Biochemical Journal

Research article

The characterization of human adenylate kinases 7 and 8 demonstrates differences in kinetic parameters and structural organization among the family of adenylate kinase isoenzymes

Christakis Panayiotou , Nicola Solaroli , Yunjian Xu , Magnus Johansson , Anna Karlsson


Differences in expression profiles, substrate specificities, kinetic properties and subcellular localization among the AK (adenylate kinase) isoenzymes have been shown to be important for maintaining a proper adenine nucleotide composition for many different cell functions. In the present study, human AK7 was characterized and its substrate specificity, kinetic properties and subcellular localization determined. In addition, a novel member of the human AK family, with two functional domains, was identified and characterized and assigned the name AK8. AK8 is the second known human AK with two complete and active AK domains within its polypeptide chain, a feature that has previously been shown for AK5. The full-length AK8, as well as its two domains AK8p1 and AK8p2, all showed similar AK enzyme activity. AK7, full-length AK8, AK8p1 and AK8p2 phosphorylated AMP, CMP, dAMP and dCMP with ATP as the phosphate donor, and also AMP, CMP and dCMP with GTP as the phosphate donor. Both AK7 and full-length AK8 showed highest affinity for AMP with ATP as the phosphate donor, and proved to be more efficient in AMP phosphorylation as compared with the major cytosolic isoform AK1. Expression of the proteins fused with green fluorescent protein demonstrated a cytosolic localization for both AK7 and AK8.

  • adenylate kinase
  • nucleotide metabolism
  • nucleotide phosphorylation


AKs (adenylate kinases, EC are phosphotransferases that catalyse the interconversion of adenine nucleotides. The AKs are important in cellular energy homoeostasis and the maintanance of high levels of ATP in a combined action of AK activity and mitochondrial oxidative phosphorylation [1]. The presence of several isoforms of AKs in mammalian tissues is already well established [2]. The presence of multiple isoforms is a specific feature of AKs that is only shared with the guanylate kinases among the nucleotide and nucleoside kinases. Seven different AK isoenzymes have been identified in human tissues, and six of them have been thoroughly characterized. The major cytosolic isoform AK1 is present at high levels in skeletal muscle, brain and erythrocytes, whereas the major mitochondrial isoform AK2 is expressed in the mitochondrial intermembrane space of tissues rich in mitochondria such as liver, heart and skeletal muscle. AK3 and AK4 are located in the mitochondrial matrix, with AK3 having its highest expression levels in the liver, heart and skeletal muscle. AK1–AK3 are expressed at higher levels compared with the other AK isoenzymes AK4–AK7, which seem to have more specialized functions in certain cell types. AK4 is expressed at low levels in brain, kidney, liver and heart tissues, and it is reported to contain an N-terminal mitochondrial import sequence that remains uncleaved after import into the mitochondria [3]. AK5 is cytosolic, or both cytosolic and nuclear depending on the transcript variants, with two separate functional domains and, in contrast with the multi-tissue expression profiles of most other AKs, it is expressed almost exclusively in brain [4,5]. As far as AK6 is concerned, fluorescence microscopy revealed a nuclear localization, but there is no confirmed expression profile available [6]. AK7 seems to have a tissue-restricted expression and its activity has been associated with cilia function [7].

The existence of several AKs with different subcellular localizations, expression profiles, substrate specificities and kinetic properties provides evidence of specialized functions in specific cellular processes. The studies of our laboratory aim to obtain a complete picture of all AKs expressed in mammalian cells, and their role in the homoeostasis and synthesis of adenine nucleotides that are required for a variety of cellular metabolic processes. The AKs are also important for DNA and RNA synthesis, and may contribute to the activity of pharmacologically active nucleoside and nucleotide analogues.

In the present study we have expressed and characterized AK7 and determined its substrate specificity, kinetic properties and subcellular localization. In addition, a novel member of the AK family with two functional domains has been identified and characterized, and it was assigned the name AK8. By the identification and characterization of AK7 and AK8 we believe that the identification of members of the AK family of isoenzymes is close to being completed. The results of the present study demonstrate that in addition to differences in tissue expression, subcellular localization and structural organization, the AKs show differences in kinetic parameters with both high and low substrate affinity enzymes.


