Selection of an aptamer and development of a genetic device to control mRNA stability in response to light

(1)

Selection of an aptamer and development of a genetic device to control mRNA stability in

response to light

Dissertation

zur

Erlangung des Doktorgrades (Dr. rer. nat.) der

Mathematisch-Naturwissenschaftlichen Fakultät der

Rheinischen Friedrich-Wilhelms-Universität Bonn

vorgelegt von

Georg Pietruschka

aus

Lüdenscheid

Bonn 2021

(2)

Angefertigt mit Genehmigung der Mathematisch-Naturwissenschaftlichen Fakultät der Rheinischen Friedrich-Wilhelms-Universität Bonn

1. Gutachter: Prof. Dr. Günter Mayer 2. Gutachter: Prof. Dr. Michael Pankratz Tag der Promotion: 17.01.2022

Erscheinungsjahr: 2022

(3)

(4)

I

Abstract

Artificial autocatalytic RNA molecules, so-called aptazymes, are utilized as genetic devices to study and influence biological processes. These genetic devices consist of the autocatalytic RNA part, the ribozyme, and a ligand-sensing part, called aptamer.

Most of these devices use small molecule ligands as a trigger to regulate the cleavage activity of the aptazyme. However, the utility of small molecules has some limitations such as potential side effects, lack of reversibility as well as limited spatial and temporal control. In contrast, visible light as a trigger has the advantage of being reversible, non-harmful, and spatially and temporally controllable. Ribozymes per se, however, have no intrinsic ability to sense light. But in the field of optogenetics, proteins with LOV domains act as light-sensitive sensors. Herein, optogenetics was combined with aptamer technology to develop a genetic device capable of regulating mRNA stability in response to light. In a rational design approach, an aptamer that can bind the LOV protein PAL was fused to the hammerhead ribozyme. The aptazyme exhibited light-dependent cleavage activity, which could be modulated by varying the stem length of the introduced aptamer. This concept created aptazymes that exhibited decreased or increased cleavage activity in the presence of light. This discovery paves the way for the utilization of this system in the field of synthetic biology or basic research.

(5)

II

Zusammenfassung

Künstliche autokatalytische RNA-Moleküle, so genannte Aptazyme, werden als genetische Werkzeuge zur Untersuchung und Beeinflussung biologischer Prozesse eingesetzt. Diese genetischen Werkzeuge bestehen aus dem autokatalytischen RNA-Teil, dem Ribozym, und einem ligandensensitiven Teil, dem Aptamer. Die meisten dieser Werkzeuge verwenden niedermolekulare Liganden als Auslöser, um die Spaltungsaktivität des Aptazyms zu regulieren. Die Verwendung niedermolekularer Liganden hat jedoch einige Limitationen, beispielsweise potenzielle Nebenwirkungen, mangelnde Reversibilität sowie begrenzte räumliche und zeitliche Kontrolle. Im Gegensatz dazu hat sichtbares Licht als Auslöser den Vorteil, dass es reversibel, nicht schädlich und räumlich und zeitlich kontrollierbar ist.

Ribozyme als solche besitzen nicht die intrinsische Fähigkeit, Licht zu perzipieren. Im Bereich der Optogenetik fungieren Proteine mit LOV-Domänen jedoch als lichtempfindliche Sensoren. In diesem Projekt wurde die Optogenetik mit der Aptamer-Technologie kombiniert, um ein genetisches Werkzeug zu entwickeln, das die mRNA-Stabilität als Reaktion auf Licht regulieren kann. In einem rationalen Designansatz wurde ein Aptamer, das das LOV-Protein PAL binden kann, mit dem Hammerhead-Ribozym fusioniert. Das Aptazym zeigte eine lichtabhängige Spaltungsaktivität, die durch Variation der Stammlänge des angebundenen Aptamers moduliert werden konnte. Mit diesem Konzept wurden Aptazyme geschaffen, die in Gegenwart von Licht eine verringerte oder erhöhte Spaltaktivität aufweisen. Diese Entdeckung ebnet den Weg für die Nutzung dieses Systems im Bereich der synthetischen Biologie oder für die Grundlagenforschung.

(6)

III

List of figures

FIGURE 1:CHARACTERISTICS OF THE HAMMERHEAD RIBOZYME. ... 1

FIGURE 2:SCHEMATIC REPRESENTATION OF A SELEX CYCLE. ... 3

FIGURE 3:STRUCTURE AND CHARACTERISTICS OF BLUF PROTEINS. ... 7

FIGURE 4:STRUCTURE AND CHARACTERISTICS OF THE CRYPTOCHROMES. ... 8

FIGURE 5:CHARACTERISTICS OF THE PHYTOCHROMES. ... 9

FIGURE 6:STRUCTURE AND CHARACTERISTICS OF THE LOV PROTEIN. ... 10

FIGURE 7:PROPOSED BINDING MECHANISM OF LOVTAP TO ITS COGNATE DNA. ... 11

FIGURE 8:DOMAIN COMPOSITION AND STRUCTURE OF PAL. ... 12

FIGURE 9:SDS-PAGE OF THE PURIFICATION OF LOV-JΑ. ... 14

FIGURE 10:SDS-PAGE OF THE PURIFICATION OF LOVTAP... 15

FIGURE 11:ASSESSMENT OF LOV PROTEIN FUNCTIONALITY IN ICB. ... 17

FIGURE 12:DOT BLOT ANALYSIS FOR DETECTION OF BIOTINYLATED LOV PROTEINS. ... 18

FIGURE 13:ASSESSMENT OF BIOTINYLATED LOV PROTEIN FUNCTIONALITY. ... 19

FIGURE 14:SCHEMATIC REPRESENTATION OF A LIGHT-DEPENDENT SELEX TARGETING PHOTORECEPTORS. 21 FIGURE 15:BINDING ANALYSIS OF THE M30 LIBRARY TO LYSOZYME,LOV-JΑ, AND LOVTAP ... 22

