Fluorescent sensor proteins for intracellular metal imaging
Citation for published version (APA):Vinkenborg, J. L. (2010). Fluorescent sensor proteins for intracellular metal imaging. Technische Universiteit Eindhoven. https://doi.org/10.6100/IR672725
DOI:
10.6100/IR672725
Document status and date: Published: 01/01/2010
Document Version:
Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)
Please check the document version of this publication:
• A submitted manuscript is the version of the article upon submission and before peer-review. There can be important differences between the submitted version and the official published version of record. People interested in the research are advised to contact the author for the final version of the publication, or visit the DOI to the publisher's website.
• The final author version and the galley proof are versions of the publication after peer review.
• The final published version features the final layout of the paper including the volume, issue and page numbers.
Link to publication
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal.
If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please follow below link for the End User Agreement:
www.tue.nl/taverne Take down policy
If you believe that this document breaches copyright please contact us at: openaccess@tue.nl
Fluorescent sensor proteins for intracellular metal imaging
PROEFSCHRIFT
ter verkrijging van de graad van doctor aan de
Technische universiteit Eindhoven, op gezag van de
rector magnificus, prof.dr.ir. C.J. van Duijn, voor een
commissie aangewezen door het College voor
Promoties in het openbaar te verdedigen
op maandag 10 mei 2010 om 16.00 uur
door
Jan Leendert Vinkenborg
Dit proefschrift is goedgekeurd door de promotor:
prof.dr. E.W. Meijer
Copromotor:
dr. M. Merkx
Cover design: Chris Weel
Printing: Wöhrmann Print Service, Zutphen
A catalogue record is available from the Eindhoven University of Technology Library ISBN: 978-90-386-2206-4
1
Chapter 1 Intracellular imaging of Zn2+ homeostasis 3
Chapter 2 Optimization of the FRET-based Zn2+ sensor ZinCh-9
yields the first genetically encoded Cd2+ sensor
33
Chapter 3 Enhanced sensitivity of FRET-based protease sensors by
redesign of the GFP dimerization interface
63
Chapter 4 Determination of cytosolic zinc levels using FRET sensors
based on conformational switching
75
Chapter 5 Imaging of zinc homeostasis in pancreatic β-cells using
protein-based FRET sensors
99
Chapter 6 Mag-FRET-1 and 2, the first genetically encoded Mg2+
sensors 125 Chapter 7 Epilogue 155 List of abbreviations 171 Summary 173 Samenvatting 175 Curriculum Vitae 179 List of publications 180 Dankwoord 181
Chapter
1
Intracellular imaging of Zn
2+homeostasis
Abstract. The transition metal Zn2+ poses an interesting dilemma to living cells, as it is essential for
numerous cellular functions, yet at the same time toxic in its free form. As a result, the cell keeps the
free concentration of cytosolic Zn2+ under tight control using zinc importers, exporters and Zn2+‐
storage proteins. This chapter gives an overview of the biological roles of Zn2+ in the human body, the
proteins involved in maintaining adequate intracellular Zn2+ levels and the pathologies emerging upon
dyshomeostasis of this metal ion. To gain more information about the mechanisms involved in zinc
homeostasis, tools are required that allow intracellular detection of the free Zn2+ concentration with
high spatiotemporal resolution. Synthetic Zn2+ sensors, consisting of a Zn2+‐chelating unit attached to
a fluorescent reporter, can be easily loaded into the cells and have been successfully used to monitor
intracellular changes in Zn2+ levels. An alternative approach involves the use of genetically encoded
sensors that consist of a Zn2+‐binding protein domain fused to a fluorescent protein domain. Such
protein‐based sensors are attractive because they do not require cell‐invasive procedures, their concentration can be tightly controlled and they can be targeted to different locations in the cell.
Attempts have already been made to create protein‐based sensors for Zn2+, but their application for
intracellular imaging has been limited. Therefore the primary goal of the research described in this
thesis was the development of a genetically encoded FRET sensor to monitor intracellular free Zn2+
levels.
Part of this work has been published in: Vinkenborg, J.L., Koay M.S., Merkx M. Curr. Op. Chem. Biol. 14: 231‐237(2010).
4
Cellular transition metal homeostasis
The unique chemical properties of transition metal (TM) ions such as iron, zinc and copper render them indispensable catalytic and structural components of numerous proteins involved in a wide variety of cellular functions, including respiration, cell signaling, electron transfer, and neurotransmission1‐3. The total cellular concentration of these ‘trace’ metal ions can easily reach 10‐100 M4, yet at the same time transition metal ions are also known to be highly toxic in their free aquatic form. Redox active metal ions such as iron and copper are potent catalysts for the formation of reactive oxygen species (ROS) leading to DNA damage and apoptosis5,6. Zinc is not redox active but its high affinity for a variety of protein side chains make Zn2+ a promiscuous protein binder that, even at nanomolar concentrations, can result in protein aggregation or enzyme inhibition7‐9. Consequently, TM homeostasis is a delicate and highly regulated balance that involves an intricate network of controlled uptake, storage, secretory and regulatory pathways8,10. Much of what we know about TM homeostasis has been inferred from the in vitro characterization of individual components such as transcription factors and chaperones. However, to understand the dynamics and adaptability of cellular TM homeostasis, tools are required that allow measuring of the actual TM concentration and changes therein at the level of a single living cell with high spatiotemporal resolution. Such tools would permit one to address key questions regarding the possible misregulation of TM homeostasis in the pathophysiology of diabetes, cancer and several neurodegenerative diseases6,11,12.The main focus of this thesis is the development and characterization of several sensors for the transition metal ion Zn2+, together with their application in live cells. In addition, we also report on genetically encoded sensors for Cd2+ and Mg2+. This chapter gives an introduction on the role of Zn2+ in health and disease, the proteins involved in Zn2+ homeostasis and on the available fluorescent tools to monitor this metal ion.
Coordination chemistry of Zn
2+Zinc is the second most abundant transition metal in the human body and is essential for hundreds of proteins by playing either a structural, catalytic or messenger role13,14. Why
5
does the human body specifically use zinc in such a variety of processes instead of other common metal ions? First, zinc is redox inactive, which is crucial for its role in nucleic acid binding15. Metal ions such as iron and copper are redox active, which could result in radical formation and subsequent damage of DNA16,17. Second, Zn2+ lacks a preferred coordination number, allowing it to readily form 4‐, 5‐ or 6‐coordinate complexes in proteins. This versatility, combined with the Lewis acid characteristics of zinc has contributed to its widespread use as a catalytic cofactor.
