Optimization of the 2-arylquinazoline-4(3H)one scaffold for a selective and potent antitrypanosomal agent: modulation of the mechanism of action through chemical functionalization

Abstract

We sought to identify a potent and selective antitrypanosomal agent through modulation of the mechanism of action of a 2-arylquinazoline scaffold as an antitrypanosomal agent via chemical functionalization at the 4-position. We wished to use the: (i) susceptibility of trypanosomatids towards nitric oxide (NO) and reactive oxygen species (ROS); (ii) capacity of the 4-substituted quinazoline system to act as an antifolate agent. Three quinazolin-based moieties that differed from each other by having at the 4-position key pharmacophores targeting the induction of NO and ROS production were evaluated in vitro against Leishmania infantum and Trypanosoma cruzi parasites and their modes of action were explored. Replacement of an oxygen moiety at the 4-position of the antifolate 2-arylquinazolin-4(3H)one by hydrazinyl and 5-nitrofuryl-hydrazinyl pharmacophores enhanced antitrypanosomatid activity significantly due to promotion of an additional mechanism beyond the antifolate response such as NO or ROS production, respectively. Among the three types of chemical functionalization, the 5-nitrofuryl-hydrazinyl moiety generated the most potent compounds. Compound 3b was a potential candidate thanks to its sub-micromolar response against the promastigotes/amastigotes of L. infantum and epimastigote of T. cruzi, moderate toxicity on macrophages (J774.1), good selectivity index (∼15.1–17.6) and, importantly, non-mutagenic effects. 2-Arylquinazoline could be an attractive platform to design new anti-trypanosomatid agents with the use of key pharmacophores.

We identified a potent and selective antitrypanosomal agent through modulation of the mechanism of action of a 2-arylquinazoline scaffold as an antitrypanosomal agent via chemical functionalization at the 4-position.

Introduction

Leishmaniasis and Chagas disease are neglected tropical diseases (NTDs) caused by obligate intracellular parasites of Leishmania spp. and Trypanosoma cruzi, respectively, which are transmitted to humans by insects.^1,2 Leishmaniasis is prevalent in 88 countries, with 350 million people under risk to acquire the disease and between 0.7 million to 1.0 million new Leishmaniasis cases. Chagas disease is localized primarily in 21 countries of Latin America with approximately 6–7 million infected and >70 million people at risk.³ In 2022, the World Health Organization revealed that 26 000 to 65 000 deaths are registered annually for Leishmaniasis and ∼14 000 deaths are reported for Chagas disease.⁴

These NTDs represent a significant challenge for medicinal chemists by the high adaptability and genetic plasticity of Leishmania spp. and T. cruzi parasites. For example, the parasite from Leishmania spp. can elude the defense mechanism by which macrophages change polarization from M1 to M2, resulting in a macrophage more focused on differentiation tasks than on defense actions.^5–7 Moreover, Leishmania spp. can adapt to hostile environments, being able to develop resistance in a short time against drugs. Also, beyond the resistance and adaptability of these parasites, the therapeutic options are limited. Only a few approved drugs are currently employed for the treatment of these diseases: (i) pentavalent antimonials (e.g. glucantime A and pentostam B, Chart 1) against Leishmaniasis in its different forms and (ii) nifurtimox F and benznidazole G against Chagas disease (Chart 1). A second line of leishmanicidal agents, including amphotericin B C, pentamidine D, and miltefosine E (Chart 1) as well as anti-Chagas drugs such as azole antifungals (e.g. posaconazole, ketoconazole, ravuconazole) are frequently used if first-line treatment fails.^8,9 However, these drugs present several disadvantages in terms of toxicity (heart, liver, kidneys), cost, therapeutic efficacy, prolonged administration treatment times (30–60 days) and parasite- resistance.

Chart 1. Structures of main approved drugs. (A) Glucantime, (B) pentostam, (C) amphotericin B, (D) pentamidine and (E) miltefosine against Leishmaniasis. (F), nifurtimox and (G) benznidazole are anti-Chagas drugs.

Consequently, because of the several drawbacks of approved drugs against NTDs, new nontoxic antitrypanosomal drugs are urgently needed for the efficacious treatment of these prevalent diseases. The knowledge gained in the biology of trypanosomatids has elicited a series of potential targets (enzyme or metabolic pathways) for the rational design of novel targeted antitrypanosomal agents. Among the therapeutic targets are: the folate pathway; inhibition of essential enzymes in parasite survival (e.g. cruzipain, superoxide dismutase, N-myristoyltransferase, trypanothione synthetase, cyclin-dependent kinases); inhibition of sterol biosynthesis.¹⁰ The folate pathway has been widely explored with the construction of active compounds. Folates are essential for several important biochemical reactions in the metabolism of parasites. The enzymes involved in the folate pathway are important targets to design antifolate agents.¹¹ Dihydrofolate reductase (DHFR) acts in conjunction with the enzyme thymidylate-synthase (TS), constituting a bifunctional DHFR–TS enzymatic system.^12,13 DHFR leads to the reduction of dihydrofolate to tetrahydrofolate, which is an important co-factor that the TS enzyme uses to catalyze the conversion from dUMP to dTMP. The latter is a vital metabolite in DNA synthesis and its low production affects replication processes, leading to cell death.¹¹ The enzymatic machinery of DHFR–TS is modulated by the enzyme pteridine reductase 1 (PTR1), acting as a “metabolic bypass” of DHFR inhibition.¹² Thus, folates are reduced by bifunctional DHFR–TS as well as by the enzyme PTR1 within the parasite.^14–18 Traditionally, pteridin and quinazolin scaffolds have been used as chemical platforms to design new antifolate agents as competitive inhibitors of DHFR and PTR1, and their derivatives have displayed significant antitrypanosomal activities with good inhibitory responses.^19–25