FIGURE 16:BINDING ANALYSIS OF LYSOZYME SELECTION USING THE M30 LIBRARY... 24

FIGURE 17:BINDING ANALYSIS OF LOVTAP SELECTION USING THE M30 LIBRARY. ... 26

FIGURE 18:BINDING ANALYSIS OF SECOND LOVTAP SELECTION USING THE M30 LIBRARY. ... 28

FIGURE 19:BINDING ANALYSIS OF THE M30 LIBRARY TO LYSOZYME AND PAL ... 29

FIGURE 20:BINDING ANALYSIS OF PAL SELECTION USING THE M30 LIBRARY. ... 30

FIGURE 21:BINDING ANALYSIS OF LOVTAP SELECTION USING THE M30 LIBRARY. ... 32

FIGURE 22:NGS ANALYSIS OF PAL SELECTION USING THE M30 LIBRARY. ... 34

FIGURE 23:SEQUENCE AND PREDICTED SECONDARY STRUCTURE OF 53.19(MOTIF 2) AND 58.18(MOTIF 3). ... 35

FIGURE 24:BINDING ANALYSIS OF SEQUENCE 53, SEQUENCE 46MU (NON-BINDING SEQUENCE), SEQUENCE 58, AND THE POINT MUTANT 58M21. ... 36

FIGURE 25:LIGHT-DEPENDENT BINDING ANALYSIS OF 58 AND 58M21 USING SPR. ... 38

FIGURE 26:A)SCHEMATIC REPRESENTATION OF THE LOCALIZATION OF THE HAMMERHEAD RIBOZYME (HHR) WITHIN THE EGFP REPORTER. ... 40

FIGURE 27:EXPRESSION OF EGFP REPORTER GENE IN HELA CELLS. ... 41

FIGURE 28:RATIONALE OF APTAMER INTRODUCTION INTO THE HAMMERHEAD RIBOZYME AND THE FUNCTIONAL INTERACTION WITH PAL. ... 42

FIGURE 29:ASSESSMENT OF APTAZYME ACTIVITY AFTER INSERTION OF THE 04 AND 53 APTAMERS INTO STEM III OF THE S. MANSONI HAMMERHEAD RIBOZYME EXPRESSED IN HELA CELLS. ... 43

FIGURE 30:EGFP-HHR-53.19 MUTATION ANALYSIS EXPRESSED IN HEK293T CELLS. ... 45

FIGURE 31:ANALYSIS OF THE HHR-53 APTAZYME CONTAINING DIFFERENT STEM LENGTHS ... 47

FIGURE 32:ANALYSIS OF THE HHR-58 APTAZYME CONTAINING DIFFERENT STEM LENGTHS. ... 49

FIGURE 33:RT-QPCR ANALYSIS SHOWING THE RELATIVE REPORTER GENE MRNA LEVEL IN HEK293T CELLS. ... 51

(10)

VII

FIGURE 34:TWO-COLOR ASSESSMENT OF EGFP-HHR-58 AND DSRED-EXPRESS-HHR-58 CONSTRUCTS IN

HEK293T CELLS. ... 52

FIGURE 35:APTAZYME PERFORMANCE OF EGFP REPORTER GENE CONSTRUCTS IN THE PRESENCE AND ABSENCE OF MCHERRYPAL IN HEK293T CELLS. ... 55

FIGURE 36:RELATIVE FLUORESCENCE OF PAL TRANSFECTED IN HEK293T CELLS. ... 56

FIGURE 37:APTAZYME PERFORMANCE OF EGFP REPORTER GENE CONSTRUCTS IN PRESENCE OF MCHERRYPAL EXPRESSED UNDER DIFFERENT PROMOTERS IN HEK293T CELLS... 57

FIGURE 38:ANALYSIS OF HHR-58 APTAZYME CONSTRUCTS EXPRESSED IN HEK293MCP STABLE CELL LINE. ... 59

FIGURE 39:SHAPE-MAP MUTATION FREQUENCIES FOR ALL HHR-58 CONSTRUCT DESIGNS TESTED IN HEK293T CELLS. ... 62

FIGURE 40:NORMALIZED EXPRESSION OF THE SECRETED LUCIFERASE METLUC2 IN HEK293 CELLS. ... 64

FIGURE 41:FOLD INDUCTION OF LUCIFERASE ACTIVITY OF THE NF-ΚB-METLUC2 REPORTER AFTER STIMULATION OF NF-ΚB PATHWAY IN HEK293T CELLS. ... 65

FIGURE 42:FOLD INDUCTION OF LUCIFERASE ACTIVITY OF THE PGL4.33[LUC2P/SRE/HYGRO] REPORTER PLASMID AFTER STIMULATION OF THE MAPK/ERK PATHWAY IN HEK293T CELLS. ... 67

FIGURE 43:CONSENSUS SEQUENCE OF MOTIF 1,2, AND 3. ... 71

FIGURE 44:SCHEMATIC REPRESENTATION OF ARTIFICIAL PISTOL RIBOZYME DESIGN ... 77

FIGURE 45:CHEMICAL PROBING REAGENTS USED TO PROBE RNA STRUCTURE IN CELLS/IN VIVO. ... 78

(11)

VIII

List of tables

TABLE 1:SUMMARY OF THE SELECTION TARGETING LYSOZYME. ... 23

TABLE 2:SUMMARY OF THE FIRST ATTEMPT FOR A LOVTAP ORIENTED SELECTION. ... 24

TABLE 3:SUMMARY OF THE SECOND ATTEMPT FOR A LOVTAP ORIENTED SELECTION. ... 27

TABLE 4:SUMMARY OF THE SELECTION TARGETING PAL. ... 30

TABLE 5:SUMMARY OF THE THIRD ATTEMPT FOR A LOVTAP ORIENTED SELECTION. ... 31

TABLE 6:ASSOCIATION AND DISSOCIATION CONSTANTS OF APTAMER 58 AND APTAMER 58M21 TO PAL IN LIGHT AND DARK USING THE 1:2 BINDING MODEL. ... 37

