SEARCH

SEARCH BY CITATION

Keywords:

  • Bombus ;
  • COI ;
  • conservation;
  • degraded DNA;
  • PCR–RFLP;
  • species identification

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results and discussion
  6. Acknowledgements
  7. References
  8. Data Accessibility

The worldwide decline and local extinctions of bumblebees have raised a need for fast and accurate tools for species identification. Morphological characters are often not sufficient, and molecular methods have been increasingly used for reliable identification of bumblebee species. Molecular methods often require high-quality DNA which makes them less suitable for analysis of low-quality or older samples. We modified the PCR–RFLP protocol for an efficient and cost-effective identification of four bumblebee species in the subgenus Bombus s. str. (B. lucorum, B. terrestris, B. magnus and B. cryptarum). We used a short partial mitochondrial COI fragment (446 bp) and three diagnostic restriction enzymes (Hinf I, Hinc II and Hae III) to identify species from degraded DNA material. This approach allowed us to efficiently determine the correct species from all degraded DNA samples, while only a subset of samples 64.6% (31 of 48) resulted in successful amplification of a longer COI fragment (1064 bp) using the previously described method. This protocol can be applied for conservation and management of bumblebees within this subgenus and is especially useful for fast species identification from degraded samples.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results and discussion
  6. Acknowledgements
  7. References
  8. Data Accessibility

Bumblebees (Bombus spp) are important pollinators of natural flowering plants and they are also widely used to pollinate various commercial crops (Goulson 2010). The recent decline of the bumblebees and local extinctions in several species worldwide (Fitzpatrick et al. 2007; Goulson 2008) have raised a need for precise identification tools for conservation and management. This is especially important for cryptic taxa, such as the B. lucorum complex (B. lucorum, B. magnus and B. cryptarum) of the Holarctic subgenus Bombus sensu stricto (s. str.), as it is extremely hard to reliably identify species within this group by using only morphological characters such as coloration (Carolan et al. 2012). As one particular species of the subgenus, B. terrestris, is widely used for pollination, it is very important to be able to distinguish the commercially used species from closely related species such as those belonging to the B. lucorum complex. This is especially relevant in areas where B. terrestris does not naturally occur, for example in most of Finland (Pekkarinen & Kaarnama 1994) and the northern parts of Sweden, as commercial strains of B. terrestris can out-compete native species (Chittka et al. 2004; Ings et al. 2006) and transmit pathogens to local pollinators (Goka et al. 2001; Colla et al. 2006; Murray et al. 2013).

Polymerase chain reaction–restriction fragment length polymorphism (PCR–RFLP) based methods are rapid, cheap and reliable tools to identify insect species when morphological characters are not sufficient or the samples are of bad quality. Murray et al. (2008) developed recently a PCR–RFLP method for distinguishing four bumblebee species (B. lucorum, B. terrestris, B. magnus and B. cryptarum) in cryptic European taxa of the subgenus Bombus s. str. Their approach uses a partial mitochondrial COI sequence (1064 bp) and primers originally developed for Apis mellifera. However, this method is not optimal for degraded DNA because then the amplification of long DNA fragments often fails. Degraded DNA is often gained when the samples are very old or otherwise damaged, for example, by suboptimal storage conditions, such as when using various trapping material, where the samples may have been in salt water and detergent for several weeks.

We modified the PCR–RFLP protocol for an efficient and cost-effective identification of four bumblebee species in the subgenus Bombus s. str. (B. lucorum, B. terrestris, B. magnus and B. cryptarum). We used a shorter partial mitochondrial COI fragment (446 bp) and three diagnostic restriction enzymes (Hinf I, Hinc II and Hae III) to identify species from degraded DNA material.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results and discussion
  6. Acknowledgements
  7. References
  8. Data Accessibility

Individuals (N = 863) of different castes (355 queens; 442 workers; 66 drones) of bumblebees in the Bombus lucorum complex were collected with yellow window traps in 2008 and 2009 in Kemiönsaari, SW Finland. These samples had first been in saturated salt water and detergent (10 mL/1 L water) for 2–4 weeks during trapping and stored in 90% ethanol for 3–4 years after collecting. DNA was extracted from individual bumblebees from one entire middle leg using a slightly modified protocol of Laird et al. (1991). To test DNA quality from our samples, we screened a subset of our trapping samples using bioanalyser (N = 11), which revealed that 66.7% of the fragments in our DNA extracts were shorter than 300 bp (data not shown). We tested the performance of our modified PCR–RFLP protocol in relation to Murray et al. (2008) approach using both relatively fresh (N = 48) B. terrestris samples (stored shortly after death in −20 °C for 1–3 months before DNA extraction) and a subset of the degraded B. lucorum trapping samples (N = 60; kept for 2–4 weeks in saturated salt water and detergent and then stored in 90% ethanol in +4 °C for 3–4 years before DNA extraction). The results of this PCR–RFLP were verified by Sanger sequencing (Macrogen Inc., Republic of Korea).

