1 Introduction

The only extant species in tribe Apini are honey bees, which are very important pollinators of wildlife and agricultural crops. Their pollinating services are considered even more important than providing honey and additional bee products (Klein et al. 2007; Hung et al. 2018). For pollination under certain scenarios, different species of bees are more beneficial than honey bees (Goulson 2003). Currently, some pollinator alternatives to managed honey bees are commercially available. Commonly used are commercial bumblebee colonies, especially the buff-tailed bumblebee Bombus terrestris Linnaeus, 1758 (de Ruijter 1997; Velthuis and van Doorn 2006; Knapp et al. 2019). B. terrestris is the best-known member of tribe Bombini, which contemporarily includes only the genus Bombus (Dehon et al. 2019). Apini are at a higher eusociality level than Bombini. An important step for understanding the nature of this difference involves the comparison of genomes of their representative species (Sadd et al. 2015). Comparison of the genomes indicated cardinal gene differences that provide patterns for understanding the differences in life history, eusociality level and adaptation to environmental pressures between honey bees and bumblebees (Sadd et al. 2015). However, open questions about the molecular nature of the “high” eusociality advances of honey bees compared to bumblebees remain.

Honey bees provide an array of unique bee products, such as beeswax, honey, royal jelly (RJ), venom, propolis and pollen/bee bread, which are utilized by the colony. Humans have learned to utilize these bee products mainly as food nutrients and medicine (Schmidt 1997). An open question is whether we can also utilize natural products of bumblebees similar to honey bees. Despite viable potential, there is gap in the level of knowledge given that apitherapy used for thousands of years relates to products derived from honey bees (Kolayli and Keskin 2020), not bumblebees.

First, we should answer the question of whether bumblebees provide the same products as honey bees. There is one important and substantial difference. Bumblebees do not produce RJ, which is specific to honey bees that use it as food provided by nurse bees for queen and young larvae (Schmitzova et al. 1998; Buttstedt et al. 2014; Altaye et al. 2019). In that relationship, there is a lack of major royal jelly proteins (MRJPs) in bumblebees, and their hypopharyngeal glands (HGs) are reduced compared to honey bees (Albert et al. 2014). Importantly, MRJPs and a number of other proteins occurring in RJ were also identified in the honey proteome of A. mellifera, as summarized in a proteomic study (Erban et al. 2019). The proteome of bumblebee honey is unknown; however, it is possible that the major matrix could provide important data about similarities and differences in gland secretions compared to honey bees.

Most people certainly know that honey bees produce honey; however, only a few people are familiar with honey produced by bumblebees. Bumblebee workers store nectar and pollen in wax pots, that is, to some extent similar to honey bees. Indeed, honey pots can be found in bumblebee nests (Sladen 1912; Crane 1972; Konzmann and Lunau 2014; Svanberg and Berggren 2018). Although little is known, bumblebee honey was important before the introduction of apiculture to some countries, and it was also used in various folk religious customs (Svanberg and Berggren 2018). In contrast to honey bees, bumblebees store only a low amount of honey that never exceeds a few ounces, and despite excellent bumblebee honey flavor, it is only rarely used for human consumption (Sladen 1912). Different chemical properties of bumblebee honey might have medical potential, which requires unique compositions to diminish the specific threat of disease. However, there is a complete lack of knowledge in this field. Data on bumblebee honey characteristics are scarce and, to our knowledge, are limited to rough information on moisture and sugar content (Crane 1972).

In addition to the candidate bioactive compounds of plants that can be found in honey (Viteri et al. 2021), interest in the secretions that honey bees provide to honey is increasing. High importance is attributed to proteins that honey bees use to process and/or protect honey. Thus, numerous studies have been devoted to better understand honey bee honey proteins/enzymes and antimicrobial peptides (e.g. Schepartz and Subers 1964; Babacan and Rand 2005, 2007; Won et al. 2008; Sojka et al. 2016; Bucekova et al. 2017; Brudzynski 2020). Proteomic studies that utilized gel-based (e.g. Di Girolamo et al. 2012; Rossano et al. 2012; Chua et al. 2015; Zhang et al. 2019) or gel-free (e.g. Erban et al. 2019, 2021; Bong et al. 2021)) proteomic approaches have provided deep insights into the complexity of this research area. Similar studies aiming to understand bumblebee honey are however lacking. Indeed, an interesting approach involves determining the difference in the honey proteomes between honey bees and bumblebees. This acquired knowledge would potentially feature functional proteome differences that were earlier inferred in the compared annotated genomes of A. mellifera and B. terrestris (The Honeybee Genome Sequencing Consortium 2006; Sadd et al. 2015). In particular, the key difference in honey proteomes may be attributed to differences in eusocial behavior connected to food storage. Unlike honey bees, bumblebees are nonperennial and lack a food-receiving/storing caste (Seeley et al. 1996; Noll 2002). Thus, it is possible that different flows of nutrients in perennial and annual colonies (Judd 2011) will affect honey composition. Key signatures reflecting the differences in honey storage could be found in the honey proteomes. Indeed, honey proteins were observed to be very important in A. mellifera honey samples. The total protein content in A. mellifera honey can differ more than tenfold between honey types, whereas major proteins are provided in a constant ratio (Erban et al. 2019). A different study suggested that worker honey bees may secrete a constant amount of each protein to preserve honey (Lewkowski et al. 2019). Another important characteristic is that some proteins that honey bees provide to honey can be unique to honey samples (Erban et al. 2019). Furthermore, it has been suggested that honey store diversity may be highly adaptive for the “social immunity” of a colony against pathogens (Erler et al. 2014).