In terms of hard‐soft acid‐base theory, Zn2+ is regarded as a borderline acid and thus Zn2+ can interact with a variety of ligands such as sulfur from cysteine, nitrogen from histidine and oxygen from glutamate and aspartate18. In this respect, Ca2+ and Mg2+ have a much ‘harder’ character, resulting in a strong preference for ligands such as the oxygen from glutamate. When we take a look at the relative strength of the interactions between metals and their ligands, the S‐M2+ bond almost has a covalent character and is relatively strong compared to the O‐M2+ bond, which is based more on electrostatic interactions. As a result S‐ M2+ interactions, such as between cysteines and Zn2+ are much more suitable for high affinity structural sites from which metal release is undesirable19,20, whereas the O‐M2+ bond (e.g. between a glutamic acid and Ca2+) allows the fast exchange required during signaling. Indeed, cysteines and histidines are the most prevalent amino acids in structural zinc sites (Figure 1A)21. In tetra‐coordinated structural sites, the Cys4 binding sequence is by far the most common, but the Cys2His2 zinc‐finger motif is probably most widely known. This sequence was discovered in the protein transcription factor IIIA (TFIIIA)22 and has been found in many other proteins since23,24, including other transcription factors25, the tumor suppressor p5326, the steroid‐thyroid hormone receptor superfamily27 and many others. Catalytic sites show a more heterogeneous distribution over the different binding sequences compared to structural sites, as the Zn2+ ion should be available for catalytic activity (Figure 1B). The first catalytical site was found in carbonic anhydrase in 194028, followed by sites in hydrolases, metalloproteases, alcohol dehydrogenases and lyases18. In cocatalytic sites, two or more metal ions are located in close proximity of each other and both ions are required for catalytical activity13. Although a few of these sites only contain zinc ions, typically zinc is found in combination with copper, iron or magnesium13. Examples of enzymes that contain a
6
cocatalytic site include phosphatases29, aminopeptidases30,31, superoxide dismutases32 and ‐ lactamases33.
A: structural proteins B: catalytic proteins
A: structural proteins B: catalytic proteins
Figure 1: Prevalence of Zn2+ binding amino acid motifs in structural or catalytic zinc binding sites. Tetra‐coordinated zinc binding patterns (ZBPs) in structural proteins (A), and catalytic proteins (B). The zinc ligands are presented in order of appearance in the sequence (from N to C terminal).
Biological functions of Zn
2+The total content of zinc in human adults is between 2 and 3 grammes34. Different amounts of daily Zn2+ intake are required depending on age, but adults on average require 10‐15 mg per day 34. The ingested zinc is absorbed by epithelial cells in the jejunum and ileum, assisted by the presence of glucose in the intestinal lumen. Zinc is excreted through the gastrointestinal tract35, secretion from the proximal tubule36 and sweat37, resulting in a total loss of 3‐6 mg per day that is replenished by dietary intake. In human blood plasma, the total zinc content is ~1 mg per liter. Most of this zinc is bound to proteins: either tightly to 2‐ macroglobulin (~30%)38 or loosely bound to other proteins such as albumin39. The loosely bound zinc is ready for uptake by individual organs or tissues when needed. Due to the widespread use of zinc in structural and catalytic sites of proteins, zinc is essential in many physiological processes in the human body, including retinal function40, spermatogenesis41, functioning of T‐cells in the immune system42, insulin secretion from the pancreas12 and neurological activity43. The role of zinc in the latter has been the subject of intense research for many years44 and will be briefly described here.
Zinc enters the brain via the blood‐brain and the blood‐cerebrospinal fluid barrier after binding to L‐histidine45, followed by uptake in neuronal cells. During neurotransmission, neuronal vesicles containing high concentrations of zinc (> 1 mM46) can
7
undergo exocytosis, thereby releasing zinc into the surrounding milieu and making it available for uptake in neighboring cells (Figure 2). Concomitantly with zinc, glutamate is also released, resulting in the term ‘gluzinergic’ neurons47,48. Such neurons are not distributed uniformly in the brain, but instead can be found mostly in the cerebral cortex and in the limbic structures of the forebrain, including the amygdala and hippocampus, from which they form a network into surrounding parts of the brain49,50. Detection of zinc using histochemical staining in these surrounding parts implies that zinc might play a role in various neuromodulatory effects, including motor coordination51, vision52, auditory stimuli47 and, cortical plasticity53,54. The latter represents changes in neuronal organization involved in neurons that are responsible for learning and memory. Various studies have focused on targets for synaptically released zinc, resulting in a number of postsynaptic channel proteins such as nicotinic receptors55, serotonin receptors56 and voltage‐gated channels57 whose activity might be altered due to Zn2+ binding. In addition, zinc has been shown to be a strong inhibitor of ‐aminobutyric acid (GABA) receptors and N‐Methyl‐D‐Aspartate (NDMA)
receptors58. However, the results of this inhitory effect are under debate, as forebrain neurons have been shown to be more excitable59, less excitable60 or unaffected58 by presynaptically released zinc. Similar controversies can be found regarding the role of zinc in cognitive function. Dietary zinc appears to have a positive effect on psychosocial functioning, yet zinc has also been suggested to reduce neurite outgrowth and branching61, which are both essential processes for synaptic plasticity62. These controversies regarding the role of zinc in neurotransmission clearly illustrate that more research is required to elucidate the exact role of zinc in synaptogenesis and dendritic branching.
8
Figure 2: Zinc trafficking at the gluzinergic synapse. Zinc enters the synaptic vesicles via the zinc
transporter ZnT3 and is stored in these vesicles together with glutamate. Neuronal stimulation results
in exocytosis of these vesicles and thus release of Zn2+ and glutamate in the synaptic cleft, where it can
act on membrane proteins on the postsynaptic cleft, thereby altering the activity of these proteins.
Postsynaptic targets of Zn2+ appear to include GABA receptors, NMDA receptors, voltage‐gated Ca2+
channels or other ion channels that have not been well defined yet. Picture obtained from 44.