Based on the mitochondrial COI sequence of Apis mellifera ligustica (NC_001566.1; Crozier & Crozier 1993) and published bumblebee sequences (Murray et al. 2008; Kozmus unpubl.; Pedersen 2002), we designed primers for the amplification of a partial mitochondrial COI fragment (446 bp): forward (COI_M2F) 5′-GAAACCTTTGGAAATTTAAGA-3′ and reverse (COI_End1R) 5′AATTGAATTTTTAATCATTTTTGA-3′. The 45 μL amplification reaction contained: 10.8 μL sterile H2O, 0.3 μm each primer, 22.5 μL 2X Qiagen multiplex PCR buffer and 9 μL of extracted total DNA (5 ng/μL). Amplification was performed using a MJ Research PTC-100 Peltier thermal cycler with the following protocol commonly used with Qiagen multiplex PCR kit: initial denaturing step of 15 min at 95 °C followed by 35 cycles of 94 °C for 30 s, 52 °C for 90 s and 72 °C for 60 s, a final extension at 72 °C for 5 min and cooling at 20 °C for 60 s. For the identification of B. lucorum, B. terrestris, B. magnus and B. cryptarum, three restriction enzymes (Hinf I, Hinc II and Hae III) were used separately and their restriction products were separated on 1.5% agarose gel (Fig. 1a–c). The fifth species of this group, B. sporadicus, is easily distinguished from the other species by morphological characters and is therefore not included in our analysis.

image

Figure 1. RFLP haplotypes of mitochondrial COI fragment (446 bp) using restriction enzyme (a) Hinf I (b) Hinc II and (c) Hae III in 1% agarose gel stained with ethidium bromide. 1, 10: GeneRuler 100 bp Plus DNA Ladder (Thermo Scientific), 2–3: B. lucorum, 4–5: B. terrestris, 6–7: B. magnus, 8: B. cryptarum, 9: uncut sample. Note that restriction is not complete for all samples.

Download figure to PowerPoint

The identification process using three restriction enzymes was performed in a stepwise manner. Initially, 15 μL of the PCR product was digested with Hinf I in R-buffer (1 × composition: 10 mm Tris-HCl (pH 8.5 at 37 °C) 10 mm MgCl2, 100 mm KC, l0.1 mg/mL BSA; Fermentas) by incubating for 2 h at 37 °C followed by an inactivation at 65 °C for 20 min. Haplotypes C, D (B. cryptarum), and E, F, G (B. magnus) have a restriction site at position 2629 when aligned with A. m. liqustica (NC_001566.1; Crozier & Crozier 1993; Table 1), where Hinf I cuts the 446-bp COI amplification into 47-bp and 399-bp long fragments (Fig. 1a, Table 2). Bombus terrestris haplotypes H/I have two restriction sites at positions 2815 and 2974 (Table 1), resulting in three fragments (53, 180 and 213 bp; Fig. 1a, Table 2). B. terrestris haplotypes K and L have one restriction site at position 2974 (Table 1), resulting in two fragments (53 and 393 bp, Table 2). In addition, B. terrestris haplotypes J and M have a restriction site at position 2815 (Table 1), where Hinf I cuts the COI amplification into 213-bp and 233-bp fragments (Table 2). The samples without the above-mentioned restriction sites are either B. lucorum or B. cryptarum haplotype B. In haplotypes H/I and G, we saw some indication of either incomplete restriction or nuclear copies of mtDNA also known as pseudogenes or NUMTs.

Table 1. RFLP haplotypes (following Murray et al. 2008) based on a short mitochondrial COI fragment (446 bp) of Bombus terrestris, B. lucorum, B. magnus and B. cryptarum
Species Positiona Restriction enzymeb Haplotypesc2617 H12629 H12752 H22815 H12910 H32966 H12974 H1
  1. a

    The beginning of the restriction site when aligned with A. mellifera liqustica sequence (Crozier & Crozier 1993).