In the present study, we performed a comprehensive proteomics analysis of honey from the bumblebee B. terrestris. In addition, we provide a comparison with a previously described A. mellifera honey proteome. Some additional basic characteristics, such as total protein content, sugars and moisture, of the bumblebee honey samples were measured.

2 Materials and methods

2.1 Biological samples

The biological samples analyzed in this study were honey samples from the bumblebee B. terrestris. Commercially available Tripol hives used for pollination consist of three complete bumblebee colonies (Koppert Biological Systems, Berkel en Rodenrijs, the Netherlands). Four Tripol hives (12 colonies overall) were placed in the Crop Research Institute (Prague-Ruzyne, Czechia) at the time of rapeseed Brassica napus and apple tree flowering. Notably, each hive/colony was supplied with a sugar water container, which was closed before placing the hives outside (5th May) for the experiment. After 15 days (20th May), the colonies were euthanized using dry ice. For this study, honey stores were manually collected into 50-mL sterile centrifuge tubes. Honey samples per colony were pooled. It was obvious (e.g., coloring) that the bumblebee honey samples were collected outside of the colony. The samples of bumblebee honey were stored in a deep freezer at − 80 °C until use in the proteomic analysis.

2.2 Sample processing for proteomics

Briefly, samples of bumblebee honeys were processed and further analyzed as previously described (Erban et al. 2019). The samples were diluted with 0.2-μm-filtered Nanopure water (Thermo, Waltham, MA, USA) at a ratio of 1 g of honey to 2 mL of H2O and purified via gel filtration using PD MidiTrap G-25 columns (Cat. No. 28–9180-08, GE Healthcare Life Sciences, Marlborough, MA, USA). The exclusion limit is an Mr of 5000, and proteins with an Mr greater than 5000 should be easily separated from those with an Mr less than 1000. The cleaned samples were lyophilized in PowerDry LL3000 (Thermo), and the total protein content in separate aliquots was determined using Bradford reagent (Cat No. B6916; Sigma–Aldrich).

2.3 1D-E electrophoresis

The samples of bumblebee honey were tested using one-dimensional gel electrophoresis (1D-E) to assess sample quality and protein purity before nano-LC–MS/MS. Separation was performed using 4 to 12% Tris–glycine SDS–PAGE gels, and the sample buffer contained SDS and dithiothreitol (DTT). Full-Range Amersham Rainbow Marker RPN 800E (GE Healthcare Life Sciences) was used. Electrophoresis was performed at a constant voltage with a MiniPROTEAN Tetra Cell (Bio–Rad, Shanghai, China). The samples were stained using PhastGel Blue R 350 Coomassie stain (GE Healthcare Life Sciences). The gel was visualized using the G:BOX documentation system (Syngene, Cambridge, UK). Selected bands were excised and destained by sonication for 30 min in 50% acetonitrile (ACN) and 50 mM ammonium bicarbonate (ABC). Furthermore, the detaining solution was removed, and the gels were dried in ACN. Disulfide bonds were reduced using 10 mm DTT in 100 mM ABC at 60 °C for 30 min. Samples were again dried with ACN, and free cysteine residues were blocked using 55 mM iodoacetamide in 100 mM ABC for 10 min at room temperature in the dark. Samples were dried thoroughly. Gel pieces were covered with digestion buffer (10% ACN, 40 mM ABC and 13 ng/µL trypsin). Proteins were digested overnight at 37 °C. After digestion, 150 µL of 50% ACN with 0.5% formic acid was added, and the samples were sonicated for 30 min. The supernatant with peptides was transferred to a new microcentrifuge tube, and 150 µL of elution solution was added. Samples were again sonicated for 30 min. The solution was removed, combined with the previous solution, and dried using a vacuum concentrator. Dried peptides were reconstituted in 2% ACN with 0.1% trifluoroacetic acid (TFA) and analyzed using nano-LC–MS/MS. Individual raw data were evaluated using the same software and database as noted in the label-free shotgun proteomic experiment (see below); however, carbamidomethyl modification was used instead of MethylThio.

To illustrate the difference in the 1D-E profile in bumblebee honey compared to honey bee honey and royal jelly, representative images that were obtained using the same procedure that was used for bumblebee honey were selected (see above). Polyfloral honey and RJ samples were from Czechia.

