Introduction

Mexico is among the top five honey producing and exporting countries worldwide. Honey quality is determined based on flavor, color and aroma^1,2, which are the result of the plant species in the zone of production. Plant species of a honey production region provides a specific botanical composition to honey, which is unique to that place of origin. Efforts on marketing classified honey based on botanical origin are increasing, particularly in Yucatan Peninsula region³. Yucatan Peninsula accounts for 30 % of total production from which, 90 % is for international markets^4,5. Honey has specific physicochemical and sensorial traits that respond to a broad diversity of plant species in each region. Approximately 2,400 plant species have been recorded nationally, 600 of which are visited by honey bees and 30 of them are important in honey production^6,7,8.

Ninety percent of annual honey production in the region depends on two main nectar sources: Viguiera dentata (sunflower goldeneye; local term is tajonal) from December to February; and Gymnopodium floribundum (dzildzilche; local term is t´sit´silche´) from March to May. Both species bloom during the driest time of year. During the rainy season, from June to November, a large variety of legumes and vines flower, but they account for only 8 % of the total annual honey harvest in the region⁹.

A widely used method to identify among the plant species from which, a honey has been produced is pollen analysis, based on the specific morphology of the pollen^10-14. However, currently there are a number of different methodologies that can complement melissopalynological analysis.

Pollen content in honey is extremely low (trace levels), however, it varies depending on the floral abundance in each zone, and their nectar and pollen contribution, which results in the under- or over-representation of some species. This highlights the need for additional methods that depend less on pollen traces to estimate honey physicochemical parameters and botanical content, such as proteomics¹⁵, optical spectrophotometry and electronic sensors¹⁶, and fluorescence spectrophotometry¹⁷, among others.

An effective method for identifying the botanical source are those based on DNA, suitable for the honey quality regulations used for standardization within the European Union and other countries¹⁸. Defining honey botanical sources with the aim of adding value to the final product requires detection of homologous nucleotide sequences using generic oligonucleotides derived from metabolic genes. These genes are common to all plant species but have amplification patterns highly specific to each plant family.

The objective of the present study was to develop and validate generic oligonucleotide amplification system using quantitative PCR and SYBR^® Green, as well as dissociation curve analysis, in order to characterize the plant species and to identify pollen sources in honey production from the Yucatan Peninsula.

Material and methods

Honey samples. Ten honeycombs with at least 80 % operculation were collected from the apiary at the Mocochá Experimental Station (21°07’N, 89°26’W; 10 m asl) of the National Institute of Forestry, Agricultural and Livestock Research (Instituto Nacional de Investigaciónes Forestales, Agrícolas y Pecuarias - INIFAP) during different flowering cycles. Honey was extracted from the honeycombs using a stainless steel extractor. The extractor was washed with water and dried with 70 % alcohol to prevent contamination between honey samples. After extraction, each honey sample was filtered and stored at room temperature in polyethylene terephthalate (PET) containers. All samples were sent to the Vegetal Tissue Culture and Genetic Engineering Laboratory at the Bajío Experimental Station (INIFAP). Five samples were collected during the 2013 season (M-1, M-3, M-4, M-7 and M-8) and five samples during the 2014 season (M -11, M-12, M-13, M-14 and M-15). Palynology analyses were done of each sample to generate floral origin reference data.

Melissopalynological analysis. The honey samples were acetolyzed following Louveaux¹⁹. Acetolyzed pollen grains identified in the honeys were fixed on slides with glycerinated gelatin. Grain structural characteristics (e.g. shape, size, openings and structural elements) were analyzed under microscope (Motic BA 410) with an image analyzer. Based on their characteristics, pollen grains were classified and identified using the published literature²⁰.

Plant Material. Leaves of five plant species known as the most important in honey production in the region were collected: Viguiera dentata (flowers December to February); Gymnopodium floribundum (flowers February to April); Piscidia piscipula (fishpoison tree; locally known as ja´bin) (flowers from March to April); Acacia angustissima (prairie acacia; locally known as xáax) (flowers from June to August); and Mimosa bahamensis (Bahama mimosa; locally known as sak káatsim) (flowers from July to November).

