A photo-inducible nano-switch for spatiotemporal controlled gene editing activation

Abstract

Genome editing allows for the manipulation of genomic DNA for biotechnology and biomedical applications, but the specificity and control of the editing process remain a challenge. This study introduces a nano-switch for spatiotemporal control of the editing process. The basic module of the nano-switch (the monomer) is composed of a gold nanorod conjugated with the catalytically dead Cas9. Based on mathematical models, we established the design and the mechanism of action of the nano-switch. Briefly, when two monomers, guided by their respective guide RNAs, form a dimer onto the DNA and get irradiated with a near-infrared pulsed laser resonant at the plasmonic properties of the dimer, they generate a localized heat that triggers a thermal break onto the DNA. The nano-switch was generated, validated, and tested in zebrafish embryos at the 1-cell stage. Molecular analysis of irradiated embryos showed targeted DNA mutations, validating the efficacy of the nano-switch as a tool for conditional gene editing that integrates the if-when-where functions.

Graphical abstract

1. Introduction

A big revolution in synthetic biology is the ability to rationally engineer intracellular switches that can trigger different signalling cascades in specific cell types within living organisms, given an input signal. The design principles of synthetic biology find important applications in the field of genome editing. This field experienced a major revolution with the discovery of the CRISPR/Cas9 system [1,2]. This system involves the Cas9 enzyme and a single guide RNA (gRNA), which allows for DNA recognition, targeting, and editing [[3], [4], [5]]. Despite its high efficiency and broad applicability [6,7], Cas9 can tolerate mismatches between the gRNA and the target sequence in the genome, resulting in off-target effects [[8], [9], [10]]. In large genomes, the risk of unintended cleavage events by Cas9 at DNA loci different from the target locus is significant, leading to unwanted mutations. To exploit the therapeutic potential of the CRISPR/Cas9 system, there is an urgent need for spatiotemporal control of the editing process to limit or prevent unwanted cuts [11,12]. This requires the design of a switch that generates the proper response when a number of conditions are met. In other words, the switch would implement multi-input AND Boolean logic gates. In this context, the CRISPR/Cas9 system can be understood as a 1-input logic gate where the output (gene editing) occurs when the input (locus recognition by the gRNA) is true. Adding more input terminals to a logic gate increases the number of input state possibilities (2^n, where n is the number of inputs). AND logic is active if and only if all inputs are on. So, when the number of possible states increases, the probability of a random activation in an unwanted condition is greatly reduced.

Some 2-input AND logic gates have already been proposed. Spatial control can be achieved by using a dimeric nuclease. This can be done by fusing the catalytically dead Cas9 (dCas9) to the FokI nuclease, which works as a dimer, meaning that cleavage requires 2 inputs to occur, i.e., the simultaneous association of two dCas9-FokI monomers to target sites that are 15–25 base pairs apart [13,14]. Temporal control has been achieved by employing light activation, such as ultraviolet–visible and far-red light, to regulate the temporal reconstruction of Cas9 in the cell [[15], [16], [17], [18]]. In this way, the Cas9 protein is split into portions that reconstitute the functional protein under light.

Here, we propose a nano-switch that implements a 3-input AND logic gate for simultaneous spatiotemporal control of gene editing (Fig. 1). The nano-switch exploits the plasmonic behavior of gold (Au) nanomaterials. It is employed as a dimer composed of the dCas9:gRNA linked to an Au nanorod (NR) (Fig. 1A). When the two AuNR-dCas9:gRNA monomers bind at proximal target sites, the dimer is constituted. When two plasmonic nanomaterials are brought into proximity, the coupling between them increases, and the optical behavior of the dimer prevails if the laser wavelength is centred on the localized surface plasmon resonance (LSPR) of the dimer rather than that of the single monomer. The nano-switch has been designed so that each monomer hybridizes to its cognate target genomic locus and, under the condition of “resonant laser on,” generates a strong confined electromagnetic field in the nanoscopic gap between the two. This strong electromagnetic field also produces rapid and localized heating of the surface of the material and a dramatic temperature jump. A thermally-induced DNA double-strand break (DSB) would occur with a time constant below milliseconds.

Fig. 1.

The nano-switch concept. Schematic illustration of (A) the nano-switch, (B) the mechanism of activation of the nano-switch by the 3-input AND logic gate: if-when-where, (C) the editing induced by the thermo-inducible double-strand DNA break by the nano-switch activation.

To summarize, the proposed nano-switch implements an if-when-where function, meaning that gene editing occurs if the ribonucleoprotein binds to the locus, when the laser is resonant, and where the dimerization occurs (Fig. 1B). We tested the nano-switch in zebrafish at early developmental stages and analysed for editing modifications occurring from the error-prone repair of the DSB at 5 h post-fertilization (Fig. 1C).

2. Results

2.1. In silico design of the nano-switch

The design of the nano-switch was based on the optimization of nanomaterial structure and parameters by numerical analysis. Commercially available COMSOL Multiphysics® software was used to evaluate the optical properties of the AuNRs (absorption, scattering, local electric field intensity, and electromagnetic power loss density) and to estimate their local heating exploited for inducing DSB on DNA. In order to obtain a plasmon resonance in the near-infrared, it is necessary for the shape to be asymmetrical rod-like [19,20]. It is the different ratio of diameter and length that determines the position of the resonance peak. Based on numerical analyses, the optimal conditions to obtain a strong temperature increment with extreme spatial localization correspond to a dimer composed of two gold nanoparticles with a nanorod shape and size of 5 × 20 nm. The irradiation conditions are ultrashort pulses (200 fs) with a repetition rate of 80 MHz and average power density impinging on the particle in the order of 2·10⁷W/m² (peak power density: Dpp = 10¹² W/m²). Fig. 2A shows the temperature profile at the surface of the two nanorods (Movie S1) at two different time points (t1, t2 are the time values at the center and end of the 200 fs pulse). According to the simulation, the temperature peak is at the center of each nanorod. The generation of two “hot spots” is believed to be particularly efficient for the generation of a gene knock-out. As the gap between the two monomers decreases, absorption tends to peak at higher wavelength values (Fig. 2B). The monomer (represented here as a gap distance of 100 nm) is substantially out of resonance at 870 nm, ensuring optimal suppression of off-target phenomena. According to our simulations, the pair of gRNAs should be designed to have an optimal distance between the nanorods in the range of 5–10 nm (Fig. 2C). As the nanorods are subjected to Brownian motions, the data in Fig.s 2A-D plot the ideal configuration in terms of Poynting vector, polarization of the incident light, dimer axis and electromagnetic field distribution. However, even if the probability that a single 200 fs pulse will be effective is negligible, we can easily assume that with irradiation lasting a few seconds, at least 1 out of 10⁸ pulses will find the nano-switch in the optimal configuration.