2.4 Label-free nano-LC–MS/MS analysis

The samples were processed for mass spectrometry analysis as previously described (Erban et al. 2019). Trypsin digests were subjected to analysis using nano-LC–MS/MS employing an Orbitrap Fusion Tribrid mass spectrometer (Thermo). The raw Thermo mass spectrometry data were evaluated with MaxQuant v1.6.17.0 software using label-free quantification (LFQ) algorithms that require MS/MS identifications of peptides (Cox and Mann 2008; Cox et al. 2014). The key criteria were as follows: a false FDR of 0.01 for proteins and peptides; minimum length of 7 amino acids; MethylThio was the fixed modification; and N-terminal protein acetylation and methionine oxidation were variable modifications. The common contaminants supplied by MaxQuant were included in the search. However, different levels of databases were used in various searches. The database search considered that the honey was produced by B. terrestris and that the dominant flowering plants at the time of the experiment were rapeseed and apple trees. By performing preliminary data searches, we found that our previously selected non-honey bee markers in A. mellifera honey (Erban et al. 2021) were also useful for the search of bumblebee honey in this study. Thus, the following protein databases from the NCBI repository were used: (i) B. terrestris — 22,118 RefSeq sequences (downloaded on 11/10/2020); (ii) Brassica napus — 123,490 RefSeq sequences (downloaded on 01/03/2021) and Malus domestica — 52,057 RefSeq sequences (downloaded on 01/03/2021); and the (iv) combined database used in our previous study (Erban et al. 2021) that consisted of plant-related marker databases (cupin, germin, berberine, nectarin, Amb_all, beta-D-xylosidase, GAPDH, MetE, nsLTP, and PL-6) and aphid- and plant-related sequences. The data were further evaluated using Perseus v1.6.14.0 software (Tyanova et al. 2016).

In the data evaluation, protein hits were verified individually using Blastp (Altschul et al. 1990). Only reliably identified and specific protein hits were selected for final presentation. Individual protein sequences were evaluated in SignalIP 5.0 for the prediction of signal peptides (Almagro Armenteros et al. 2019). Furthermore, the protein sequences of B. terestris were searched in NCBI, and the first/best result related to A. mellifera was selected. The corresponding sequences were individually compared. Gene names were selected for both B. terrestris and A. mellifera accessions. In addition, conserved domains (CCDs) (Lu et al. 2020) were identified for B. terrestris accessions. In addition, honey bee–related protein identifications in A. mellifera honey from our previous studies (Erban et al. 2019, 2021) were individually evaluated to assess consistency with bumblebee-derived proteins found to be similar in B. terrestris honey. Further, among the similar proteins were identified those denoted in our previous study (Erban et al. 2019) as 71 major or trace protein hits.

2.5 pH, water and protein content, and sugar analysis

To determine the pH of honey, 1 g was diluted in 10 mL of 0.2-μm-filtered Nanopure water. The analysis was performed immediately after dilution using an Orion Star A111 pH meter (Thermo). The water content was determined using the lyophilization technique. Honey samples were placed into 50-mL sterile centrifugal tubes and lyophilized overnight in PowerDry LL3000 (Thermo). The water content was calculated as the difference between the tube and the sample before lyophilization and after lyophilization. The protein content per g of honey sample was calculated from the analyses in aliquots, which were analyzed using the Bradford reagent (Sect. 2.2). Sugar analysis was performed using the HPLC-ELSD technique as described previously (Erban et al. 2019) with a slight modification of the isocratic elution that was performed with 85:15 ACN/H2O.

3 Results and discussion

In this study, we provide the first proteomic analysis of honey from the bumblebee B. terrestris. The difference in the proteomes between B. terrestris and A. mellifera honey is visually illustrated with the 1D-E SDS–PAGE profile (Figure 1). Our comprehensive shot-gun label-free proteomic analysis reliably (i.e., after applying the cutoff threshold and individual verification of results; see Table S1) identified more than one hundred bumblebee-derived proteins in the honey samples (Table I and Figure 2). This number is greater than that of the A. mellifera honey proteome analyzed using the same shotgun proteomics method in our previous studies (Erban et al. 2019, 2021). Again, we stress that the same protein content was used to analyze each sample using nano-LC–MS/MS (Erban et al. 2019, 2021). Furthermore, according to our data, plant-derived proteins (Figure 2) are more abundant in the bumblebee honey proteome than in the honey bee proteome (Erban et al. 2021). The top-hit plant-derived proteins ranked approximately 10th in abundance based on intensity and 6th in abundance based on MS/MS counts. The array of the top-hit plant-derived proteins is listed in Figure 2, and the complete list is provided in Table S2. Overall, the array of predominantly identified plant-derived proteins agrees with the honey collection for our analyses at the time of rapeseed flowering.

Figure 1.
figure 1

Differences in the proteomes between B. terrestris and A. mellifera honey based on the Coomassie-stained 1D-E SDS–PAGE profile. In addition, the example of 1D-E profile of royal jelly produced exclusively by honey bees is shown for comparison. 1D-E profiles of B. terrestris honey collected from 12 colonies overall (four Tripol hives 1–4) are shown. Nano-LC–MS/MS of three bands (①, ②, ③) excised from the gel confirmed the results of label-free shotgun analysis (Table I, Figure 2) that the most abundant protein in B. terrestris honey is alpha-glucosidase (GenBank: XP_012164669). The nano-LC–MS/MS analysis results of the three bands showed that many proteins can be identified in the two strongest bands (Tables S3–S5). Note that the proteins of 12 bumblebee honey samples were separated in two gels. Thus, the two images are separated by white space between 2c and 3a samples.