Beyond enzymatic targets, these parasites are highly susceptible to small chemical species such as nitric oxide (NO)^26–27 and reactive oxygen species (ROS).²⁸ NO production can be achieved via drug decomposition or immunologic activation with the use of NO-donors and NO-inductors, respectively.^29–38 Use of NO-donors is beneficial because they can generate NO in situ without altering immunological pathways. Some NO-modulators have been reported to be potential antitrypanosomatid agents.^29–33 Most of them are represented by metallic complexes containing NO ligands, which release NO through cleavage from the metal–NO bond.³⁴ Other examples are organic NO-donor heterocycles, which can release NO via opening of the hetero-ring-masked NO-releasing moiety. Previously, we described the anti-infective activities of a series of furoxans against T. cruzi, Leishmania spp. and Mycobacterium spp. without toxic effects, absence of mutagenicity, and a relationship between bioactivity and ability to produce NO.^35–37 More recently, we found that the appropriate location of hydrazine in an appropriate electron-deficient heterocycle (e.g. phthalazine) was crucial to generate a NO-donor via hydrazine decomposition.²⁵

Promotion of ROS production is a secure way to obtain active compounds.²⁸ In general, ROS-promoters are based on heterocycles bearing (among others) a nitroheterocyclic moiety or N-oxide moieties.^38–39 Recently, we showed that a 5-nitrofuryl-moiety connected to a highly π-electron-deficient conjugation (4-chloro-1-phthalazinyl system) was essential to display high biological activity against parasites from Leishmania spp. and T. cruzi.^40–42 However, the high toxicity of this type of chemical system obligates the proposal of new strategies to improve biological selectivity. We have taken into account that trypanosomatids are significantly more susceptible to oxidative stress than host cells. Herein, we proposed that use of highly reducible pharmacophores (i.e. 5-nitrofuryl groups) connected to an active heterocyclic scaffold with a highly electron-deficient character could be an interesting way to design new active and selective antitrypanosomal agents targeting oxidative stress. Then, we planned the synthesis and biological evaluation of a series of 2-arylquinazolin-4-(furfurylidene)hydrazones 3 (Scheme 1) to establish a comparison with their previous active versions of 2-arylquinazolin-4(3H)-ones 1 and 2-arylquinazolin-4-hydrazines 2 (Fig. 1) from biological and mechanistic viewpoints. The designed 2-arylquinazolin-4-(furfurylidene)hydrazones may act through at least two mechanisms of action: (i) inhibition of the biosynthesis of folates by 2-arylquinazoline recognition as the folate substrate;²³ (ii) oxidative stress induced via bio-reduction of a nitro moiety.^38–41 Compounds were examined biologically against Leishmania infantum and T. cruzi parasite strains. In addition, we undertook a detailed mechanistic study focused on three tentative mechanisms: (i) antifolate activity; (ii) promotion of NO generation; (iii) stimulation of ROS production. Further biological studies (parasite proliferation in vitro and mechanistic assays) were carried out in the presence of metal cations, which sought to favor the nitro-reduction and water-solubility of 2-arylquinazolin-4-(5-nitrofurfurylidene)hydrazones. The effect of metal or metal-complexation represents a special topic in the design of antitrypanosomatid agents. The antitrypanosomal activity of organic systems or their physicochemical properties can be improved via ligand–metal complexation.⁴³

Scheme 1. Synthesis of 2-arylquinazolin-4-(furfurylidene)hydrazones 3a–h. Conditions: 2-arylquinazolin-4-hydrazines 2a–g (1.0 equiv.), corresponding furylcarbaldehyde (1.2 equiv.), HCl (aq) (1 M), at 60 °C for 15 min.

Fig. 1. Rational design of 2-arylquinazolin-4-(furfurylydene)hydrazones (3) from previous versions of 2-arylquinazolin-4(3H)ones (1) and 2-arylquinazolin-4-hydrazines (2).^23,25.

Results and discussion

Design and synthesis

The design of the target 2-arylquinazolin-4-(5-nitro-2-furfurylidene)hydrazones 3a–g (Scheme 1) was inspired by the molecular structure of a series of active 2-arylquinazolin-4(3H)-ones (1)²³ and 2-arylquinazolin-4-hydrazine (2),²⁵ which possess recognized activity against trypanosomatids (Leishmania spp. or T. cruzi). To construct compounds 3a–g, the oxygen atom or a hydrazine moiety at the 4-position of the 2-arylquinazoline core was replaced by the 4-(5-nitro-2-furfuylidene)hydrazinyl group. Selection of 2-aryl-functionalities (Scheme 1) was focused on those that led to the best activity/toxicity profile in the previous 2-arylquinazolin-4(3H)-ones 1 and 2-arylquinazolin-4-hydrazine 2: phenyl, 4-Cl-, 4-Br, 4-F-, and 4-Me-phenyl.^23,25 Other electron-deficient aryl-functionalities such as 3-chlorophenyl (3f) or 3-bromophenyl (3g), as aryl substitution at the 2-position, were also considered because an electron-deficient aryl group could favor the hypothesized bio-reduction of the nitro-moiety thanks to its ability to increase the electron-deficient character of the quinazoline core.^40–41 In addition, a 5-denitrofuryl derivative was synthesized to study the relevance of the nitro-group in the antitrypanosomal activity and mechanism of action (3h, Scheme 1). With the derivatives 3a–h in hand, a comparison with their relatives from families 1 and 2 (Fig. 1) was established to evaluate the role of 4-functionalization in modulation of the mechanism of action. It was essential to identify a potent and selective compound.