Fig. 2.

Simulations of the dimer's efficiency regarding temperature distribution, absorption wavelength shift,and ΔΤ increase. (A) ΔT profile at time instants t1 and t2. 200 fs laser pulse. Dpp = 10¹² W/m²@200 fs; 80 MHz repetition rate. (B) Dimer absorption profile as a function of the gap between the monomers. The wave vector, polarization of the incident light, and the dimer axis used during the simulation are schematically shown. (C) ΔT as a function of the gap between the two monomers (mean value). (D) Simulation of the dimer electromagnetic field distribution.

2.2. Optimization of the synthesis of the monomer

As the next step, we synthesized the nano-switch monomer consisting of a gold nanorod (5 × 20 nm) conjugated with the enzyme dCas9, named AuNR-dCas9. We needed to demonstrate that the conjugation method does not alter the protein activity. Since dCas9 is a mutated protein without the ability to induce DSB on DNA, we first conjugated the Cas9 protein onto gold nanorods because its endonuclease activity can be easily monitored. In this way, cleavage activity induced by the Cas9 enzyme served as an indicator of Cas9 integrity, and allowed us to assess which conjugation chemistry best preserves Cas9 functionality. The tested conjugation methods were either N-Hydroxysuccinimide (NHS) chemistry or Nitrilotriacetic acid (NTA)-Ni²⁺-Histidine tag affinity, and the nanoformulation was named AuNR.NHS-Cas9 or AuNR.NTA-Cas9, respectively (Fig. 3A). While NHS chemistry potentially targets all amines localized in the lateral chains of protein amino acids, the histidine tag offers the advantage of unidirectional and reproducible binding. The different functionalizations are reported in the Supplementary Table S1. The endonuclease activity of the nanoformulations was tested in vitro on a linear DNA fragment. Despite the fact that the same concentrations of protein (50, 100, and 150 nM) were tested for both constructs, it was noticed that the AuNR.NTA-Cas9 was more effective than the AuNR.NHS-Cas9 in the two highest concentrations (Fig. 3B and C). Indeed, AuNR.NTA-Cas9 showed similar cleavage efficiency with the non-conjugated Cas9 at the concentrations 100 and 150 nM (Fig. 3D). This may suggest that the orientation of the bond between the Cas protein and the AuNR plays an important role in enabling Cas9 to maintain its catalytic activity after conjugation. Thus, for further experiments, the AuNR.NTA-Cas9 was preferred and named simply AuNR-Cas9.

Fig. 3.

In vitro cleavage analysis of AuNR-Cas9 with different chemical backgrounds; NHS (AuNR.NHS-Cas9) or NTA (AuNR.NTA-Cas9). (A) Schematic illustration of the synthesis of the nanoformulation AuNR-Cas9 with the two different chemical backgrounds. (B) Invitro cleavage of the AuNR.NHS-Cas9 the first week upon functionalization. (C) Invitro cleavage efficiency of the AuNR.NTA-Cas9 the first week upon functionalization. (D) Quantification of endonuclease efficiency of the AuNR-Cas9 nanoformulations. Statistical analysis was performed by 2-way ANOVA (Tukey's multiple comparisons test), Interaction: p = 0.0173, Row Factor (dose): p = 0.0602; Column Factor (treatment): p < 0.0001. For 150 nM: comparison between Cas9 and AuNR.NTA-Cas9 p = 0.9840, between Cas9 and AuNR.NHS-Cas9 p < 0.0001, and between AuNR.NTA-Cas9 and AuNR.NHS-Cas9 p = 0.0002. For 100 nM: comparison between Cas9 and AuNR.NTA-Cas9 p = 0.6265, between Cas9 and AuNR.NHS-Cas9 p < 0.0001, and between AuNR.NTA-Cas9 and AuNR.NHS-Cas9 p = 0.0002. For 50 nM: comparison between Cas9 and AuNR.NTA-Cas9 p = 0.0003, between Cas9 and AuNR.NHS-Cas9 p < 0.0001, and between AuNR.NTA-Cas9 and AuNR.NHS-Cas9 p = 0.1301, n≥2. As a positive control, the invitro cleavage efficiency of unconjugated Cas9 was also tested.

2.3. Characterization of the protein activity in vivo after functionalization

Next, we assessed the gene editing capability of the AuNR-Cas9 in zebrafish, by injecting the nanoformulation in the zygote targeting the tyrosinase gene (tyr) and checking the larvae after 3 days (Fig. 4A). The AuNR-Cas9 in complex with the gRNA0 induced embryo depigmentation (Fig. 4B and C), with most larvae exhibiting mild depigmentation at 3 days post-fertilization (dpf), whereas injection with the Cas9:gRNA0 resulted in strong and mild depigmentation phenotypes. Non-functionalized AuNR and AuNR-Cas9 without gRNA failed to induce gene editing in zebrafish, as their melt curves were similar to the control group (Fig. 4D). Whereas, gene editing for AuNR-Cas9:gRNA0 was confirmed by High-Resolution Melting (HRM) analysis (Fig. 4E). Furthermore, the DNA of 6 AuNR-Cas9:gRNA0-injected zebrafish larvae was analysed by Sanger sequencing and ICE software analysis, showing a mutagenesis efficiency ranging between 28 and 95 %, and inducing as predominant indel mutation a deletion of 4 nucleotides, D4 (GAG----GATA) (Fig. 4F). The sequences of the zebrafish larvae after the ICE software analysis are provided as supplementary information (Fig. S1).

Fig. 4.