Full size image
Table I 107 bumblebee-derived protein hits that were reliably identified in the 12 honey samples analyzed using nano-LC–MS/MS
Full size table
Figure 2.
figure 2

The top-hit plant-derived proteins identified among the bumblebee-derived proteins in B. terrestris honey. The first 70 protein hits were visualized and selected based on MS/MS counts. Legend: P – plant-derived proteins.

Full size image

An important parameter of honey is total protein content (Rossano et al. 2012; Chua et al. 2013; Erban et al. 2019). Our analysis shows that this factor is similar in bumblebees given that the average accounted for 0.4 mg proteins per g of honey with approximately 11% water content as determined by the lyophilization technique (Table II). However, the protein content is thought to differ based on botanical origin, which is similar to that noted for honey bee honey (Erban et al. 2019). Honey moisture will likely be affected by botanical origin and weather/season. It is necessary to consider that bumblebees have not been observed collecting water (Crane 1972); hence, the supposition is that nectar is the source of water for bumblebees.

Table II Characteristics of the bumblebee honey samples on protein, water, pH and sugars
Full size table

We suggest two general explanations for the different honey proteome structure of the bumblebee compared to honey bee. Bumblebee honey has higher total number of bee-derived proteins and increased relative abundance of plant-derived proteins. The most important difference is the lack of high-abundance proteins that have been identified mutually in RJ and honey of honey bees (Erban et al. 2019). These proteins mainly include MRJPs, glucose oxidase (GOx) and alpha-amylase. The lack of these proteins is consistent with the absence of RJ production, reduced HGs in bumblebees (Albert et al. 2014), and differences in the gene arsenal between the species (Sadd et al. 2015). Furthermore, there are different storage and flow of nutrients in the perennial and annual colonies of honey bees and bumblebees (Judd 2011). Overall, it appears that secretion of proteins to honey is more precisely regulated in colonies of honey bees compared with bumblebees. Indeed, honey bees have evolved a food-receiving/storing caste that has not been reported in bumblebees (Seeley et al. 1996; Noll 2002). Our results indicate that compared to bumblebees, honey bees invest more energy in honey in the form of a higher proportion of self-produced proteins within the proteome structure. We suggest that providing honey/food stores with more self-produced proteins may represent a further adaptation for the “highly eusocial” perennial colonies.

3.1 Low array of similar proteins in A. mellifera and B. terrestris honey

Sequence matches between A. mellifera- and B. terrestris-derived proteins in honey samples that were similar between the two species were limited (Table I). Specifically, approximately 40 similar (likely homologous) proteins were reliably identified in the honeys. However, the number of unique proteins in B. terrestris honey (see Table I) prevailed over those that we identified to be similar given that 107 proteins were reliably identified (Table I). We must consider that while some protein hits are major (high-intensity) in A. mellifera or B. terrestris honey, the homologs in the honey from the second organism were only present at trace levels or were missing in the analysis. Thus, important examples include increased relative abundances of vitellogenin, lipases, peroxiredoxin and regucalcin in bumblebee honey. Some of the different features, such as the lack of MRJP, could be revealed by a simple comparison of their genomes (Sadd et al. 2015), but the focus on differences in honey proteins was not noted until the present study. Indeed, the lack of alpha-amylase and GOx in honey is also key newly observed features. Thus, here, we provide novel insights into eusociality differences by comparing honey proteomes as representatives of gland secretions.

The similar sequences in honey proteomes of the two species (Table I) are likely true homologs (Krishna and Grishin 2004) that remained in the honeys of the two species after divergence from a common ancestor (Bossert et al. 2019; Porto and Almeida 2021). Future functional studies are necessary to uncover how these proteome similarities relate to the level of eusociality. For instance, the similarities as well as differences in protein secretions into honey could explain the need for the preservation and processing of honey (Erban et al. 2019; Lewkowski et al. 2019) between the annual and perennial colonies. Further understanding may be obtained regarding different brood rearing, immunity and/or signal transduction mechanisms.

3.2 Bumblebee-derived proteins among ~ 30 top hits

Due to the high number of bumblebee-derived proteins identified, for simplification, we focus in this section on those that were among the ~ 30 top-hit results (based on LFQ intensity in Table I). These relatively high-abundance proteins constitute an important difference between the honey of bumblebees and honey bees.

According to label-free shot gun proteomic analysis (Table I, Figure 2) and analysis of bands from 1D-E (Figure 1), the dominant protein in bumblebee honey is alpha-glucosidase (XP_012164669). The alpha-glucosidase homolog in honey bees (NP_001011608; Hbg3 gene) is secreted from hypopharyngeal glands (Kubota et al. 2004), and it is among the most abundant and key proteins in honey bee honey (Di Girolamo et al. 2012; Rossano et al. 2012; Erban et al. 2019; Zhang et al. 2019). Our analysis of sugar composition (Table II; Figure S1) showed that fructose and glucose are the main sugars in bumble bee honey, and the resulting fru/glu ratio (1.5) also appears similar (compare with Table II in Erban et al. (2019)). Due to the lack of major nectar disaccharide sucrose (Wolff 2006), it is likely that the alpha-glucosidase added by bumblebees to honey is invertase that functions in the same manner as honey bee Hbg3 (Kubota et al. 2004).