2-Arylquinazolin-4-(furfurylidene)hydrazones 3a–h were prepared from their corresponding 2-aryl-quinazolin-4-hydrazines 2a–g following reported procedures with a few modifications^25,40,41 (Scheme 1). Then, hydrazine compounds 2a–g were reacted with the corresponding aldehyde, 5-nitrofurfural, or furfural (1.2 equiv.), under acid aqueous media to give the desired 2-arylquinazolin-4-(furfurylidene)hydrazones 3a–h in excellent yields (83–93%) (Scheme 1). The final products were purified and characterized. The experimental details and spectroscopic information can be found in ESI.† A comparative analysis based on nuclear magnetic resonance (¹H-NMR, correlated spectroscopy (COSY) and nuclear Overhauser effect (NOE) experiments) spectra to assign the prevalence of the geometric isomer, (Z)-isomer or (E)-isomer was done. For example, nuclear Overhauser effect spectroscopy (NOESY) experiments on derivative 3f confirmed spatial correlations between the tautomeric NH-hydrogen and the ylidenic proton, which supported the presence of the E-isomer in the mixture. The ylidenic proton in the E-isomer showed a typical chemical shift of ∼8.5 ppm for most of the studied compounds 3a–h, whereas a typical chemical shift at 8.9 ppm was found for the Z-isomer of these 2-arylquinazolin-4-(furfurylidene)hydrazones. A discussion based on 1H-NMR and NOE experiments can be found in section 7.1 of ESI.† For compounds 3a–h, a specific isomeric mixture was found (Scheme 1, section 2 of ESI†). In particular, derivatives 3a and 3e displayed a remarkable E/Z isomeric mixture having percentages of 31% and 69% for the Z- and E-isomers of 3a, respectively, and by about 32% and 68% for the Z- and E-isomers of 3e, respectively. Other derivatives, such as 3b, 3c and 3d, showed a discrete mixture with predominance of the E-isomer by about 87–90%. Derivatives 3f, 3g, and 3h showed a clear predominance of the E-isomer (>95%). The 2-arylquinazolin-4-hydrazines 2a–g were prepared as described previously²⁵ from 2-aryl-quinazolin-4(3H)-ones 1.⁴⁴

Biological evaluation

All new synthesized compounds 3a–h were initially tested against the promastigote of L. infantum and against the epimastigote of T. cruzi (Table 1). In general, compounds showed higher activities against T. cruzi parasites than against L. infantum. In particular, compounds 3a and 3f displayed the best antitrypanosomal responses, with half-maximal inhibitory concentration (IC₅₀) values of ∼800 nM against L. infantum parasites and 600 nM against T. cruzi parasites. Compound 3b also displayed good antitrypanosomal activity against L. infantum and T. cruzi, with IC₅₀ values of 4.56 and 4.08 μM, respectively. Biological activity (IC₅₀ > 25 μM) was not found for the remaining nitro-compounds 3c, 3d, 3e and 3g for both parasites. Importantly, no antitrypanosomal activity of compounds 3c, 3d, and 3g seemed to be associated with the observed poor solubility, which was low in aqueous solution and common organic solvents such as DMSO. Meanwhile, compound 3h, which represented the unique non-nitro derivative in the series, displayed moderate biological responses, giving IC₅₀ values of 9.07 and 21.0 μM against L. infantum and T. cruzi parasites, respectively. These results suggested that the mode of action of compound 3h may be substantially different from that derived from the active compound featuring nitro-furyl derivatives 3a, 3b, or 3f. Nitro-derivatives 3a and 3f were 10-fold more active than reference drugs, whereas compound 3b was near 2-fold more active than reference drugs. Compound 3h was in the same range as Miltefosine (10.10 μM) and Nifurtimox (7.7 μM).

In vitro anti-trypanosomatid activity of compounds 1a–g, 2a–g and 3a–h against L. infantum promastigote and T. cruzi epimastigote.

Entries	R
		IC50 (μM)				IC50 (μM)		Log Pe	S f
		L. infantum	T. cruzi	L. infantum	T. cruzi	L. infantum	T. cruzi
A	–C6H5	13.32 ± 0.93	>25.0	12.67 ± 0.74	>25.00 (42.7)a	0.88 ± 0.04	0.63 ± 0.03	3.19	7.99
B	–p-C6H4F	8.67 ± 0.51	22.67 ± 1.32	5.01 ± 0.31	17.95 ± 1.21	4.56 ± 0.31	4.08 ± 0.26	3.51	5.11
C	–p-C6H4Cl	4.88 ± 0.32	18.05 ± 1.20	1.56 ± 0.09	11.48 ± 0.49	> 25.0	> 25.0	3.73	3.21
D	–p-C6H4Br	6.89 ± 0.31	19.78 ± 1.45	2.13 ± 0.12	14.32 ± 0.87	> 25.0	> 25.0	3.82	2.67
E	–p-C6H4Me	5.23 ± 0.30	>25.0	1.20 ± 0.08	17.99 ± 1.11	> 25.0	> 25.0	3.40	7.23
F	–m-C6H4Cl	>25.0	>25.0	4.55 ± 0.29	>25.00 (38.0)a	0.77 ± 0.03	0.69 ± 0.03	3.73	7.11
G	–m-C6H4Br	>25.0	>25.0	21.14 ± 1.21	>25.00 (21.4)a	> 25.0	> 25.0	3.82	5.67
H	–p-C6H4Fb	—	—	—	—	9.07 ± 0.45	21.0 ± 1.21	3.94	7.87
9	Nifurtimoxc	—	7.7 ± 0.5
10	Miltefosined	10.10 ± 0.46	—

^aIn parenthesis, corresponding PGI (percentage growth inhibition of parasite cell) values at 25 μM are shown.