Zinc related diseases
Considering the wide variety of functions in which Zn2+ plays a role, it is not surprising that Zn2+ deficiency can lead to a broad range of pathologies63. Limited dietary zinc has been shown to lead to growth retardation64,65, diarrhea66, eye and skin lesions67, impotence68 and delayed sexual maturation69. In addition to deficiency, zinc overload can also adversely affect the human body, leading to neuronal excitotoxity, oxidative stress and ischemic cell death43,44. Both zinc deficiency and overload have been reported to induce apoptosis70‐72, showing that to maintain proper cell functioning, zinc homeostasis must be under tight control (Figure 3).
9
Zinc overload
Zinc deficiency Apoptosis
Amyloid plaque formation Excitotoxicity Ischemia Oxidative stress
Apoptosis
Amyloid plaque formation Retarded growth Diarrhea
Skin and eye lesions Impotence
Delayed sexual maturation
Figure 3: A wide range of pathologies are related to zinc overload and zinc deficiency, emphasizing the
importance of tight control over intracellular zinc homeostasis by the human body. Picture adapted from 44.
Zinc has also been implicated to be directly or indirectly involved in the development of numerous neurological diseases including schizophrenia, amyotrophic lateral sclerosis, depression, epilepsy, ischemia and Alzheimer’s disease (AD)1,6. Zinc appears to play an important role in AD, as high concentrations of zinc are observed in neuritic plaques and amyloid deposits in both humans and mice73,74. In addition, the interactions between zinc and amyloid ‐protein (A) have been shown to lead to aggregation into amyloid plaques in
vitro75. However, zinc might also have a protective role in the development of AD, as A
accumulation appears to be enhanced by oxidative stress76,77, while the redox activity of A peptides is quenched in presence of zinc 78,79. The total amount of Zn2+ in the AD brain is a matter of debate, as several studies reported increased zinc levels78,80‐82, while others observed unchanged83 or even decreased73,84 levels of zinc in the AD brain. Overall, zinc appears to be connected to AD, but the exact role requires further investigation.
Zinc homeostasis
The pathological consequences of both zinc deficiency and excess zinc suggest that tight control over the amount of Zn2+ is required for proper functioning of the cell. Indeed, the total zinc content of E. coli is in the millimolar range , but the femtomolar affinities of the
10 zinc sensitive transcription factors Zur and ZntR have been used to argue that almost no free zinc is present in this organism85. Similarly human cells can reach total zinc concentrations of up to 100 M34, but most of this zinc seems buffered, as free intracellular zinc concentrations have been suggested to be in the pico‐ to nanomolar range86,87. Zn2+ MT ZIP ZnT Zn2+ MT ZIP ZnT Figure 4: Cytosolic Zn2+ levels in mammalian cells are tightly controlled. Transport of zinc into the cytoplasm from either organelles or the extracellular matrix is directed by the ZIP transporter family. The ZnT family is responsible for the export from the cytosol. In the cytosol, metallothioneins (MTs) can bind up to 7 Zn2+ ions per protein and function as a buffering system. The active transport and buffering mechanisms lead to a free cytosolic Zn2+ concentration that is estimated to be well below a nanomolar level. Adapted from 88. In the cytosol, Zn2+ is buffered by metallothioneins (MTs), which are part of a whole machinery of proteins involved in zinc homeostasis (Figure 4). MTs contain at least three classes of binding sites with different affinities, allowing it to act as an efficient Zn2+ buffer over a wide concentration range. One zinc ion is bound relatively weak (log K = 7.7), two zinc ions are bound with an intermediate affinity (log K = 9.9 and 10.4) and four ions are bound tightly (log K = 11.8) at pH 7.489 to MT, resulting in control over nanomolar to picomolar free Zn2+ concentrations, depending on the amount of apoprotein thionein and its redox state90. Based on these findings, Maret and coworkers suggested that MTs can keep Zn2+ concentrations in the pico‐ to nanomolar range, which allows Zn2+ to be incorporated into
11
structural and catalytic sites of proteins to enable essential enzymatic activity, while at the same time these concentrations prevent zinc to bind to inhibitory sites on certain proteins7,91.
Next to being buffered by MTs, part of the intracellular Zn2+ is also stored in organelles such as endosomes, mitochondria, the endoplasmatic reticulum (ER) and secretory vesicles. In these compartments, total and potentially free zinc concentrations can become very high, as millimolar Zn2+ levels were reported in secretory vesicles of pancreatic ‐cells92. To allow these differences in Zn2+ concentrations in the cytosol and various organelles to occur, active transport between the cytosol and individual compartments is required. Two families of Zn2+ transporters are known that transport zinc in and out of the cytosol93. Zinc importer proteins (ZIP) are responsible for zinc import from either the extracellular fluid or from various organelles. Fourteen members of this family have been identified thus far, their activity being demonstrated either by uptake of 65Zn or using membrane permeable synthetic probes that change in fluorescence intensity upon binding of Zn2+. ZIP proteins are mostly observed at the plasma membrane, although ZIP7 was also localized at the Golgi apparatus94. The mechanism of Zn2+ uptake by ZIP transporters is poorly understood, but since the activity of ZIP1 and ZIP2 does not require ATP and transport activity is induced by HCO3‐, transport might be through a symport mechanism95. However, elucidation of the crystal structure of one of the ZIP variants is required to gain more insights into the exact transport mechanism of this protein family.
Export of zinc from the cytosol is carried out by proteins from the ZnT family. This family consists of 10 mammalian variants (ZnT1‐10) and their transporter activity has been confirmed either by measuring cell survival in medium containing high zinc levels96 or by measuring zinc efflux from transfected mammalian cells or accumulation in mutated cells97,98. Most ZnTs contain 6 transmembrane domains and all have their N‐ and C‐ termini on the cytoplasmic side of the membrane99. The localization of each variant is different, with ZnT1 being exclusively present at the plasma membrane100, while the other variants localize at vesicular membranes, in the nucleus or in the ER, sometimes in combination with the plasma membrane93. ZnT3 has been found to be localized in the membrane of secretory vesicles of neuronal cells, enabling release of large quantities of zinc upon exocytosis during neurotransmission44. In the past few years, ZnT8 has raised a lot of interest. This transporter
12
localizes almost exclusively on the membrane of secretory vesicles in pancreatic ‐ and ‐ cells101 and in humans, two polymorphisms of ZnT8 are known that either have an arginine or a tryptophan at position 325. The arginine variant generates a 16% increased risk of developing type II diabetes102, suggesting a direct link between zinc homeostasis and this disease. No mechanism for zinc transport by ZnTs is known, but since homologous proteins function as antiporters exchanging Zn2+ for H+ or K+ 103,104, the ZnTs might also function as antiporters. Recent elucidation of the crystal structure of the bacterial Zn2+ exporter Yiip is consistent with this hypothesis, as this protein also functions as a Zn2+/H+ antiporter and shows 25‐30% homology with mammalian ZnT3 and ZnT8 variants105.
