Macrocyclic RGD-peptides with high selectivity for α_vβ₃ integrin in cancer imaging and therapy

Abstract

Integrins, particularly the α_vβ₃ subtype, are critical receptors involved in cell adhesion, migration, and signaling, playing a significant role in tumor progression and metastasis. Despite extensive research into integrin-targeted therapies, challenges remain in developing ligands that exhibit high selectivity for α_vβ₃ over other integrin subtypes, such as α_vβ₅. This study employs a one-pot sortase A-mediated on-resin peptide cleavage and in situ cyclization method to synthesize two generations of macrocyclic RGD-peptide libraries. Systematic screening through surface plasmon resonance and cell-based competition assays identified the lead compound, c-(G5RGDKcLPET), which demonstrated high affinity and selectivity for α_vβ₃. Additionally, the optimized cyclic peptide was functionalized with a fluorescent dye (Cy5) and the cytotoxic drug monomethyl auristatin E (MMAE), enhancing its potential for cancer imaging and targeted therapy. This work contributes a novel platform for developing integrin-targeted diagnostics and therapeutics, highlighting the importance of macrocyclic peptides in cancer treatment strategies.

We developed a novel macrocyclic RGD-peptides (2-c) with high selectivity for α_vβ₃ Integrin in specific tumor imaging and therapy via one-pot Srt A-mediated on-resin in situ cyclization strategy.

Introduction

Integrins are heterodimeric transmembrane receptors composed of alpha (α) and beta (β) subunits, essential for mediating interactions between cells and the extracellular matrix (ECM) and facilitating cell–cell interactions.^1,2 These receptors play pivotal roles in biological processes such as migration, proliferation, differentiation, and apoptosis, influencing diverse cellular functions.^3,4 Notably, integrin signaling is critical in tumor progression and metastasis.^5,6 Among the integrin subtypes, α_vβ₃ integrin is prominently expressed in several cancers, including melanoma, breast, prostate, pancreatic, ovarian, cervical, and glioblastoma.^7,8 Extensive research indicates that α_vβ₃ integrin facilitates cancer cell migration, invasion, and distant metastasis formation, while promoting angiogenesis, thereby supporting tumor growth and survival.^4,9,10 Furthermore, α_vβ₃ integrin expression is associated with resistance to certain cancer therapies, making it a compelling target for developing therapies that inhibit its function and impede tumor progression.¹¹

The arginine–glycine–aspartic acid (RGD) tripeptide sequence, a naturally occurring integrin-binding motif found in ECM proteins including vitronectin, fibronectin, fibrinogen, and osteopontin, has been extensively investigated.^8,12–14 In recent decades, numerous RGD-based peptides targeting α_vβ₃ integrin have been developed to illuminate its regulatory mechanisms.^15–20 These RGD peptides are also amenable to conjugation with various therapeutic agents, including chemotherapeutics, toxins, and nanoparticles, thereby enhancing therapeutic efficacy in cancer treatment and diagnosis.^21–23 For example, cilengitide [c(RGDf(NMe)V)], a cyclic RGD peptide antagonist of α_vβ₃ and α_vβ₅ integrins, has undergone clinical investigation.²⁴ Integrin expression has been correlated with disease progression in specific cancers, with α_vβ₃ and α_vβ₅ integrins exhibiting distinct expression profiles and biological functions.^25,26 Therefore, the development of ligands exhibiting selectivity for α_vβ₃ over α_vβ₅ offers significant therapeutic potential in cancer treatment through the inhibition of pathological angiogenesis. However, achieving high selectivity and affinity for distinct integrin subtypes, such as α_vβ₃, remains challenging owing to the conserved nature of the RGD-binding site among integrins.²⁷