Cloning and sequencing

The IMAGE clone 4828427 (GenBank® accession number BC035256) for AK7 was purchased from RZPD and the IMAGE clone 5744517 (GenBank® accession number BC050576) for AK8 was purchased from the American Type Culture Collection.

AK7 cDNA was cloned into NdeI/BamHI sites of the pET16b plasmid vector (Novagen) using the primers AK7 forward, 5′- GAAGGTCGTCATATGGCTGAAGAAGAGGAAACTGCTGCTCTCA-3′ with an NdeI site, and AK7 reverse, 5′-CGCGGATCCTCACTGTGCTTCAGGATTGTTCTTGAAGAGATATT-3′ with a BamHI site, and into NheI/BamHI sites of pEGFP-N1 plasmid vector (Clontech) using the primers AK7GFP (GFP is green fluorescent protein) forward, 5′-TCGAGCTAGCATGGCTGAAGAAGAGGAAACTGCTGCTCTCACG-3′ with an NheI site, and AK7GFP reverse, 5′-GTGAGGATCCCGCTCCCAGCGAGCTATCTTC-3′ with a BamHI site.

AK8, AK8p1 and AK8p2 cDNAs were cloned into the NdeI site of pET16b using the primers AK8p1 forward, 5′-GAAGGTCGTCATATGGACGCCACTATCGCCCCGCACCGTATC-3′ and AK8p1 reverse, 5′-GAAGGTCGTCATATGTCACGGGGTGAACGGGGCATTAGTACGATGGTT-3′, where a stop codon (underlined) was introduced; AK8p2 forward, 5′-GAAGGTCGTCATATGGCCCCGTTCACCCCGAGGGTGCTGCTGCTC-3′, where a start codon (underlined) was introduced, and AK8p2 reverse, 5′-GAAGGTCGTCATATGTCAGGGGATTTTCTTGGGCAGGGGATTAAT-3′. AK8p1 forward and AK8p2 reverse primers were used for the cloning of the full-length AK8. For cloning of full-length AK8 into XhoI/EcoRI sites of pEGFP-N1, the oligonucleotides AK8GFP forward, 5′-AGATCTCGAGCTGGAAGTTCTGTTCCAG-3′ with a XhoI site, and AK8GFP reverse, 5′-GTGAGAATTCGGGGGATTTTCTTGG-3′ with an EcoRI site were used. The plasmids were sequenced by MWG Biotech to verify the DNA sequences.

Expression and purification

The plasmids were transformed into the Escherichia coli strain BL21 (DE3) pLysS (Stratagene) and single colonies were inoculated into LB (Luria–Bertani) medium supplemented with 100 μg/ml ampicillin and 34 μg/ml chloramphenicol. Bacteria were grown at 37 °C and protein expression was induced at a D600 of 0.9 with 1 mM IPTG (isopropyl β-D-thiogalactopyranoside) for 12 h at 27 °C. The expressed proteins were purified using TALON Metal Affinity Resin (Clontech) as described by the manufacturer. The size and purity of the recombinant proteins were verified by SDS/PAGE and the protein concentration was determined using the Bradford protein assay (Bio-Rad) using BSA as a standard. The proteins were divided into 50 μl aliquots and stored at −80 °C.

In vitro translation

AK8, AK8p1 and AK8p2 were synthesized with the PURExpress® In Vitro Protein Synthesis Kit (NEB) in the presence of [35S]methionine (PerkinElmer), following the manufacturer's instructions. Then, 2 μl of each reaction mixture and 5 μl of Rainbow 14C-methylated protein molecular mass marker (Amersham) were loaded on to a NuPAGE 4–12% Bis-Tris gel (Invitrogen). The gel was washed with Gel-Dry drying solution, dried with Dry Ease Mini Cellophane (Invitrogen) and used for autoradiographic detection.

Enzyme assays

The substrate specificity of the recombinant AK7, the full-length AK8 and its two domains were assayed by TLC as described previously [8]. The nucleoside monophosphates and triphosphates were purchased from Sigma. [γ-32P]ATP/GTP and [α-32P]UTP/CTP/TTP were purchased from PerkinElmer. All assays were performed in 10 μl volume reactions containing 50 mM Tris/HCl (pH 7.6), 5 mM MgCl2, 1 mM unlabelled NTPs, 0.1 μCi/μl radiolabelled NTPs (3000 Ci/mmol), 1 mM NMPs or dNMPs, and 0.5 μg of recombinant enzyme. After incubation for 1 h at 37 °C, 2 μl of the reaction products were separated on poly (ethyleneimine)–cellulose F chromatography sheets (Merck), soaked in methanol prior to use. The nucleotides were separated in 0.5 M ammonium formate (pH 3.5) for 2 h. TLC sheets were then autoradiographed by phosphorimaging plates (BAS 1000, Fuji Photo Film).