An important regulator of zinc homeostasis in mammalian cells is metal responsive element binding factor 1 (MTF‐1)106. MTF‐1 is a 72.5 kDa protein that contains 6 zinc finger domains that are highly conserved107,108. Upon binding of zinc to MTF‐1, this protein is translocated to the nucleus, where it binds to promoters containing a metal responsive element (MRE)108. This results in expression of metallothioneins108, ZnT1109 and other proteins involved in metal homeostasis or oxidative stress110. Indirectly, MTF‐1 is also responsive to excess cadmium or copper. At elevated cytosolic concentrations, these metal ions can bind to MTs, thereby displacing Zn2+ that subsequently activates MTF‐1111,112. The exact mechanism by which MTF‐1 senses and responds to changes in intracellular zinc concentrations is still a matter of debate. Circular dichroism and NMR measurements by Giedroc et al. were used to propose a model in which zinc finger 5 and 6 form the metalloregulatory part that mediates DNA binding113,114. However, other data suggest that zinc fingers 1 and 2 are important for metalloregulatory function115,116, whereas zinc fingers 5 and 6 are necessary for transcriptional functions of MTF‐1 that do not require zinc117.
Fluorescence imaging
The importance of Zn2+ in neurotransmission and neurological disorders illustrate the need for tools to image the free Zn2+ concentration in living cells in real time. Because fluorescence microscopy allows intracellular imaging of live cells at submicrometer
13
resolution, several fluorescent sensor systems for Zn2+ have been developed over the past decades, using either synthetic molecules or proteins as building blocks.
Synthetic Zn
2+sensors
The development of synthetic fluorescent sensor dyes for zinc imaging has become a very active area in chemical biology118‐122. A large amount of synthetic dyes is now available, and each year new dyes are reported that show improved properties for intracellular imaging relative to older variants. The sensors usually consist of a fluorophore that is modified with a Zn2+‐chelating group and binding of Zn2+ generates a change in emission intensity and/or wavelength. In this paragraph, we do not attempt to give a full overview of the available dyes, but briefly sketch the history of synthetic Zn2+ sensors and mention some of the key sensors that were developed.
The first reported fluorescent Zn2+ sensor was 6‐methoxy‐8‐p‐toluenesulfonamide quinoline (TSQ), which selectively binds Zn2+ using a quinoline moiety and showed an increase in intensity at 495 nm upon Zn2+ binding123. As TSQ is poorly soluble in water, derivatives have been synthesized that display improved solubility. One of these derivatives, Zinquin (Scheme 1), displayed a Kd for Zn2+ of 7 M and has been used to image Zn2+ in living cells70. However, quinoline derivatives generally have the disadvantage of a low quantum yield and require excitation at UV wavelengths, which is potentially toxic to the cells. SO2 N NH O HO2C N N N R DPA Zinquin Scheme 1: Structure of Zinquin and the commonly used DPA moiety.
14
Subsequently, many Zn2+ sensors were developed that can be excited by visible light. A commonly used fluorophore in these sensors is fluorescein, which displays a high quantum yield in aqueous solution and has an excitation maximum around 488 nm. The group of Nagano generated a series of Zn2+ dyes called ZnAF’s by combining fluorescein with di‐2‐picolylamine (DPA) moieties (Scheme 1), one of the most frequently used chelating moieties in Zn2+ sensors124,125. These sensors showed affinities between 0.78 nM and 5.5 nM and could be used to monitor release of Zn2+ from hippocampal mossy fibers121. Lippard et al. also used fluorescein derivatives in combination with DPA moieties to create multiple sensors, called Zinpyrs126. Among these, Zinpyr‐4 displayed a 6‐7 fold increase in emission at 515 nm and a high Kd for Zn2+ of 0.65 nM that allowed visualization of the zinc pool in the hippocampal area of the brain from rats127.
One of the most frequently used fluorescein‐based dyes is FluoZin‐3 (Invitrogen). This probe has a Kd of 15 nM and shows a large increase in emission at 515 nm upon binding of Zn2+ 128. FluoZin‐3 was originally used to monitor Zn2+ release from pancreatic ‐cells during insulin secretion (Figure 5) and has been applied in multiple cell‐types since129‐131.
A
B
O O F KO F N(H)CH2CO2K OCH2CH2O N(CH2CO2K)2 OMeA
B
O O F KO F N(H)CH2CO2K OCH2CH2O N(CH2CO2K)2 OMe Figure 5: (A) Chemical structure of FluoZin‐3. (B) Imaging of Zn2+ secretion from pancreatic β‐cells using FluoZin‐3. Images show fluorescence intensities that were normalized using the fluorescence ofthe starting image. The temporal response of Zn2+ secretion was analyzed using the four regions of
interest (ROI) as indicated in the first image. The traces from top to bottom correspond to the ROI 1,
2, 3, 4, respectively. Picture reproduced from 128.
15
The natural tendency of rhodamine and Texas Red to accumulate in mitochondria has allowed the development of a Zn2+ dye targeted to this organelle. Sensi et al. developed Rhodzin‐3 (Kd for Zn2+ = 65 nM), a rhodamine‐based dye that showed an increase in emission at 575 nm upon Zn2+ binding and was used to detect changes in mitochondrial Zn2+ levels (Figure 6)132. A B O N(CH3)2 (H3C)2N N(H)CH2CO2-AM OCH2CH2O N(CH2CO2-AM)2 OMe (AM=CH2OC(O)CH3) + A B O N(CH3)2 (H3C)2N N(H)CH2CO2-AM OCH2CH2O N(CH2CO2-AM)2 OMe (AM=CH2OC(O)CH3) + Figure 6: (A) Chemical structure of RhodZin‐3. (B) Fluorescence microscopy image showing
RhodZin‐3 AM (red fluorescence) and the mitochondrial selective probe MitoTracker Green (green
fluorescence)132. Yellow fluorescence indicates co‐localization of the two fluorescent dyes. Bar is 10
μM. Picture reproduced from 132.