^bNitro-moiety is replaced by hydrogen atom for an furyl ring.

^cReported antichagasic.⁴⁵

^dLeishmanicidal.²⁵

^eLog P values were calculated from Swiss-ADME website.⁵³

^fSolubility (μM) determined from UV-Vis spectroscopy.

A comparison of these 2-arylquinazolin-4-hydrazones 3a–h with their analogues of 2-arylquinazoline functionalized at the 4-position with oxo (1a–g) or hydrazinyl (2a–g) moieties revealed that incorporation of the 5-nitrofuryl-hydrazinyl pharmacophore promoted significant enhancement of antitrypanosomal activity. Against parasites from Leishmania spp., most of the active 5-nitrofuryl-hydrazinyl showed IC₅₀ values <1 μM, whereas hydrazinyl- and hydroxyl-analogues showed IC₅₀ values ranging from 1.2 to 5.0 μM and from 5.0 to 10.0 μM, respectively. With regard to anti-T. cruzi activity, 4-hydroxylquinazoline displayed IC₅₀ values >18 μM, whereas replacement of the 4-oxygen by the hydrazinyl or 5-nitrofuryl-hydrazinyl moieties at the 4-position promoted improvement in activity, giving IC₅₀ values of 11–18 μM and <1 μM for their respective derivatives. Importantly, parasites from Leishmania spp. were more susceptible to the action of 2-aryl-quinazolin-4(3H)ones (1a–g) and 4-hydrazinyl (2a–g) than T. cruzi parasites, whereas the 5-nitrofuryl-hydrazinyl analogue (3a–g) was barely more active against T. cruzi parasites than against Leishmania spp. parasites.

Another important structural feature to discuss was the influence of aryl-substitution. In hydroxyl- and hydrazinyl-quinazolines derivatives, the electron-deficient aryl-group (4-Cl and 4Br) generated, in general, the most active compounds in these compound-families, whereas unsubstituted and 3-Br-phenyl derivatives were the least active. Meanwhile, in 5-nitrofuryl-hydrazinyl analogues, the effect of the aryl-substituent seemed to have a role in biological activity, but it also seemed to be affected by the solubility of the compound. For example, incorporation of a halogen (4-Br, 4-Cl) and 4-Me at the 2-phenyl moiety generated the least soluble agents, which significantly compromised their biological activities (Table 1). A comparison between the apparent most soluble 2-arylquinazolin-4-(furfurylydene)hydrazones 3a, 3b, 3e, and 3f revealed that the presence of electron-withdrawing moieties at the aryl ring (e.g. H-, 4-F- and 3-Cl-, respectively) was essential to generate active compounds. It seemed that a highly electron-deficient core, such as quinazoline, connected to the 5-nitrofurylidene motif through a hydrazinyl linker could be highly convenient to generate a more susceptible bio-reductive nitro group. Previously, we found that a 5-nitrofuryl moiety connected to highly electron-deficient fused-heterocyclic, such as 4-chlorophthalazine, generated compounds highly active against trypanosomatids (e.g. Leishmania braziliensis or T. cruzi)^40–41 involving a tentative redox mechanism. Hence, a highly electron-deficient fused-heterocyclic could be an attractive platform to connect to a 5-nitrofurfurylidenehydrazinyl chain to generate potent antitrypanosomal agents. Then, we thought that the significant biological response of the active compounds 3a–h against L. infantum and T. cruzi parasites was associated with the aryl substitution to enhance the electron-deficient character of the nitro group, which is essential for its effective bio-reduction. More details are discussed in the “mechanism of action” section.

Next, the most active derivatives (3a, 3b, 3f, and 3h) against both parasites were selected for the cytotoxic evaluation on murine J774.1A macrophages. Selectivity indices were calculated as a ratio between the half-maximal cytotoxicity concentration (CC₅₀) value from macrophages and IC₅₀ values from parasites (L. infantum or T. cruzi). Cytotoxicity and selectivity index results are listed in Table 2. Compounds 3b and 3h were the least toxic compounds, with CC₅₀ values of 69.06 and 119.15 μM, whereas derivatives 3a and 3f displayed moderate toxicities, with CC₅₀ values of 15.06 and 41.55, respectively. Compounds 3a and 3f displayed the best selectivity index values of 17.1 and 54.0 with regard to the Leishmania model, respectively, being at least two-times more selective than miltefosine under identical conditions whereas, against T. cruzi, the selectivity index values were 23.9 and 60.2, respectively, with the derivative 3f being two-times more selective than nifurtimox. Meanwhile, the derivative 3b displayed good selectivity index values of 15.1 and 16.8 against L. infantum and T. cruzi, respectively. The non-nitro-derivative 3h was more selective than Miltefosine against L. infantum (near 1.5-times), but less selective than Nifurtimox against T. cruzi. Compared with 4-hydroxyl and 4-hydrazine-derivatives, 5-nitrofuryl-hydrazinyl functionalization was 10- to 20-fold more selective than the corresponding 2-arylquinazolin-4(hydroxyl/hydrazinyl) derivatives.^23,25 Hence, 2-arylquinazolin-4-(2-furylidene)hydrazones could be interesting scaffolds for further biological evaluations.

In vitro cytotoxicity using macrophage J774.1A, antiamastigote activity, and the relative selectivity index of compounds 3a–h.