The second most abundant protein was pancreatic lipase-related protein 2 (XP_003398432), but three additional similar protein hits with the same name were identified among the top protein hits (XP_003398436; XP_003398435; XP_012168805). The latter sequence XP_012168805 had the highest similarity to A. mellifera lipase member H-A (XP_026299638; query cover 93%; identity 51.43%). Thus, these lipases are relatively similar in honey bee and bumblebee honey, but our results indicate that more isoforms are present as top hits in B. terrestris honey. The question is whether lipases can produce sugar esters (Siebenhaller et al. 2018).

Another high-abundance protein was uncharacterized protein LOC100644055 (Gdh; XP_020718843), which is similar to honey A. mellifera glucose dehydrogenase [FAD, quinone] (XP_006567694/XP_016767722; Gld2 gene; Query cover 94%, identity 77.8%) (Erban et al. 2019, 2021). Importantly, our proteomic data in this paper do not indicate that bumblebees secrete GOx into honey, which is one of the key enzymes secreted from honey bee HGs into honey (Bucekova et al. 2014). The question is to what extent the putative bumblebee “Gdh” (XP_020718843) substitutes or compensates for the function of GOx in the production of D-glucono-1,5-lactone/gluconic acid (Ferri et al. 2011; Milton et al. 2013; Erban et al. 2019). Gluconic acid is a major acid in honey (Mato et al. 2003; Ramachandran et al. 2006) and contributes to acidity that prevents the growth of bacteria and spoilage. The pH in our bumblebee honey samples was also acidic. The average pH of 4.6 (Table II) was approximately in the middle of the common pH range (between 3.5 and 5.5) for honey bee honey (Sereia et al. 2017). Thus, it is possible that bumblebees adjust the pH of honey, but whether gluconic acid contributes is unknown. The absence of GOx is also important because it is one of the key enzymes that maintains H2O2 in honey bee honey (Bucekova et al. 2014). Based on that connection, the relatively high abundance of peroxiredoxin 1 (XP_012172464) in B. terrestris honey appears to be important because as an antioxidant enzyme, it should be associated with reducing H2O2 levels (Neumann et al. 2009). Due to the high abundance, peroxiredoxin may play an important role in cell signaling (Neumann et al. 2009). A different situation has been observed in A. mellifera honey, in which peroxiredoxin 1 (XP_003249289) was only present in trace amounts (Erban et al. 2019, 2021). Secretion regulation of proteins that function to maintain H2O2 and their antagonists, such as peroxiredoxin, is an open question. It is also possible that these regulations occur due to environmental exposure, and one possible trigger may be microbes in the food sources of bees (Brudzynski 2021). Overall, bumblebees may utilize a different mechanism to preserve honey than bumblebees, i.e., H2O2 level maintenance.

We also identified vitellogenin (XP_012163499), which is somewhat similar to A. mellifera vitellogenin (NP_001011578; query cover 99%, identity 51.94%) among the top-hit proteins of B. terrestris honey. However, vitellogenin was identified only among the trace protein hits of A. mellifera honey (Erban et al. 2019, 2021). In the A. mellifera colony, vitellogenin was found to be multifunctional. Various functions of honey bee vitellogenin have been recognized (Amdam et al. 2012). Thus, this protein might function in immunity, oxidative stress and lifespan regulation. Although vitellogenin is known to interact with juvenile hormone in honey bees, a similar function was not indicated in bumblebees (Amsalem et al. 2014); instead, social interactions affect vitellogenin levels more strongly than a worker’s reproductive physiological state (Amsalem et al. 2014). Thus, we suggest that the difference in vitellogenin abundance in B. terrestris honey and the lack in A. mellifera honey is an important attribute corresponding to the intermediate stage of eusociality evolution in bumblebees.

The 7th top-hit protein (yellow-e3-like) in Table I is discussed in a separate section (Sect. 3.3).

Our results show that regucalcin (XP_003401730) is the fundamental B. terrestris honey protein component, which is similar to that noted for A. mellifera (XP_026298038; unch. protein LOC413627) honey (Erban et al. 2019, 2021). Regucalcin is interesting because it is a senescence marker in insects and mammals (Nakajima and Natori 2000). Regucalcin may function in honey to modulate biological processes by affecting signal transduction, and it affects cell growth by regulating apoptosis/proliferation (Izumi and Yamaguchi 2004; Yamaguchi 2013; Feng et al. 2015; Yamaguchi and Murata 2015).