DNA extraction. Plant DNA extraction was done using a commercial kit (DNeasy Plant Mini; Qiagen ^®) following manufacturer instructions. Honey samples DNA was extracted from 50 g of honey per sample. This amount was distributed evenly among four 50 mL tubes (NALGEN^®), approximately 30 mL double-distilled sterile water added to each, and incubated in water bath at 40 °C for 15 min. Each tube was then agitated by inversion, and centrifuged at 10,000 xg for 15 min until a sediment (i.e. pollen traces) settled. Once pollen traces were obtained, the pollen DNA extraction was performed using protocol published by the European Union Federal Office of Consumer Protection and Food Safety (2011), and modified by the Vegetal Tissue Culture and Genetic Engineering Laboratory. DNA concentration ranged from 40 to 200 ng/L, depending on pollen content.

Primers. Seven previously published consensus sequence primers were used to establish plant species amplification patterns¹⁸, and then to identify these amplification patterns and their dissociation curves in the honey samples: 1) non-coding tRNA-Leu chloroplastic DNA sequence of trnL (UAA) intron, consensus of 19 species including algae, bryophytes, pteridophytes, gymnosperms and angiosperms^18,21 (Plant 1); 2) nested Plant 1 sequence (Plant nest)¹⁸; 3) canola and sunflower actin (Act)¹⁸; 4) sunflower profilin (Helli-all)¹⁸; 5) canola lipase (Brass-lip)¹⁸; 6) sorghum, rice and rye alcohol dehydrogenase (Adh1)¹⁸; and 7) cotton, pepper, tomato, potato and tobacco 3-hydroxymethyl glutaryl CoA (Hmg2)¹⁸.

qPCR Amplification. SYBR^® Green system amplification was done with the following reaction buffer: 1X Fast Master mix SYBR^® Green buffer (Applied Biosystems No. Cat. 4385612); 150 nM generic marker derived primers; and 100 ng DNA. Amplification conditions (equipment: ABI PRISM 7000, Applied Biosystems) were: 1 cycle at 95 °C for 20 sec (pre-denaturation); 40 cycles at 95 °C for 3 sec (denaturation); and 1 cycle at 60 °C for 30 sec (alignment and extension). After amplification, a dissociation curve (fusion temperature curve [Tm]) analysis was run to verify the presence of specific and non-specific products according to the fusion temperature of each reaction^22,23. Each plant and honey DNA sample was amplified in triplicate to establish amplification pattern consistency.

In order to verify homogeneous DNA concen-trations TaqMan^® 18S (human18s rRNA No. Cat. 4319413E, Applied Biosystems) commercial probe was used as internal control, together with the TaqMan^® Universal PCR Master Mix II with UNG (No. Cat. 4440038, Applied Biosystems) mixture. Reaction elements without DNA template were used as a negative control.

Results

The internal control (18S) amplification curves validated the standard DNA concentration and quality for executing the following reactions with the generic metabolic gene primers. The established concentration was 50 ng/reaction, which resulted in amplification between cycles 17 and 20, the least amount of separation between replicates, increasing reliability among reactions (Figure 1).

Figure 1 qPCR amplification curves for the 18S constitutive gene (internal control) of plant DNA (Piscidia piscipula, Gymnopodium floribundum, Viguiera dentata, Mimosa bahamensis and Acacia angustissima) (7000 SDS v 1.2.3 software, Applied Biosystems) 

Amplification of generic genes in studied plant species

Primer Plant 1 exhibited amplification in all the plant DNA samples, but not in the negative control (Figure 2A). Amplification of the Plant nest primer was also observed in all five species. However, differences in amplification cycles were present between different species; that is, Plant nest primer exhibited greater similarity with M. bahamensis, G. floribundum and A. angustissima than with P. piscipula and V. dentata (Figure 2B). Primers Act, Brass-Lip and Helli-all amplified in all plant DNA samples in a range of different cycles (Figure 2C, 2D AND 2F). Primer Hmg2 amplified in M. bahamensis, V. dentata, G. floribundum and A. angustissima, but did not amplified in P. piscipula (Figura 2E). No amplification was observed with the Adh1 primers in any of the five studied plant species.

Figure 2 qPCR amplification curves for the five studied plant species (Piscidia piscipula, Gymnopodium floribundum, Viguiera dentata, Mimosa bahamensis and Acacia angustissima) using the SYBR^® Green system. Primers: A) Plant 1; B) Plant nest; C) Actin; D) Brass-lip; E) Hmg2; and F) Helli-all. NTC = negative control without DNA (7000 SDS v 1.2.3 software, Applied Biosystems) 

PCR amplification product and dissociation curve analysis in plant species

A single curve for each species was observed in the analysis of dissociation curves derived from the Plant 1 amplification, when compared to the fluorescence derivative (maximum curve peak corresponded to product fusion temperature - Tm). Variation below the curve was associated with the number of gene copies in each species (Figure 3A). Temperature and fluorescence values also varied between species with the Plant nest primer and single dissociation curves for each of the five plant species were observed (Figure 3C). This confirms that use of Plant nest reduced the spectrum of similarity between species, making detection more accurate. However, even though all five species showed amplification, the negative control amplified as well. This non-specific amplification may have been caused by primer dimer formation, which made Plant nest primer unsuitable for this purpose.

