smritisingh96.files.wordpress.com€¦  · Web viewConversely, adrenomedullin has higher affinity than CGRP when binding to the AM1 and AM2 receptors, all three of which induce specific

Title Page Template for Biology 395H

Title: Investigation of Apelin Interactions with RAMPs 1, 2, and 3

Student: Smriti Singh

Signature: _________________

Date: December 5, 2017

These signatures verify that a research paper was written and submitted in the semester of record. As required by the university, this paper will be kept on file in the department for 4 years.

Biology Faculty Sponsor: Dr. Gidi Shemer

Signature:_____________________________

Date: _____________________________________

Faculty Research Mentor (PI of lab): Dr. Kathleen Caron

Signature: __________________________

Mentor Department: Cell Biology, Physiology, and Genetics

Date: December 5, 2017

Introduction into the investigation of apelin interactions with RAMP via biochemical and

cellular-based methodologies:

The Caron Lab has had a long-standing interest in studying a specific family of proteins

called Receptor-Activity Modifying Proteins, or RAMPs (1). RAMPs function by modifying the

activity of other receptors, specifically working to regulate the ligand specificity, translocation,

and coupling of G-protein coupled receptors (GPCRs) at the plasma membrane of cells. They are

single-transmembrane proteins that interact with GPCRs and work to manipulate the manner in

which a ligand is able to effectively bind to the GPCR receptor of interest (2). G-proteins, which

couple to GPCRs, function as critical signaling components – almost 30% of all known drugs

target GPCRs, indicating their significance in pharmaceutical applications (3). There are

currently eleven known GPCR-RAMP interactions confirmed (2). The RAMP proteins are of

particular interest due to their heavy involvement in various disease conditions, such as

imbalances in hypothalamic-pituitary systems, asthma, cardiac failures, and skeletal wasting (1).

Many studies have been conducted to confirm the significance of RAMP interactions in

additional pathways in life cycles. GPCRs and RAMPs interact in distinct conformations,

making them ideal to examine as possible drug targets due to their ability to activate a host of

various signaling pathways (ex. Gs, Gq, and Gi), their ability to bind several ligands, and their

ubiquitous expression in the body (2). These sites can be exploited in regulating the function of

therapeutically relevant RAMP-interacting GPCRs (4). The Caron Lab has focused its work on

the lymphatic system, elucidating the mechanism of the action of the adrenomedullin peptide

(ADM) as a lymphangiogenic factor, an endogenous peptide ligand that facilitates the

development and expansion of lymph branches upon a main lymph stalk (5). This process is

similar to that of angiogenesis, a function we are interested in connecting to the apelin receptor

(APJ).

In an effort to understand the effects of RAMP 1, RAMP 2, and RAMP 3 on GPCR

signaling, several labs have begun to uncover different functions through the study of

adrenomedullin and its binding to the calcitonin receptor-like receptor (CLR) (6). The interaction

of RAMPs 1, 2, and 3, with CLR, generate three distinct GPCRs named calcitonin gene-related

peptide receptor, AM1 receptor, and AM2 receptor, respectively. In a CGRP receptor, RAMP1 is

present, and this receptor has a great affinity for the peptide ligand CGRP. Conversely,

adrenomedullin has higher affinity than CGRP when binding to the AM1 and AM2 receptors, all

three of which induce specific effects (the RAMP 2 pathway undergoes Gq and Gs signaling,

whereas the RAMP 3 pathway involves a mechanism for recycling the receptor itself on top of

activation of G-protein signaling). RAMPs are integral in the proper functioning of many GPCR

pathways. Mutants or reduced functioning of RAMPs leads to a host of diseases, including ones

involving regulation in skeletal structures, dysfunction of the endocrine system,

hyperprolactinemia, and fertility reduction (7).

Similar to adrenomedullin as a lymphogenic factor, apelin is another peptide ligand that

functions as an angiogenic factor; its signaling pathway in the cardiovascular system in particular

is of interest (21). Apelin is a member of the apelinergic family of proteins (Galphai, Galphaq,

and Galpha13) and induces multiple signaling pathways, including but not limited to Gi, Gq,

G13 (7, 10). Two peptide ligands – apelin and Elabela/Toddler – are the endogenous apelin

receptor agonists (8, 9). These two ligands have been shown to induce different signaling

cascades through the same receptor, APJ (11). In theory, a pharmaceutical drug could be

produced to affect the affinity of endogenous ligands, decreasing or increasing their ability to

bind to a RAMP-bound APJ receptor. The RAMP-APJ interaction would provide a druggable

target with greater specificity than targeting the endogenous ligands orthosteric pocket. To

determine if RAMPS and APJ are interacting, we will utilize the biochemical assay BRET. Then,

we will utilize immunofluorescence confocal microscopy to confirm their interaction.