To determine the kinetic properties of AK1, AK7 and AK8, the non-radiolabelled products were separated and quantified by reversed-phase HPLC using a Chromolith™ column (RP-18e, 100–4.6 mm) (Merck) as described previously [9]. All assays were performed in 25 μl volume reactions containing 5 mM phosphate donor, 50 mM Tris/HCl (pH 7.6), 10 mM MgCl2, 5 mM DTT (dithiothreitol) and 0.5 mg/ml BSA. To determine the Km and Vmax values, 25 ng of AK7, 20 ng of AK8 and 10 ng of AK1 and different concentrations of substrate (from 1 μM to 10 mM) were used, and thereafter the Vmax/Km ratios were calculated. Human AK1 was purchased from Abcam.

Cell culture and transfection

HeLa cells were seeded in μ-Slides (Ibidi) and cultured in Dulbecco's Modified Eagle's Medium with 10% (v/v) fetal calf serum (Gibco BRL), 100 units/ml penicillin and 0.1 mg/ml streptomycin. Cells were grown at 37 °C in a humidified incubator with 5% CO2. Plasmids were transfected into cells using FuGENE™ HD transfection reagent (Roche) as described in the manufacturer's protocol. The cells were stained for 30 min with 100 nM MitoTracker Red (Molecular Probes) 48 h after transfection. Cell fluorescence was imaged using a Leica DMI6000B microscope with Leica Application Suite AF software v1.8.


Expression of AK7 and the different forms of AK8

The cDNA predicted to encode AK7 is longer than previously identified sequences of single-domain AKs (Figure 1A). The AK7-encoding cDNA contained an N-terminal 368 amino acid sequence with no clear homology with any known sequence. The full-length AK7 cDNA was expressed and shown to produce a 656 amino acid protein with a molecular mass of ~74 kDa. According to the databases, the gene product was expected to be 723 amino acid residues, but the presence of a stop codon at the end part of the gene product cloned in the present study suggests that there may exist more than one transcript variant of AK7. The gene is localized on chromosome 14 and consists of 18 exons with a total length of 96.67 kb. The AK domain of AK7 showed high conservation with the other single-domain AKs, AK1–4 and AK6 (Figure 1B).

Figure 1 Human AK7

(A) Map of human AK7. (B) Alignment of human AK7 with the other one-domain human AKs (black boxes indicate completely conserved amino acid residues and different shades of grey indicate different levels of conserved residues).

The cDNA of AK8 was identified in the same genomic library in open reading frame 98 of chromosome 9 and its expression produced a 479 amino acid protein. Since the AK8 polypeptide contains two p-loops, we expressed its possible two domains by inserting: (i) a stop codon before the second p-loop to translate only the first domain (268 amino acids) and (ii) a start codon before the second p-loop to translate only the second domain (216 amino acids) (Figure 2A). Our previous studies of other AKs suggest that a short N-terminal amino acid fragment before the p-loop structure is crucial for the correct folding and activity of the enzyme. A start codon was therefore introduced in a way that an N-terminal fragment, consisting of amino acid residues MAPFTP, is present before the p-loop of AK8p2. The two-domain AK protein sequence of AK8 is similar to the previously characterized two-domain sequence of AK5 (Figure 2B). The full-length AK8 and its two domains were synthesized with a cell-free in vitro translation system in order to verify their expression and their sizes that were ~53, 33 and 27 kDa for full-length AK8, AK8p1 and AK8p2 respectively (Figure 3). A phylogenetic tree analysis suggests that AK7 and AK8 are more distantly related to AK1 than the previously characterized AKs (Figure 4).

Figure 2 Human AK8

(A) Map of the two-domain human AKs, AK8 and AK5. (B) Alignment of AK8 and AK5 (black boxes indicate completely conserved amino acid residues and different shades of grey indicate different levels of conserved residues).