Compds (X, Y)	CC50a (μM)	S.I.b		IC50c (μM) (S.I.)
	Macrophage	Promastigote	Epimastigote	Axenic amastigote
	J774.1a	L. infantum	T. cruzi	L. infantum
3a (H, NO2)	15.06 ± 0.11	17.1	23.9	0.93 ± 0.04 (16.2)
3b (4-F, NO2)	69.06 ± 3.99	15.1	16.8	3.52 ± 0.35 (19.6)
3f (3-Cl, NO2)	41.55 ± 1.87	54.0	60.2	0.89 ± 0.03 (46.7)
3h (4-F, H)	119.15 ± 8.12	13.1	5.7	11.12 (10.7)
Miltefosined	89.21 ± 7.01	8.9	—	8.69 ± 0.03 (10.1)
Nifurtimoxe	280.00 ± 4.00f	—	36.3	—

^aThe results are the mean of three independent experiments.

^bS.I.: selectivity index calculated as CC₅₀/IC₅₀ (promastigote or epimastigote).

^cSelectivity index was calculated as CC₅₀/IC₅₀ (axenic amastigote).

^dMiltefosine as antileishmanial reference drug (from ref. 25).

^eNifurtimox as antichagasic reference drug.

^fFrom ref. 45.

Finally, the antiamastigote activity of the most active 2-arylquinazolin-4-(furfurylidene)hydrazones, 3a, 3b, 3f, and 3h, was evaluated using an axenic amastigote of L. infantum (Table 2). In general, these compounds showed a significant antiamastigote response, which was in the same range as the antipromastigote response (Table 1). Compounds 3a and 3f showed the most significant response, displaying IC₅₀ values of 930 and 890 nM, whereas compounds 3b and 3h displayed IC₅₀ values of 3.52 and 11.12 μM, respectively. As a function of these antiamastigote responses and macrophage-cytotoxicities, compounds 3f, 3b, and 3a emerged as potential candidates for further evaluation thanks to their good selectivity index (macrophages/amastigote) values of 46.7, 17.6 and 16.2, respectively. In particular, compounds 3b and 3f were selected for further evaluation (including the Ames test) to analyze mutagenic effects (see the Drug-like profile section).

Mechanism of action

We wished to elucidate the role of chemical functionalization at the 4-position in the promotion of a specific type of mechanism of action. Three tentative mechanisms were explored: (i) antifolate response; (ii) production of NO; (iii) promotion of ROS. Key (most active) compounds were selected for each compound-family: (i) 1b and 1c from the 2-arylquinazolin-4(3H)ones 1a–h; (ii) 2c and 2d from the 2-arylquinazolin-4-hydrazines 2a–h; (iii) compounds 3a, 3b, 3f, and 3h from 2-arylquinazolin-4-(furfurylidene)hydrazones.

Antifolate activity was determined analyzing the change in the IC₅₀ value in the presence or absence of d,l-folic acid (FA) (0, 50, 200, or 1000 nM) for both models of promastigote L. infantum and epimastigote T. cruzi (Fig. 2A and B).^23–25 An increase of the IC₅₀ in the presence of FA with regard to FA-free cultures implied that the compound acted as an antifolate, being a substrate of any specific enzyme in the folate pathway. An appreciable increase in IC₅₀ values was found for 2-arylquinazolin-4(3H)ones (1b and 1c) and 2-arylquinazolin-4-hydrazines (2c and 2d), with an increase in the FA amount for both parasites. Against promastigote Leishmania, the relative increase in the IC₅₀ value was less significant than that found against the T. cruzi parasite, but appreciable for both chemical systems in the presence of 1000 nM of FA, increasing from 8.67, 4.88, 1.56 and 2.13 μM to 9.78, 6.22, 2.05 and 2.83 μM for compounds 1b, 1c, 2c, and 2d, respectively, representing relative increases of 14, 27, 31 and 33%, respectively (Fig. 2A). Meanwhile, within 2-arylquinazolin-4-(furfurylidene)hydrazones, a modest increase in IC₅₀ values (<10%) was found for the nitro-derivatives 3a and 3f, and only compounds 3b and 3h showed a very relevant antifolate response (increases in IC₅₀ values of 31 and 51%, respectively). On the other hand, for T. cruzi epimastigotes, the IC₅₀ values clearly increased in the presence of 1000 nM of FA, from 22.67, 18.05, 11.48 and 16.78 μM to 33.21, 24.18, 19.12 and 25.23 μM for compounds 1b, 1c, 2c, and 2d, representing relative increases of 47, 34, 42 and 48%, respectively (Fig. 2B). The IC₅₀ values increased in the presence of 1000 nM of FA for compounds 3b and 3h from 4.08 and 21.0 μM to 6.54 and 33.13 μM, representing relative increases of 38 and 37%, respectively. That last result suggested that 2-arylquinazolin-4-(5-nitrofurfurylidene)hydrazones also presented an antifolate response.

Fig. 2. Antifolate activities of 2-arylquinazolin-4(3H)ones 1b and 1c, 2-arylquinazolin-4-hydrazine 2c and 2d, and 2-arylquinazolin-4-[2-(5-nitrofurfurylidene)]-hydrazones 3a–h against the promastigote of L. infantum (A) and epimastigote of T. cruzi (B). Antifolate response was tested using FA at 0, 50, 200, or 1000 nM. Molecular docking on the active site of PTR1–Leishmania (PDB ID: 1E7W) (C) and DHFR–TS/T. cruzi (3INV) (D) for the most active 1c (cyan), 2c (green), 3f (orange), and 3h (white) and reference antifolates (MTX 1292 is yellow in C; C1029 is yellow in D).