  2. b

    H1 = Hinf I, H2 = Hinc II, H3 = Hae III.

  3. c

    Haplotypes assigned by Murray et al. (2008) with some suggestions for corrections/clarifications.

  4. d

    New GenBank Accession nos, confirmed by sequencing.

B. lucorum KC192047 dA       
AY694095 A       
AY530010 A       
AY530009 A       
AY181119 A       
EF523366 A       
EF523364 A       
EF523362 A       
EF523363 A       
B. cryptarum KC192045 d B    x  
AY181123 B    x  
AY181124 B    x  
AY530011 C x     
AY694069 C x     
AY530011 C x     
EF362727 D xx    
B. magnus AY530015 E xx    
AY530014 E xx    
EF362733 F xx    
AY533014 F xx    
KC192046 d G xx   x
B. terrestris KC192043 d H/I   x  x
AY181170 H (corrected)   x  x
EF362742 I (same as H)   x  x
AY181171 J   x   
EF362743 K      x
EF362744 L      x
AY181169 M   x   
Table 2. The expected fragment lengths of the RFLP haplotypes of Bombus terrestris, B. lucorum, B. magnus and B. cryptarum based on enzyme restriction sites of the 446-bp mitochondrial COI fragment
SpeciesHaplotypeaFragment lengths (bp)
  1. a

    Haplotypes assigned by Murray et al. (2008).

B. lucorum Restriction enzymeHinf IHinc IIHae III
B. cryptarum A446446446
B. magnus B446446118/328
C47/399446446
D47/399170/276446
E/F/G47/399170/276446
B. terrestris H/I53/180/213446446
J213/233446446
K53/393446446
L53/393446446
M213/233446446

Second, 15 μL of PCR product was digested with Hinc II in Tango buffer (1 × composition: 33 mm Tris-acetate (pH 7.9 at 37 °C), 10 mm Mg-acetate, 66 mm K-acetate, 0.1 mg/mL BSA; Fermentas) by incubating for 2 h at 37 °C followed by an inactivation at 65 °C for 20 min. B. magnus haplotypes E, F and G are cut into two fragments (170 bp and 276 bp; Fig. 1b, Table 2). There is also one possible haplotype of B. cryptarum (D; Table 1) that has an identical restriction pattern with B. magnus according to GenBank sequence EF362727 (Murray et al. 2008). There are no Hinc II restriction sites in either B. lucorum or B. terrestris. In haplotype G, we found some indication of incomplete restriction.

Finally, 15 μL of the PCR product was digested with Hae III (BsuRI) in Tango buffer by incubating for 2 h at 37 °C followed by an inactivation at 80 °C for 20 min. B. cryptarum haplotype B has one Hae III recognition site at position 2910 (Table 1) producing two fragments of sizes 328 and 118 bp (Fig. 1c, Table 2), and we also saw some indication of incomplete restriction. There are no sites in B. lucorum, B. terrestris and B. magnus.

Results and discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results and discussion
  6. Acknowledgements
  7. References
  8. Data Accessibility

We compared our modified protocol to the PCR–RFLP approach of Murray et al. (2008) using both relatively fresh (N = 48) Bombus terrestris samples from commercial colonies and a subset of the degraded (N = 60) B. lucorum trapping samples. None of the degraded B. lucorum trapping samples amplified the longer 1064-bp COI fragment used by Murray et al. (2008), and from the relatively fresh B. terrestris samples, only 64.6% amplified successfully. On the other hand, 100% of both the fresh and degraded samples amplified the shorter 446-bp COI fragment using our newly designed primers. From 863 trapping samples, total DNA was extracted from 810 individuals of which 667 (82.3%) amplified with our method. Of these 667 individuals belonging to the Bombus s. str. group, four individuals were identified as B. terrestris, nine as B. cryptarum, none as B. magnus and the majority (654 ind.) as B. lucorum.