TABLE 7:AFFINITY CONSTANTS OF APTAMER 58 AND 58M21 TO PAL USING THE 1:2 BINDING MODEL. ... 38

TABLE 8:LIST OF COMPONENTS FOR SDS-PAGE ... 85

TABLE 9:LIST OF COMPONENTS FOR AMPLIFICATION OF DNA USING GOTAQ POLYMERASE. ... 87

TABLE 10:PCR PROGRAM TABLE FOR PCR USING GOTAQ POLYMERASE. ... 88

TABLE 11:LIST OF COMPONENTS FOR AMPLIFICATION OF DNA USING PFU POLYMERASE. ... 88

TABLE 12:PCR PROGRAM TABLE FOR PCR USING PFU POLYMERASE. ... 88

TABLE 13:LIST OF COMPONENTS FOR AMPLIFICATION OF DNA USING PHUSION FLASH POLYMERASE. ... 89

TABLE 14:PCR PROGRAM TABLE FOR PCR USING PHUSION FLASH POLYMERASE. ... 89

TABLE 15:LIST OF COMPONENTS FOR UREA PAGE. ... 89

TABLE 16:LIST OF COMPONENTS FOR IN VITRO TRANSCRIPTION. ... 92

TABLE 17:INCUBATION TEMPERATURE USED FOR BINDING ANALYSIS OF PROTEINS USED IN THIS THESIS. ... 94

TABLE 18:LIST OF COMPONENTS FOR PCR AMPLIFICATION FOR NGS. ... 95

TABLE 19:PCR PROGRAM FOR NGS PREPARATION. ... 95

TABLE 20:LIST OF COMPONENTS FOR ANALYTICAL DIGEST. ... 97

TABLE 21:LIST OF COMPONENTS FOR AMPLIFICATION OF PLASMID. ... 98

TABLE 22:PCR PROGRAM FOR AQUA CLONING. ... 98

TABLE 23:LIST OF COMPONENTS USED FOR PHOSPHORYLATION OF 5’-ENDS. ... 99

TABLE 24:LIST OF COMPONENTS FOR DNA LIGATION. ... 100

TABLE 25:PRIMING MIX COMPONENTS FOR REVERSE TRANSCRIPTION OF MRNA. ... 103

TABLE 26:REACTION PREMIX FOR REVERSE TRANSCRIPTION ... 103

TABLE 27:LIST OF COMPONENTS FOR QPCR PROTOCOL. ... 104

TABLE 28: QPCR PROGRAM PROTOCOL. ... 104

TABLE 29:CONSTRUCTS USED FOR SHAPE. ... 105

(12)

IX

List of abbreviations

ANTAR AmiR and NasR transcription antitermination regulators AQUA Advanced quick-assembly

BLUF Blue light-using flavin

cDNA Complementary DNA

ddH2O Ultra-pure water

DMEM Dulbecco‟s modified Eagle medium DMSO Dimethyl sulfoxide

DNA Deoxyribonucleic acid

dNTPs Deoxyribonucleotide triphosphates dsDNA Double-stranded DNA

DTT 1,4-dithiothreitol

EDTA Ethylenediaminetetraacetic acid EGFP Enhanced green fluorescent protein

EtOH Ethanol

FBS Fetal bovine serum fwd Forward-side primer HCl Hydrochloric acid

HEK293 Human embryonic kidney cell line

HEK293mCP HEK293 cells stably expressing mCherryPAL

HEK293T HEK293-derived cell line, expressing the SV40 large T antigen

HeLa Henrietta Lacks cell line

HEPES 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid ICB Intracellular buffer

iPP Inorganic pyrophosphatase KOAc Potassium acetate

LOV Light-Oxygen-Voltage NaCl Sodium chloride NaOAc Sodium acetate

NGS Next-generation sequencing NHS N-Hydroxysuccinimide

NTPs Ribonucleotide triphosphates PAGE Polyacryl gel electrophoresis

PAL PAS-ANTAR-LOV

PAS Per-Arnt-Sim

PCR Polymerase chain reaction PYP Photoactive yellow protein rcf Relative centrifugal force rev Reverse-side primer RNA Ribonucleic acid

SELEX Systematic evolution of ligands by exponential enrichment SHAPE Selective 2′-hydroxyl acylation analyzed by primer

extension

SOC Super optimal broth with catabolite repression ssDNA Single-stranded DNA

Tris 2-Amino-2-hydroxymethyl-propane-1,3-diol

WT Wild-type

(13)

1

1 INTRODUCTION

1.1 Ribozymes

RNA in biology is not usually associated with catalytic activity. Nevertheless, this paradigm changed with the first discovery of catalytically active RNA molecules such as the self-splicing group I intron [1], RNase P [2], and even the ribosome [3]. Of importance for molecular biology, particularly in the field of bioengineering, became the discovery of a ribozyme found in a group of atypical plant pathogens with small circular RNA (circRNA) genomes such as viral satellite RNAs and viroids: the hammerhead ribozyme [4, 5]. To date, at least nine classes of naturally occurring self-cleaving ribozymes have been discovered: the hammerhead (HHR) [4, 5], hairpin [6], human hepatitis-δ (HDV) [7], Varkud-satellite (VS) [8], glmS [9], twister [10], twister sister, hatchet and pistol [11] ribozyme.

Figure 1: Characteristics of the hammerhead ribozyme. A) Schematic of the hammerhead ribozyme cleavage mechanism. The mechanism is divided into three cleavage steps: Enzyme- Substrate complex, Transition-State, and Enzyme-Product complex. G12 (red) acts as a general base, while G8 (blue) plays a possible role in acid catalysis. Deprotonation of G12 by G8 results in abstraction of 2’-H of C17, creating a nucleophile that attacks the phosphate. Image is taken from [12]. B) Consensus sequence of the minimal hammerhead ribozyme. C) Naturally

(14)

2

occurring ribozymes exist in three types named after the open stem. Images B) and C) are taken from [13].

The reaction catalyzed by these systems is an SN2-like nucleophilic attack of the 2’- oxygen to the adjacent 3’-phosphate, resulting in cleavage of the phosphodiester bond to form a 2’-3’-cyclic phosphate as well as a phosphate and a 5’-hydroxyl RNA (Figure 1 A) [14, 15]. The ribozyme, which has been extensively studied since its discovery, is the hammerhead ribozyme. This ribozyme has been used as a model RNA to study and understand the structure, biological and biochemical function of ribozymes [14]. The catalytic center consists of 15 highly conserved nucleotides flanked by three double-stranded stems (I–III) (Figure 1 B). The open end of the helix identifies the type of ribozyme; therefore, three possible types are defined, type I, II, and III (Figure 1 C) [14, 15]. Since then, more than 20 hammerhead ribozymes with a variety of structural and biochemical peculiarities have been identified and described [4, 5, 16-22]. The structure of the hammerhead ribozyme was solved at high resolution in the mid-1990s by Kim et al. using the sTRSV ribozyme [23].

