Comparative Proteomics of Cannabis sativa Plant Tissues
Abstract
Comparative proteomics of leaves, flowers, and glands of Cannabis sativa have been used to identify specific tissue-expressed proteins. These tissues have significantly different levels of cannabinoids. Cannabinoids accumulate primarily in the glands but can also be found in flowers and leaves. Proteins extracted from glands, flowers, and leaves were separated using two-dimensional gel electrophoresis. Over 800 protein spots were reproducibly resolved in the two-dimensional gels from leaves and flowers. The patterns of the gels were different and little correlation among the proteins could be observed. Some proteins that were only expressed in flowers were chosen for identification by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry and peptide mass fingerprint database searching. Flower and gland proteomes were also compared, with the finding that less then half of the proteins expressed in flowers were also expressed in glands. Some selected gland protein spots were identified: F1D9.26-unknown prot. (Arabidopsis thaliana), phospholipase D beta 1 isoform 1a (Gossypium hirsutum), and PG1 (Hordeum vulgare). Western blotting was employed to identify a polyketide synthase, an enzyme believed to be involved in cannabinoid biosynthesis, resulting in detection of a single protein.
The biosynthesis of cannabinoids, a class of terpenephenolic compounds found in Cannabis sativa, is not yet fully known. Cannabinoids are found in all tissues of the C. sativa plant, but the amount in which they are present differs considerably among the tissues.1 Cannabinoids are most abundant in flowers, especially in the glands. This raises the question of whether biosynthesis of cannabinoids occurs in all tissues but in different quantities, or only in one tissue and is then transported to the others. In both scenarios it is assumed that the expression level of the genes involved in the cannabinoid biosynthesis is different among the tissues. In any case, the differential expression of cannabinoid biosynthesis may be used to further clarify this pathway by comparing, on the level of proteins or mRNAs, the tissues with varying amounts of cannabinoids with the tissues that do not produce cannabinoids.
Gene expression can be studied by measuring mRNA levels using methods such as microarrays, serial analysis of gene expression, and real-time polymerase chain reaction. However, studies in yeast revealed the absence of a strong correlation between the abundance of the protein and the corresponding mRNA.2Alternative methods of study involve the use of enzyme assays or proteome analysis to identify expressed proteins. The enzymes known to be involved in cannabis biosynthesis are olivetolic acid prenylase, tetrahydrocannabinolic acid synthase (THCA synthase), cannabidiolic acid synthase (CBDA synthase), and cannabichromenic acid synthase (CBCA synthase).3,4 Though assays are available for several of the enzymatic steps of the cannabinoid biosynthesis, it would be an immense task to purify each of these enzymes for sequencing. Moreover, not all of the steps are known. Thus, proteome analysis (proteomics) seems to be preferable to enzyme assaying in obtaining sequence information from all proteins connected with cannabinoid biosynthesis. The use of proteomics in the study of secondary metabolite biosynthesis has been reviewed by Jacobs et al.5
Proteomics is a new tool used to identify and characterize all proteins expressed in an organism or cell.6Single proteins can be separated using column chromatography or two-dimensional (2D) electrophoresis prior to mass spectrometric (MS) analysis.7 Advanced MS allows ionization of macromolecules such as proteins and peptides.8 Proteins can be identified by matching peptide mass fingerprinting with database sequences or by sequencing whole-length proteins with tandem MS. Peptide fingerprints can be obtained by ionization of the peptides that result from enzymatic digestion, usually by trypsin. Accurate peptide masses of peptide fingerprints can be used for searching matching proteins in the databases resulting in molecular weight search (Mowse) score.9 The peptides themselves can be fragmented using tandem MS resulting in the amino acid sequences.