Compared with earlier techniques, our modified PCR–RFLP method is more suitable for analysis of degraded DNA for identification of cryptic species within the subgenus Bombus s. str. Such materials are often obtained from samples that have been collected with various window or pan traps using liquids and detergent to avoid surface tension and subsequently stored in ethanol, from samples captured alive and after killing stored in a freezer for many years and from old museum samples (Dean & Ballard 2001). Our method can also be applied in combination with nondestructive sampling, as it uses only one tarsus of each bumblebee. Compared with Murray et al. (2008) approach where two restriction enzymes, EcoNI and Hinf I, were used simultaneously, our method is slightly more time-consuming as three restriction enzyme digestions are carried on separately. On the other hand, this increases the overall hands-on laboratory time only slightly, while the agarose gel electrophoresis of individual restriction enzyme products results in an easier interpretation of the RFLP patterns. Murray et al. (2008) approach identifies more intraspecific variation because it uses a longer COI fragment with a greater number of diagnostic restriction enzyme sites. Our modified method is not able to separate B. magnus and haplotype (D) of Irish B. cryptarum, and therefore, all samples showing the B. magnus pattern should be verified by sequencing. In addition, the following within species haplotypes cannot be separated with our modified protocol: B. terrestris K and L, as well as J and M; all B. magnus (E, F and G). As our modified method identifies one species (B. lucorum) by an absence of a restriction site, it is important to use positive controls for all taxa.

To validate the gel electrophoresis results gained with our modified protocol, we used Sanger sequencing for a subset of the whole data set (= 96). Nuclear copies of mitochondrial DNA, or NUMTs, are generated through the natural transfer of DNA from mitochondria to the nucleus (Hazkani-Covo et al. 2010). When universal primers are used for amplification of mtDNA fragments, NUMTs may co-amplify with the mitochondrial orthologue which can cause incorrect species identification (Moulton et al. 2010). However, the presence of NUMTs can be detected by carefully examining sequence characteristics such as indels, in-frame stop codons, increased polymorphisms and nucleotide composition from Sanger sequencing (Song et al. 2008). The frequency of NUMTs increases if sample material is in poor condition because the extracted DNA is often degenerated and recombines more easily (Thalmann et al. 2004; Schitzas 2012). We suspected that there might be either incomplete restriction or NUMTs especially in haplotypes H/I and G, as we occasionally saw additional bands in the electrophoresis gel. The Sanger sequencing of a subset of samples confirmed the presence of NUMTs in 23 of 92 samples (25%), of which B. cryptarum had NUMTs in 9 of 15 samples (60%), B. lucorum in 10 of 32 samples (31%) and B. terrestris in 5 of 41 samples (12.2%). The growing evidence shows NUMTs to be a common phenomenon among insects (Moulton et al. 2010) and causing various problems especially in rapid DNA barcoding methods (Song et al. 2008). Considering that the honeybee genome has a very high frequency of NUMTs among metazoans (Behura 2007; Pamilo et al. 2007), bumblebee samples yielding inconsistent restriction bands on agarose gel should be sequenced to confirm their species identification. Also, we recommend that a subset of samples should always be sequenced to validate the observed PCR–RFLP patterns.

In conclusion, the newly developed PCR–RFLP approach provides a useful tool for species identification from degraded DNA, which can be applied for conservation and management within the subgenus Bombus s. str. Our modified approach is most useful in distinguishing B. lucorum and B. terrestris but also enables to separate most B. magnus and B. cryptarum. Hence, our protocol provides a quick and cost-effective means for species identification from degraded DNA within the subgenus Bombus s.str.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results and discussion
  6. Acknowledgements
  7. References
  8. Data Accessibility

We would like to thank Maria Kakko, Erkki Kaarnama and the staff at TEGlab (University of Turku) for their much needed assistance. Mark Brown and James Carolan kindly provided us with B. magnus reference samples. Our research was supported by Maj and Tor Nessling Foundation, University of Turku Graduate School, and Societas Entomologica Fennica (to S.-R. V.); Estonian Science Foundation (Grant Number 8215 to A.V.); Kone Foundation and Eemil Aaltonen Foundation (to J.S.).

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results and discussion
  6. Acknowledgements
  7. References
  8. Data Accessibility

S.-R.V. designed sampling, designed the study system and the laboratory protocol, performed molecular genetic analysis and wrote the manuscript. A.V. designed the experiment and the laboratory protocol and provided comments on the manuscript. J.S. designed the study system and provided comments on the manuscript. All authors read and approved the final manuscript.

Data Accessibility

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results and discussion
  6. Acknowledgements
  7. References
  8. Data Accessibility

DNA sequences: GenBank Accession nos KC192043 and KC192045KC192047.

Sampling locations and dates for trapping samples used in the PCR–RFLP protocol and sequenced samples; alignment data: DRYAD entry doi: 10.5061/dryad.59g56.