Gene editing efficiency of the AuNR-Cas9 in zebrafish. (A) Schematic illustration of the experimental process. (B) Representative images of zebrafish larvae injected with the AuNR-Cas9 without gRNA and the AuNR-Cas9 in complex with the gRNA0 targeting tyr gene, showingthe phenotype outcomes 'Absent', 'Mild', or 'Low'. Scale bars for the eye images correspond to 200 μm and the whole zebrafish larvae at 1000 μm. (C) Quantification of depigmented phenotypes in zebrafish larvae injected with 750 pg of Cas9: gRNA0 (n = 148) or AuNR-Cas9: gRNA0 (n = 111). A χ2 test was performed p < 0.0001. The microinjections of the zebrafish were performed in duplicate. (D–E) Melting analysis performed by High Resolution Melt Software (Applied Biosystems); for the AuNR-Cas9: gRNA0 (n = 21), for AuNR (n = 10), and AuNR-Cas9 without gRNA (n = 10). (F) Sanger sequencing analysis by ICE software from Synthego. Indel mutations for each tested zebrafish larva are shown as percentages. WT stands for wild type sequence same as in the control untreated zebrafish larvae. D stands for deletion followed by the number of deleted bases.

2.4. Design of the nano-switch

After demonstrating that the catalytic activity of Cas9 is not impaired by the functionalization, we conjugated the dCas9 protein onto gold nanorods following the pre-selected method, via affinity binding. The mechanism of action of our nano-switch requires the dimerization of two gold nanorods of AuNR-dCas9:gRNA. To select the pair of gRNA at desired distances and orientation, we designed the I-GENEMatcher software that allows for selection of pair of gRNAs in target genes in 3 available genomes (Homo sapiens, Danio rerio, Mus musculus) upon analysis by commonly used software for the design of gRNAs (CHOPCHOP and CRISPRscan) (https://i-gene.d4science.org/group/i-genepublic/i-gene-tool). Using the I-GENEMatcher, we selected a pair of gRNAs targeting the first exon of the tyr gene in zebrafish, Danio rerio (danRer10 or danRer11 assembly), consisting of gRNA1 and gRNA2 in a distance of 43 nt from the cutting sites (3 nt upstream of PAM sequence) of the respective gRNAs in a PAM OUT configuration. The AuNR-dCas9:gRNA1 and AuNR-dCas9:gRNA2, representing the nano-switch, were incubated with a DNA linear fragment containing the first exon of the tyr gene. We estimated the coupling distance between the new nanorods by considering the following parameters: the maximum size of the nanorods (20 nm), the maximum size of Cas9 (10 nm) [21], the maximum span of the flexible linker (21 nm, Fig. 5A1), and the distance between the gRNAs (14 nm). Based on these values, we expect that, in a liquid environment, the distance between the centers of mass of the two nanorods can theoretically range from 20 nm (when touching) to 106 nm (when diverging) (Fig. 5A2).

Fig. 5.

Characterization of the nano-switch and its dimerization on DNA. (A1) Simulation showing the relative dimensions of a dCas9 molecule interacting with DNA when covalently linked to an AuNR via a flexible linker. (A2) The image shows the maximum distance (213.4 Å) that can be spanned by the linker. Due to its flexibility, the distance between the two monomers can vary in the range of 20–107 nm (see explanation in the main text). (A3) Representative TEM image showing the formation of gold nanorod dimers on a linear DNA fragment, mediated by the respective monomers dCas9:gRNA1 and dCas9:gRNA2. (B1) Representative AFM image showing the formation of a gold nanorod dimer on a linear DNA fragment, also driven by dCas9:gRNA1 and dCas9:gRNA2. (B2) Zoomed-in view of panel B1. Cyan arrows indicate the distance between the two nanorods. (B3) Height profile corresponding to the sections (red and black) shown in panel B2. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

The sample was dried and analysed by transmission electron microscopy (TEM). Fig. 5B reveals the formation of dimers of AuNR onto a linear DNA fragment driven through the respective dCas9:gRNA1 and dCas9:gRNA2 ribonucleoproteins. Then, we validated the sample by atomic force microscopy (AFM) that enable reliable and reproducible imaging of DNA of various structures and its interaction with nanomaterials [22]. AFM provided direct visualization and measurement of DNA fragments (Fig. S2 A1-2). The height profile of DNA in air was 0.72 ± 0.10 nm (n = 5, Fig. S2 A1–2), in line with previous measurements [23]. The height profile of AuNRs in air was 3.94 ± 0.26 nm (n = 5, Fig. S2 B1–2). This deficit in height is usually attributed to adhesion effects [24]. Fig. 5B1–2 shows an example of an AuNR:gRNA dimer lying on the DNA, as confirmed by the height profiles (Fig. 5B3), with a distance of 104 nm between the centers of mass of the two AuNRs.

2.5. AuNR-dCas9 has no side effect on embryonic development

To evaluate the efficacy of the nano-switch system, we injected zebrafish embryos at the zygote stage with the AuNR-dCas9 in complex with the gRNA1 and gRNA2 to form the nano-switch in the targeted DNA area, and monitored different features up to 120 h post fertilization (hpf) to assess induced toxicity. As control, we analysed embryos injected with only PBS solution, which is the resuspension buffer for the functionalized nanorods. We found a mild reduction of survival in microinjected embryos compared to uninjected embryos (WT), an expected result of embryo manipulation at that early stage, but no statistically significant differences in terms of the embryo survival have been observed between AuNR-dCas9:gRNA-injected embryos and PBS-injected ones (Fig. 6A). A low percentage of aberrant embryos in AuNR-dCas9:gRNA -injected group has been observed, but it is not statistically different from the one of the control group and WT embryos (Fig. 6B). The zebrafish larvae were divided into three groups depending on the phenotype; Normal, Mild, and Strong aberrant. Normal is considered a zebrafish larvae without any malformation. Zebrafish larvae presenting one malformation were considered as Mild aberrant. Zebrafish larvae presenting two or more malformations were considered as Strong aberrant. The malformation phenotypes are presented in Fig. 6C and involve pericardial edema (PE), spinal cord malformation (SC), yolk sac edema (YSE), and tail malformation (TM). At the developmental stage of 24 hpf we performed a Tail flick test, to evaluate the possible impact of AuNR-dCas9:gRNA on this early locomotory behaviour (Fig. 6D1, D2). We quantified the percentage of burst activity as well as the total time of burst activity but we did not find any significant differences when compared to WT embryos. As positive control, we included embryos treated with 1 % DMSO, that induced a hypercomotory activity, demonstrated by an increase of both parameters. Finally, we did not find a delay in hatching from the chorion in all the analysed groups (Fig. 6E).

Fig. 6.