Thousands of proteins occur in the cell, and to choose and separate the protein responsible for a particular function is not an easy task. Using 2D electrophoresis proteins are separated based on pI and molecular weight (MW), which results in a proteome pattern of the cells or tissues under a certain condition. Proteins involved in the production of metabolites can be studied by comparing producing with nonproducing conditions of the cells or tissues: Proteins that are present in the producing conditions but not in the nonproducing conditions might be involved in the production of the compounds of interest. This comparison can be performed more easily with cell cultures, as they tend to have a less complex matrix than plant tissues. Unfortunately, cannabinoids are not produced by cell cultures. Another option is to compare high-producing tissues, such as flowers, with low-producing tissues, such as leaves. The pI and MW of THCA synthase, CBDA synthase, and CBCA synthase are available (Table 1?1).). Therefore, these proteins might be identified from the tissues of flowers and glands using 2D electrophoresis and confirmed by MS analysis.
Western blot analysis using antibodies against a group of proteins known from other plants, e.g., cytochrome P450 or chalcone synthase, can be combined with 2D electrophoresis in order to choose one or more spots for MS analysis. In the case of cannabinoid biosynthesis, a polyketide synthase called stilbene synthase carboxylate-like (STCSL) enzyme is predicted to catalyze the first step of cannabinoid biosynthesis, the reaction between a molecule of n-hexanoyl-CoA with three molecules of malonyl-CoA yielding olivetolic acid. A serum antibody of chalcone synthase, also a polyketide synthase enzyme, was available to try to identify STCSL in C. sativa.
Here we report a comparison of the proteomes of leaves, flowers, and glands of C. sativa for analyzing the biosynthesis of cannabinoids. We also tried to identify the polyketide synthase involved in the biosynthesis of cannabinoids, STCSL, by means of a combination of 2D electrophoresis and Western blot analysis followed by MS analysis.
MATERIAL AND METHODS
Plant material
Plants of the C. sativa “four-way” cultivar (The Sensi Seed Bank, Amsterdam The Netherlands) were legallly grown from seeds in a protected greenhouse. The flowers were harvested from 18-week-old female plants and stored at −20°C. The glands were isolated from the harvested flowers according to the method developed by Hammond and Mahlberg.10
Protein extraction
The proteins were extracted by trichloroacetic acid/acetone precipitation based on the method developed by Granier.11 Frozen flowers (0.5 g) cooled with liquid nitrogen were ground in a mortar with a pestle. Five milliliters of cold (−20°C) 10% trichloroacetic acid in acetone containing 0.07% β-mercaptoethanol was added to the powdered sample. The sample was kept at −20°C to allow complete precipitation. The sample was centrifuged (15 min at 4000 rpm) in a Heraeus centrifuge, than washed three times with 5 mL acetone containing 0.07% β-mercaptoethanol. The precipitate was lyophilized for 1 h and suspended with 2 mL lysis buffer (9.5 M urea, 2% CHAPS (3[(3-cholamidopropyl)dimethylammonio] propanesulfonic acid), 1% dithiothreitol, and 0.8% carrier ampholytes from Amersham Biosciences, Uppsala, Sweden). The protein concentration was determined according to Peterson.12 A smaller amount of the gland (0.1 g) and proportionally smaller amounts of solvent were used for protein extraction with the same method as for flowers.
2D Electrophoresis
Approximately 50 μg and 150 μg of protein were brought to a total volume of 125 μL and 375 μg and loaded on 7-cm and 18-cm IPG strips (Amersham Biosciences) consecutively. Two types of IPG strips were used, pH gradient 4–7 and 3–10. Isoelectric focusing was performed at 20°C for 80 kVh on an IPGphor (Amersham Biosciences). For SDS-PAGE, first IPG strips were equilibrated for 2 × 10 min by 6 M urea, 30% glycerol, 2% SDS in 0.05 M Tris-HCl containing 1% DTT and 4% iodoacetamide. SDS-PAGE (12% T, 2.6% C) in Laemmli buffer system13 was performed at 20°C and 100 V with MINIPROTEAN II system (Bio-Rad, Hercules, CA, USA) using Power Pac 300 (Bio-Rad) power supply. Low molecular weight marker proteins (Amersham Biosciences) were applied on the gel via a small piece of filter paper. The gels were silver-stained according to a modification of Blum et al.14 After staining, the gels were scanned with a UMAX Powerlook III-scanner. The gel images were analyzed with the software package ImageMaster 2D Elite v3.01 (Amersham Biosciences) in order to calculate the number of spots and for comparison analysis.