Additional important proteins in A. mellifera honey are proteases (Di Girolamo et al. 2012; Rossano et al. 2012; Erban et al. 2019, 2021). Our proteome results indicate differences and similarities in protease proteins between B. terrestris and A. mellifera. Venom acid phosphatase Acph-1 (XP_012176421) and venom serine protease precursor (NP_001267823) represent significant and reliably identified proteases in bumblebee honey. In A. mellifera honey, homologs of these two proteases are reliably identified venom-like proteins (XP_001122458 and XP_006560620, respectively) (Erban et al. 2019, 2021). However, a protein (XP_016771183) homologous to B. terrestris transmembrane protease serine 9 (XP_020721948) was detected in only trace amounts in A. mellifera honey (Erban et al. 2019). A relatively substantial difference in chymotrypsin proteases was observed. Specifically, a high level of chymotrypsin-1 (XP_003393535) was identified in B. terrestris honey and was only rarely detected in A. mellifera (XP_394370) honey (Erban et al. 2019); additionally, chymotrypsin-2-like (XP_003396094; XP_003396093) homologous proteins (XP_393127) were not identified in A. mellifera honey (Erban et al. 2019, 2021). However, our results indicate the presence of proteins/peptides having potential to inhibit protease activity in B. terrestris honey, such as cysteine-rich venom protein 6 (XP_003401255), which is similar to A. mellifera chymotrypsin inhibitor-like (XP_006563421) (Erban et al. 2019, 2021).

Among the relatively high-intensity proteins identified were additional proteins whose homologs were not identified in A. mellifera honey, but these proteins are also of interest. Based on the domain architecture of multiple Ins_allergen domains in the uncharacterized protein LOC100651249 (XP_020719676), this protein might be linked to the chemical interaction of bumblebees with plants. For instance, similar proteins may be found in brassicaceous-feeding insects (Fischer et al. 2008). Notably, the bumblebee honey examined in this study was produced at the time of rapeseed B. napus flowering, which supports our interesting supposition that it might have evolved as an adaptation to detoxify certain plant compounds.

Other top-hit proteins are also of interest. The uncharacterized protein LOC100644337 (XP_012169771) contains a JHBP s.f. domain. The domain architecture of mucin-5AC (XP_012171910) has multiple chitin-binding peritrophin-A domains, serine-rich domains and a herpes_BLLF1 domain. Mucin-5AC alters host-microbiome interactions, bacterial adhesion and colonization (Quintana-Hayashi et al. 2015; Ramsey et al. 2017). Esterase FE4 (XP_003397297) contributes to the transformation of various compounds and the detoxification and elimination of environmental stressors (Ma et al. 2018). Carbonic anhydrase 2 (XP_003401445) is an enzyme that reversibly catalyzes carbon dioxide transformation. The enzyme can affect various physiological properties of honey, including pH regulation (Soydan et al. 2020). LOW QUALITY PROTEIN: RNA polymerase-associated protein LEO1 (XP_012169491) is classified as a prolipoprotein diacylglyceryl transferase based on its domain architecture (Lu et al. 2020). We did not detect any similar sequence in honey bees; therefore, this protein may be specific to the B. terrestris genome.

3.3 Presence of yellow-e3-like in the absence of MRJPs

The lack of MRJPs has individual consequences. For instance, the lack of MRJP1 and its “true” homolog has consequences for the antimicrobial peptide arsenal given that jelleins are derived from MRJP1 and contribute to antimicrobial functions in honey (Brudzynski and Sjaarda 2015; Brudzynski et al. 2015). Furthermore, bumblebees lack the functional potential of MRJP3 that participates in RNA uptake in honey bees (Maori et al. 2019). Therefore, an interesting question is whether some MRJP homologs(s) are present in bumblebee honey.

Importantly, a detailed investigation of our data showed that MRJP sequence homolog (XP_020719667.1| uncharacterized protein LOC100651683), which is a yellow-e3-like protein with two MRJP domains (CCD analysis (Lu et al. 2020)), is present among the high-abundance (top ten protein hits in Table I) bumblebee honey proteins. This protein has been denoted BtRJPL (XP_020719667/ADW82102). A nonnutritive function has been suggested for BtRJPL (Kupke et al. 2012), and its expression in the bumblebee brain has been reported (Kupke et al. 2012; Albert et al. 2014). In addition, this yellow-e3-like protein has been reported in B. terrestris venom (GenBank removed record gi|340,716,434|ref|XP_003396703.1–100% identity to XP_020719667). However, in the reporting study (Van Vaerenbergh et al. 2015), its expression was presented relative to MRJP9 in A. mellifera. According to our analysis, the most similar protein related to BtRJPL in A. mellifera is yellow-e3 (XP_006558991; Query cover 94% and Identity 77.25%). Our comparison with the A. mellifera honey proteome (Erban et al. 2019, 2021) showed that yellow-e3 is not a honey protein in A. mellifera. Therefore, these proteins with similar sequences likely have different functions in the context of eusociality level adaptations.

In general, the primary protein structure (sequence) of all MRJPs is highly similar to that of the yellow gene family (Buttstedt et al. 2014). Conceivably, BtRJPL would be replaced by MRJPs in evolution, or these proteins would have no direct evolutionary connection. Incidentally, our hypothesis may be related to the substantial change in the morphology of HGs of honey bees due to specific food production linked to RJ production (Kupke et al. 2012; Albert et al. 2014).