Apelin and its role in angiogenesis:

In the heart, the apelin receptor functions like a mechanosensor, essentially allowing

stretching to occur in an apelin-dependent/G-protein manner, through the recruitment of -

arrestin (12) (a protein that works to internalize receptors as well as promote downstream

signaling); studies show that these complexes confer an alternative method of receptor activation

and signaling (13). The APJ receptor is a 7-transmembrane GPCR present in many kinds of cells,

including smooth muscle cells (14), cells of the central nervous system (15, 16), and vascular

endothelial cells (15); its ubiquitous expression in the body makes it a useful protein to study

further.

In mouse models, studies have indicated that apelin knockout mice contain normal heart

morphology and blood pressure (4, 17), but they lack fully-functioning heart contractility (4).

Knockout mice have impaired cardiac contractility with age (4), a notable decrease in exercise

capacity (1), and demonstrate worsened pulmonary hypertension under hypoxia (18). Severe

heart failure can result under high amounts of pressure in these mice, but can be rescued by the

activation of the AT1 receptor, or by infusing angiotensin, a protein that causes the

vasoconstriction of blood vessels to maintain blood pressure in the body (17, 19).

Apelin knockout-mice reveal increased mortality rates in acute myocardial infarction,

with carotid ligation areas becoming reduced (9). Data from knockout mice reveals apelin as a

significant receptor maintaining several aspects of cardiovascular health, as well as promoting

protective mechanisms/pathways when organisms are facing stress. Activating the apelin

receptor promotes many physiological processes, like vasoconstriction and dilation,

strengthening of heart muscle contractility, angiogenesis, and regulation of energy metabolism

and fluid homeostasis (20).

Materials and Methods:

Many techniques were utilized to conduct this study of interactions between RAMPs and

the apelin receptor, including cloning, colony polymerase-chain reactions, bacterial

transformations, extraction of apelin DNA via mini-prep, cell culture, bioluminescence

resonance energy transfer (BRET) saturation examination, and site-directed mutagenesis.

Cloning APJ receptor cDNA into mammalian expression plasmids:

Plasmids containing apelin DNA were created using a renilla luciferase (rLuc) backbone

to generate the APJ receptor-rLuc fusion protein (Figure 1). The rLuc protein serves as the

donor in the BRET experiment. Myc-tag was incorporated into the rLuc vector backbone for

future detection methods. This cloning will result in the expression of the Myc-APJ-rLuc fusion

protein. RAMP plasmids were generated with a human influenza hemagglutinin (HA tag) to

serve as an epitope for antibodies to detect RAMP expression by immunofluorescence confocal

microscopy. The acceptor protein in the BRET assay will be a yellow-fluorescent protein (YFP,

purchased as an eYFP-N1 plasmid from Addgene). The RAMP cDNA was cloned into the YFP

backbone to generate the HA-RAMPX-YFP fusion protein. All cloned vectors were confirmed

through Sanger sequencing by Eton Biosciences. This process is repeated for all three RAMPS

of interest – 1, 2, and 3.

Figure 1. Renilla luciferase (rLuc) was pulled from a jellyfish and forms a fusion protein with apelin at a length of approximately 7200 base pairs. In the plasmid, the apelin gene (of about 1200 base pairs) was inserted with cuts from BamHI and NotI as restriction enzymes. A Myc tag was added prior to the apelin gene sequence to be detected by antibodies in confocal microscopy. Subsequently, a CD33 tag was added before Myc to promote the signaling of this pathway. The plasmid also contains an ampicillin-resistant section, effectively distinguishing these plasmids as ones containing the apelin DNA when plated upon bacteria. For ligations, T7 primer was added (1L), as well as 1L each of the BamHI and NotI restriction enzymes. T4 ligase was added (1L) to seal the components successfully into the plasmid. Once the ligation was effectively generated, PCR’s were conducted at temperatures of about 55 degrees Celsius to promote maximum annealing between the primers and the strands of replication.