Protein Digestion
Gel pieces containing the protein of interest were cut out from the stained gel. They were cut into small pieces (1 mm2) and transferred to 0.5-mL tubes. Samples were then digested in-gel with trypsin according to the procedure of Shevchenko et al.7 with slight modifications. The excised gel pieces were washed with 100 μL of 100 mM NH4HCO3 for 5 min and then dehydrated in 100 μL of acetonitrile for 10 min. One hundred microliters of reduction buffer (10 mM dithiothreitol in 100 mM NH4HCO3) was added and incubated for 30 min at 56°C. After cooling to room temperature, 150 μL of alkylation buffer containing 55 mM iodoacetamide in 100 mM NH4HCO3 was added and incubated at ambient temperature in the dark for 20 min. The gel pieces were washed with 100 μL of 100 mM NH4HCO3 for 5 min and dehydrated with 100 μL of acetonitrile for 10 min; this procedure was repeated a second time. After drying under vacuum, the gel pieces were rehydrated in a digestion buffer containing 50 mM NH4HCO3, 5 mM CaCl2, and 12.5 ng/μL trypsin (Promega, Madison, WI, USA) on ice. After 45 min, the supernatant was removed and replaced with 10–20 μL of the same buffer but without trypsin, and incubated for 16 h at 37°C. The resulting tryptic peptides were subsequently extracted for 10 min in an ultrasonic bath by addition of 10–15 μL of 25 mM NH4HCO3, acetonitrile, 5% formic acid, and acetonitrile, respectively. Pooled extracts were dried using a SpeedVac and the extracts were re-dissolved in 10 μL of 0.1% trifluoroacetic acid and purified with Zip-Tip C18 pipette tips (Millipore, Billerica, MA, USA), using the procedure recommended by the manufacturer. The peptides were eluted from the tip directly onto the matrix-assisted laser desorption/ionization (MALDI) plate with matrix solution of cyano-4-hydroxycinnamic acid saturated in 50% acetonitrile, 0.1% trifluoroacetic acid.
MALDI-ToF-MS Analysis and Database Searching
Matrix-assisted laser desorption/ionization time-of-flight (MALDI-ToF) experiments were performed on a Voyager-DE STR mass spectrometer (Applied Biosystems, Framingham, MA, USA) equipped with delay ion extraction. Data were acquired in the delayed ion extraction mode using a 20-kV bias potential, a 6-kV pulse, and a 150-ns pulsed delay time. Mass spectra were obtained over a mass range of 600–4,000 Da.
For identification of proteins, the peptide mass fingerprinting data were used to search in databases NCBInr.20030905 for viridiplantae using Mascott program (http://www.matrixscience.com/cgi/index.pl?page=../home.html). The peptide mass fingerprinting of the proteins were scored with the Mowse score.