3.4 Lack of alpha-amylase in bumblebee honey

One important component of honey bee honey is alpha-amylase. Alpha-amylase activity is expressed as a diastase number (DN) (Bodganov et al. 1997; Pasias et al. 2017), which has become an internationally valid measure of freshness and quality, enabling the marketing of honey (Bogdanov et al. 1999; FAO and WHO 2001; Council of the European Union 2002). However, we did not identify homologous sequences to alpha-amylase in B. terrestris honey despite the fact that the expected sequence was in our search database. Thus, alpha-amylase does not appear to be functionally required in honey for bumblebee colonies. This shortage can again be explained by the lack of RJ production (Albert et al. 2014). Indeed, similar to MRJPs and GOx, alpha-amylase is among the main A. mellifera honey components that are also found in RJ (Erban et al. 2019).

3.5 Note on other proteins processing carbohydrates

As mentioned above (Sect. 3.2), in B. terrestris honey, we identified alpha-glucosidase (XP_012164669), which is homologous to A. mellifera Hbg3. The homologous proteins/enzymes of Hbg3 alpha-glucosidase are major protein components in honeys from both species. Further, we reliably identified maltase 1 (XP_003395914) and maltase A1 (XP_003402225), which are similar to A. mellifera Hbg1 and Hbg2 (Table I), respectively. Although the two alpha-glucosidases are of interest in A. mellifera (Kimura et al. 1990; Takewaki et al. 1993; Kubota et al. 2004), they were not identified to be components of honey proteome (Erban et al. 2019, 2021). It is in agreement with that they are not secreted from honey bee HGs (Kubota et al. 2004), which are reduced in bumblebees (Albert et al. 2014). On the other hand, Hbg1 and Hbg2 are present in ventriculus and hemolymph of honey bees (Kubota et al. 2004). Taken together, we suggest that identification of maltase 1 and maltase A1 in B. terrestris honey can be due to different ways of secretion to honey.

Overall, important differences exist in the detection of alpha-amylase and alpha-glucosidases in A. mellifera and B. terrestris honey. Although alpha-amylase is absent in B. terrestris honey, homologous sequences to all three alpha-glucosidases (Hbg1, 2 and 3) that are of interest in honey bees were reliably detected.

3.6 The differences in antimicrobial peptides and protease inhibitors

One of the important components of A. mellifera honey is antimicrobial peptides (AMPs). Based on our previous conclusion from a study of the A. mellifera honey proteome, serine protease inhibitors might also be involved in antimicrobial activity (Erban et al. 2019). Interestingly, a genome analysis showed that B. terrestris only contains a single copy of the defensin gene compared to two copies in A. mellifera. On the other hand, serine protease inhibitors are expanded in the B. terrestris genome compared to the A. mellifera genome (Sadd et al. 2015). We were not able to identify homologs of defensin-1, which is an important and well-known AMP in A. mellifera honey (Kwakman et al. 2010; Di Girolamo et al. 2012; Sojka et al. 2016; Bucekova et al. 2017; Erban et al. 2019, 2021). In addition, we did not identify a hymenoptaecin homolog that was recently identified to be commonly present in A. mellifera honey (Erban et al. 2019, 2021). These results indicate lack of AMPs in B. terrestris honey. On the other hand, a striking resemblance to the A. mellifera honey proteome showed identification of homologs of B. terrestris serine protease inhibitors, such as serine protease inhibitor 88Ea (XP_003398424) and cysteine-rich venom protein 6 (XP_003401255).

Further, an immune-related function was suggested for the B. terrestris peptidoglycan-recognition protein (PGRP) SC2 (XP_012170795) (Barribeau et al. 2015), which we identified in B. terrestris honey. This protein might be of interest because an additional PGRP was identified in B. impatiens (XP_003487752), which is missing in both B. terrestris and A. mellifera (Barribeau et al. 2015).

Finally, we identified a “probable salivary secreted peptide” (XP_003394058) that has a homolog in A. mellifera but was not detected in its honey. This peptide belongs to the MBF2 superfamily. In Bombyx mori, this peptide acts as a transcriptional activator through interaction with TFIIA (Li et al. 1997; Liu et al. 1998, 2000), which functions in defense against reactive oxygen species (Kraemer et al. 2006).

3.7 Plant proteins in bumblebee honey

Based on our analyses, plant-derived proteins appear more abundant (Figure 2) in the bumblebee honey proteome than in the honey bee (Erban et al. 2021) honey proteome; however, the total protein content appears in a similar quantities, i.e., tenths mg of protein per g (Table II). Providing honey with more secreted proteins is associated with a higher energy cost of protein synthesis. Such an energy investment must be highly beneficial. Based on the structure of the honey proteome, honey bees invest more in protein secretion than bumblebees. Notably, RJ-related proteins, mainly MRJPs, are also dominant in honey (Won et al. 2008; Di Girolamo et al. 2012; Chua et al. 2015; Erban et al. 2019, 2021). The lower proportion of bumblebee-derived proteins in honey might be due to a simpler flow of nutrients in the annual colonies (Noll 2002; Judd 2011).