1.1.1 General considerations about aptamers

Aptamers are regularly single-stranded short RNA- and DNA- oligonucleotides, folded in a three-dimensional structure that can bind their ligand with high affinity and specificity [24-26]. The term aptamer was defined by Ellington and Szostak in 1990 and is derived from the Latin word “aptus” (to fit) [24]. The method used to select aptamers was denoted Systematic Evolution of Ligands by EXponential enrichment (SELEX) by Tuerk and Gold, which describes the iterative process of binding, elution, and amplification of ligand binding species (Figure 2) [25].

(15)

3

Figure 2: Schematic representation of a SELEX cycle. The selection of an aptamer starts with a diversified DNA or RNA library incubated with the target molecule. Partitioning is used to isolate the bound sequences from the unbound ones. The bound sequences are subsequently amplified to prepare for the next selection cycle. After several cycles of selection, sequences are analyzed by cloning and sequencing. Image is taken from [27].

The in vitro selection process starts with a combinatorial library of sequences, comprising 10¹⁴ – 10¹⁵ unique sequences. This library consists of a random region of 16 – 100 nt, flanked by a constant 5’ and 3’ region, usually chemically synthesized as a DNA template [28]. This double-stranded library (dsDNA) serves as a template to generate single-stranded RNA upon transcription [24, 25]. The target of interest is subsequently incubated with the library, followed by the partition process. The remaining sequences bound to the ligand are eluted, reverse transcribed, and afterward amplified to serve as a DNA template for the next selection cycle [28]. After several cycles of selection ranging from 5 – 15 cycles [29], the cycles are tested for binding to the ligand, and the cycles in which sequences were enriched are cloned into a plasmid that is sent for Sanger sequencing [24-26, 28, 30]. The typical number of plasmids sent for sequencing varies between 10 – 100 sequences. These sequences are analyzed, categorized, and the most abundant sequences are characterized. However, with Illumina’s emerging high-throughput sequencing technology, millions of sequences can now be analyzed in parallel [31, 32], allowing an in-depth analysis of sequences [33, 34]. The overall goal is to select aptamers with high affinity and specificity. Aptamers have been selected against a plethora of

(16)

4

targets [35] such as peptides [36, 37], proteins [38, 39], small molecules [40, 41], carbohydrates [42, 43], viruses [44, 45], and even whole cells [46-48]. An important aspect for the selection of any kind of ligand is the SELEX method itself. Over the past 30 years, a variety of SELEX methods have been established [35, 49, 50].

1.1.2 Ligand controlled ribozymes

The understanding of the functionality of ribozymes and how to manipulate them enabled the utility for basic and medical research as well as synthetic biology. The technology of small molecule-dependent ribozymes is very advanced and has even reached the stage of in vivo application [51-54]. The development of the first rationally designed artificial ligand-dependent ribozymes began in 1997 by Tang and Breaker [55]. In this study, the cleavage of the hammerhead ribozyme was set under the control of an ATP binding aptamer, providing a blueprint of how to rationally design ligand-dependent ribozymes. These allosterically regulated ribozyme systems were called aptazymes. Subsequently, these systems were implemented among others in synthetic biology [56] and medical research [53, 54]. However, aptazymes controlled by a protein-ligand are still rare. In vitro, the concept of a protein-ligand- dependent ribozyme has been demonstrated several times [57]. But as of now, only Kennedy et al. reported the first protein-ligand-controlled ribozyme to function in mammalian cells [58].

Furthermore, the complexity of a ligand-dependent aptazyme can be increased when an additional trigger is implemented such as light. Light offers the advantages that it is non-invasive and it can be used with spatial and temporal precision [59]. Therefore, aptamers were selected targeting photoswitchable small molecules. Subsequently, these aptamers were introduced into a ribozyme creating a ligand-based ribozyme controlled by light. Lee et al. selected an aptamer targeting a dihydropyrene derivative, which was subsequently implemented in a ribozyme presenting the first light-controllable ligand-dependent ribozyme [60]. Thereby, the aptazyme reached a

>900-fold difference in ribozyme activity between the two states of the photoswitchable dihydropyrene derivative in vitro. In another study, Rotstan and co- workers selected an aptamer targeting the trans isoform of a stiff-stilbene. The resulting aptamer was capable of binding the trans isoform 100 times better compared to the cis form of the stiff-stilbene. This aptamer was implemented in an

(17)

5

artificial small-molecule sensing platform, termed riboswitch, to modulate the protein translation of a firefly luciferase reporter in E. coli in a light-dependent manner [61].

1.2 Brief excursion into optogenetics

The era of “optogenetics” started in 2006 when Deisseroth et al. established the study of neural circuits using technologies from optics, genetics, and bioengineering [62]. This method opened the path of controlling and monitoring biological functions in cells, tissues, organs, and organisms in a light-dependent manner [63-65]. With its potential to control biological processes using encodable proteins, optogenetics was elected as the method of the year in 2010 by Nature [66]. Different types of proteins are used in optogenetics, but the most prominent ones belong to the class of rhodopsins. These proteins are used in the field of neuroscience [67] and cardiovascular science [68], and recently the first human clinical trials of optogenetic vision restoration using channelrhodopsin have been initiated [69, 70]. However, the field of optogenetics does not only encompass channelrhodopsins, as of particular interest are also cytoplasmic photoreceptor proteins, which are used to study and control biological functions in cells [71, 72]. There are a variety of cytoplasmic light- dependent proteins and these proteins cover the entire visible light spectrum, including UV and near-infrared [73]. The family of cytoplasmic proteins consists of UV receptors [74, 75],Photoactive Yellow Protein (PYP) [76, 77], BLUF proteins [78-80], LOV proteins [81, 82], Cryptochromes [83], and Phytochromes [84]. Their use in synthetic biology includes manipulations of protein conformation/activity, alterations in gene expression, changes in protein localization, and changes in protein complexes [85].

1.3 Family of UV-sensors

The UV Resistance Locus 8 (UVR8) protein in Arabidopsis thaliana is a UV-B specific signaling component that orchestrates the expression of a variety of genes with UV protective functions [74, 86-91]. UVR8 is localized in the cytoplasm as a homodimer.

Upon UV-B irradiation, structural changes lead to the translocation of the active monomer to the nucleus where it is functional [74, 89, 90]. Subsequently, the active monomer interacts with constitutively photomorphogenic 1 (COP1), which is part of the E3 ubiquitin ligase complex, resulting in regulation of key genes in UV-B response [74, 75, 87, 92, 93].