No toxicity effect in zebrafish larvae induced by the nano-switch (AuNR-dCas9:gRNA). (A) Survival profile of zebrafish larvae at 72 h post-fertilization. A Fisher's exact test was performed for every pair of groups; WT vs CTRL PBS p < 0.0001, WT vs AuNR-dCas9:gRNA p = 0.0201, CTRL PBS vs AuNR-dCas9:gRNA p = 0.0512, for WT n = 190, CTRL PBS n = 165, and AuNR-dCas9:gRNA n = 223. (B) Malformation profile of zebrafish larvae at 72 h post-fertilization. The zebrafish larvae were divided into three groups depending on the phenotype: normal, mild, and strong aberrant. Normal is considered a zebrafish larvae without any malformation. Zebrafish larvae presenting one malformation were considered as Mild aberrant. Zebrafish larvae presenting two or more malformations were considered as Strong aberrant. A Fisher's exact test was performed for every pair of groups; WT vs CTRL PBS p = 0.6320, WT vs AuNR-dCas9:gRNA p = 0.5412, CTRL PBS vs AuNR-dCas9:gRNA p = 0.1947, for WT n = 160, CTRL PBS n = 159, and AuNR-dCas9:gRNA n = 164. (C) Representative images of zebrafish larvae at 72 h post-fertilization with different malformation phenotypes. Malformations are considered PE = pericardial edema, SC = spinal cord malformation, TM = tail malformation, YSE = yolk sac edema. (D) Tail flick activity of zebrafish at 24 h post-fertilization. Scale bar is 1 mm. (D1) Burst activity was analysed by Kruskal-Wallis test, WT vs AuNR-dCas9:gRNA p > 0.9999, WT vs DMSO p < 0.0001, DMSO vs AuNR-dCas9:gRNA p < 0.0001. (D2) Total burst duration was analysed by Kruskal-Wallis test, WT vs AuNR-dCas9:gRNA p > 0.9999, WT vs DMSO p < 0.0001, DMSO vs AuNR-dCas9:gRNA p < 0.0001. For the analysis, WT n = 84, DMSO n = 95, and AuNR-dCas9:gRNA n = 162. (E) Hatching profile of zebrafish larvae at 53 h post-fertilization. A binomial test was performed for every pair of groups; WT vs CTRL PBS p (one-tailed) = 0.4322 and p (two-tailed) = 0.7751, WT vs AuNR-dCas9:gRNA p (one-tailed) = 0.2470 and p (two-tailed) = 0.4642, for WT n = 149, CTRL PBS n = 82, and AuNR-dCas9:gRNA n = 152. The microinjections of the zebrafish were performed in triplicate. WT = zebrafish larvae uninjected, CTRL PBS = zebrafish larvae injected with PBS solution, AuNR-dCas9:gRNA = zebrafish larvae injected with functionalized nanorods with the dCas9 protein and in complex with the pair of gRNAs.

2.6. Irradiation setup in zebrafish embryos

Based on the peak energy density values suggested by simulations, we developed a specific experimental strategy aimed at improving the effectiveness of in vivo irradiation experiments. To this aim, we decided to implement a spiral motion of the sample stage in order to reduce the effects of error-prone repair by extending the activation-irradiation time and to prevent the need for an exact co-localization of the DNA-nano-switch site with the focal spot of the laser. Moreover, we increased the laser power to bring above the activation threshold a sample volume of the order of 1 nl, while remaining below the limit for photothermal bubble generation [25,26] or photo-toxicity effects. Firstly, we evaluated the induction of DSBs in vitro by employing a sensitive approach: post irradiation, the DNA samples were incubated with terminal deoxynucleotidyl transferase (TdT) and biotin-UTP to label-free 3′-OH ends generated by double-strand breaks, followed by dot blot detection of biotin. To validate the method, the DNA was digested by the restriction enzyme EcoRI to induce DSBs and subsequently detect the presence of biotin in the sample after labeling (Fig. 7A). Three different concentrations of DNA were spotted; 30 ng, 15 ng, and 7.5 ng, resulting in dose-responsive signal intensity of the biotin (ctrl+ 1 to 3). DNA samples incubated with the nano-switch and irradiated at different laser power densities showed a dose-responsive biotin signal validating the presence of free 3′ OH ends caused by DSBs (Fig. 7B b,c, and d). On the contrary, DNA (Fig. 7B ctrl-) and irradiated DNA incubated with AuNR-dCas9 without the gRNAs (Fig. 7B a) did not present any strong signal.

Fig. 7.

Irradiation setup and validation invitro and invivo.(A) Dot blot detecting biotin in DNA incubated with EcoRI for 1 h at 37 °C. Ctrl-represents circular DNA 4 kb, and ctrl + represents circular DNA 4 kb incubated with the EcoRI for 1 h at 37 °C. Samples 1 to 3 are spotted DNA of different concentrations; 1 corresponds to 30 ng, 2–15 ng, and 3–7.5 ng of DNA. (B) Dot blot detecting biotin in irradiated DNA samples. 4 μl of DNA 3 μg/ml were spotted in a nitrocellulose membrane after labeling with terminal deoxynucleotidyl transferase (TdT) and biotin-UTP. Sample ctrl- is circular DNA 4 kb, a is DNA incubated with the AuNR-dCas9 without any gRNA and irradiated, b,c,d are DNA incubated with the nano-switch (AuNR-dCas9:gRNA1/gRNA2) and irradiated, respectively, and sample ctrl+ is DNA incubated with EcoRI for 1 h at 37 °C. Sample ctrl + represents 4.64E+7 number of 3′-OH ends. The membrane was baked at 80 °C for 2 h, blocked with 3 % milk in TBS-T 0.05 %, and incubated overnight with the primary antibody anti-biotin 1:1000 at 4 °C. After two washes with TBS-T 0.05 %, the blot was incubated with the secondary antibody anti-mouse 1:5000, and in the end, it was developed with the iBright instrument. (C) Schematic illustration of the experimental work-flow. (D) Control uninjected zebrafish embryos at the 1-cell developmental stage showing laser ablation after irradiation at 3 × 10¹⁵ W/m² (E) ΔΤ of the bulk temperature of control uninjected embryos during irradiation. Embryos irradiated at 3 × 10¹⁵ W/m² at the 1-cell stage, at 1 × 10¹⁵ W/m² and at 8 × 10¹⁴ W/m² in different cell stages. Each line shows the ΔΤ profile of one zebrafish embryo. (F) Survival profile of control uninjected zebrafish embryos 1-day post-irradiation. (G) Survival profile 1-day post-irradiation of zebrafish embryos injected with the AuNR-dCas9:gRNA1/gRNA2.