Figure 3 Comparison of dissociation curve patterns of amplification products generated with five generic gene plant primers in plant species and honey samples. Plant species: A) Chloroplast trnL (UAA) intron (Plant 1); C) Chloroplast trnL (UAA) intron (Plant nest); E) Actin (Act); G) Lipase (Brass lip); I) 3-hidroxymethyl glutaryl CoA (Hmg2); and K) Profilin (Helli all). Honey samples: B) Chloroplast trnL (UAA) intron (Plant 1); D) Chloroplast trnL (UAA) intron (Plant nest); F) Actin (Act); H) Lipase (Brass lip); J) 3-hidroxymethyl glutaryl CoA (Hmg2); and L) Profilin (Helli all) (7000 SDS v 1.2.3 software, Applied Biosystems) 

Dissociation curves obtained with Act primer exhibited a single product in all five plant species, with variations in temperature and fluorescence. This eliminates the possibility of crossed amplification (Figure 3E). Brass-lip primer on the other hand, gave single products for M. bahamensis and A. angustissima (Figure 3G). Curves for G. floribundum, V. dentata and P. piscipula exhibited two or more non-specific curves due to the non-specific binding of the primer to the genome sequences in these three species.

When using Hmg2 primers single curves were observed for A. angustissima and G. floribundum (Figure 3I). The fluorescence values produced for Hmg2 gene were relatively lower than the other genes, reducing the size of the area below the curve. This may have been a result of a considerably lower number of Hmg2 gene copies in the plant species under study, even though there was specificity in the sequences.

Finally, Helli-all primers produced single curves for G. floribundum and V. dentata (Figure 3K). Fusion temperature values differed for M. bahamensis, P. piscipula and A. angustissima, and non-specific or single curves were recorded for these species.

Discussion

Characterization of floral origin in honeys produced in Mexico using primers from generic metabolic gene consensus sequences and dissociation curve analysis provides new analytical support for melissopalynology as part of an initial detection stage.

SYBR^® Green fluorophore tends to adhere to any amplified product, either the target sequence or a non specific DNA sequence, adding up all these signals and producing a single amplification curve. The amplification pattern cannot be used directly to determine specificity, but dissociation curve analysis provides information of a specific amplification. In a dissociation curve, fluorescence decreases gradually as temperature increases²³. By analyzing these dissociation temperature (or fusion) curves, specific products can be identified since curve shape is related to GC (guanine and cytosine) content, and amplicon size and sequence²².

The seven sets of primers from generic gene sequences used here, generated specific amplification patterns even though these sequences are highly conserved in multiple biological families and different kingdoms. The chloroplast trnL (UAA) intron gene, the base for the Plant 1 and Plant nest primers, amplified the DNA samples from all five studied plants. However, this result did not agree with the dissociation curve analysis, which showed non-specificity with the target sequence in the honey samples DNA. One possible explanation is that since pollen is a masculine germinal cell it has low probabilities of containing plastids (cells from the maternal line), thus limiting identification of DNA chloroplastic sequences from honey samples²⁴. These results contrast with Laube et al¹⁸, who established that this multicopy sequence found in the chloroplast gene was adequate for detecting low DNA levels in plants and is therefore apt for detecting plant DNA in honey.

Amplification and dissociation curve analysis with Act primer identified V. dentata in honey sample M-1 and M. bahamensis in sample M-3. Viguiera dentata blooms between December and January, which coincides with the collect date of sample M-1 (24 January 2013), and the palynological analysis (7.5 % V. dentata). This proportion of pollen in honey sample M-1 indicates “minor important pollen” according to the International Apiculture Botanical Commission (Comisión Internacional Botánica Apícola)¹⁹. However, the fact that the Act primer detected such low levels of pollen content indicates that this primer is highly efficient; being difficult to detect these amounts using traditional methods. Two species were detected in the M-3 sample, M. bahamensis (Act and Brass-lip) and G. floribundum (Hmg2 and Helli-all). Nonetheless, it was classified as a M. bahamensis monofloral because this species accounted for 65 % of the pollen grains, making it the predominant pollen in the melissopalynological analysis. Its dominance in this sample coincides with this species’ flowering period and the collect date of honey sample. Brass-lip primer confirmed the presence of M. bahamensis in M-8 with fusion temperatures similar to those for this species, but the area below the curve was less than that detected for M-3, indicating a lower pollen content.