Additionally, experimentation has been conducted with the chemokine receptor family to

demonstrate further GPCR interactions between chemokine receptors and RAMPs. Chemokines,

when activated, traffic cells to desired locations within the body – a process called chemotaxis

(21). Furthermore, calcatonin receptor-like receptor (CLR) is utilized as a positive control for the

GCPR-RAMP interaction, interacting successfully with RAMP1, 2, and 3. The beta-2 adrenergic

receptor functions as a negative control, displaying no significant interactions with any of the

RAMP proteins. These controls were established by previous experimentation by the Caron Lab.

Colony Polymerase Chain Reaction (PCR):

While cloning the apelin gene, work was conducted to clone additional GPCR sequences,

including many from the chemokine family, to test these alongside the APJ receptor. Initially,

many chemokine receptors did not successfully ligate – many false positives were received;

chemokine receptors of interest were not successfully inserting into plasmids. Thus, colony PCR

was utilized to confirm the successful cloning of each GPCR of interest, ensuring they were

present in vectors, and further examining where discrepancies were appearing. Additionally,

sequencing was conducted on these samples to confirm the accuracy of the ligation product.

Transforming Bacteria

Once plasmids were successfully created with the apelin receptor, these vectors were

inserted into E. coli bacterial cells through transformations. Tubes with 200 microliters of E.coli

were thawed on ice. Fifty microliters of E.coli was aliquoted into 1.5 microliter tubes; then, 2

microliters of apelin/rLuc or chemokine receptor/rLuc DNA ligation were added to each tube.

The tubes were incubated on ice for approximately 20 minutes. Tubes containing DNA and

bacteria were heat shocked at 42 degrees Celsius for 30 seconds exactly to promote the uptake

the DNA by E. coli cells. After 2-3 minutes of incubation on ice, 300L of SOC outgrowth

medium (containing many beneficial ions to supplement the storage of these cells, including 2%

Tryptone, 0.5% Yeast Extract,10 mM NaCl, 2.5 mM KCl. 10 mM MgCl2, 10 mM MgSO4, and

20 mM glucose) were pipetted into each tube, minimizing air exposure to SOC (3).

Bacterium were then shaken at 37 degrees Celsius at 260rpm for 1 hour. While shaking

occurred, two ampicillin-resistant agar plates were prepared for each transformation. These

plates were labeled lid-down upon the lab bench. Each plate was labeled with the DNA ligation

name, vector, date, and High or Low label (i.e., Apelin ligation, rLuc, 11/01/17, high/low). These

bacteria were plated on both high (180L) and low (70L) plates of luria broth (LB) agar. The

solutions were spread around the plate with sterile glass pipettes specifically blown with a curved

tip, allowing for convenient spreading. Plates were incubated for eighteen hours to induce

growth at 37 degrees Celsius. Then, plates were viewed for colonies; three colonies were picked

for the APJ receptor, grown in liquid LB culture overnight, and Qiagen mini-prep’s were

conducted upon these colonies. Purified DNA concentration was determined using a NanoDrop

Spectrophotometer.

Growing, maintaining, and plating cells

HEK293T cells were grown at 37, 5% CO2, and 5% humidity. Cells were maintained in

complete DMEM with 10% FBS. Cells were passaged at 90% confluence and not used beyond

passage 30.

A 96-well plate was coated with poly-D-lysine, a solution that promotes the adhesion and

successful growing of cells upon the surface. 75,000 HEK293T cells were seeded into a 96-well

plate, each with 8 rows labeled “A-H” and 12 columns. Three plates were grown – one

containing the positive control, CLR, one containing the negative control, Beta-2 adrenergic, and

one containing the gene of interest, apelin - overnight at 37 degrees Celsius, with 5% CO2 and

5% humidity (standard growth conditions).

The following morning, media was changed to 50L of low serum optimem and

incubated at standard growth conditions for two to four hours. Optimem functions as the most

efficient media for lipofectamine – an agent that promotes the formation of vesicles that contain

the RAMP and plasmid DNA – for cellular uptake. The apelin rLuc plasmid is tested for

interaction with the RAMPs (1, 2, and 3) which are sorted in columns 1-4, 5-8, and 9-12,

respectively. Half of each of these sets of columns have the presence of coelontrozine, which

functions as a chemical substrate for the rLuc donor protein, which in turn, releases photons that

can excite the YFP acceptor protein when these two moieties are in close proximity. Yellow

fluorescent protein’s electrons are excited in this process, and the falling of these electrons down

from a higher state of energy to a lower state of energy emits energy in the yellow light

wavelength range (22). Columns containing coelontrozine (columns 1-2, 5-6, and 9-10) will

demonstrate rLuc and YFP reads, whereas columns containing no coelontrozine (3-4, 7-8, 11-12)

will read only YFP emissions. This allows for the determination of a total YFP signal.