Cyclic peptides have garnered considerable attention for their unique properties, including enzymatic stability, binding specificity, and low toxicity.²⁸ Previously, we synthesized three macrocyclic peptides incorporating mono-, di-, and tri-RGD motifs using sortase A-mediated ligation (SML) in solution.²⁹ The macrocyclic peptide c-(G7RGDLPET), containing a mono-RGD motif, exhibited moderate potency and good selectivity for α_vβ₃ integrin in in vitro cell competition assays. Recently, we developed a one-pot strategy for synthesizing head-to-tail cyclopeptides and complex cyclotides via sortase A-mediated on-resin cleavage and in situ cyclization.^30,31 This method streamlines the synthesis process, yielding macrocyclic peptides with high efficiency and minimal oligomeric side products. In this study, we utilized the cyclic peptide c-(G7RGDLPET) as a lead compound to create a macrocyclic RGD-peptide library and investigated the biological activities of these compounds toward integrins α_vβ₃ and α_vβ₅ using surface plasmon resonance (SPR) and in vitro cell-based competition binding assays. The ligand exhibiting high affinity and selectivity for α_vβ₃ integrin was further functionalized with the fluorescent Cy5 dye and the anti-cancer drug monomethyl auristatin E (MMAE) for potential applications in cancer imaging and therapy in α_vβ₃-positive cancer cells.

Results and discussion

Design and synthesis of macrocyclic RGD-peptide libraries

We previously reported the macrocyclic peptide c-(G7RGDLPET) as a selective antagonist of α_vβ₃ integrin over α_vβ₅ integrin. To further investigate the structure–activity relationship (SAR) and develop novel RGD-containing macrocyclic peptides with improved α_vβ₃ integrin affinity and selectivity, we employed c-(G7RGDLPET) as a template for designing a cyclic RGD-peptide library. Prior studies have demonstrated that the orientation of arginine (R) and aspartic acid (D) residues, and the conformation of the RGD loop, within RGD-containing peptides significantly influence binding affinity and selectivity.^32,33 This is particularly dependent on the two amino acids flanking the RGD motif at its C-terminus (–RGDX₁X₂–).³⁴

To systematically investigate this, we designed a C-terminal dimer motif (X₁X₂) for the RGD sequence, generating a library of cyclic peptides with the general structure c-(G5RGDX₁X₂LPET) (Fig. 1). For the first-generation library, we fixed X₂ as valine (V), based on prior literature,^16,35 while varying X₁ across all natural l- and d-amino acids to assess their impact on integrin binding. Following screening of this library for α_vβ₃ and α_vβ₅ integrin binding affinity, we fixed X₁ in the optimal cyclic peptide candidate and systematically varied X₂ across all natural l- and d-amino acids to generate a second-generation library for further screening.

Fig. 1. Methodology for the design of macrocyclic RGD-peptides libraries. The RGD and lead motifs are shown in red. The capital letters represent l-type amino acids and lowercase letters represent d-type amino acids.

We synthesized these cyclic peptides using a one-pot, sortase A (SrtA)-mediated strategy involving on-resin cleavage and in situ cyclization. This approach enabled highly efficient production of the target cyclic peptides in a one-bead-one-compound format. The linear peptide (H₂N-GGGGGRGDX₁X₂LPETGGS-CO₂H), incorporating a C-terminal LPETGGS sequence and an N-terminal penta-glycine motif, was assembled onto PEGA resin using standard Fmoc solid-phase peptide synthesis (SPPS). Subsequently, side-chain protecting groups were removed using TFA/i-Pr₃SiH/H₂O (95 : 2.5 : 2.5, v/v/v) while the peptide remained resin-bound.

Subsequently, the polyethylene glycol-polyacrylamide (PEGA) resin was incubated with SrtA in a Tris-HCl buffer (pH 7.5) containing 150 mM NaCl, 5 mM CaCl₂, and 0.5 mM mercaptoethanol for 1 hour. Under these conditions, SrtA cleaved the linear peptide containing the LPETG sequence from the resin, resulting in the formation of a peptide-SrtA thioester intermediate. This was followed by intramolecular head-to-tail cyclization, yielding the RGD cyclic product. Finally, the PEGA resin was removed via simple filtration, and the filtrate was purified using preparative RP-HPLC to obtain the cyclic peptides. This in situ cyclization on PEGA resin by Srt A reduced undesired oligomeric side-products and purification steps.³¹ Utilizing this method, we successfully synthesized first- and second-generation libraries comprising 39 and 38 cyclic peptides, respectively. The yield was 32–84% (see Tables S1 and S2†). All compounds were characterized using MALDI-TOF MS (see Fig. S1 and S2†), and their purities were assessed via HPLC (above 97%, see Fig. S3 and S4†).