(18)

6

In 2021, Ouyang et al. highlighted various applications in the field of optogenetics using UV-receptors, which proved to successfully induce gene expression, protein translocation, chromosomal looping, and protein secretion [94]. However, these tools are rarely used due to the hazard potential of UV-B light, which may eventually lead to cell death and the low tissue penetration of UV-light [95].

1.4 Photoactive yellow protein (PYP)

Photoactive yellow protein (PYP) is a globular photoreceptor protein first discovered in the photosynthetic purple bacterium Halorhodospira halophila [76]. PYP is a small photoreceptor comprising 125 amino acids, with its α/β-fold representing the structural archetype of the Period-ARNT-Singleminded (PAS) domain superfamily [96, 97]. As chromophore, PYP contains p-coumaric acid covalently attached to a cysteine residue via a thioester bridge [98-101]. Upon irradiation with light, the chromophore undergoes trans-cis isomerization, resulting in a substantial change in the structure and dynamics of the protein [102, 103]. It has to be noted that p- coumaric acid is not regularly available in most organisms, which complicates the use of this protein in optogenetics because precursor compounds must be provided or biosynthetic enzymes must be expressed. Nevertheless, an attempt with PYP as an optogenetic tool was initiated when it was fused to the basic zipper protein GCN4 [104, 105]. This placed the DNA binding ability of GCN4 under light control. In another attempt, the transcription factor CREB was regulated using a dominant- negative inhibitor (A-CREB) fused to PYP [106]. In this way, the binding of CREB to its cognate DNA was significantly abolished. However, due to the weak performance and the supply of the specific chromophore p-coumaric acid, these systems are difficult to use in optogenetics.

1.5 Family of BLUF proteins

Blue light-using flavin (BLUF) proteins are photoreceptors that contain a flavin chromophore that responds to blue light (Figure 3) [107]. BLUF proteins are found in bacteria [80, 108] and eukaryotes [79], but not in plants [73]. BLUF proteins are involved in physiological processes such as photosynthesis-related gene expression, biofilm formation, or the phototaxis response [78, 109, 110].

(19)

7

Figure 3: Structure and characteristics of BLUF proteins. A) Structure of the BLUF domain of the rhodobacterial protein AppA, PDB: 2IYI. B) Hydrogen bond network and molecular mechanism of photoactivation within BLUF domain. Image is taken from [111].

In optogenetics, Masuda et al. developed a technique using the BLUF protein PixD to modulate the activity of transcription factors [112]. Recently, a sophisticated optogenetic toolkit was designed using BLUF proteins to generate protein condensates that dissociate upon blue light irradiation, creating a spatial memory of protein phase separation in cells [113]. More recently, metabolic flux could be controlled using BLUF proteins to trigger assembly and disassembly of metabolically active enzyme clusters in yeast [114].

1.6 Family of Cryptochromes

Cryptochromes (CRY) are photoreceptors that belong to the flavoprotein superfamily and are found in all three kingdoms of life [115-117]. Like BLUF proteins, cryptochromes harbor the flavin cofactor capable of sensing light (Figure 4 B) [118].

Cryptochromes are structurally related to photolyases (PHY) [119], however, unlike photolyases, cryptochromes have lost their ability to repair DNA [120, 121].

Additionally, cryptochromes display a divergent C-terminal domain that is intrinsically unstructured [120]. Cryptochromes are primarily involved in the regulation of the circadian clock, however, several other signaling functions have also been reported, ranging from growth and development in plants [122] to magnetoreception in animals [123, 124]. In animals, cryptochromes can be divided into two protein classes: light- responsive type 1 (in invertebrates) and light-insensitive type 2 (vertebrates and

(20)

8

some insects) [115]. The first type is involved in clock entrainment, while the second type acts as a transcriptional repressor in the central clock mechanism [115]. In recent years, the family of cryptochromes/photolyases has been expanded by the identification of new CRY/PHY types [125].

Figure 4: Structure and characteristics of the Cryptochromes.A) Structure of the PHR domain of Cryptochrome 1 from A. thaliana. PDB: 1U3C. Redox forms of the FAD chromophore in cryptochromes. Of these redox forms, only the oxidized (orange) and the anion radical semiquinone form (red) absorb a significant amount of blue light. Image is taken from [126].

For optogenetic switches in the class of cryptochromes, the photoreceptor cryptochrome 2 CRY2 derived from Arabidopsis thaliana is used [127]. The developed switches regulate a variety of cellular functions such as translocation of transcription factors, inactivation of proteins, and DNA transcription [128].

1.7 Family of Phytochromes

Phytochromes are a family of proteins present in bacteria, cyanobacteria, fungi, algae, and land plants [129, 130]. Phytochromes were first discovered in 1959 as photoreceptors that control plant growth and development in response to long- wavelength (red to far-red) visible light [131]. Among photoreceptors, phytochromes possess a special characteristic of photoconversion enabling the switching between two states according to the wavelength applied (Figure 5) [84]. Upon irradiation with red light, the phytochrome switches to its active state Pfr, however, when irradiated with far-red light (photoconversion) or in the dark (thermal reversion), it switches back to its ground state Pr (Figure 5 B) [84]. Since phytochromes are obligate dimers, three different species – Pr-Pr, Pr-Pfr, Pfr-Pfr – can be adapted dependent on the

(21)

9

ratio of red to far-red light (Figure 5 A) [132]. The chromophore required for light- responding activity is bilin, which is covalently bound to the protein [133, 134]. In plants, phytochromes regulate important biological processes such as germination, de-etiolation, and shade avoidance [135, 136].

Figure 5: Characteristics of the phytochromes. A) Activity controlling factors of plant phytochromes. Phytochromes are dimers that exist in two forms: an active form Pfr and an inactive form Pr. Red light (R) triggers the active form, while far-red light (FR) causes inactivation. Apart from FR light, temperature (T) triggers phytochrome inactivation induced by thermal reversion. B) Absorption spectra of plant phytochrome in the Pr and Prf conformation.

Image is taken from [84].

Unlike members of the flavoprotein family, phytochromes are of special interest since these photoreceptors respond to the red light that penetrates deep into tissues.

Therefore, phytochromes form a large group of optogenetic switches [127]. One elegant example was shown in 2016 by Reinhardt et al. who demonstrated the control of receptor tyrosine kinases (RTK) using red light. When the phytochrome module CPH1 was fused to the RTK mFGFR1, MAPK/ERK signaling pathway could be triggered in a light-dependent manner [137].