Two stocks, lipofectamine 2000/optimem or DNA/optimem, were generated for the

completion of transfecting the cells. The first is an optimem stock containing apelin plasmid

DNA, created to ensure that constant 0.5g of apelin DNA is added into each of the 96 wells.

Three sets of nine tubes were prepared for eight rows in the 96-well plate (an extra tube was

present to account for serial dilutions) (Figure 2).

Figure 2. 96-well plate set-up. 34 microliters of apelin-rLuc DNA were mixed in 866 microliters of optimem (900 uL final volume). 200 microliters of optimem/apelin-rLuc DNA were added to the first tube, and 100uL was added to each subsequent tube in the dilution series. A serial dilution was conducted by moving 100uL from tube 1 to tube 2 and repeating for each tube (ending with tube 7) ensuring that tube 8 would only contain apelin-rLuc and no RAMP-YFP DNA. Calculations were made to determine the volume of each RAMP (1, 2, and 3) to have a final amount of 2.5g of each RAMP-YFP per tube A, and a serially diluted amount ending with 0.039 micrograms per tube G. RAMP 1’s volume was at 23.8 microliters; RAMP 2’s at 22 microliters; and RAMP 3’s at 32.9 microliters. These amounts were added in each of the three sets of eight tubes, to tube 1. This allowed for the distribution of apelin-rLuc DNA and RAMPs down each row, creating a high to low concentration of RAMP-YFP. The final row, H, served as an internal control, demonstrating the “base” level of luminesce from rLuc only (and not the luminescence produced by both rLuc and YFP in conjunction).

A second optimem stock was created to contain lipofectamine2000; 32 microliters were

added to 768L of optimem yielding 1L of lipofectamine2000 stock per well. Lipofectamine

enables vesicle formation to contain apelin DNA, to be absorbed via endocytosis into HEK293T

cells. After 20 minutes, 100 microliters of the lipofectamine/optimem stock was added to each

DNA/optimem tube (8 tubes total, from previously-created apelin-rLuc/optimem stock tubes),

bringing each optimem-DNA tube to 200 microliters. The apelin-rLuc/lipofectamine/optimem

mixture was incubated for 20 minutes at room temperature. Then, 50L of the transfection

mixture were added per well per row. Cells were grown overnight at standard growth conditions

in an incubator.

Reading Plates

The following day, plates were removed from incubation, where growth media was

aspirated from each well. A 1.5mL microcentrifuge amber tube, functional for two plates, was

prepared to create 950L of ice-cold 200-proof ethanol and 50L of coelentrozine were mixed in

the 1.5mL amber microcentrifuge tube. White tape was added to the bottom of each plate, and

upon addition of the coelentrozine, the plates were incubated at room temperature for 10 minutes

in the dark to avoid photobleaching. 90L of DPBS (Dulbecco’s phosphate-buffer saline) was

added into each well. Wells absent of coelentrozine (-) had an additional 10L of DPBS added to

them, to ensure the volumes of all the wells were consistent (100L). Wells marked as

coelentrozine (+) had 10L of coelentrozine added to them. Each plate was monitored for rLuc

and YFP activation using a Mithras microplate reader with the proper band pass filters.

BRET

Resonance energy transfer or BRET utilizes a bioluminescent luciferase enzyme (renilla

luciferase or rLuc) as a donor protein and a fluorescent acceptor protein. The rLuc and YFP

proteins are genetically fused to each of the candidate proteins – in our case, the GPCR apelin-

rLuc and the RAMP-YFP. Interactions between the two fusion proteins can bring the renilla

luciferase and yellow fluorescent protein close enough for resonance energy transfer to occur,

thus changing the color of the bioluminescent emission. In BRET readings, apelin-DNA is held

constant, while the RAMP protein amount increases from rows H-A. This promotes a

“saturation” of RAMP-YFP to GPCR-rLuc protein, as indicated from row H-A (i.e., increasing