In order to assess the light activation of the nano-switch in vivo using zebrafish embryos, we first optimized the irradiation set-up. Zebrafish embryos control (uninjected) or injected with the nano-switch (AuNR-dCas9:gRNA1/gRNA2) in the zygote were exposed to a 200 fs laser at 870 nm and the survival of the irradiated zebrafish was evaluated 1 day post-fertilization (Fig. 7C). To determine the optimal laser irradiation protocol, increasing peak power densities 8 × 10¹⁴, 1 × 10¹⁵, and 3 × 10¹⁵ W/m² were tested and delivered in a focal spot with a waist of 5 μm, all above the value suggested by the simulations in order to compensate for actual local optical properties of the sample, e.g., nanorod alignment with respect to laser polarization, diffusion, and scattering effects of the material. At the highest power density, the embryos were adversely affected, with cells likely destroyed due to the photothermal bubble generation (Fig. 7D). The bulk temperature increase (ΔT) during irradiation reached up to 5 °C (Fig. 7E), and none of the embryos irradiated at 3 × 10¹⁵ W/m² survived until 1-day post-irradiation (1 dpi) (Fig. 7F). Testing the power densities 1 × 10¹⁵ W/m² and 0.8 × 10¹⁵ W/m² in control embryos, the bulk temperature during irradiation increased approximately to 1 and 0.5 °C, respectively (Fig. 7E), and the embryos survived till 1 dpi (Fig. 7F). Consequently, we tested the lower power densities on zebrafish treated with the nano-switch, to assess irradiation-induced toxicity. Injected embryos irradiated at 1 × 10¹⁵ W/m² at the first developmental stages; 1-, and 2-cell stage, also did not survive until 1 dpi (Fig. 7G). Therefore, we selected the lowest power density, 8 × 10¹⁴ W/m², which resulted in negligible ΔT below 0.7 (Fig. 7E) and high survival rates (Fig. 7G). In detail, when irradiated in the 1-cell stage 11 embryos out of 14 survived 1 dpi, in the 2-cell stage 12 out of 16 embryos, in the 4-cell stage 23 out of 24, and in the 8-cell stage 24 out of 26.

2.7. Irradiation schemes

As we mentioned earlier, we employed two different irradiation schemes, referred to as "3D-spiral" and "2D-spiral" (Fig. 8A). In both methods, the laser power density at the focal spot was maintained at 8 × 10¹⁴ W/m², with the spot moving in a spiral pattern to cover an area with a 150 μm radius over 60 s. The primary distinction between the two protocols is the movement of the focal plane in the "3D-spiral" scheme, which affects the distribution of laser energy within the irradiated volume. In the "2D-spiral" scheme, the focal plane remains fixed at a set distance below the surface of the sample, typically around 50 μm, which approximately corresponds to the Rayleigh range under the experimental conditions. Conversely, in the "3D-spiral" scheme, the focal plane shifts within the cell volume during irradiation. The total irradiation time of 60 s is divided into four equal phases, with the laser spot moving along a spiral path in each phase. After completing a spiral, the focal plane shifts by 10 μm, and the spiral movement is repeated (Fig. 8A).

Fig. 8.

Geneeditinganalysisonthetyrosinasetargetgeneinzebrafishuponon-demandlight-activationofthenano-switch. (A) Schematic illustration of the two irradiation schemes; “2D-spiral”, and “3D-spiral”, showing the movement of the laser focal spot in 60 s of irradiation. (B) Schematic illustration of the experimental workflow. (C) Mutagenesis percentages of the tested zebrafish embryos injected with the nano-switch AuNR-dCas9:gRNA1/gRNA2 (Dimer) or only AuNR-dCas9:gRNA1 (Monomer) and then irradiated (IRR) with 2D-spiral (2D) or 3D-spiral (3D) scheme. As control, zebrafish zygotes were only injected with the nano-switch but not irradiated (Dimer_NO IRR). The analysis was performed by the Mann-Whitney test. The ns difference is not represented in the plot. n for Dimer_NO IRR was 24, n for Monomer_3D_spiral_IRR was 24, n for Dimer_3D_spiral_IRR was 32, n for Monomer_2D_spiral_IRR was 20, n for Dimer_2D_spiral_IRR was 36. The analysed groups were Dimer_NO IRR vs Dimer_3D_spiral_IRR (p = 0.1937) and Dimer_2D_spiral_IRR (p = 0.0846), Monomer_3D_spiral_IRR vs Dimer_3D_spiral_IRR (p = 0.0208), and Monomer_2D_spiral_IRR vs Dimer_2D_spiral_IRR (p = 0.0316). (D) Indel mutations of the tested zebrafish. Data were analysed by χ² test. The ns difference is not represented in the plot. n for Dimer_NO IRR was 28, n for Monomer_3D_spiral_IRR was 24, n for Dimer_3D_spiral_IRR was 37, n for Monomer_2D_spiral_IRR was 20, n for Dimer_2D_spiral_IRR was 52. The analysed groups were Dimer_NO IRR vs Dimer_3D_spiral_IRR (p = 0.0253) and Dimer_2D_spiral_IRR (p = 0.0016), Monomer_3D_spiral_IRR vs Dimer_3D_spiral_IRR (p = 0.1492), and Monomer_2D_spiral_IRR vs Dimer_2D_spiral_IRR (p = 0.0293).

2.8. Gene editing by light activation of the nano-switch

To assess the gene editing efficiency of the nano-switch on the target DNA locus, we injected the AuNR-dCas9:gRNA1/gRNA2 in the zebrafish zygote. After approximately 15 min, the injected zebrafish underwent the irradiation process with either of the two previously mentioned schemes, “3D-spiral” or “2D-spiral”, at different developmental stages (1-, 2-, 4- or 8-cell stage). After the irradiation, the zebrafish were placed at 29 °C and were lysed at 5 hpf, at the stage of 50 % epiboly (Fig. 8B). The hypothesis is that upon irradiation at 870 nm, the dimer of the gold nanorods is activated leading to extremely high localized heat that would be subsequently responsible for a thermo-inducible DSB. Then, the break has to be repaired, resulting in indel mutations induced by error-prone DNA repair mechanisms.