Both Hmg2 and Helli-all primers confirmed the presence of G. floribundum in honey sample M-3. This validates the presence of this species in this honey sample, as well as identifying two generic metabolic genes that produce consistent results, providing additional primers for verification of honey botanical source. Although these genes come from distant plant families (Asteraceae for Helli-all, and Solanaceae for Hmg2), there may be high sequence similarity between the profilin (Helli-all) and 3-hydroxylmethyl glutaryl CoA (Hmg2) genes and their homologues in G. floribundum. For instance, profilin (PRF4) from the Arabidopsis profilins family exhibit greater similarity to the profilins from other species than to the other three Arabidopsis profilins (PRF1, PRF2, PRF3)²⁵. At the amino acid level, PRF4 was 78.6 to 82.4 % identical to the pollen-specific profilins of monocotyledons such as Zea mays (corn), Phleum pratense (Timothy grass), Betula verrucosa (silver birch), Nicotiana tabacum (tobacco) and Triticum aestivum (wheat).

Hmg2 showed a single dissociation curve in honey samples M-14 and M-15, however, this specific product was not observed in any of the studied plant species; its temperature in the dissociation curves was far higher than those for the five plant species. These two honey samples were collected at the end of the flowering period of the principal plant species used in honey production. The sequence similarity of these primers are probably associated with secondary species in this cycle such as Mimosa pudica, Acacia gaumeri, Bursera simaruba, Pithecellobium albicans, Senna pallida, Guazuma ulmifolia, Lysiloma latisiliquum and Leucaena leucocephala, among others.

Overall, the most frequent species in the studied honey samples were V. dentata, G. floribundum and M. bahamensis, which bloom for the longest periods among the five analyzed plant species. Acacia angustissima and P. piscipula were not detected in any of the honey samples, most likely because they bloom for only a very short period during the flowering season.

The results indicate that amplification patterns were successfully established for the five studied plant species based on their florescence derivatives. These patterns could then be compared to the patterns detected in honey samples for pollen source identification. Three of the main species present during the season of honey production were identified. Four of the genes (profilin, lipase, 3-hydroxylmethylglutaryl CoA, and actin) contained useful sequences for plant species identification based on amplification patterns which were then validated on honey samples to estimate the presence of one or more plant species.

None of the five plant species were identified in samples M-7, M-11 and M-12, even though the samples were collected during these species’ flowering period. This was confirmed by the palynological analysis results in which Asteraceae, Malvaceae and Fabaceae species were identified. Similarly, honey samples M-14 and M-15, which exhibited a distinct amplification pattern possibly related to species not included in the present analysis, showed the presence of Lysiloma bahamensis, Mimosa pudica and Acacia gaumeri after the palynological.

In the present results, V. dentata was identified by the genes actin and profilin, M. bahamensis by the genes actin and lipase, and G. floribundum by the genes Hmg2 and profilin. Acacia angustissima was detected via palynological analysis at very low percentages in honey samples M-4 (6.2 %), M-8 (2.0 %) and M-12 (0.5 %). These levels were probably below the detection limit of the markers used in the genetic analysis.

Use of primers other than those used in the present analysis, with a broad similarity in metabolic function, could help to establish amplification patterns for P. piscipula and A. angustissima. In addition, design of new gene amplification primers that do not amplify any of the studied plant species or honey samples (e.g. Adh1) could be part of a strategy to identify other plant species on the Yucatan Peninsula.

Conclusions and implications

The amplification results as well as the analysis of the dissociation curves for the studied plant species and honey samples, demonstrated that a single gene could be used to detect more than one species (e.g. actin and profilin) in the same sample. Two different genes can also detect the same species in the same sample (e.g. actin and lipase for M. bahamensis, Hmg2 and profilin for G. floribundum), thus confirming and reinforcing detection results. Finally, actin and profilin gene sequences linked to metabolic activity of cell wall function can independently identify the same species (e.g. V. dentata) in honey samples harvested at different times. The detection system presented here, using DNA amplification with generic gene consensus sequences, is an effective protocol that can support melissopalynology analysis in identifying pollen content of specific plant species in honey.