and later with the Agilent 2100 Bioanalyzer system (Agilent Technologies, Palo Alto, California, USA) before the library preparations. C. waltheriae samples were divided into five replicates: two containing only one individual in each (samples collected in 2013); and three containing the pooled RNA of two individuals (collected between 2015 and 2016). Library preparation and sequencing of the first two replicates (2013 samples) were performed by Macrogen (Macrogen Inc., Seoul, South Korea), along with T. diversipes samples from³³, using the Illumina® HiSeq 2000 platform. The other three replicates of C. waltheriae (sampled from 2015–2016) were sequenced at LaCTAD (Unicamp, Campinas, Brazil), also using an Illumina® HiSeq 2000 sequencer.

C. waltheriae RNA sequencing resulted in 285,806,598 raw reads. After cleaning, read number decreased to 169,961,662. For T. diversipes, 343,059,264 raw reads were used, resulting in 190,339,784 cleaned reads after trimming.

Transcripts assembly

Reads quality assessment was performed using the FastQC 0.11.5¹⁸⁴. Trimmomatic 0.38¹⁸⁵ was used to quality trimming the raw reads (options: SLIDINGWINDOW:4:30 TRAILING:3 MINLEN:80 AVGQUAL:30 HEADCROP:14). Cleaned reads were then digitally normalized (20 × coverage) using a script (insilico read normalization) implemented within the Trinity toolbox^186,187. Transcriptome assembly was performed differently for each species. C. waltheriae transcriptome was assembled following Trinity 2.10.0^186,187 de novo assembly protocol with default parameters. For T. diversipes, a draft genome was available (Santos et al., unpublished), so two different approaches were used: (1) a genome-guided transcriptome assembly, using STAR 2.7.3¹⁸⁸ and Cufflinks 2.2.1¹⁸⁹; and (2) a reference guided de novo assembly with Trinity, using the bam file generated by the STAR software. The transcripts resulting from these two assemblies were clustered into SuperTranscripts¹⁹⁰ by using CD-Hit¹⁹¹, Corset¹⁹², and Lace¹⁹⁰. This clustering approach was also applied to the de novo assembly of C. waltheriae, to make the datasets more comparable between the two species and to reduce transcript redundancy in C. waltheriae. Downstream differential expression analyses were performed using the SuperTranscripts non-redundant datasets. Quality parameters of the transcriptomes were analyzed with Transrate 1.0.3³⁸ and BUSCO 5.2.2¹⁹³ (using hymenoptera_odb10 database).

Annotation and expression analysis of orthologs

Non-redundant transcripts were annotated with Annocript 2.0.1¹⁹⁴ pipeline using the UniProt Reference Clusters (UniRef90)¹⁹⁵ and UniProtKB/Swiss-Prot¹⁹⁶ databases from February 2021. Transcripts with significant blast hits (e-value < 1e−5) against possible contaminants (i.e., Acari, Alveolata, Archaea, Bacteria, Fungi, Rhizaria, Rhodophyta, Viridiplantae, and Viruses) in the UniRef90 database were removed from the final data set using custom scripts (https://github.com/PauloCseri/Annotation.git).

To identify orthologs between the two species the amino acid sequences of the resulting ORFs predicted in the Annocript pipeline were used as input to Orthofinder 2.1.1^40,41. Then, we selected the best matching transcripts per species in each orthogroup to get one-to-one ortholog data. This selection was performed by filtering the transcripts of the highest bitscore value (better alignment) between sequences of the two species in the same orthogroup. These alignments were performed using the blastp algorithm^42,43 with default parameters. Differential expression (DE) analyses were then performed using only the orthologs identified in both species.

To estimate gene expression levels, we used Bowtie 2.3.5.1¹⁹⁷ to align the cleaned reads to their respective ortholog transcripts set, i.e., C. waltheriae sample reads were aligned to C. waltheriae ortholog transcripts and T. diversipes reads to its corresponding set of ortholog transcripts. Then, we used RSEM 1.3.3¹⁹⁸ for read counting. The significant DE orthologs between the two species were identified by combining the following approaches: (1) using the edgeR 3.34.0⁴⁵ with TMM normalized counts⁴⁴; and (2) using NOISeq 2.36.0^47,48 with the normalization factor calculated through the SCBN method to normalize read counts⁴⁶. This normalization strategy is optimized for cross-species DE analyses⁴⁶ and is implemented in the Bioconductor package SCBN 1.10.0. Only significant DE orthologs (|Log2FC|≥ 2, adjusted p-value < 0.05) commonly identified as so in these two analyses were selected for the resulting set of DEs. It is worth mentioning that as the samples were not sequenced in the same batch (different years), we additionally adjusted these counts for batch effects using the ComBat-seq method¹⁹⁹ available on the Bioconductor package sva 3.38.0²⁰⁰. In detail, for the batch parameter, we used the different sequencing times as factors (0: initial sequencing of C. waltheriae and T. diversipes samples; 1: later sequencing of C. waltheriae samples only) and for the group parameter (biological condition), we used the respective species (0: C. waltheriae; 1: T. diversipes).

GO functional analysis

Assuming that sequences from the same orthogroup descend from a single gene in the last common ancestor^40,41 and hence they likely have similar functions, the ensemble of functional GO annotations from all SuperTranscripts belonging to the same orthogroup were used for functional analyses. The Bioconductor package PloGO2 1.4.0²⁰¹ was used to plot and visualize the GO annotations of DE orthologs. To test whether any GO term was enriched among the DE orthologs in comparison to all other orthologs identified we used the Bioconductor package TopGO 2.20.0²⁰². Redundant GO enriched terms were summarized in a two-level hierarchical GO set using the REVIGO web server²⁰³ for simplification, and these hierarchical sets were represented in charts generated by the CirGO 2.0 software²⁰⁴.

Author contributions

P.C.R.: Methodology, investigation, data analyses and interpretation. N.S.A.: Conceptualization, methodology, sampling, assistance in data analyses and interpretation. M.C.A.: Conceptualization, supervision, fund acquisition. All authors contributed to the preparation of the manuscript.

Funding

This study was financed in part by Coordenação de Aperfeiçoamento de Pessoal de Nível Superior—CAPES, Brasil (scholarship to PCR, Proc. 1783961 and 88882.377416/2019-01, Finance Code 001), CNPq—Conselho Nacional de Desenvolvimento Científico e Tecnológico (research sponsorship to MCA—306932/2016-4) and Fundação de Amparo à Pesquisa do Estado de São Paulo—FAPESP (Proc. 2013/12530-4, 2016/24669-5 and 2019/23186-9). NSA is supported by the Fonds de la Recherche Scientifique—FNRS (Grant ID 40005980).

Data availability

The raw sequence reads have been deposited in the NCBI Sequence Read Archive (SRA) under the following accession numbers: SRR24187037, SRR24187038, SRR24187039, SRR24187040, SRR24187041, SRR24187042, SRR24187043, SRR24187044, SRR24187045, SRR24187046. The corresponding BioProject accession number is PRJNA955762. The transcriptome assemblies, as well as the sets of SuperTranscripts, are available on FigShare (10.6084/m9.figshare.23264771.v1).

Competing interests

The authors declare no competing interests.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-024-56261-5.