To obtain more information on antifolate activity as a function of 4-functionalization, molecular docking for the key compounds 1c, 2c, 3f, and 3h (the most active in their respective families) was done on proteins related to the folate pathway, such as PTR1 of Leishmania major (PDB code: 1E7W)^47a and DHFR–TS of T. cruzi (PDB code: 3INV).^47b Experiments were undertaken following reported docking studies with some modifications.^23,42 Both proteins showed two cavities, which located methotrexate (MTX) or C50 ligand into their corresponding active binding sites, PTR1–Leishmania (cavities of MTX301 and MTX1292), and DHFR–TS/T. cruzi (cavities of C50 601 and C50 1029), respectively. The ligand C50 is an antifolate agent based on purine heterocycles.¹¹ Docking results are illustrated and listed in Fig. 2C and D and Table 3, respectively. The specific hydrogen-bonding interactions of compounds with essential residues into the binding sites of PTR1–Leishmania protein and DHFR–TS/T. cruzi protein are shown in Fig. S5 and S6,† respectively. Beginning our analysis with Leishmania–PTR1 with a special focus on the MTX 1292 active site that showed the highest affinities, compounds 1c (−10.13 kcal mol⁻¹) and 2c (−10.16 kcal mol⁻¹) displayed significantly higher binding affinities than MTX (−8.87 kcal mol⁻¹), whereas 2-arylquinazolin-4-(furfurylidene)hydrazones 3f (−8.87 kcal mol⁻¹) and 3h (−9.01 kcal mol⁻¹) showed lower binding energies that were comparable in magnitude to that given by MTX (9.01 kcal mol⁻¹). These findings were in concordance with the location of compounds in the active site of the protein. Compounds having the highest binding affinities (1c and 2c) were located inside the active site with a similar spatial disposition of the quinazoline core to the MTX reference (Fig. 2C, right). Meanwhile, compounds with lower protein affinities, 3f and 3h, were placed partially outside the active site with a spatial disposition of the quinazoline core in an opposite orientation to that of MTX. Importantly, despite compounds 1c and 2c showing a similar spatial orientation in active site to that of MTX, the test compounds presented different contacts into the active site with residues compared with MTX (Table 3).

Binding affinities of selected compounds 1c, 2c, 3f, and 3h with essential residues into the active site of PTR1–Leishmania protein and DHFR–TS/T. cruzi.

Entries	Compd.	Energies in binding sites (kcal mol−1)				H-interactiona
		PTR1–Leishmania		DHFR–TS–T.cruzi
		MTX 1292b	MTX 301c	C601d	C1029e	MTX 1292	C1029
1	MTX	−8.87	−7.85	−9.11	−9.54	Tyr-455 (2.85 Å), Tyr-452 (2.67 Å), Ser-385 (2.64 Å)	Asp-561 (2.83 Å), Ile-667 (2.70 Å)
2	1c	−10.13	−8.44	−10.18	−10.65	No hydrogen bonding	Arg-566 (two bindings at 2.87 Å; 2.72 Å)
3	2c	−10.16	−8.93	−10.33	−11.14	Arg-300 (2.79 Å)	Arg-566 (two bindings: 2.30 Å; 2.78 Å), Phe-601 (2.99 Å)
4	3f	−8.87	−8.79	−11.24	−10.09	No hydrogen bonding	Arg-566 (2.42 Å), Ser-553 (2.90 Å)
5	3h	−9.01	−8.89	−10.33	−10.06	Tyr-452 (2.99 Å), Tyr-452 (2.25 Å)	Arg-566 (2.41 Å), Ser-533 (2.90 Å)

^aHydrogen bond interactions.

^bBinding site in cavity 1 for methotrexate.

^cBinding site in cavity 2 for methotrexate.

^dBinding site in cavity 1 for antifolate C50 and.

^eBinding site in cavity 2 for antifolate C50.

With regard to DHFR–TS/T. cruzi, key compounds showed the highest binding affinities for the binding sites of C50–1029 and C50–601 ligands. Focusing on binding site of the C50–1029 ligand, compounds 1c and 2c also exhibited slightly higher binding affinities than 2-arylquinazolin-4-(furfurylydene)hydrazones 3f and 3h. Similar to PTR1–Leishmania, the higher binding affinities could be attributed to the good entrance and partial entrance of 1c and 2c inside the active site. Also, all studied 2-arylquinazolines showed higher affinities with docking energies between −10.06 and 11.14 kcal mol⁻¹ than the reference C50 compound (−9.54 kcal mol⁻¹ in the C50–1029 site). Importantly, most of the studied compounds showed similar intermolecular contact, having hydrogen bonding with residues (Arg-566 and Ser-553) into the active site and non-covalent π-interaction (π–π, Ar–H–π) with the phenyl ring of the Phe-601 residue (Fig. S6†). Then, we demonstrated that all tested chemical systems (1, 2, and 3) had an antifolate response which could be derived from an effective interaction with key residues into the active sites of PTR1/Leishmania or DHFR/T. cruzi. The antifolate activity seemed to affect the cell viability of L. infantum and T. cruzi discretely, being appreciable for compounds with IC₅₀ values between 10 and 15 μM. If the parasite viability was severely affected by other aggressive effectors (i.e. NO or ROS), as evidenced for the nitro-compounds 3a and 3f (see below), the effect of the antifolate activity on the antitrypanosomal response was not noticeable.