The first/most abundant non-bumblebee protein hit in our analysis was identified as 5-methyltetrahydropteroyltriglutamate-homocysteine methyltransferase (MetE); based on the total MS/MS counts, this protein hit was 6th among all proteins (Figure 2, Table S2). In A. mellifera honeys, MetE was also observed to be the most abundant plant protein group, whereas it was 24th based on the MS/MS counts in a dataset (Erban et al. 2021). MetE and number of other plant-derived proteins that we identified in the honey of B. terrestris according to taxonomy identity assigned to rapeseed origin. This result is consistent with the fact that rapeseed flowered at the time of sampling and that it was near our experimental colonies. Of note, some rather rarely identified proteins were also specific to apple trees, which also flowered at the sampling time.

These results regarding plant-derived proteins are useful for the future development of methods to analyze honey proteins that are not provided by bumblebees. The applicability of this proteomic approach is similar to that used for the honey of honey bees, which is elucidation of honey properties and analyses of differences in honey composition and origin (Erban et al. 2021). Finally, some proteins provided by bumblebee workers are potentially affected by the plant source; e.g., the aforementioned results suggested an interaction between a protein with multiple Ins_allergen domains and brassicaceous-defensive compounds (Fischer et al. 2008; Lu et al. 2020).

3.8 Markers for future interdisciplinary research

We mainly targeted similarities and differences in honey bee and bumblebee honey proteins. Although important differences have been previously inferred from comparison of genomes (Sadd et al. 2015), the proteome comparison of honey provided functional markers specific to proteins that the two species of bees provide to honey. Importantly, some of the markers that indicate interspecies difference at qualitative proteome level (present/absent in a sample) in honey were found in genomes (Sadd et al. 2015) of both A. mellifera and B. terrestris. Thus, our proteome study provided differentiation markers that could not be identified from comparison of genomes. A nice example is AMPs defensin and hymenoptaecin, which were identified only in A. mellifera honey proteome (Erban et al. 2019, 2021). This difference clearly relates to immunity adaptation. The most important qualitative markers are due to the absence of RJ production in bumblebees. Indeed, results of this study show that the difference in RJ production is projected to honey proteomes. This difference has multiple direct consequences such as nutritional adaptation (e.g., MRJPs), immunity (e.g. MRJP1 is the source of AMPs Brudzynski and Sjaarda 2015; Brudzynski et al. 2015)), and physical properties (e.g., GOx). An open question is if bumblebees substitute the missing proteins by other proteins in their honey or whether the protein functions are not required at the lower eusociality level. This hypothesis may be valid also vice versa, because some adaptations could be lost due to higher eusociality level of honey bees. Whether the interspecies differences are qualitative or quantitative, both can have roots at the level of evolutionary adaptations. Overall, the comparison and acquired markers are useful for future research in bee biology and evolution. Furthermore, the identification of bee-derived and other proteins is useful for the determination of protein flow in honeys of different bee species depending on environmental and other biological factors.

The new complex knowledge acquired in this study may also be applied to pharmacological research. Honey bee honey has become a subject of increasing interest due to its broad medial applications (Molan 2001; Hidaka et al. 2008; Othman 2012; Ahmed and Othman 2013; Majtan 2014; Ranzato and Martinotti 2014; Martinotti et al. 2017; Samarghandian et al. 2017) in addition to its use as a common first aid measure or for the prevention of cold and upper respiratory infection in general (Samarghandian et al. 2017; The Lancet 2018; Abuelgasim et al. 2021). Honey is also a very important matrix for mining markers for potential use to treat various diseases (Majtan 2014; Samarghandian et al. 2017; Erban et al. 2019). However, there is a lack of knowledge in this research area regarding the honey of bumblebees.

4 Conclusions

In conclusion, we provide the first in-depth insights into honey from the bumblebee B. terrestris. We also provide pilot data on total protein, moisture, pH and sugar (fru/glu ratio) characteristics. The results reveal similarities and differences in proteins secreted into honey by bumblebee and honey bee colonies. Similarities and differences in bee-derived proteins are useful in future studies related to the determination of eusociality differences. Furthermore, the approach and data are useful to study seasonal and regional variances within bumblebee species and variances due to different bumble species. The major difference reported here at the proteome level is the lack of major RJ proteins, mainly MRJPs, in bumblebee honey. A dispute exists whether the yellow-3-like protein might represent the prediversification stage. However, a number of other proteins that show differences in these honeys were identified. Importantly, with the exception of results for ~ 40 proteins, we were not able to identify A. mellifera honey-related homologs in the majority of the overall array of 107 reliably identified B. terrestris honey-related proteins. The key missing proteins include alpha-amylase and GOx, which are also important components of honey bee RJ and honey. An important open question is how bumblebees regulate the physical properties of honeys in the absence of various proteins, such as GOx. In this connection, the presence of carbonic anhydrase 2 in bumblebee honey may be important. However, further studies are needed to confirm the functions of these and other proteins. The results of this study and future studies of the bumblebee honey proteome should help to better understand how secretion of these proteins is regulated in bumblebees, which have substantially reduced HGs.