Screening macrocyclic RGD-peptides libraries for α_νβ₃-binding

To evaluate the binding activity of the macrocyclic RGD-peptides to integrins, a preliminary SPR assay was conducted. This involved the immobilization of α_vβ₃ or α_vβ₅ integrin onto the chip surface at a peptide concentration of 50 μM. The lead macrocyclic RGD-peptide c-(G7RGDLPET) and cilengitide served as controls. As illustrated in Fig. 2, the X₁ residues in the first-generation macrocyclic peptides significantly influenced their binding activities to α_vβ₃ integrin. Five compounds (1-A, 1-K, 1-v, 1-m, 1-f), with X₁ replaced by l-alanine (A), l-lysine (K), d-valine (v), d-methionine (m), and d-phenylalanine (f), exhibited slightly stronger binding to α_vβ₃ compared to the lead macrocyclic peptide c-(G7RGDLPET). In contrast, the remaining 34 variants in the first-generation library exhibited weaker binding activities to α_vβ₃ integrin. Notably, the first-generation cyclic peptides generally exhibited better binding to α_vβ₅ than the lead com-pound, with the exception of compound 1-m . All integrin binding activities of the first-generation library were lower than those of cilengitide, a standard integrin inhibitor.

Fig. 2. Result of the integrin binding SPR assay for 1st library. The y-axis shows the relative binding intensities of the RGD cyclic peptide library to integrin α_vβ₃ (black columnar) and α_vβ₅ (red columnar) immobilized on the chip. The x-axis represents different RGD cyclic peptides including 1st library, and the control cilengitide and c-(G7RGDLPET). Green and yellow dotted line reveals α_vβ₃ and α_vβ₅ binding intensities of the lead compound c-(G7RGDLPET), respectively.

To determine the binding affinities of the macrocyclic RGD-peptides for α_vβ₃ and α_vβ₅ integrins, we calculated the equilibrium dissociation constant (K_D) for the six best binders (1-A, 1-K, 1-v, 1-m, 1-f, 1-e) using the SPR assay. As shown in Table 1, compounds 1-K and 1-f exhibited significantly higher binding affinities for α_vβ₃ integrin compared to the other four candidates (1-A, 1-v, 1-m, 1-e), although their affinities remained lower than those of the positive controls. Concerning binding selectivity, 1-K demonstrated a favorable preference for α_vβ₃ over α_vβ₅, showcasing the highest selectivity index among the six cyclic and control peptides (see Table 1, α_vβ₅/α_vβ₃ ratio = 2.16 : 1).

Table 1. K _D of best RGD cyclic peptides of 1st and 2nd generation libraries (mean ± SD, n = 3).

Compounds	K D(μM)		Selectivity (αvβ5/αvβ3)
	αvβ3	αvβ5
Cilengitide	0.11 ± 0.02	0.15 ± 0.03	1.36 : 1
c-(G7RGDLPET)	0.90 ± 0.26	1.90 ± 0.35	2.11 : 1
1-A	5.16 ± 0.65	6.20 ± 0.30	1.19 : 1
1-K	1.93 ± 0.15	4.10 ± 0.30	2.16 : 1
1-m	5.10 ± 0.21	6.30 ± 0.20	1.24 : 1
1-f	1.93 ± 0.65	1.83 ± 0.15	1 : 1.06
1-e	2.67 ± 0.21	4.23 ± 0.61	1.56 : 1
1-v	6.26 ± 0.86	7.66 ± 1.35	1.24 : 1
1-A	0.29 ± 0.02	0.31 ± 0.02	1.07 : 1
2-C	0.32 ± 0.03	0.35 ± 0.03	1.09 : 1
2-c	0.14 ± 0.01	0.29 ± 0.03	2.07 : 1
2-y	0.27 ± 0.04	0.47 ± 0.03	1.74 : 1
2-t	0.52 ± 0.03	0.14 ± 0.02	1 : 3.71
2-k	2.73 ± 0.45	2.90 ± 0.26	1.07 : 1
2-n	1.10 ± 0.21	1.20 ± 0.11	1.09 : 1