The synthetic dyes that are discussed thus far display an increase in intensity upon binding of Zn2+. As these increases are often several‐fold, they are easy to detect. However, the emission intensity can also be affected by factors such as photobleaching, changes in the probe concentration and changes in excitation light energy. One way to circumvent these artifacts is by using probes in which Zn2+ binding induces a change in the fluorescence intensities at two different wavelengths. The ratio between these two emissions can then be used as a measure of the Zn2+ concentration that is less sensitive to changes in sensor concentration, photobleaching and the intensity of excitation light. FuraZin (Scheme 2) and IndoZin (Invitrogen) are two examples of commercially available ratiometric sensors with micromolar affinities that are derived from the synthetic Ca2+ dyes Fura (Scheme 2) and Indo respectively. They both use the BAPTA (O,O’‐bis(2‐aminophenyl)ethylene‐glycol‐N,N,N’,N’‐ tetraacetic acid) structure to bind Ca2+. By using the DPA moiety instead of BAPTA, Nagano
16
recently, the same group combined the DPA moiety with iminocoumarin to create the ratiometric dye ZnIC with high affinity and specificity for Zn2+ (Kd = 1.3 pM) that was applied to HEK293 and neuronal cells134. Kikuchi et al. fused the DPA moiety to a coumarin derivative to obtain a ratiometric sensor with affinity similar to ZnIC (Kd = 3.6 pM), but a smaller ratiometric change upon Zn2+ binding and a lower specificity135. Nevertheless, this dye was also able to report on changes in Zn2+ concentration in live cells. O N O COOH OMe N HOOC HOOC O N O COOH O N O N COOH COOH HOOC HOOC N N O N O COOH OMe N H N A B C Scheme 2: Schematic representations of Fura‐2 (A), FuraZin (B) and ZnAF‐R2 (C). As illustrated by all the examples mentioned above, synthetic Zn2+ dyes are attractive tools to monitor Zn2+ levels in live cells with high spatiotemporal resolution. Depending on the application, one can choose a dye that best suits a particular experiment from a wide range of colors and affinities. Although solubility issues that resulted in poor loading of the cells have been a problem, dyes can now be functionalized with an acetoxymethyl(AM)‐ester that allows easy uptake in the cells. Following uptake, the ester is hydrolyzed, effectively preventing efflux of the dye from the cell. However, synthetic dyes come with a few intrinsic limitations. Although RhodZin‐3 displays targeting to mitochondria, localization in an organelle of interest is rather the exception than the rule and usually the result of trial and error instead of rational design. In addition, although the majority of a ‘nonlocalized’ dye will accumulate in the cytosol, a certain amount will also enter organelles such as the nucleus, ER or mitochondria, excluding their ability to report on a specific location in the cell136. Furthermore, both leakage and extreme accumulation of dye have been shown to influence the fluorescence readout during intracellular measurements, even for ratiometric sensors137,138.
17
Genetically encoded sensors
Many of the intrinsic limitations of synthetic probes can be overcome by using proteins as building blocks to create a sensor in which a native ligand binding protein is used in combination with fluorescent protein domain(s). Such protein‐based sensors are attractive because they do not require cell‐invasive procedures, their concentration can be tightly controlled and they can be targeted to different locations in the cell. One important breakthrough that allowed generation of genetically encoded sensors was the discovery and subsequent development of green fluorescent protein (GFP)139‐141, for which Osamu Shimomura, Martin Chalfie and Roger Tsien received the Nobel prize in 2008. GFP is a 27 kDa protein with a so‐called ‐barrel structure that consists of 11 ‐strands (Figure 7A)142. Its fluorescent properties arise from a spontaneous autocyclization of the serine, tyrosine and glycine at positions 65, 66 and 67 to result in a chromophore (Figure 7B) that emits green fluorescent light at 509 nm after excitation at either 395 or 475 nm143. Since its discovery, many mutant variants of GFP have been generated, resulting in a color palette of fluorescent proteins displaying fluorescence from the blue to the red range and showing improvements in e.g. brightness, maturation time or photostability144. A B N N O R2 O H R1 A B N N O R2 O H R1 Figure 7: (A) Crystal structure of GFP. (B) Structure of the chromophore of GFP.
The primary use of GFP and its variants has been in protein localization studies by using it as a genetically encoded fluorescent tag. The group of Roger Tsien pioneered the use
18
of fluorescent proteins (FP’s) to create protein‐based sensors in which Ca2+ binding induces a change in fluorescence145,146. Two types ofprotein‐based fluorescent sensors using FP variants can be distinguished. In the first category, the fluorescence of a single fluorescent protein is made sensitive to binding of a specific ligand. One way to achieve this is by introducing mutations in close proximity of the chromophore that lead to a change in fluorescence upon ligand binding. A second approach involves destabilization of the ‐barrel via insertion of a ligand binding domain. Ligand binding stabilizes the barrel structure, thereby changing the fluorescence intensity and/or wavelength. This way, Tsien et al. were able to create multiple Ca2+ sensors by inserting the Ca2+ binding domain calmodulin in various variants of GFP such as enhanced yellow fluorescent protein (EYFP)147. Since then, several protein‐based sensors based on a single fluorescent domain have been developed, e.g. to image hydrogen peroxide148 or the intracellular ratio of ADP to ATP149. A more modular protein‐based sensor was made possible by the availability of different color variants of GFP, as this allowed the development of biosensors that use fluorescence resonance energy transfer (FRET). FRET is a process that can occur between two fluorophores that have overlap between the emission spectrum of the first fluorophore (donor) and the absorption spectrum of the second fluorophore (acceptor). If the acceptor fluorophore is close to the donor fluorophore, excitation of the latter can lead to non‐radiative transfer of energy to the acceptor fluorophore as a result of long range dipole‐dipole interactions150. The rate of energy transfer kT(r) from a donor to an acceptor is given by 6 0 1 ) ( r R r k D T
(1) where D is the decay time of the donor in absence of the acceptor, r is the donor to acceptor distance and R0 is the Förster distance, The efficiency of this energy transfer E is the fraction of photons absorbed by the donor that are transferred to the acceptor, as given by T Tk
k
E
D
1
(2) which represents the ratio of the transfer rate to the total decay rate of the donor in presence of the acceptor. Using equation (1), equation (2) can be rearranged to yield the widely known Förster equation:19 6 6 0 6 0
r
R
R
E
(3)Equation (3) shows that the efficiency of energy transfer is strongly dependent on the distance between the donor and acceptor and also that the efficiency is 50% if r equals R0. Besides the distance, equation 3 also shows that the spectral overlap is dependent on R0,
which can be calculated using 6 / 1 2 0
0
.