Figure 3 SDS/PAGE of the in vitro synthesized proteins

Lane 1, AK8p1; lane 2, AK8p2; and lane 3, full-length AK8; lane M, molecular-mass markers (sizes in kDa).

Figure 4 The human AK isoenzymes

A phylogenetic tree of the human AKs shows that AK7 and AK8 are more distantly related to AK1 as compared with AK2–AK6.

Substrate specificity of AK7 and the different forms of AK8

An initial screening of phosphotransferase activity was performed using TLC. AMP, CMP, GMP, UMP, dAMP, dCMP, dGMP, dTMP and dUMP were tested as phosphate acceptors at 1 mM concentrations with 1 mM of the phosphate donors ATP, CTP, GTP, TTP or UTP. Similar substrate specificity was observed for all of the enzymes investigated. With ATP as the phosphate donor AK7 (results not shown), AK8, AK8p1 and AK8p2 phosphorylated AMP, CMP, dAMP and dCMP, and in addition to that they phosphorylated AMP, CMP and dCMP when GTP was used as the phosphate donor (Figure 5). There was no phosphorylation of substrates using CTP, TTP or UTP as phosphate donors. The phosphorylation reactions were confirmed by using the in vitro translated enzymes to exclude the possibility of bacterial contamination of the recombinant enzyme preparations (results not shown).

Figure 5 Substrate specificity of human AK8 and its two domains

Screening of ribonucleoside and deoxyribonucleoside monophosphate specificity of full-length AK8, AK8p1 and AK8p2. NMPs and dNMPs (1 mM) were used as substrates with 0.5 μg of enzyme. (A) Unlabelled ATP and [γ-32P]ATP was used as the phosphate donor. (B) Unlabelled GTP and [γ-32P]GTP were used as phosphate donors. C1, control without enzyme or substrate; C2, control without substrate. Figures shown here are representative of at least three repeats of each experiment.

Kinetic properties of AK7 and AK8

The substrates and phosphate donors identified to be most efficiently utilized by AK7 and AK8 in the initial TLC screening were used to determine the kinetic properties of the enzymes. AK7 showed the highest affinity for AMP with ATP as the phosphate donor with a Km of 1 μM and a Vmax of 1130 pmol/μg per min (Table 1). The affinity for dAMP and CMP was also high, with a Km of 28 μM and a Vmax of 860 pmol/μg per min for dAMP and a Km of 1.2 μM and a Vmax of 150 pmol/μg per min for CMP. The affinity of dCMP for AK7 with ATP as the phosphate donor was too low for kinetic determinations. The best substrate for AK8 was AMP when ATP was used as the phosphate donor, followed by CMP (Table 2). AK8 was less efficient in dAMP phosphorylation, as compared with AK7, with a Km of 630 μM and a Vmax of 1360 pmol/μg per min. As judged by the TLC assays, AMP was the preferred substrate for both AK7 and AK8 also when using GTP as the phosphate donor, but the affinity of the substrates was too low to obtain accurate values for kinetic determinations.

View this table:
Table 1 HPLC-based kinetic properties of the recombinant human AK7 with ATP as the phosphate donor

Values are presented as the means±S.D. from three independent experiments. NM, not measurable.

View this table:
Table 2 HPLC-based kinetic properties of the recombinant human AK8 with ATP as the phosphate donor

Values are presented as the means±S.D. from three independent experiments. NM, not measurable.

A comparison of the kinetic parameters of AMP as the substrate for AK7 and AK8 with the more abundant AK1 showed that the Km of AK1 was in the millimolar range, whereas the Km values of AK7 and AK8 were in the micromolar range (Table 3). Also, the Vmax of AK1 was in a different range as compared with the Vmax of AK7 and AK8. The ratio Vmax/Km, which is used to define the overall efficiency of an enzyme, was higher for AK7 and AK8 as compared with AK1.

View this table:
Table 3 HPLC-based kinetic properties of the recombinant human AK1 with ATP as the phosphate donor

Values are presented as means±S.D. from three independent experiments. NM, not measurable.

Subcellular localization of AK7 and AK8

The subcellular prediction software used did not indicate any targeting signal within the amino acid sequences of the enzymes investigated. Nevertheless, the highest probability, according to the software, was a cytosolic localization for both enzymes. This was confirmed by in vivo expression of AK7 and full-length AK8 as fusion proteins with GFP, demonstrating that both proteins had a cytosolic localization (Figure 6).