To further confirm the binding activity and selectivity of compounds 1-K and 1-f towards α_vβ₃ and α_vβ₅ integrin receptors, cell-based competition inhibition assays were performed. Human embryonic kidney cells (HEK-293), which overexpress α_vβ₃ integrin,²⁹ and human ovarian cancer cells (SKOV-3), which overexpress both α_vβ₃ and α_vβ₅ integrins, were selected as model systems.³⁶ Fibrinogen served as a natural ligand for α_vβ₃ integrin, while HT-29 cells, which overexpress α_vβ₅ were used with vitronectin as the ligand.³⁷ As shown in Table 2, both compounds 1-K and 1-f demonstrated more potent inhibition in cell experiments compared to the lead compound c-(G7RGDLPET). The IC₅₀ value of compound 1-K increased by 2.9-fold (3.5 μM vs. 10.2 μM) for HEK-293 cells and by 1.3-fold (28.1 μM vs. 37 μM) for SKOV-3 cells. Notably, 1-K exhibited a weak binding affinity for α_vβ₅ integrin in HT-29 cells, with an IC₅₀ of 50 μM. Therefore, compound 1-K is identified as a selective antagonist of α_vβ₃ over α_vβ₅ and has been selected for further investigation.

Table 2. Cell adhesion inhibition activity of macrocyclic RGD-peptides in different cell lines (mean ± SD, n = 3).

Compounds	Fibrinogen (IC50, μM)		Vitronectin (IC50, μM) HT-29(αvβ5)
	HEK293(αvβ3)	SKOV-3(αvβ3/αvβ5)
Cilengitide	0.08 ± 0.08	0.12 ± 0.02	0.13 ± 0.03
c-(G7RGDLPET)	10.2 ± 0.91	37 ± 6.59	239 ± 8.47
1-K	3.5 ± 0.24	28.1 ± 5.43	50 ± 9.05
1-f	6.6 ± 1.36	28.1 ± 6.39	35.4 ± 4.79
2-A	2.39 ± 1.15	3.89 ± 0.91	22.38 ± 3.91
2-C	2.88 ± 0.93	15.84 ± 4.60	26.3 ± 8.00
2-c	0.91 ± 0.29	2.45 ± 1.07	12.3 ± 5.21
2-y	3.16 ± 0.71	8.12 ± 1.11	35.48 ± 9.49
2-t	9.77 ± 3.51	25.11 ± 4.78	31.62 ± 7.01
c-G7RGDLPK(N3)T	10.0 ± 0.78	38 ± 5.62	237 ± 7.57
2-c(N3)	0.90 ± 0.31	2.51 ± 0.43	12.7 ± 4.42

To enhance integrin-binding affinity, a second-generation library was designed based on compound 1-K and synthesized using a one-pot enzymatic on-resin in situ cyclization strategy. In this second-generation library, X₁ was fixed as l-lysine (K), while X₂ was substituted with various amino acids, including both d- and l-residues. The synthesis and characterization data are summarized in Table S2.†

The integrin binding profiles of the second-generation library were characterized by SPR, following established procedures. When X₂ was substituted with natural l-amino acids, only compounds 2-A, 2-F, and 2-C exhibited slightly stronger binding activities to α_vβ₃ compared to 1-K (Fig. 3). In contrast, when X₂ was substituted with d-amino acids, 63% of the cyclic peptides displayed stronger binding to α_vβ₃ compared to 1-K . Notably, three of these compounds (2-k, 2-y, and 2-c) outperformed the standard integrin antagonist cilengitide. These findings indicate that the configuration of the amino acid at the X₂ position significantly impacts α_vβ₃ integrin binding, consistent with previous results.^33,34

Fig. 3. Result of the integrin binding SPR assay for 2nd library. The y-axis shows the relative binding intensities of the RGD cyclic peptide library to integrin α_vβ₃ (black columnar) and α_vβ₅ (red columnar) immobilized on the chip. The x-axis represents different RGD cyclic peptides including 2nd library, and the control cilengitide and 1-k . Green and yellow dotted line reveals α_vβ₃ and α_vβ₅ binding intensities of the lead compound 1-k, respectively. The purple dotted line represents the α_vβ₃ binding intensities of positive control cilengitide.

The seven strongest binders from the second-generation library were selected to determine their kinetic parameters (K_D). As shown in Table 1, the K_D values for five cyclic RGD peptides (2-A, 2-C, 2-c, 2-y, and 2-t) for α_vβ₃ were approximately one order of magnitude lower than those of 1-K, indicating an enhancement in binding affinity of roughly 10-fold. Interestingly, their binding affinities for α_vβ₅ were also improved. Among these, compound 2-c exhibited the best selectivity, binding to α_vβ₃ approximately 2-fold stronger than to α_vβ₅. Conversely, compound 2-t displayed an inverse binding profile, favouring α_vβ₅ over α_vβ₃.