211
[
nQ
J
(
)]
R
D (4)in which QD is the quantum yield of the donor, n is the refractive index of the medium (typically set at 1.4 for biomolecules in aqueous media), J() is the spectral overlap integral
between the emission spectrum of the donor and the absorption spectrum of the acceptor, expressed in M‐1 cm‐1 nm4 (Figure 8A). 2 is the orientational factor describing the relative
orientation of the transition dipole moments. When the fluorophores can rotate freely in solution, 2 averages out to 2/3, but for fixed conformations 2 can be calculated via
2
2
(cos
3
cos
cos
)
A D T
(5) or 2 2(sin
sin
cos
2
cos
cos
)
A D A D
(6) In equations (5) and (6), θT is the angle between the emission transition dipole of the donorand the transition absorption dipole of the acceptor, θD and θA are the angles between these
dipoles and the vector joining the donor and acceptor, and φ is the angle between the planes (Figure 8B). Values for 2 can vary between 0 and 4, depending on the orientation of the
dipoles. Head‐to‐tail parallel transition dipoles result in a 2 value of 4 and parallel dipoles
20 D A θA θD θT φ
κ
2= 4
κ
2= 1
κ
2= 0
A
B
CFP YFP Donor-Acceptor Spectral Overlap Region absorption spectra emission spectra Fl uo re sc enc e I n te nsi ty 400 450 500 550 Wavelength (nm) D A θA θD θT φκ
2= 4
κ
2= 1
κ
2= 0
A
B
CFP YFP Donor-Acceptor Spectral Overlap Region absorption spectra emission spectra Fl uo re sc enc e I n te nsi ty 400 450 500 550 Wavelength (nm) Figure 8: (A) Fluorescence spectra showing the absorption and emission spectra of cyan and yellowfluorescent protein (CFP and YFP, respectively). The area filled in grey shows the spectral overlap
between CFP emission and YFP absorption. (B) Dependence of the orientational factor κ2 on the
direction of the emission dipole of the donor and the absorption dipole of the acceptor. Adapted from150.
Genetically encoded sensors based on FRET typically consist of a ligand binding domain that is flanked by a donor and acceptor fluorescent protein. The advantage of a FRET sensor over sensors using a single fluorescent domain is that the emission ratio of the donor and acceptor fluorophore generates an output that is independent of the sensor concentration. The most commonly used FRET pair is cyan fluorescent protein in combination with yellow fluorescent protein (CFP and YFP, respectively), but reports of other combinations also exist144. The first genetically encoded FRET sensor was created by flanking calmodulin and an M13 peptide by CFP and YFP and used to detect intracellular changes in Ca2+ levels146. Since then, many other genetically encoded FRET sensors have been developed, allowing intracellular detection of e.g. ATP151, glucose152, estrogen153 and protein phosphorylation154.
Protein‐based Zn
2+sensors using a single domain
Compared to Ca2+ signaling, where cytosolic Ca2+ levels are in the range of 10‐7‐10‐6 M155, Zn2+ is present at much lower free concentrations. These low concentrations put additional constrains on the design of Zn2+ sensor proteins in terms of metal affinity and specificity. Sensor proteins need to effectively compete with other high affinity Zn2+ binders
21
in the cell such as metallothioneins and bind Zn2+ in the presence of a large excess of metal ions such as Ca2+ and Mg2+. Furthermore, since the sensor concentration will be orders of magnitude higher than the free metal concentration, potential disturbance of Zn2+ homeostasis by the sensor protein needs to be considered.
One approach to obtain metal responsive fluorescent proteins is to reengineer a single GFP domain such that the chromophore becomes sensitive to metal ion binding. Barondeau
et al. reported a variant of GFP in which the tyrosine that is part of the chromophore was
replaced by a metal‐coordinating histidine (Y66H)156. Upon creation of additional space for metal binding near the chromophore, a sensor (BFPms1) was obtained that showed a 2‐fold increase in fluorescence in the presence of μM concentrations of Zn2+. However, the same protein bound Cu2+ with similar affinity, resulting in quenching of fluorescence. A different approach to affect the fluorescent properties of GFP is to make use of circularly permuted variants of GFP that are sensitive to ligand induced folding in an adjacent metal binding domain. Mizuno et al. used this strategy to construct 191cpGFP190‐IZ‐H, a circularly permuted GFP variant in which Zn2+ binding induces the formation of a coiled coil and a concomitant increase in GFP emission157. This sensor showed an affinity of 570 nM for zinc, but unfortunately binding of Cu2+ and Ni2+ resulted in a similar increase in fluorescence (Kd = 60 nM and 130 nM, respectively). The major drawback of these sensor designs, besides their current lack of metal specificity, is that they are intensity‐based. This makes them less suitable for quantitative intracellular applications, where fluorescence intensity will also depend on sensor expression levels and can be affected by photobleaching.