RAMP concentrations to constant apelin protein concentrations). A ratio is produced to establish

BRET interactions – the ratio of the fluorescence of the YFP protein over the luminescence of

the rLuc protein in the ceolentrozine (+) wells is used. The ratio reveals how much RAMP-YFP

acceptor protein is being activated by the GPCR-rLuc donor protein. An asymptote in the graph,

as shown by positive control CLR, indicates a binding maxima or Bmax of the RAMP-YFP:

GPCR-rLuc interaction i.e., no additional RAMP-YFP added, will bind the GPCR-rLuc

effectively saturating the protein: protein interaction.

Results:

Figure 3. Successful colony PCR. Lanes 1 and 2 are of colony PCR products from individual bacteria colonies, a chemokine gene CXCR3, at about 1200 bp; lane 8 is the control PCR.

Successful cloning of the APJ receptor cDNA into mammalian expression plasmids

occurred, as well as creation of RAMP-YFP plasmids. Colony PCR was also conducted

successfully to clear up distinguishing cloning errors (Figure 3). HEK293T cells were

successfully plated, grown, and maintained according to instruction upon 96-well plates, with

experimental set-up completed effectively (Figure 2). The reading of the plates via the Mithras

microplate reader also was conducted successfully to generate data to produce isotherm graphs.

Isotherms were run on one set of plates (n=1), though further runs will be conducted in due time.

Discussion:

The first plate’s readings indicated a lack of meaningful interaction between APJ and

each of the three RAMPs (Figure 4G-I). This is compared alongside calcatonin receptor-like

receptor (CLR), utilized as a positive control for the GCPR-RAMP interaction, which interacts

successfully with RAMP1, 2, and 3 (Figure 4A-C). The beta-2 adrenergic receptor functions as

a negative control, displaying no significant interactions with any of the RAMP proteins (Figure

4D-F). These controls were established by previous experimentation by the Caron Lab. No

substantial isotherm, with a rectangular hyperbola, was observed for apelin with any of the 3

RAMPs. Due to this result, the apelin-rLuc DNA construct was sent for additional sequencing

and it was discovered there is a single base deletion that resulted in a frame shift error in the

coding sequence. This frame shift was at the end of the apelin reading frame before the rLuc

sequence and inserted a stop codon, thus truncating the fusion protein.

Due to this occurrence, hopes to repeat this experiment are in place, and site-directed

mutagenesis will be conducted upon the apelin-rLuc expression vector. This process will insert

the missing base of interest and allow for proper expression of the apelin-rLuc fusion protein to

form for the investigation of the RAMP: apelin interaction.

Figure 4. BRET isotherms as shown. CLR (panels A, B, and C) acts as a positive control, showcasing saturated isotherms that indicate thorough interactions between CLR and all three RAMPs. Beta-2-adrenergic data

A

B

C

D

ED

FD

GD

HD

ID

(panels D, E, and F) showcase linear relationships – no isotherm curves are detected for any of the three RAMPs. In these runs of APJ (panels G, H, and I), no significant isotherm is produced.

Future BRET data can be interpreted to furthering questioning. Saturation isotherms that

may indicate a weak interaction, for example, could indicate that the APJ receptor protein

successfully interacts with a specific RAMP of interest, but that the signal is weak due to the

exact conformation of each protein (they may not interact in an optimized, fitting manner). Much

of primary literature points to apelin functioning as a strategic ligand interacting with RAMPs,

but this has not been confirmed in this demonstration, likely due to the mutation that has caused

truncation in the rLuc portion of the apelin-rLuc protein. Further runs of apelin-rLuc with

corrected sequencing will be conducted alongside RAMP-YFP to see if a conclusive isotherm

curve will be produced to support the hypothesis suggested primarily.

Works Cited

1. Klein KR, Matson BC, Caron KM. The explanding repertoire of receptor activity modifying protein (RAMP) function. Crit Rev Biochem Mol Biol. 2016;51(1):65-71.

2. Parameswaran, Narayanan and Spielman, William (2012). RAMPs. New York, NY: Landes Bioscience and Springer Science+Business Media, LLC.

3. Wootten DL, Simms J, Hay DL, Christopoulos A, Sexton PM. Receptor activity modifying proteins and their potential as drug targets. Prog Mol Biol Transl Sci. 2010;91:53-79. doi: 10.1016/S1877-1173(10)91003-X. Review. PubMed PMID: 20691959.