Cell-based competition inhibition assays were conducted to further evaluate the binding affinity and selectivity of the five strongest binders identified in the SPR experiments. As shown in Table 2, all tested compounds (2-A, 2-C, 2-c, 2-y, and 2-t) demonstrated stronger binding activities than the control compound c-(G7RGDLPET) in α_vβ₃-expressing cells. Among them, compound 2-c exhibited the lowest IC₅₀ values in HEK-293 (0.91 μM) and SKOV-3 (2.45 μM) cells expressing α_vβ₃ integrin, indicating the highest binding activity. The binding potency of 2-c was approximately 11-fold greater in HEK-293 cells and 15-fold greater in SKOV-3 cells compared to the control compound c-(G7RGDLPET). In HT-29 cells expressing α_vβ₅, 2-c demonstrated a potency approximately 19.4-fold higher than c-(G7RGDLPET). Interestingly, compound 2-c maintained strong selective binding for α_vβ₃ over α_vβ₅ (0.91 μM vs. 12.3 μM), with a selectivity ratio of 14-fold. These results suggest that systematically mapping the X₁X₂ motif at the C-terminus of cyclic RGD peptides provides an effective approach to optimizing integrin binding affinity and selectivity. Consequently, compound 2-c is identified as a preferred candidate for further development as a promising subtype integrin antagonist for functionalization with imaging probes and drug carriers.

Furthermore, there are certain discrepancies between the two experiments, SPR and cell-based competition inhibition assays. Cilengitide exhibited the lowest IC₅₀ values in all the three cel lines. However, the response signal of cilengitide binding with integrin, as determined by SPR, was comparable to those of other 2nd RGD peptides (Fig. 3). Moreover, the compound 2-t exhibited an inverse binding profile in SPR and cell experiments. These discrepancies may arise from the differing experimental conditions. The SPR assay is a method for detecting the interaction of integrin receptors with RGD cyclic peptide ligands in vitro molecular level. Although this method facilitates high-throughput screening, it does not mimic the real cellular interaction environment. Consequently, it should be complemented with adhesion inhibition assays at the cellular level to further clarify the affinity of RGD cyclic peptide.

Molecular docking studies of 2-c /integrin complex

To further explore the interaction mechanism between 2-c and integrin (α_vβ₃/α_vβ₅), molecular docking was performed. The three-dimensional structure of integrin α_vβ₃ was retrieved from the Protein Data Bank (PDB ID: 1L5G). For the integrin α_vβ₅, as no three-dimensional structure is available, SWISS-MODEL was used to construct the model based on the crystal structure of integrin α_vβ₃.³⁸ Molecular docking of 2-c with integrin (α_vβ₃/α_vβ₅) were performed by AutoDock Vina 1.2.0 (Fig. 4).³⁹ The binding energy of 2-c towards integrin were ranked as follows: 2-c /α_vβ₃ (−8.8 kcal mol⁻¹) <2-c /α_vβ₅ (−6.9 kcal mol⁻¹), consistent with the results of previous cell adhesion experiments. As shown in Fig. 4, there are more polar interactions between integrin α_vβ₃ and 2-c than between integrin α_vβ₅ and 2-c, including Arg: D217, Lys: N175, Glu: S90, and Thr: D84 (Fig. 4A). In contrast, there is only one hydrogen bond formed between the α_v subunit of integrin α_vβ₅ and compound 2-c (Q150: Glu), while the T199 and N200 on β₅ subunit established polar interactions with Arg and Asp, respectively (Fig. 4B). These results indicate that the binding affinity of 2-c toward integrin α_vβ₅ is weaker than that toward integrin α_vβ₃.

Fig. 4. Structural model of cyclic peptide 2-c and binding detail with integrin receptor α_vβ₃ (A) or α_vβ₅ (B). Detailed bonding between 2-c (cyan) and integrin receptor (α_v: orange, β₃/β₅: green), interaction forces including hydrogen bonding and van der Waals. Shadow refers to the solvent accessibility area, digits refer to the residue number, and three-letter residues refer to residues from cyclic peptides. One-letter residues are from receptor.