1.8 LOV protein family

Light-Oxygen-Voltage (LOV) domains represent a subset of the broader family of Period-ARNT-Singleminded (PAS) domains [138, 139] and were initially discovered in phototropins [81, 82], which are responsible for phototropism in higher plants and algae. Furthermore, LOV proteins are also found in fungi and bacteria, including serine/threonine kinases, histidine kinases, and transcription factors [140, 141].

Sequencing analysis revealed that over 7000 LOV domains have been identified to

(22)

10

date [141]. Since LOV domains belong to the PAS domain family they share the same structural folding comprising a mixed α/β protein fold, with several α-helices located on one face of an antiparallel β-sheet (Figure 6 A) [138].

Figure 6: Structure and characteristics of the LOV protein.A) Structure of the AsLOV2 domain from A. sativa. PDB: 2V1B. B) Photocycle of FMN embedded in the LOV domain. Blue light induces a transient covalent bond between the C4a of the isoalloxazine ring of flavin and the γ- sulfur atom of an invariant cysteine residue. Darkness reverses this reaction. C) Example of a typical FMN spectrum in LOV proteins, taken from [142]. Upon irradiation with blue light, the characteristic flavin absorbance between 400 and 500 nm is eliminated.

The photosensory functionality of the LOV domains is mediated by a non-covalently bound flavin chromophore [143]. In this process, the C4a atom of the flavin isoalloxazine ring forms a covalent adduct with a photochemically triggered γ-sulfur atom from an invariant cysteine residue (Figure 6 B) [144]. This reaction can be observed using absorption spectroscopy, typically with the elimination of the characteristic flavin absorbance between 400 and 500 nm and a simultaneous increase in absorbance at around 390 nm (Figure 6 C) [145]. Most of the LOV proteins contain an effector domain on the C-terminal part of the LOV domain. The photochemistry between the flavin and the cysteine results in a conformational change, of which the Avena sativa LOV2 (AsLOV2) domain is the best studied. In AsLOV2, the light-triggered allosteric change results in a reversible unfolding of a C- terminal α-helix denoted Jα [146]. This striking mechanism led to the start of the development of artificial photoreceptors [147].

(23)

11

1.8.1 Artificial photoreceptors using LOV domains

The usage of artificial photoreceptors expanded the toolbox of optogenetics capable of modulating cellular processes such as apoptosis [148], cell motility [149], protein degradation [150, 151], protein localization [152], kinase activity [152-155], and epigenetic editing [156]. Additionally, light-dependent gene expression has been achieved in prokaryotes [153, 157], yeast [158], as well as mammalian cells [156, 159-161].

1.8.2 Modulating DNA binding by light: LovTAP

One interesting example of an artificial photoreceptor is LovTAP. LovTAP is a fusion protein between the LOV protein and the Escherichia coli trp repressor (TrpR) [162].

Thereby, the LOV domain act as a light-sensitive input module, while TrpR acts as an output module capable of binding to DNA when irradiated with light (Figure 7).

Figure 7: Proposed binding mechanism of LovTAP to its cognate DNA. Upon irradiation with light, the sensory domain of LovTAP is activated, releasing the Jα-helix and opening the TrpR domain, which can bind to its cognate DNA. In dark, this reaction is reversed. Picture adapted from [163].

This design was realized using the Jα domain of the AsLOV2 domain, which was introduced into an amino-terminal helix domain of the TrpR protein. With this approach, Strickland et al. could demonstrate that this rationally designed fusion protein can bind its cognate DNA 70-times better in light compared to the binding in dark [163].

1.8.3 A novel photoreceptor with RNA binding capabilities: PAL

Another remarkable example of a LOV photoreceptor is the newly discovered protein composed of the PAS-ANTAR-LOV domains, which was named PAL according to

(24)

12

the domain arrangement (Figure 8 A) [164]. A sequence database research identified this protein in the gram-positive actinobacterium Nakamurella multipartita [165, 166].

So far, this photoreceptor has been used to control cellular processes in eukaryotic cells such as protein translation [164], transcription [167], and miRNA maturation [168] in a light-dependent manner, however, the natural function of this photoreceptor in Nakamurella remains elusive.

Figure 8: Domain composition and structure of PAL. A) Primary protein structure domain from N-terminus to the C-terminus. B) Structure of the RNA-binding photoreceptor PAL. PDB: 6HMJ

The unique composition of PAL comprising of the PAS ANTAR-LOV domain allowed the speculation to bind RNA in a light-dependent manner (Figure 8 A and B) [164]. Of interest is the ANTAR domain, which is known to bind RNA in order to regulate gene expression in several bacteria [169]. However, no LOV domain has yet been discovered that is c-terminally located to the ANTAR domain. This arrangement would potentially allow the binding of RNA in response to light. In light of this finding, the in vitro selection technique SELEX was used to identify RNA aptamers capable of binding to PAL in a light-dependent manner [164].

(25)

13

2 AIM OF THIS STUDY

Current optoribogenetic devices lack modalities that allow the control of mRNA stability in cis in response to light. Therefore, a fully genetically encodable device that allows more precise control of mRNA stability is needed. This study aimed to develop such a device capable of regulating mRNA stability in a light-dependent manner. In this regard, an aptamer that binds to the photoreceptor PAL was introduced into the hammerhead ribozyme of S. mansoni. This genetic device was then inserted into the 3’-UTR of the EGFP reporter plasmid to test the activity of the ribozyme in response to light. From a design perspective, the stem length of the aptamer was varied to optimize activity. Lastly, the scope was extended towards the general expression of aptamers using the newly developed Tornado expression platform for future implementation of the light-responsive aptamers into this system. Herein, the developed genetic device has the potential to be used to study biological processes with spatial and temporal precision.

(26)

14

3 RESULTS

The result section is divided into five subsections. It starts with the expression and purification of LOV proteins used for subsequent selection (section 3.1). The next section introduces the SELEX method used to identify sequences targeting LOV proteins (section 3.2). In the following section, the aptamer is identified and characterized using next-generation sequencing and surface plasmon resonance, respectively (section 3.3). Section 3.4 examines the introduction of LOV protein binding aptamers into the hammerhead ribozyme dedicated to controlling mRNA stability in a light-dependent manner. The final section assesses the usage of an expression system to express RNA in high concentrations inside cells (section 3.5).