4. Charo, D.N. et al. (2009). Endogenous regulation of cardiovascular function by apelin-APJ. Am. J. Physiol. Heart Circ. Physiol. 297, H1904–H1913.

5. Saaristo, A., Veikkola, T., Tammela, T., Enholm, B., Karkkainen, M. J., Pajusola, K., Alitalo, K. (2002). Lymphangiogenic Gene Therapy With Minimal Blood Vascular Side Effects. The Journal of Experimental Medicine, 196(6), 719–730.

6. 6. Shindo T, Sakurai T, Kamiyoshi A, Ichikawa-Shindo Y, Shimoyama N, Iinuma N, Arai T, Miyagawa S. Regulation of adrenomedullin and its family peptide by RAMP system--lessons from genetically engineered mice. Curr Protein Pept Sci.

2013Aug;14(5):347-57. Review. PubMed PMID: 23745699.

7. 7. Kadmiel M, Fritz-Six K, Pacharne S, Richards G, Skerry T and Caron KM Haploinsufficiency of receptor activity modifying protein 2 (Ramp2) causes reduced fertility, hyperprolactinemia, skeletal abnormalities and endocrine dysfunction in mice. Molecular Endocrinology 2011 Jul; 25(7):1244-53.

8. Yang, Peiran, Maguire, Janet J., and Davenport, Anthony P. (September 2015).Apelin, Elabela/Toddler, and biased agonists as novel therapeutic agents in the cardiovascular system. Cell Press, Volume Trends in Pharmacological Sciences, Vol. 36. pp. 560-567

9. Chapman, Nigel A., Dupré, Denis J, and Rainey, Jan K (2014). The apelin receptor: physiology, pathology, cell signalling, and ligand modulation of a peptide-activated class A GPCR1. Biochem. Cell Biol. Vol. 92. pp. 431-440.

10. APJ Pathway. Kegg Pathway Maps. http://www.kegg.jp/kegg-bin/search_pathway_text?map=map&keyword=APJ&mode=1&viewImage=true

11. Scimia, M.C. et al. (2012) APJ acts as a dual receptor in cardiac hypertrophy. Nature 7411, 394–398.

12. Rajagopal, S. et al. (2010) Teaching old receptors new tricks: biasing seven-transmembrane receptors. Nat. Rev. Drug Discov. 9, 373–386.

13. Kleinz, M.J. et al. (2005) Immunocytochemical localisation of the apelin receptor, APJ, to human cardiomyocytes, vascular smooth muscle and endothelial cells. Regul. Pept. 126, 233–240.

14. Medhurst, A.D. et al. (2003) Pharmacological and immunohistochemical characterization of the APJ receptor and its endogenous ligand apelin. J. Neurochem. 84, 1162–1172.

15. Pope, G.R. et al. (2012) Central and peripheral apelin receptor distribution in the mouse: species differences with rat. Peptides 33, 139–148.

16. Kuba, K. et al. (2007) Impaired heart contractility in Apelin gene- deficient mice associated with aging and pressure overload. Circ. Res. 101, e32–e42.

17. Chandra, S.M. et al. (2011). Disruption of the apelin–APJ system worsens hypoxia-induced pulmonary hypertension. Arterioscler. Thromb. Vasc. Biol. 31, 814–820.

18. Sato, T. et al. (2013) Apelin is a positive regulator of ACE2 in failing hearts. J. Clin. Invest. 123, 5203–5211.

19. Kojima, Y. et al. (2010) Upregulation of the apelin–APJ pathway promotes neointima formation in the carotid ligation model in mouse. Cardiovasc. Res. 87, 156–165.

20. Walsh, David, Lalli, Mark, Kassas, Juliette, Asthagiri Anand, Murthy, Sashi K. Cell Chemotaxis on Paper for Diagnostics (2015).

21. Hanahan, D. Studies on transformation of Escherichia coli with plasmids (1983). J Mol Biol.166(4). pp. 557–580.

22. Xu Y, Piston DW, Johnson CH. A bioluminescence resonance energy transfer(BRET) system: application to interacting circadian clock proteins. Proc Natl Acad Sci U S A. 1999 Jan 5;96(1):151-6. PubMed PMID: 9874787; PubMed Central PMCID: PMC15108.