Application of RGD-based peptide 2-c for cancer imaging and therapy

To demonstrate the potential application of the RGD-based peptide 2-c in imaging integrin α_vβ₃ expressed on cancer cells, we synthesized the cyclic peptide 2-c(N3), where the lysine residue in LPKTG motif on 2-c was replaced with azido-modified lysine (Fig. S7, S8, S11 and S12†). The parent cyclic peptide c-G7RGDLPK(N₃)T (Fig. S5, S6, S9 and S10†) was used as a control. Azido modification of lysine on 2-c and c-G7RGDLPK(N₃)T did not affect the integrin binding capacity (Table 2). In vitro imaging experiments were conducted by incubating α_vβ₃-positive SKOV-3 cells with either 2-c(N₃) or c-G7RGDLPK(N₃)T. DBCO-tagged Cy5 was used to stain the cell surfaces bound by 2-c(N₃) and c-G7RGDLPK(N₃)T via click chemistry. The fluorescence microscopy imaging results are presented in Fig. 5. The optimized RGD cyclic peptide 2-c(N₃) exhibited significantly higher staining levels than the lead compound c-G7RGDLPK(N₃)T. This result indicates that cyclic peptide 2-c(N₃) has a stronger integrin-binding capability, making it suitable for imaging integrin-positive cancer cells.

Fig. 5. Cell imaging assays of c-G7RGDLPK(N3)T (A) and 2-c(N3) (B) and followed by stain with DBCO-tagged Cy5. The images were processed via ImageJ (LUT: Fire). The contrast is shown in arbitrary units (au) 0, no fluorescence; 50, maximum fluorescence.

Peptide–drug conjugates (PDCs) have demonstrated significant promise in cancer treatment due to their ability to deliver potent anticancer drugs specifically to tumour cells, reducing systemic toxicity and enhancing therapeutic outcomes.⁴⁰ MMAE, a potent cytotoxic agent, is commonly used as a payload in antibody-drug conjugates (ADCs) for cancer therapy.⁴¹ In this study, we synthesized the PDC 2-c(MMAE) from DBCO-PEG4-VC-PAB-MMAE and 2-c(N₃) using copper-free click chemistry (Fig. 6A and S13 and S14†). The incorporation of the PEG4 chain enhances the aqueous solubility and stability of the drug. Conversely, the VC-PAB linker promotes the intracellular release of MMAE in the target cell. We selected α_vβ₃-positive SKOV-3 cells to evaluate the cytotoxicity of 2-c(MMAE), with MMAE used as a control. As shown in Fig. 6B, the IC50 values for 2-c(MMAE) and MMAE were 0.25 μM and 1.34 μM, respectively, indicating that 2-c(MMAE) is more effective at inhibiting cancer cell growth and inducing cell death. This result also demonstrates that cyclic peptide 2-c(N₃) could serve as a novel carrier for targeted drug delivery for therapeutic purposes.

Fig. 6. Cytotoxicity assays of the MMAE and 2-c(MMAE). (A) Chemical structure of the 2-c(MMAE). (B) The toxic effects of compounds MMAE and 2-c(MMAE) on SKOV-3 cells.

Conclusions

In this study, RGD cyclic peptides libraries were synthesized in good yield via one-pot Srt A-mediated on-resin in situ cyclization strategy. The α_vβ₃/α_vβ₅ binding activities of RGD cyclic peptides libraries were successfully screened by SPR and cell adhesion assays, eventually yielding a high-affinity RGD cyclic peptide 2-c with good selectivity for α_νβ₃ over α_νβ₅ (IC₅₀ = 0.91 μM vs. 12.3 μM, 14-fold). In the cytotoxicity assays and cell imaging experiment, we validated the targeting effect of compound 2-c to integrin over-expressed tumor cells. This peptide represents an attractive structural platform to target integrin for biomaterial functionalization, therapeutic applications, or as tracers.

Data availability

The authors confirm that the data supporting the findings of this study are available within its ESI.†

Author contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Xiaozhong Cheng: writing, methodology, conceptualization, synthesis. Chen Li: synthesis. Haofei Hong: expression of Sortase A. Zhifang Zhou: data analysis. Zhimeng Wu: writing – review & editing, supervision, funding acquisition.

Conflicts of interest

There are no conflicts to declare.