3.1 Expression of LOV-Jα and LovTAP

LOV-Jα and LovTAP were expressed in BL21(DE3) bacteria under general bacterial culture conditions (section 5.2.1). Bacteria were cultured under antibiotics and the protein expression was induced at OD600 = 0.400 with anhydrotetracycline (AHT) at a final concentration of 200 ng/ml. and bacteria cultures were incubated at 20°C at 130 rpm overnight in the dark. To preserve the functionality of the LOV proteins, protein purification was performed in absence of light. Protein purification was monitored by SDS-PAGE and concentration was measured by absorption spectroscopy. Furthermore, absorption spectroscopy was used to test the light- dependent switching of LOV proteins.

3.1.1 SDS-PAGE of LOV-Jα and LovTAP purification

Figure 9: SDS-PAGE of the purification of LOV-Jα.Separation of samples was performed using a 15% SDS gel. Samples were separated in Tris-Glycine buffer (25 mM Tris (pH 8.3), 192 mM Glycine, 0.1% SDS) for 60 min at 150 V. SDS polyacrylamide gel was loaded with following samples: before induction with AHT (Pre-Ind.), 3 hours after induction (3h-Ind.), induction

(27)

15

overnight (O/N-Ind.), supernatant (SN), flow-through (FT), wash fraction 1 and 2 (W1, W2), elution fractions (E1-E6), and the final fraction after buffer exchange (FIN). M = PageRuler™

Prestained Protein Ladder, 10 to 180 kDa (ThermoFisher). The black triangle highlights the expected protein.

The quality of LOV-Jα purification was evaluated by SDS-PAGE. The result of the purification process is given in Figure 9. Of all the elution fractions separated on the gel, a protein band was detected only in E2. The expected size of LOV-Jα was calculated with ~18 kDa. However, it appeared that LOV-Jα migrated faster than the predicted protein size resulting in an apparent protein size of ~13 kDa (Figure 9). In the last lane, the protein is depicted after buffer exchange to intracellular buffer (ICB) containing 12 mM HEPES (pH 7.4), 135 mM KCl, 10 mM NaCl, and 10% Glycerol (Figure 9, lane “FIN”) migrating at the same apparent protein size of ~13 kDa. This discrepancy in size is commonly observed and the reason behind this can be explained by differences in the ratio of SDS to protein or the by the increased number of acidic amino acid residues [170-172].

Similar to the purification of LOV-Jα, the purification of LovTAP was evaluated by SDS-PAGE (Figure 9). Again, the major protein fraction was eluted in the elution fraction E2, exhibiting the same faster migration of the protein on the gel, compared to the calculated protein size of ~28 kDa. The apparent protein size was ~23 kDa. In the last lane, the protein is depicted after buffer exchange to ICB (Figure 10, lane

“FIN”) migrating at the same apparent protein size of ~23 kDa

Figure 10: SDS-PAGE of the purification of LovTAP. Separation of samples was performed using a 15% SDS gel. Samples were separated in Tris-Glycine buffer (25 mM Tris (pH 8.3), 192 mM Glycine, 0.1% SDS) for 60 min at 150 V. SDS polyacrylamide gel was loaded with following samples: before induction with AHT (Pre-Ind.), 3 hours after induction (3h-Ind.), induction overnight (O/N-Ind.), supernatant (SN), flow-through (FT), wash fraction 1 and 2 (W1, W2), elution fractions (E1-E6), and the final fraction after buffer exchange (FIN).

(28)

16

M = PageRuler™ Prestained Protein Ladder, 10 to 180 kDa (ThermoFisher). The black triangle highlights the expected protein.

A special feature of LOV proteins is that the incorporated flavin molecule can be measured by spectrometric means and thus conclusions can be drawn concerning the concentration of the proteins (see section 1.8). Therefore, the absorption maximum of flavin at 447 nm was used to determine the concentration of LOV proteins. Upon 447 nm, the concentration of purified LOV-Jα and LovTAP in ICB was 87 µM and 202 µM, respectively.

3.1.2 Light-responsive functionality of expressed LOV proteins

LOV proteins are photoreceptors that induce a conformational change upon a light stimulus. The photosensory functionality of LOV domains is achieved by a non- covalently bound flavin chromophore. Upon irradiation with light, a photochemical reaction is initiated between the C4a atom and an invariant cysteine residue [144].

This reaction leads to the formation of a transient covalent bond, which can be monitored by absorption spectroscopy. This eliminates the typical characteristic flavin absorbance between 400 and 500 nm and results in a simultaneous increase in absorbance at about 390 nm (see section 1.8).

(29)

17

Figure 11: Assessment of LOV protein functionality in ICB. A) Absorbance spectra of purified LovTAP and LOV-Jα, respectively. B) LOV protein recovery measurement. Samples were irradiated with blue light (λ = 465 nm) for 1 min and subsequently, the absorbance at 447 nm was monitored for 300 min. C) Relative spectra of LOV-Jα and LovTAP, respectively.

This particular characteristic can be used to assess the functionality of LOV proteins using absorption spectroscopy. The absorbance was measured between 400 nm and 500 nm to obtain the typical FMN spectrum embedded in the LOV-domain. Two maxima are observed in the range of ca. 450 nm and ca. 470 nm, respectively, as well as a shoulder at ca. 420 nm, indicating the incorporation of FMN into the LOV- domain [82]. Functionality was tested upon irradiation of the proteins with blue light and subsequent monitoring of the absorbance at 447 nm. Upon exposure to blue light of 465 nm, the absorbance initially decreases but increases over time due to thermal reversion of the protein to the dark state. In this regard, samples were irradiated with blue light for 30 seconds and the absorbance at 447 nm was measured over a time frame of 300 seconds. The resulting recovery curve was used to calculate an approximate half-life of the light state (t1/2), giving an estimated guess of how quickly LOV-Jα and LovTAP reverse to the dark state in the respective buffer

(30)

18

system. For both proteins, a t1/2 of 27 s was calculated indicating a fast thermal reversion of the protein.

3.1.3 Biotinylation of LovTAP and LOV-Jα

For immobilization on the SELEX affinity matrix, the purified proteins were functionalized with a biotin tag to allow immobilization on streptavidin. As a biotinylation reagent, EZ-Link Sulfo-NHS-LC-Biotin was used in a 4-fold molar excess over the protein according to the manufacturer’s instructions. The remaining reagent was removed using Zeba Desalting spin columns and the success of biotinylation was assayed using the DotBlot method (Figure 12).