Figure 6 Cytosolic localization of human AK7 and AK8

Fluorescent microscopy of HeLa cells expressing GFP (control), fusion protein of AK7 with GFP and fusion protein of full-length AK8 with GFP. The mitochondria were counterstained with MitoTracker.


Including the results of the present study, the human AK family of enzymes now has eight members. These enzymes have high sequence similarity with distinguishable conserved AK domains and catalyse similar reversible phosphoryl transfer reactions between adenine nucleotides. In the present study, AK7 and AK8 were expressed and characterized. Both AK7 and AK8 were shown to have the highest affinity for AMP as substrate, but both of these enzymes also recognized dAMP, CMP and dCMP as substrates. The preferred substrate of all AKs is AMP and their main phosphate donor is ATP, although some can phosphorylate several other substrates and use other NTPs as phosphate donors. AK1 and AK2 can utilize all NTPs as phosphate donors, but they specifically recognize AMP as substrate and, as we show in the present study, AK1 also phosphorylates dAMP [10]. AMP is also the main phosphate acceptor of AK3, but the phosphate donor specificity is restricted to GTP and ITP [11]. AK4 phosphorylates AMP, dAMP, CMP and dCMP with ATP and GTP as phosphate donors, a pattern of substrate and phosphate donor specificity that AK4 shares with AK5, AK7 and AK8 with some differences in kinetics [3,5]. AK6 can use all phosphate donors to phosphorylate AMP, dAMP, CMP and dCMP [6]. Taken together, the common feature of the different AKs is to participate in AMP conversion and ATP homoeostasis. Whether the additional substrate affinities have any physiological importance has not yet been shown.

There are significant differences in the phosphorylation efficiency among the AKs demonstrated both in the present and earlier studies. When the kinetic properties of AK7 and AK8 were determined, AK1 was included as a reference enzyme since AK1 is the major AK in human cells. Both AK7 and AK8 showed a higher affinity for AMP as substrate, as compared with AK1, but a much lower maximal catalytic rate. When both substrate affinity and maximum velocity of AMP phosphorylation were combined and expressed as Vmax/Km, both AK7 and AK8 showed higher efficiency in AMP phosphorylation as compared with AK1. This feature could compensate for the low expression levels of these enzymes compared with the expression profile of AK1. The high AMP affinity of AK7 and AK8 may relate to specific functions in certain cell types, such as kinocilia-bearing cells where AK7 was initially identified [7]. There may be a special demand on high local ATP production in kinocilia-expressing cells as well as in other cells with similar high-energy-demanding functions. Of the previously characterized human AK4, AK5 and AK6, all of these enzymes have different kinetic parameters that may indicate that they contribute to adenosine nucleotide homoeostasis in different microenvironments and have specific cellular functions [3,5,6].

All previously reported AKs with a single AK domain (AKs 1–4 and AK6) have a size of approx. 200 amino acids, whereas the presently studied AK7 is substantially larger. However, the AK domain of AK7 has approximately the size of the single-domain AKs. The first N-terminal 368 amino acid domain of AK7 does not contain any of the conserved AK sequences, and other indications of its potential function were not suggested by BLAST searches of this N-terminal sequence. Whether this domain encodes a functional polypeptide or not, there is no evidence for any contribution to the AK activity of AK7. In a previous study, human AK5 was identified to have two enzymatically active AK domains and we now demonstrate in the present study the existence of a second human AK with two complete AK domains. This enzyme was named AK8 because it was the eighth human AK isoenzyme identified. Although AK8 has a similar two-domain structure as found in AK5, AK8 is different from AK5 in that it apparently has only one transcript, whereas the AK5 open reading frame has two transcripts, which differ in a 27 amino acid Tag sequence [4]. However, when the two AK domains of AK8 were expressed separately they both showed similar AK activity. The AK8 protein thus may harbour two functionally active domains in the same polypeptide chain, something that is supported by the activity data in our present study. Whether different AK isoenzymes can enhance their combined activity by formation of functional complexes with multiple sequential catalytic domains is an important issue for future investigations. As of today only monomer forms of AKs are known, but future studies may reveal if oligomerization is of physiological significance for regulating the activity of the AK isoenzymes. According to databases and the literature there are at least two AKs in lower species with more than one possible catalytic domain: a Drosophila melanogaster predicted AK that consists of 562 amino acids with two domains and a sea urchin sperm flagellar AK (920 amino acids) with triplicated domains [12]. With the picture becoming complete regarding the members of the human AK family, the question addressed was how many AKs exist in other organisms. E. coli is known to encode one single AK and interestingly, when searching the databases, the number of AK isoenzymes appeared to increase with more complex organisms. According to the databases, Saccharomyces cerevisiae encodes three AKs, Caenorhabditis elegans encodes four AKs and Drosophila melanogaster encodes six different AK isoenzymes.