Zebrafish zygotes were injected with the nano-switch, and were irradiated with either the scheme of “3D-spiral” or “2D-spiral" which both led to indel mutations and/or fragment deletions in the target DNA. In detail, 9 out of 32 zebrafish larvae injected with the nano-switch and irradiated with the “3D-spiral” scheme were edited with a mutagenesis efficiency ranging between 1 and 18 % per sample (Fig. 8C). The editing mutagenesis included mostly mutations corresponding to fragment deletions, deletion of −45 to −50 and −90 to −95 bases (Fig. 8D). On the other hand, 12 out of 36 zebrafish larvae that were injected with the nano-switch and irradiated with the scheme “2D-spiral” were edited and showed higher efficiency of mutagenesis per sample than with the scheme “3D-spiral” reaching up to 24 % (Fig. 8C). In this case, it was noticed a high variety of indel mutations per sample, large fragment deletions of 97 bases, fragment deletions of 28, 29 bases or smaller deletions of 1–12 bases (Fig. 8D). All the sequencing results are provided as Supporting Information (Fig. S3–S7).

Zebrafish that were injected with the monomer and irradiated at 870 nm sporadically presented indel mutations (Fig. 8C and D). This result is not surprising because even the monomer shows some residual optical absorption at the used wavelength, and it can be rarely activated with a very low percentage of mutagenesis. Indeed, comparing the percentage of mutagenesis between the monomer and the nano-switch (dimer) for both irradiation schemes, there is a statistical difference of p = 0.0208 for the “3D-spiral” irradiation scheme and of p = 0.0316 for the “2D-spiral" irradiation scheme (Fig. 8C). Furthermore, it was shown that there is a significant presence of indel mutations in zebrafish injected with the nano-switch and irradiated with either the “3D-spiral” (p = 0.0253) or the “2D-spiral” (p = 0.0016) compared to the zebrafish that were injected with the nano-switch but not irradiated (Fig. 8D) despite the presence of small indel mutations in the latter group.

3. Discussion

In this paper, we provide a proof-of-concept of a new technology for spatiotemporal control of inducible-genome editing (I-GENE technology). Specifically, we generated a nano-switch composed of a sensor and a nanotransducer. The sensor is the DNA recognition element, i.e., the dCas9:gRNA that is responsible for guiding the nano-switch in the desired genomic location. The nanotransducer is a plasmonic nanoparticle highly tuned to absorb a specific optical wavelength and efficiently convert it into heat. In this system, the actuator is the heat generated by the nano-switch under irradiation that is used for a thermo-inducible DNA DSB. Numerical analysis pointed out that heat generation can be more efficient when the metallic core is in the form of nanorods. Moreover, gold nanorods of 5 nm diameter and 20 nm length have an optical absorption peak of 780–810 nm in the form of a monomer and 850–910 nm in the form of a dimer. Indeed, irradiating at the wavelength of 870 nm, the dimer (but not the monomer) is activated, leading to an extreme increase in temperature (>150 °C) at the zeptoliter volume surrounding the dimer, sufficient for determining a thermo-inducible DSB [[27], [28], [29]] (Fig. S8). The monomer was generated by conjugated Cas proteins to the AuNR by His-tag affinity. This functionalization scheme preserves the biological activity of the protein. Using a pair of guides having their DNA target sequence respectively on each filament, spaced 43 nt considering the cutting sites of the respective gRNAs and in a PAM OUT configuration, we validated the ability of the nano-switch to localize as a dimer on the target DNA locus guided by the pair of gRNA1/gRNA2. The nano-switch was validated in zebrafish embryos injected with the nano-switch and irradiated with a wavelength of 870 nm in pulsed mode (200 fs pulses at 80 MHz repetition rate) and an average power density of 2 × 10⁹ W/m². We found that the dimer induces a statistically significant increase in the indel mutation in comparison to the irradiated monomer and the non-irradiated dimer. In conclusion, inspired by the principle of synthetic biology, we generated a nano-switch that implements a multi-input AND logic gate that produces a single binary output (gene editing) conditioned by a number of binary inputs (a resonant light, loci recognition of the gRNAs, dimerization).

A foreseen limitation of this study was the inability to measure the local temperature increase at a molecular level. We attempted to synthesize an intracellular temperature sensor [30] but we were unsuccessful in obtaining reproducible measurements of the temperature increase within the zeptoliter volume of the particles. Another limitation of this study is that we did not directly demonstrate the ability of the nano-switch to cross the nuclear envelope; although previous studies support this possibility [31,32], experimental confirmation will be required in future investigations.

The I-GENE technology holds promise for highly controlled gene editing applications in both research and therapeutic strategies including biotechnology, gene therapy, and cancer research. Due to its simplicity and operator-controllable modulation, the nano-switch adds a significant concept in the growing field of controlled and inducible strategies as emerging safer therapeutic tools, and its future advancement should focus on refining activation in multiple models from synthetic human tissue to living organisms.

4. Methods

4.1. Simulations

Commercially available COMSOL Multiphysics® software was used to evaluate the optical properties of the AuNPs (absorption, scattering, local electric field intensity, and electromagnetic power loss density) and estimate their local heating exploited for DNA strand break. The refractive index and extinction coefficient of gold were taken from the work of Rakić et al. [33],and the refractive index of the environment (water) was set to 1.33.

4.2. Nanotransducer synthesis

Gold nanorods were bought from Nanopartz, C12-5-780-TC-DRY-2.5 and C12-5-780-TNT-PBS-50-1. The dCas9 protein was bought from IDT, Alt-R® S.p. dCas9 protein V3, # 1081067. For the NHS chemistry protocol, a solution containing 2 × 1013 nanoparticles (C12-5-780-TC-DRY-2.5) is incubated for 20 min at room temperature with 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC) (PG82073, Thermo Scientific) and N-hydroxysulfosuccinimide (Sulfo-NHS) (PG82071, Thermo Scientific) in 1 ml of 2-(N-morpholino)ethanesulfonic acid (MES) buffer at pH 6. The mixture is transferred into a VIVAspin 500 centrifugal column (30000 MW, VS0122, Sartorius) and washed twice in PBS. The solution is then concentrated to 500 μl and placed in a glass vial to incubate with 50 μg of protein for 1 h at room temperature. After protein conjugation, 50 mM Tris at pH 8.8 is added for 10 min. The functionalized nanoparticles are subsequently washed three times by centrifugation, resuspended in 40 μl of PBS, and stored at 4 °C. For the NTA chemistry protocol, a solution containing 1.5 × 10¹² nanoparticles (C12-5-780-CUS-PBS-50-1) is incubated with 50 μg of protein in PBS for 1 h at room temperature in a glass vial. Following protein conjugation, the mix is incubated with 25 μl of HisPur^TM Ni-NTA resin (88221, Thermo Scientific) for 30 min at 4 °C. The resin was previously equilibrated with 100 μl of PBS +1 mM Imidazole. In the end, the resin is eliminated using the Pierce^TM Centrifuge column (89868, Thermo Scientific) and by centrifugation at 500g for 1 min. The functionalized nanoparticles were stored at 4 °C.