Figure 12: Dot blot analysis for detection of biotinylated LOV proteins. 50 pmol LOV protein with and without biotinylation was spotted on a nitrocellulose membrane. The membrane was blocked with PBS/5% BSA. After blocking, the membrane was incubated with Streptavidin-HRP dissolved in PBS/5% BSA. As an HRP substrate, ECL Western Blotting Substrate was used.

In order to test successful biotinylation, 50 pmol of biotinylated protein and non- biotinylated protein as control was spotted onto a nitrocellulose membrane with subsequent detection of biotinylation using streptavidin-HRP. A luminescence signal of a spot indicates a biotinylated protein. The absence of luminescence indicates that no biotin tag was attached to the proteins. In conclusion, both proteins were successfully biotinylated (Figure 12) and samples were subsequently challenged for light-responsive functionality.

(31)

19

Figure 13: Assessment of biotinylated LOV protein functionality. A) Absorbance spectra of biotinylated LovTAP-biotin and LOV-Jα-biotin, respectively. B) LOV protein recovery measurement. Samples were irradiated with blue light (λ = 465 nm) for 1 min and subsequently, the absorbance at 447 nm was monitored for 300 min. C) Relative spectra of LOV-Jα-biotin and LovTAP-biotin, respectively.

Any modification of the protein might disturb or destroy the functionality. In this regard, the light-responsive functionality was tested again to exclude aberrant behavior due to biotinylation. Therefore, absorption spectra were recorded and the switching ability was tested as mentioned above. In general, the absorbance spectra of the biotinylated proteins resemble the spectra of the non-biotinylated proteins (compare Figure 13 A with Figure 11 A) Next, the thermal reversion was measured and the t1/2 was calculated.

The thermal reversion half-time of both purified proteins was found to be slightly shorter upon biotinylation. For LOV-Jα, the t1/2 upon biotinylation was 22.25 sec, whilst it was measured to be 26.82 sec for the native protein. Similarly, the thermal reversion half-time of LovTAP decreased from 27.38 sec to 25.09 sec. The indicated results suggest a slightly accelerated thermal reversion for the proteins modified with

(32)

20

a biotin-tag. More importantly, the results demonstrated that the modification did not disturb the light-responsive functionality of the photoreceptors LOV-Jα and LovTAP.

3.2 Selection of aptamers targeting LOV proteins

Selection is usually performed by immobilizing the target protein on a solid matrix such as sepharose, agarose, or other resins. However, for the selection of proteins that respond to light, a light source needs to be used to activate the light state of the photoreceptors. If a solid matrix is used, there is a risk that not all proteins are present in the light conformation, which could result in a bias during the selection.

Therefore, selections targeting the light state of LOV proteins were performed in a clear streptavidin-coated well of a 96-well plate. This approach was already successfully conducted by Weber et al. [164]. Among the two LOV proteins expressed and purified, LovTAP was selected first due to its ability to bind to DNA in a light-dependent manner (section 1.8.2) [162]. LovTAP consists of the E. coli transcription factor repressor TrpR and the AsLOV2 domain, where the Jα domain was integrated into an N-terminal α-helix of the TrpR. Therefore, it was sought to serve as a good starting point for the selection of novel aptamers targeting LOV proteins. Figure 14 shows the SELEX scheme used for the selection of aptamers targeting the LOV proteins.

(33)

21

Figure 14: Schematic representation of a light-dependent SELEX targeting photoreceptors. The selection starts with the transcription of RNA out of a randomized DNA library. The RNA library is then incubated together with the target photoreceptor, e.g. LOV protein, under blue light to enable binding of the RNA to the target. In the next step, the unbound RNA is washed off and the remaining bound RNA is eluted in dark. The eluted RNA is harvested and reverse transcribed, serving as a DNA template for PCR amplification and as a template for subsequent transcription for a new selection cycle.

3.2.1 The M30 library was capable of enriching target binding sequences

Prior to selection targeting LOV proteins, the M30 library was assessed for the general capability to enrich target binding sequences. Therefore, a test selection was conducted targeting lysozyme, a benchmark protein in selections. The binding of the

(34)

22

M30 RNA library was assessed to elucidate the non-specific binding of the library to lysozyme and the LOV proteins LOV-Jα and LovTAP. Figure 15 shows the binding profile of the library to lysozyme, LOV-Jα, and LovTap, respectively.

Figure 15: Binding analysis of the M30 library to lysozyme, LOV-Jα, and LovTAP, respectively using RiboGreen™ based fluorescence assay. Biotinylated lysozyme in PBS and biotinylated LOV-Jα and LovTAP in ICB were immobilized on streptavidin-coated plates. As a control, biotinylated ERK2 in PBS was immobilized. For interaction analysis 500 nM M30 RNA library in ICB/1 mM MgCl2 was incubated with LOV-Jα and LovTAP at 25°C for 30 min, respectively. For interaction analysis with lysozyme, 500 nM M30 RNA library in PBS/3 mM MgCl2 was incubated at 25°C for 30 min. As a control, the aptamer C5.71 was incubated with ERK2 in PBS/3 mM MgCl2 in the final concentration of 500 nM at 25°C for 30 min. The bound RNA was detected using RiboGreen™ dissolved in TE (10 mM Tris-HCl pH 7.6, 1 mM EDTA) buffer.

As an internal control, the C5.71 aptamer was used targeting ERK2. In general, a phosphate buffer (PBS) supplemented with 3 mM MgCl2 was used for ERK2 and Lysozyme, based on previous selection protocols [173-175]. However, a different buffer was used for the photoreceptors in view of a final application in a cellular environment. Therefore, an intracellular buffer (ICB) supplemented with 1 mM MgCl2

was used as reported in Weber et al. [164]. Figure 15 illustrates the binding analysis of the M30 library to lysozyme, LOV-Jα, and LovTAP in relation to the C5.71 aptamer binding to EKR2. The M30 library binds to lysozyme to a substantial amount suggesting an increased non-specific binding. However, no binding of the M30 library to LOV-Jα or LovTAP was observed (Figure 15). The low binding of the M30 library to Lysozyme was accepted and a test selection was started with 1000 pmol RNA in phosphate buffer saline pH 7.4 (PBS), supplemented with 3 mM MgCl2 incubated for 30 min at 29°C (Table 1). In total 7 selection cycles were performed with continually

Selection of an aptamer and development of a genetic device to control mRNA stability in response to light