Promotion of NO production was analyzed in terms of the NO quantity in parasite cultures under compound treatment. This amount was determined using the Griess protocol^25,40,41 (Fig. 3A–B). NO production seemed to be dependent on the nature of the chemical function at the 4-position. An absence of production of NO in both parasites was recognized from the action of the 2-arylquinazolin-4(3H)one 1c, whereas marked NO production was observed from the 2-arylquinazolin-4-hydrazine 2c (Fig. 3A–B). With respect to 2-arylquinazolin-4-(furfurylidene)hydrazones, only discrete production was appreciated for the nitro-derivative 3f, whereas its de-nitro-derivative 3h did not induce production of an appreciable amount of NO in parasite cultures. The NO production derived from compounds 2c and 3f was dose-dependent, reaching >4 and >2 μM of NO for the highest concentration of 2-arylquinazolin-4-hydrazine 2c, and almost 2 and 1 μM for the highest concentration of compound 3f, against L. infantum and T. cruzi parasites, respectively. This observation suggested that the hydrazinyl or 5-nitrofurylhydrazinyl moieties at the 4-position seemed to have important roles in NO production in parasite cultures. Previously, we demonstrated from fluorescence and NMR measurements that 2-arylquinazolin-4-hydrazine acts as a NO-donor through oxidative cleavage of N–N in the hydrazine moiety, which releases NO and 2-arylquinazoline metabolites.²⁵ This phenomenon explains the origin of NO levels in parasite cultures derived from the action of 2-arylquinazolin-4-hydrazine, because incorporation of the hydrazine moiety is essential for it. Consequently, NO release derived from chemical decomposition was not suitable for the 2-arylquinazolin-4(3H)one 1c and 2-arylquinazolin-4-(furfurylidene)hydrazine 3h. The appreciable NO production derived from the 2-arylquinazolin-4-(5-nitro-furfurylidene)hydrazine 3f was an exceptional case. The NO production could have originated from coupling between ROS (see below) with nitrogenated sources (e.g. arginine) in culture which, in the presence of oxygen, could promote the indirect formation of NO.^48a

Fig. 3. NO production derived from promatigote of L. infantum (A) and epimastigote of T. cruzi (B), and ROS production derived from promastigote of L. infantum (C) and epimastigote of T. cruzi (D) of 2-arylquinazolin-4(3H)ones 1b and 1c, 2-arylquinazolin-4-hydrazine 2c and 2d, and 2-arylquinazolin-4-[2-(5-nitrofurfurylidene)]-hydrazones 3a–h. Note: NO and ROS measurements for compounds 1c and 2c were undertaken at 25, 50, and 100 μM concentrations, whereas compound 3f was assayed at 1, 5 and 10 μM.

Finally, the level of oxidative stress in parasite cultures was measured through a fluorometric measurement using a 2,7-dichlorodihydrofluorescein diacetate (DCFH-DA) probe.⁴⁶ Oxidative stress was evaluated for key compounds (1c, 2c, 3f, and 3h) (Fig. 3C and D). The non-nitro compounds 1c, 2c, and 3h did not induce appreciable production of ROS according to discrete or unappreciable changes in fluorescence measurements compared with the negative control. The nitro compound 3f showed a distinguished increase in the ROS level as evidenced from a significant enhancement of fluorescence from cultures derived from the L. infantum promastigote or T. cruzi epimastigote, and it increased with an increase in the concentration of 3f. These results revealed that the nitro-functionality of compounds 3a–g could be essential for promoting appreciable ROS production by parasite metabolization, with most of the active nitro-compounds 3a–g being ROS promoters.

We wished to obtain a clearer vision of modulation of the mechanism as a function of the chemical functionalization at the 4-position. Hence, a general scheme was elaborated (Scheme 2). All 2-arylquinazoline systems (1, 2 and 3) act as antifolate agents, which affects the viability of the parasite. This response is derived from the quinazoline core, which acts as a substrate model of any enzyme in the folate-biosynthesis pathway.^19–25 Then, the biological response of 2-arylquinazolin-4(3H)ones can be attributed to the intrinsic antifolate response of the quinazoline core. Meanwhile, replacement of the oxo moiety at the 4-position by a hydrazinyl moiety enhances the antitrypanosomal response of the chemical system, which can be attributed to the additional mechanism of promotion of NO production derived from oxidative hydrazinyl cleavage. NO is a toxic molecule for trypanosomatids.^26–27 Results for 2-arylquinazolin-4-hydrazine suggest that Leishmania-spp. parasites are more sensitive to the toxic action of NO than that observed for T. cruzi parasites.

Scheme 2. General representation of modulation of the mechanism of action as a function of the 4-position into 2-arylquinazoline. Hydrazine cleavage from ref. 48a and superoxide-anion formation via nitro-reduction from ref. 48b. NR II: nitro reductase II enzyme.

Next, the 2-arylquinazolin-4-hydrazines act through a dual mechanism as antifolates and as NO-donors. Importantly, the level of antifolate response of 2-arylquinazolin-4-hydrazine was in a similar range to that of 2-arylquinazolin-4(3H)ones. Hence, the 2-arylquinazolin core seems to be essential to promote the antifolate response of compounds of the families 1 and 2. Finally, with respect to the 2-arylquinazolin-4-(furfurylidene)hydrazones described herein, 5-nitrofuryl-hydrazinyl confers to the chemical system the ability to promote oxidative stress beyond the intrinsic antifolate activity derived from the quinazoline core. Then, active 2-arylquinazolin-4-(5-nitrofurfurylidene)hydrazones act through at least a triple mechanism. The remarkable antifolate response found for the non-nitro 2-arylquinazolin-4-(furfurylidene)hydrazine 3h suggests that 2-arylquinazolin-4-(5-nitrofurfurylidene)hydrazones also displayed an antifolate response, but it was not the significant response seen with the nitro-derivatives. This observation, in combination with discrete NO production from nitro-compound 3f, suggests that the sub-micromolar antitrypanosomal activity of the nitro-derivatives was attributed mainly to ROS production with a minimal contribution of NO production and antifolate activity. The NO production seen in 2-arylquinazolin-4-(furfurylidene)hydrazones was promoted only by their nitro-compounds, which indicated that the discrete NO production in these compounds was dependent on nitro-substitution and not on the hydrazine chain.