4.3. Protein concentration calculation

After functionalization, the conjugated protein was validated and calculated by Dot blot assay. 2 μl of the functionalized nanoparticles were spotted on a nitrocellulose membrane (10600003, GE Healthcare Life Science) in duplicate. A standard curve was generated using known concentrations of (d)Cas9 protein and spotting 2 μl. Two membranes were created in parallel per experiment. The membrane was blocked using 3 % milk in TBS-T 0.05 % solution (SC-2324, Chem-Cruz) for 30 min at room temperature. Then, it was incubated with the primary monoclonal Cas9 antibody 7A9-3A3 (Ab191468, Abcam) in a dilution 1:1000 in the blocking solution overnight at 4 °C. The primary antibody was washed three times with TBS-T 0.05 %, and the membrane was incubated with the secondary antibody anti-mouse IgG conjugated with peroxidase (A9044, Sigma-Aldrich) in a dilution 1:2000 in the blocking solution for 1 h at room temperature. After three washes to remove the secondary antibody, the detection was carried out using an ECL chemiluminescent substrate (1705062, BioRad), and the signal acquisition was performed with the ChemiDoc™ XRS + Imaging System (BioRad). Image analysis was completed using ImageLab software version 6.1. First, the bands detecting each dot were automatically detected by the software. With the function “Quantity tools” and then, “Absolute Quantity”, it was entered manually the quantity for the standard samples, and thus, it was generated a standard curve. The absolute quantity of the unknown samples corresponding to the spots of the functionalized nanoparticles was calculated automatically by the software.

4.4. Invitro cleavage assay

The DNA linear fragment included the entire first exon of the tyrosinase gene of zebrafish (Danio rerio) and was synthetized as part of a plasmid ordered by IDT. Using the CHOPCHOP software, the gRNA0 with sequence GGGCCGCAGTATCCTCACTC (5′-3′) was identified and used for the in vitro studies. The protocol was provided by IDT https://sfvideo.blob.core.windows.net/sitefinity/docs/default-source/protocol/alt-r-crispr-cas9-protocol-in-vitro-cleavage-of-target-dna-with-rnp-complex.pdf?sfvrsn=88c43107_30. The cleavage sample was combined with a 6X loading buffer (R0611, Thermo Scientific) and loaded onto a 1.5 % agarose gel. A 100 bp DNA ladder (G210A, Promega) served as a reference. The signals were visualized using the ChemiDoc™ XRS + Imaging system (BioRad), and the images were obtained using ImageLab 6.1 software.

The efficiency of cleavage activity was calculated using the ImageLab 6.1 software. First, the bands in each column of the gel were automatically detected by the software. With the function “Quantity tools” and subsequently, “Relative Quantity”, the band corresponding to the control sample at 1274 bp was selected as reference. Automatically, the software calculated the Relative Quantity of the band at 1274bp for the rest of the samples. The obtained values were plotted and analysed using the GraphPad software (version 9.0).

4.5. DNA thermal degradation

150 ng of linear DNA 1.3 kb was resuspended in 20 μl of dH₂O in a glass vial and incubated for 5 min at different temperatures; 20 °C, 90 °C, 120 °C, 150 °C, 180 °C, 210 °C, 240 °C. After heating, the DNA was cooled at room temperature and if needed, resuspended in 20 μl of dH₂O. The total volume was loaded in a 1.5 % agarose gel which was visualized by ChemiDoc™ XRS + Imaging system (BioRad).

4.6. Animal use - zebrafish

All animal procedures were conducted in strict accordance with protocols approved by the Italian Ministry of Public Health and the local Ethical Committee of the University of Pisa (authorization no. 99/2012-A, dated April 19, 2012), in compliance with Directive 2010/63/EU. Zebrafish were bred in the animal facility of the University of Pisa (Authorization Number DN-16/43 on 19/01/2015, renewal Authorization Number 1695 on 12/10/2023N).

Zebrafish were microinjected according to the protocol https://sfvideo.blob.core.windows.net/sitefinity/docs/default-source/user-submitted-method/crispr-cas9-rnp-delivery-zebrafish-embryos-j-essnerc46b5a1532796e2eaa53ff00001c1b3c.pdf?sfvrsn=52123407_4, using the Pneumatic Picopump PV820 air microinjector (World Precision Instruments). For the AuNR-Cas9 experiments, the same gRNA0 as for the in vitro studies was used. The pair of gRNAs to form the dimerization of the nanotransducers was selected using the I-GENEMatcher software, gRNA1: TGTCCAGTCTGGCCCGGCGA, and gRNA2: CCCCAGAAGTCCTCCAGTCC.

The zebrafish embryos were kept in E3 water (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl₂, 0.33 mM MgSO₄) at 29 °C. The development was controlled till 5 dpf. Zebrafish larvae were anesthetized with tricaine 1X (A5040, SIGMA-ALDRICH) for acquiring images at 2,3, and 5 dpf using the Nikon stereomicroscope SMZ1500 n. At 24 hpf, zebrafish larvae were analysed for tail flick activity using the DanioScope software (Noldus). As a positive control, zebrafish larvae were incubated from 6 hpf with 1 % DMSO. At 3dpf, when needed, zebrafish were lysed in 20 μl Thermopol buffer (B90004S, New England Biolabs) and heated at 95 °C for 10 min. Then, 5 μl of Proteinase K solution (20 mg/ml) was added to each sample, incubating them at 55 °C for 1 h and then at 95 °C for 10 min to inactivate the enzymatic activity.

For the High-Resolution Melting, the following protocol was used https://assets.thermofisher.com/TFS-Assets/LSG/manuals/cms_069853.pdf. The primers used were Forward primer (5′-3′): AGTCAGGTCGAGGGTTCTGT, and Reverse primer (5′-3′): AACCCCATGTAGTTTCCGGC. In the Difference plots, a threshold was indicated at −5 considering the profile of the control groups. Significant lower melting profiles were considered presenting a difference < -5.