Western Blot Analysis
Proteins separated by 2D electrophoresis were blotted onto an Immobilon-P transfer membrane (Millipore). The blotting was performed using a BioRad TransBlot Electrophoresis Cell Apparatus for 1 h at 4°C at 100 V, according to the manufacturer’s instructions. After blotting, the dried membrane was incubated for 1 h in blocking solution consisting of 1% BSA in phosphate-buffered saline tween (PBST: 10 mM NaH2PO4, 150 mM NaCl adjusted to pH 7.2, and 0.5% Tween-20). The chalcone synthase antibody was bound to the membrane by incubation for 1 h using a 1/1000 dilution in PBST solution. The membrane was then washed for 2 × 10 min in PBST solution and for 2 × 10 min in blocking solution. Detection of the protein on the membrane was done using alkaline phosphatase conjugated anti-rabbit IgG antibody (Promega) at a 1:5000 dilution in PBST solution. After 30 min incubation, the membrane was washed 2 × 10 sec. The membrane then was exposed with staining solution until the signal reached the desired contrast. Staining solution consists of 200 μL NBT/BCIP stock solution [18.75 mg/mL 5-bromo-4-chloroindoxyl phospate (Sigma-Aldrich, St. Louis, MO, USA), 9.4 mg/mL 4-nitro blue tetrazolium (Sigma-Aldrich) in 67% DMSO], and 250 μL 1 M MgCl2 in 10 mL TBS buffer (0.1 M Tris, 0.1 M NaCl, pH 9.5). All the reactions were performed in sealed plastic bags using 5 mL solution.
RESULTS AND DISCUSSION
Comparison of Proteins Expressed in Cannabis Leaves and Flowers
Cannabinoids are more accumulated in flowers than in the leaves.1 Some proteins are expressed more in flowers than in the leaves as well. These proteins might be involved in cannabinoid biosynthesis, and identified by using MS in combination with genome database searching.15,16 No studies have yet been performed on the C. sativa genome, but the genomes of the model plants Arabidopsis thaliana and rice (Oriza sativa) have been completely sequenced and could be helpful.
The proteins of flowers and leaves were separated using 2D electrophoresis. We started using 18-cm IPG strips, pH 3–10, for the first dimension and then followed with SDS-PAGE (19 × 15 cm) for the second dimension. Using this broad pH range we expected to obtain an overall picture of all major proteins.
In Figure 1?1 we observe that the pattern of leaf proteins in the gel is more crowded than that of the flower proteins. Both gels show more than 800 individual proteins, but it seems that the leaves contain a greater variety of proteins than do the flowers. We found several protein spots that appear in the flowers but not in the leaves. We checked some of these proteins using MALDI-MS and obtained peptide fingerprints of these proteins.
For peptide fingerprint database searching we set the following criteria: Coverage of the mature protein by the match must reach a minimum of 15%, at least four independent peptides should match, mass tolerance is 1 Da, and maximum number of missed tryptic cleavages is 1. The modification parameters were oxidation of Met and modification of Cys. Table 2?2 shows the results of searching the peptide fingerprint database. We can see that none of the analyzed proteins is related to known cannabinoid biosynthesis proteins. Even the Mowse scores are at less than significant (![[less-than-or-eq, slant]](https://www.ncbi.nlm.nih.gov/corehtml/pmc/pmcents/les.gif)
64) levels. Most proteins that have the highest score are proteins found in A. thaliana or O. sativa, two plants for which the genome has been sequenced. This proves that genomic data are essential for protein identification using proteome analysis.
Comparison of Proteins Expressed in Cannabis Flowers and Glands
In our comparison of the proteome of leaves and flowers we did not find proteins involved in cannabinoid biosynthesis. Therefore, we moved to a more specific part of the flower, the glands. This part of the plant has the highest levels of cannabinoids and the biosynthesis of cannabinoids is thought to take place in the glands. We compared the 2D gel proteome pattern of flowers with that of the glands. We expected that some proteins in the flowers would overlap with proteins of the glands, as we used flowers that included the glands.
The number of protein spots in 2D gels of the flowers and the glands are significantly different (Fig. 2?2).). Fewer than 100 proteins were detected in the pI 3–10, 7.4 × 6.8-cm gel of the glands, compared with more than 300 proteins detected for the flower in the same size gel. The lower number of proteins present in the glands helped us study the majority by MS. The proteins of the glands primarily have a pI from 4 to 7. Therefore, for MALDI-MS, most proteins were excised from a pI 4–7 gel to provide optimum separation.