FRET‐based Zn
2+sensors
In our lab, FRET‐based Zn2+ sensors with intrinsically large changes in emission ratio have been developed by simply tethering ECFP and EYFP via a flexible peptide linker, followed by introduction of Zn2+ coordinating amino acids (Ser208Cys and Tyr39His) directly at the surface of both fluorescent domains158,159. The sensor thus created, ZinCh‐9, acts as a large macromolecular Zn2+ chelator that displays a 4‐fold increase in emission ratio upon Zn2+ binding. The presence of these two de novo Zn2+ binding sites results in a biphasic response
22
and sensitivity over a large range of Zn2+ concentrations between 20 nM and 500 M (Figure 9). In a similar approach, a sensor was developed by incorporation of His‐tag sequences at the N‐ and C‐termini of an ECFP‐linker‐EYFP fusion protein. This sensor, CLY9‐2His showed a Zn2+ affinity comparable with ZinCh‐9 (Kd of 47 nM), albeit with a smaller, 1.6‐fold increase in emission ratio160. Kd~ 10-7 M Zn2+ Zn2+ Kd~ 10-4 M
A
B
C
Kd~ 10-7 M Zn2+ Zn2+ Kd~ 10-4 MA
B
C
Figure 9: (A) Design of the Zn2+‐chelating ECFP‐EYFP chimera ZinCh‐9, consisting of ECFP fusedto EYFP by a flexible (GGSGGS)9‐linker. Introduction of Zn2+ coordinating residues using Y39H and
S208C mutations at the surface of both fluorescent domains resulted in biphasic Zn2+ binding, yielding
a 4‐fold increase in emission ratio. (B) Emission spectra in absence (solid line) and presence (dotted
line) of 0.7 mM Zn2+. (C) The EYFP/ECFP emission ratio, as a function of the Zn2+ concentration
shows a biphasic response that allows Zn2+ detection between 10 nM and 1 mM. Picture obtained from
159.
Although ZinCh‐9 and CLY9‐2His show substantial ratiometric changes and specificity over other biologically relevant metal ions, their moderate affinities probably restrict their use to intracellular compartments that are rich in Zn2+. Several groups have therefore developed FRET sensors based on native metal binding proteins that have affinities more compatible with intracellular Zn2+ levels. Interestingly, the highest affinity sensor, displaying a Kd for Zn2+ of 170 fM at pH 7.5, was created by van Dongen and coworkers using
23
metal binding domains derived from proteins (ATOX1 and WD4) involved in copper homeostasis161. The ECFP‐ATOX1‐Linker‐WD4‐EYFP (CALWY) sensors using these domains were designed to detect Cu+, but the total of four cysteine residues at the metal binding site also provided an excellent tetrahedral Zn2+ binding site with femtomolar affinity (Figure 10). Unfortunately, the sensor suffered from a small change in energy transfer efficiency upon zinc binding that challenged intracellular use162. -15 -14 -13 -12 -11 0.80 0.85 0.90 0.95 log [free Zn2+], (M) R a ti o ( EYF P /EC F P ) A B -15 -14 -13 -12 -11 0.80 0.85 0.90 0.95 log [free Zn2+], (M) R a ti o ( EYF P /EC F P ) A B Figure 10: (A) FRET‐based Zn2+ sensor proteins consisting of ECFP, ATOX1, a flexible peptide
linker consisting of 9 GGSGGS repeats, WD4, and EYFP. Binding of Zn2+ to Atox1 and WD4
resulted in a decrease in energy transfer. (B) Zn2+ titration for ECFP‐ATOX1‐linker‐WD4‐EYFP.
Titrations were done in 50 mM Tris, 100 mM NaCl, 1 mM DTT, and 10% glycerol (pH 7.5) using
EDTA and HEDTA as buffering systems. Data could be fit using a 1:1 binding model, yielding a Kd
for Zn2+ of 140 fM.
Pearce et al. reported a sensor based on human metallothionein IIa (hMTIIa), which was flanked by ECFP and EYFP163. This sensor was developed to study nitric oxide induced metal‐release from MT, however, and the zinc affinity and specificity were not thoroughly characterized. Metallothioneins are probably not the optimal Zn2+ binding domains for sensor applications, as they also bind Cu+ and other TMs with high affinity. Thompson and coworkers have extensively explored fluorescent sensors based on carbonic anhydrase CA164,165, an enzyme that contains a single Zn2+ binding site with pM zinc affinity and high specificity over other metal ions. A series of adaptations to CA allowed its use as a fluorescent biosensor based on FRET from a zinc‐bound aryl sulfonamide to an Alexa Fluor 594 that is covalently coupled to the protein (Figure 11). Because the use of a synthetic
24
fluorescent label prevents intracellular expression of the sensor, a transactivator of transcription (TAT) peptide was attached that permitted uptake of the sensor protein in the cytosol of PC‐12 cells. No changes in emission ratio could be induced using either excess Zn2+ or strong Zn2+ chelators, preventing accurate calibration of the sensor in situ. The value of 5 pM for the free Zn2+ concentration that was reported in this study should therefore be treated with some care, even more so because this value is at the lower detection limit of the CA‐ sensor (Kd = 70‐130 pM).
Figure 11: Schematic representation of a carbonic anhydrase‐based Zn2+ sensor. In the absence of Zn2+,
the free fluorescent donor dapoxyl sulfonamide exhibits weak fluorescence emission at 600 nm (F < 0.01). Upon binding of zinc to carbonic anhydrase (CA), a heteromeric complex is formed between CA,
Zn2+, and the dapoxyl sulfonamide donor moiety. Binding of zinc to dapoxyl sulfonamide leads to a
dramatic increase and a blue shift in the fluorescence emission, resulting in fluorescence resonance energy transfer (FRET) from the dapoxyl donor to the AlexaFluor 594 acceptor dye and high
fluorescence emission at 617 nm. Adapted from 164.