Furthermore, a standard PCR protocol was used to amplify the genomic DNA and the amplicon was purified by Monarch® PCR & DNA Cleanup Kit from New England BioLabs Inc (#T1030S). The primers for the PCR were Forward primer (5′-3′): GTGAAGCCTCTCACTCTCCTCGAC, and Reverse primer (5′-3′): GCCAGATTTAGGTACGAGATGAACC. The Sanger sequencing was performed by Genewiz. Analysis of the sequencing data was performed with the ICE software by Synthego. To define the presence of indel mutations by analyzing the ICE results, indel mutations in the range of −5< x <+5 were considered as “false-positive” mutations and were excluded from the category of indel mutations. Indel mutations >+5 and <-5, and Fragment deletions (as indicated from the analysis with the ICE software) were considered in a common category.

4.7. Transmission electron microscopy

The formation of dimers of gold nanorods onto a linear DNA fragment morphology was analysed through transmission electron microscopy (TEM), with a Jeol JEM 1011 electron microscope using an acceleration voltage of 100 kV and equipped with a 2 Mp charge-coupled device (CCD) camera (Gatan Orius SC100). A drop of the samples was deposited on a copper grid of 150 mesh, previously coated with an amorphous carbon film and plasma treated to remove hydrocarbon residues. The samples were then stained, treating the grids with a 1 % uranyl acetate solution in water for 30 s before the specimen dried.

4.8. Atomic force microscopy

A stock solution of 5 μg/ml grafted DNA was diluted 1:1 with a 50 mM MgCl₂ buffer. For DNA adsorption, a freshly cleaved mica substrate was rinsed with the same MgCl₂ buffer. Subsequently, 10 μl of the diluted DNA solution was deposited onto the rinsed mica surface and incubated for 20 min to allow adsorption. Following incubation, the surface was thoroughly rinsed with Milli-Q water and gently dried using paper and a stream of nitrogen. The sample was then air-dried for 30 min at room temperature.

AFM imaging was performed in air using a Park XE-100 operating in tapping mode with standard OMCL-AC160TS cantilevers. Imaging parameters, including amplitude oscillation and tip-sample interaction, were optimized during scanning. Stronger tip-sample interactions (i.e., lower set-points and larger oscillation amplitudes) yielded improved image quality.

4.9. Irradiation

The used laser source was a tunable ultrafast Ti:Sapphire laser, model Sprite-XT from the M2 company. Power was adjusted by a rotatable waveplate/polarizer assembly and measured by a Scientech 361 power meter with a 10 % accuracy. The laser beam was expanded by a 1:2 telescope, coupled to a Nikon Eclipse Ti2-E microscope by the rear port and focused onto the sample with a 45-degree mirror and a Nikon CFI E-Plan 10x, 0.25 NA objective.

One zebrafish embryo was deposited as a drop in E3 water on a glass coverslip which was located on a Nikon Ti2-S-SS-E motorized scanning stage controlled by NIS Element AR software and a National Instruments NI-PCIe-6321 card. An infrared temperature detector (Melexis MLX90614xCI) was positioned above the drop to monitor and record the bulk temperature of the entire sample during irradiation. Irradiation begins at the center of the sample, which is automatically identified, and the stage moves in a spiral pattern.

4.10. Invitro irradiation dsDNA breaks assay

150 ng of DNA were incubated with AuNR-dCas9 in complex with the gRNA1 and gRNA2 in a ratio 1 DNA copy to 2 AuNR-dCas9:gRNA molecules in a volume of 100 μl and irradiated in a cuvette at different power densities as mentioned in the text. Afterwards, the DNA was labeled with biotin-UTP using the terminal deoxynucleotidyl transferase (TdT) with the protocol named “Biotin 3′ End DNA Labeling Kit” from ThermoFisher Scientific. 4 μl of DNA 3 μg/ml were spotted in a nitrocellulose membrane. The membrane was baked at 80 °C for 2 h, blocked with 3 % milk in TBS-T 0.05 %, and incubated overnight with the primary antibody anti-biotin (03–3700, Invitrogen) 1:1000 at 4 °C. After two washes with TBS-T 0.05 %, the blot was incubated with the secondary antibody anti-mouse (A9044, Sigma) 1:5000, and in the end, it was developed with the iBright instrument. Analysis was done by the iBright software version 5.1.

4.11. Statistical analyses

Each graph includes information about the statistical tests applied and the reporting of values in the legend. The raw data were analysed using GraphPad software (version 9.0). Normality of distribution was assessed using the D'Agostino & Pearson test, Shapiro-Wilk test, or Kolmogorov-Smirnov test. Statistical significance was defined as p ≤ 0.05 for all tests.

CRediT authorship contribution statement

Soultana Konstantinidou: Writing – original draft, Validation, Supervision, Methodology, Investigation, Formal analysis, Data curation. Marta d’Amora: Writing – review & editing, Supervision, Methodology, Investigation, Formal analysis, Data curation. Mafalda A. Rocco: Validation, Investigation. Francesco Nocilla: Validation, Investigation. Doriana Debellis: Supervision, Resources. Matteo Lorenzoni: Investigation, Methodology. Francesco De Angelis: Supervision, Software, Resources, Methodology, Funding acquisition, Conceptualization. Chiara Gabellini: Writing – review & editing, Supervision, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Francesco Fuso: Writing – review & editing, Supervision, Resources, Methodology, Investigation, Data curation, Conceptualization. Francesco Tantussi: Writing – review & editing, Supervision, Resources, Methodology, Data curation, Conceptualization. Vittoria Raffa: Writing – original draft, Supervision, Resources, Project administration, Methodology, Investigation, Funding acquisition, Data curation, Conceptualization.

Materials & correspondence

Correspondence and requests for materials should be addressed to Vittoria Raffa, vittoria.raffa@unipi.it.

Extended data figures

The PDF file includes Fig. S1–S8.

Funding

This research was funded by the European Union's Horizon 2020 Research and Innovation Programme under grant agreement No 862714 (I-GENE project).

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following are the Supplementary data to this article:

Data availability statement

The datasets generated during and/or analysed during the current study are available in the ZENODO repository (10.5281/zenodo.11640533).