Fifty-five proteins spots of the gland were targeted for identification. In Table 3?3 we can see the results of peptide fingerprint database searching (only spots with a Mowse score greater than 40 are shown). Only three protein spots have a Mowse score greater than 64 (significant). They are spots 3, 11, and 21, and identify as F1D9.26-unknown prot. (A. thaliana), phospholipase D beta 1 isoform 1a (Gossypium hirsutum), and PG1 (Hordeum vulgare), respectively. Unfortunately, these proteins are not known to be involved in or related to cannabinoid biosynthesis. Spot numbers 46–55 are located in the pI and MW range of known cannabinoid biosynthesis proteins. Once again, the highest Mowse scores are obtained from proteins of A. thaliana and O. sativa; however, in most cases these are proteins with unknown functions. None of the peptide fingerprints of the selected spots matches with peptide fingerprints of THCA synthase, CBDA synthase, and CBCA synthase. This result might be caused by the fact that the proteins involved in the cannabinoid biosynthesis are expressed at very low levels even in the glands. Most of the protein spots that we identified are involved in primary metabolism. Studying proteins involved in the secondary metabolism such as cannabinoid biosynthesis remains a major challenge because of the species-specific enzymes (and genes) involved.
Western Blot-Guided Protein Identification
To determine if any cannabinoid biosynthesis protein is present we decided to detect it by means of an antibody against chalcone synthase. We expected to identify one or more polyketide synthase proteins in the gels of the leaf, flower, and gland proteomes.
The first step of cannabinoid biosynthesis is olivetolic acid formation.17 This step is thought to be catalyzed by STCSL enzyme, a polyketide synthase. Many sequences of polyketide synthase enzyme have been submitted to the databases both on the gene and protein level, such as chalcone synthase, stilbene synthase,15 and stilbene carboxylate synthase.18 Among the polyketide synthases, the sequence homology is high (70%).15 Therefore, the antibody of one polyketide synthase might be used as a tool for identifying other polyketide synthases in the 2D gel electrophoresis.
We used proteins extracted from leaves and a Pinus sylvestris chalcone synthase antibody in order to identify polyketide synthase proteins in C. sativa. We believe that in C. sativa more than one polyketide synthase is present. Therefore we expected to detect more than one spot in the 2D gel electrophoresis with the antibody.
Figure 3?3 shows the 2D gel electrophoresis of leaf protein using IEF pH 4–7 and SDS-PAGE (7.4 × 6.8 cm). Only one spot gave a positive test in the Western blot. We performed the same experiment using proteins extracted from the gland, and again we detected only the same protein. This protein has a MW of approximately 45 kDa and pI 6.5. The size of this protein is similar to that of other polyketide synthases such as chalcone synthase19 and valerophenone synthase20 of Humulus lupulus.
The protein spot in the gel was subject to trypsin digestion and the peptides analyzed by MALDI-ToF. Table 4?4 shows the result of database searching of peptide mass fingerprinting data. There was no polyketide synthase matching with this protein spot. Because we used a very narrow gel (7 cm in width), this result may be due to overlapping with another much more prominent protein. We also divided the spot into four parts prior to trypsin digestion and did MS analysis separately. All parts of the spot gave the same result.
Overall, it appears that the cannabinoid biosynthesis genes are not easily found by using proteomics. The first difficulty is the lack of the genome sequence of C. sativa. But even with this sequence available, it is not certain whether the levels of the proteins are sufficiently high to be observed. The chalcone synthase-reactive protein spot is an indication of this problem, as within that spot no peptides were detected that match with polyketide synthase proteins. Most likely it is overlapped with a much more abundant primary metabolism protein.
Acknowledgments
The authors would like to thank to Prof. J. Schröder (University Freiburg, Germany) for his generous gifts of antibodies against P. sylvestris chalcone synthase. We thank Mr. W. Snoeijer for growing the cannabis plant. This work was supported financially by the Quality Undergraduate Education project, Chemistry Study Program, Gadjah Mada University, Department of National Education Republic of Indonesia.