There are correlations found between AK expression and certain human disorders, although conclusive confirmation of genetic AK alterations as the cause of the disease may still be lacking. In humans, AK1 deficiency caused by a nonsense homozygous mutation has been associated with mild chronic haemolytic anaemia and psychomotor impairment [13]. Reticular dysgenesis, the most severe form of human inborn immunodeficiencies, is characterized by complete absence of granulocytes and lymphocytes and has been linked to mutations of the AK2 gene [14]. The mitochondrial localization of AK2 implies a central role in providing the energy required for the proliferation of haematopoietic precursors and in controlling cell apoptosis [15,16]. Increased AK4 protein levels have been detected in cultured cells exposed to hypoxia and in an animal model of amyotrophic lateral sclerosis, a neurodegenerative disease in which oxidative stress is implicated [17]. A previous study has shown that pancreatic β-cells express AK1 and AK5 which play an important role in regulating K-ATP channel activity and thereby control insulin secretion [18]. Autoantibodies targeted against AK5 have been traced in two cases of non-viral immune-mediated limbic encephalitis [19]. The identification of the nuclear-localized AK6 provided evidence on nucleotide metabolism in the nucleus and the energy supply routes between the mitochondria and the nucleus [20]. The involvement of AK7 in human disease has been reported in recent studies where it was related to the expression of AK7 in tissues rich in epithelium with cilia [7]. The AK8 gene (C9ORF98) was identified to be one of the genes that has a negative regulatory role in epithelial cell migration [21]. Both AK1 and AK7 have been deleted in mouse models and found to have specific phenotypic alterations. In a study of AK1-deficient mice they were shown to have low AMPK (AMP-activated protein kinase) phosphorylation in skeletal muscle, consistent with limited AMP production [22]. In addition, deletion of the AK1 gene accelerated the loss of cardiac contraction on ischaemic challenge [23]. An AK7-deficient mouse model presented pathological signs of microtubular defects, decreased ciliary beat frequency, hydrocephalus, abnormal spermatogenesis, mucus accumulation and acute respiratory responses upon allergen challenge [7]. In summary, the vast literature on AK-associated alterations and disorders supports the importance of AK activity for many vital processes in humans and in other organisms.

The number of AK isoenzymes present in different subcellular localizations and with different tissue and cell expression illustrates the importance of maintaining a sufficient AK activity for proper cell and organism function. All of the eight human AKs that have now been characterized catalyse the reversible reaction ATP+AMP↔2ADP, and their contribution to energy transfer and metabolic signalling has been demonstrated to be central for cell viability and function. The present study contributes to complete the picture of the family of human AK isoenzymes and should be an important contribution to clarify their role in human health and disease.


Christakis Panayiotou was involved in project design, cloning of AK7, AK8, AK8p1 and AK8p2, protein expression and purification, in vitro translation, enzymatic assays and manuscript preparation. Nicola Solaroli was involved in project design, identification of AK7, AK8, AK8p1 and AK8p2 in databases, primer design, cell culture and transfection for the subcellular localization of AK7 and manuscript preparation. Yunjian Xu was involved in project design, technical advice, participation in cloning of AK8, in cell culture and transfection for subcellular localization of AK8 and manuscript preparation.

Magnus Johansson was the co-supervisor of the study and was involved in project design, technical advice and manuscript preparation. Anna Karlsson was the supervisor and was involved in project design, technical advice, manuscript preparation and obtaining funding.


This work was supported by the Swedish Cancer Society [grant number 090432]; the Swedish Research Council [grant number K2011-66X-12162-15-3]; and the Karolinska Institute.

Abbreviations: AK, adenylate kinase; GFP, green fluorescent protein


View Abstract