Two groups have reported the use of zinc finger (ZF) domains to generate fluorescent Zn2+ sensors. In an elegant study, Qiao et al. created two sensors by fusing ECFP and EYFP to ZF1/ZF2 or ZF3/ZF4, two pairs of zinc finger domains from the yeast transcriptional activator Zap1 that have nanomolar affinities for Zn2+ 166. Both sensors could be used to study the kinetics of Zn2+ binding and release in suspension of yeast cells, showing a 1.3‐fold ratiometric change upon Zn2+ binding. This study did not include experiments at the single cell level and conversion of the emission ratio into a free Zn2+ concentration was not reported. In a more recent example, Dittmer et al. developed similar FRET sensors using a single zinc
25
finger domain derived from Zif268 and used them to detect changes in cytosolic and mitochondrial zinc levels in single mammalian cells167. Subcellular targeting to the mitochrondria and the extracellular membrane was achieved by attachment of specific targeting sequences and release of zinc from mitochrondria was observed upon stimulation with glutamate. Although a large 2‐4‐fold ratiometric change was observed upon zinc binding in vitro, the intracellular changes were significantly smaller (~0.25). Furthermore, while the original Zif268 domain contained a site with nanomolar affinity (Kd = 10 nM), the FRET sensors displayed surprisingly weak Zn2+ affinities of 1.7 M for Cys2His2 and 160 M forHis4, respectively, which made reliable determination of free intracellular metal concentrations difficult. Table 1: Properties of genetically encoded Zn2+ sensors. Sensor name Ref (nr) Metal binding domain Kd (M)a Signal change (fold) Application in cells BFPms1 156 ‐ 5∙10‐5 (pH 8.0) ~ 2b ‐ 191cpGFP190‐ IZHH
157 Peptide 5.7∙10‐7 (pH 6.8) N.A.b E. coli
ZinCh‐9 159 De novo 2.1∙10‐7 (pH 8.0)
5∙10‐5 (pH 8.0)
~ 4 ‐
CLY9‐2HIS 160 His‐tag 4.7∙10‐8 (pH 8.0) ~ 1.5 ‐
CALWY 162 Atox and WD4 1.4∙10‐13 (pH 7.5) ~1.15 ‐
FRET‐MT 163 Metallothionein N.A ~ 1.6 Endothelial cells
CA‐based 164 Carbonic Anhydrase 7∙10‐11 (pH 7.5) ~ 1.6 PC‐12 cells; CHO‐
cells ZF1/ZF2c 166 ZAP1 2∙10‐10 (pH 7.5) 4∙10‐9 (pH 7.5) ~ 1.3 ZF3/ZF4c 166 ZAP1 2∙10‐10 (pH 7.5) 4∙10‐9 (pH 7.5) ~ 1.3 S. cerevisiae (suspension) Cys2His2 167 Zif268 1.7∙10‐6 (pH 6.8) ~ 2.2 (in vitro) ~ 0.25 (in situ)
His4 167 Zif268 1.6∙10‐4 (pH 6.8) ~ 4 (in vitro)
~ (0.25 (in situ) Hippocampal neurons; targeting to mitochondria and extracellular membrane a) The pH at which the Kd was determined is listed between brackets b) Sensor is intensity‐based and thus the signal change indicates the increase in intensity. c) ZF1/ZF2 and ZF3/ZF4 show similar affinities, but the Zn2+ binding and release kinetics of ZF3/ZF4 are much slower, both in vitro and in situ.
26
Aim and outline of this thesis
Although a variety of synthetic and protein‐based sensors for Zn2+ have been developed (Table 1), no fluorescent sensors have been reported that were able to reliably measure the cytosolic free Zn2+ concentration. Development of such a sensor would not only facilitate research on Zn2+ homeostasis, but it would also provide the first evidence that such low free concentrations of transition metals can be measured using genetically encoded sensors. Therefore, the primary aim of the work described in this thesis was to develop genetically encoded FRET sensors for Zn2+ that can report cytosolic free Zn2+ levels.
As a starting point, two existing protein‐based Zn2+ sensors were used. The first existing sensor was ZinCh‐9, a protein‐based FRET sensor developed by Evers et al., which was shown to give a large signal change upon Zn2+ binding, but also displayed a relatively weak affinity of 210 nM159. In Chapter 2, rational attempts to improve the intracellular imaging properties of ZinCh‐9 are described. Unfortunately, none of the mutant variants containing a (Cys)4 metal binding site displayed an improved affinity for Zn2+. Cd2+ titrations were performed to test whether the (Cys)4 pockets were too large for Zn2+, leading to the serendipitous discovery of a Cd2+ sensor with 2500‐fold specificity for Cd2+ over Zn2+.
The second Zn2+ sensor that we sought to improve was CALWY, a Zn2+ sensor with a high, femtomolar Zn2+ affinity162. However, only a 15% change in emission ratio was observed upon Zn2+ binding, which prevented intracellular application of this sensor. The small signal change was caused by a distribution of conformations prior to Zn2+ binding that reduced the average change in distance between the fluorophores upon binding of Zn2+. We reasoned that by promoting complex formation between the fluorescent domains prior to Zn2+ binding to the metal binding domains, we could greatly enhance the signal change upon addition of Zn2+. By using an ECFP‐linker‐EYFP protein as a model system, we discovered that introduction of hydrophobic mutations S208F and V224L at the dimerization interface of ECFP and EYFP was sufficient to promote complex formation (Chapter 3). Therefore, the S208F and V224L mutations were used in Chapter 4 to improve the ratiometric change upon Zn2+ binding of CALWY, creating enhanced CALWY‐1 (eCALWY‐1). In addition, we systematically attenuated the affinity of eCALWY and used the resulting toolbox of sensors
27
to determine the cytosolic free zinc concentration in HEK293 cells. Chapter 5 describes further application of the eCALWY sensors in pancreatic ‐cells, in which Zn2+ plays an important role in the storage and secretion of insulin. Zinc levels were monitored in the cytosol of ‐cells, both in resting cells and during insulin secretion. In addition, the sensors were also targeted to insulin‐containing granules of these cells, where zinc levels were found to be orders of magnitude higher than in the cytosol.
The insights gained from rationally designing and improving protein‐based Zn2+ sensors could also be used to develop a sensor for Mg2+, a metal ion that is much more abundant in the cell. Despite the abundance of Ca2+ probes, no sensors exist that can be used to specifically measure intracellular Mg2+ levels. In Chapter 6, we used two approaches to develop FRET based sensors for Mg2+. The first approach involved introduction of Mg2+ binding residues at the dimerization interface of improved variants of ECFP and EYFP to create a macromolecular Mg2+ chelator. However, none of the created mutant variants showed complex formation of the fluorescent domains upon addition of Mg2+. In the second approach, a native Mg2+ binding domain was flanked with improved variants of ECFP and EYFP, resembling a more classical FRET sensor design. Both designs that were created via the second approach were able to detect intracellular changes in Mg2+ levels and showed specificity for Mg2+ over Ca2+, even during signaling. Chapter 7 contains a general discussion regarding the imaging of transition metal homeostasis and provides some strategies to optimize the ratiometric change of FRET sensors using ‘sticky’ fluorescent proteins.