Finally, for active nitro-derivatives 3a, 3b, and 3f, we analyzed the effect of metal ions [copper(ii), iron(iii) and zinc(ii) salts at 1 μM] on parasite proliferation. We sought to: (i) improve the relative solubilization of the compound in a culture milieu; (ii) enhance ROS production. Both issues could be favored by the complexation ability of the compound as a bidentate ligand through an amidine moiety to form a stable complex with transition metals (Scheme 2). Susceptibilities of 2-arylquinazolines 3a, 3b, and 3f against both parasites (promastigote of L. infantum and epimastigote of T. cruzi) in the presence or absence of metallic solutions of copper(ii), iron(iii), or zinc(ii) salts are shown in Fig. 4A and B as IC₅₀ values. Importantly, the metal-ion concentration (1 μM) did not affect the cell viability of studied trypanosomatids. In general, significant changes were not detected against T. cruzi parasites. However, distinguished changes were found against L. infantum for the three studied compounds, with an appreciable decrease in their IC₅₀ magnitudes, more in particular, in the presence of copper(ii) and iron(iii). To obtain deeper understanding of the effect of metal addition, the ROS level was measured for parasite cultures in the presence of each of the metal cations (Zn²⁺, Cu²⁺, Fe³⁺). In general, for a range of compound concentrations, the ROS production derived from culture in the presence of metal cations was similar to that derived from a control culture in the absence of metal cations (see examples in Fig. 3C and D). From these evidences, the increase in the antitrypanosomal response by metal addition seemed to be associated with improvement in compound solubilization via an effective metal–compound complexation in aqueous solution. The partial coloration of the culture solution containing compounds from yellow (color of the medium) to green (in culture) was obtained with the addition of a metal cation (Fe³⁺ or Cu²⁺). To interpret the effect of ligand–metal complexation, the binding of compound 3a (as a model) with Zn²⁺, Cu²⁺ and Fe³⁺ was calculated using the CAM-B3LYP/6-31G(d,p) approach.^48–49 This approach allowed us to make an excellent estimation of geometry optimization²⁵ and binding complexation.⁵⁰ Then, theoretical calculations demonstrated that the complexation free energies for 3a–metal decreased in the order Zn²⁺, Cu²⁺, and Fe³⁺ with values of −312.54, −357.79, and −914.81 kcal mol⁻¹, respectively. This was in good correlation with the reduction trend of anti-Leishmania IC₅₀ in the presence of a metal: Zn²⁺ < Cu²⁺ < Fe³⁺. Further theoretical analysis based on LUMO+1 showed that iron(iii) could promote a remarkable CT process from ligand to metal (Fig. 4C–E). This phenomenon was not observed in theoretical Zn(ii)- and Cu(ii)-complexes. A more feasible reduction of the nitro group can occur in the presence of the iron(iii) cation. Electron-attraction to the metal from a ligand is favored, which is essential for reduction of a nitro-group by reception of an external electron. Thus, the effect of iron addition to promote ROS production could not be excluded. Further images of structures and HOMO–LUMO are shown in Fig. S8 and S9.†

Fig. 4. In vitro effect of 2-arylquinazolin-4-hydrazones 3a, 3b, and 3f against the promastigote of L. infantum (A) and epimastigote of T. cruzi (B) in the absence and presence of metallic ions (Cu²⁺, Fe³⁺, and Zn²⁺) at 1 μM. LUMO+1 orbital graphics for modelled 3a–Zn²⁺ (C), 3a–Cu²⁺ (D) and 3a–Fe³⁺ (E). Significant LMCT for 3a–Fe³⁺ complexation. Free binding complexation energies for the ligand–cation interaction were calculated at CAM-B3LYP/6-31G(d,p).

Drug-like profiles

Drug-like properties for the key 2-arylquinazolin-4-(furfurylidene)hydrazones developed herein (3a, 3b, and 3f) were studied in silico using the Swiss-ADME platform.⁵¹ Physicochemical properties (lipophilicity, water solubility, pharmacokinetic properties, and other druglikeness predictors) are summarized in Table 4. In general, the compounds showed good physicochemical properties within Lipinski,⁵² Ghose, Veber, Egan and Muegge rules.⁵³ Meanwhile, the compounds displayed a discrete predicted aqueous solubility of 5.28, 3.68 and 1.35 μM for compounds 3a, 3b, and 3f, which was in concordance with their appreciable relative solubilities in a culture milieu (Table 1). Within pharmacokinetic properties, compounds showed a predicted high gastrointestinal absorption index, good skin permeation (log K_p of −5.15, −5.19, and −4.91 cm s⁻¹ for 3a, 3b, and 3f, respectively), especially relevant for cutaneous Leishmaniasis, and a negative ability to be substrate of P-gp. Compounds had the potential to be substrates for cytochrome p450 (CYP)1A2, CYP2C19 and CYP2C9, but not for CYP2D6 or CYP3A4. Within medicinal chemistry, non-Pan-assay interference compounds (PAINS) moieties⁵⁴ were identified from 2-arylquinazolin-4-hydrazine. According to the Brenk filter (related to toxicity, chemical reactivity, and metabolic instability), the nitro- and imine-groups were alerted. To evaluate the security of the selected compounds, we carried out the Ames test to exclude the mutagenic effect derived from the interaction of compounds with biological systems. The mutagenic assay was done using the genetically modified Salmonella typhimurium TA 98 strain for derivatives 3b and 3f (Table 5).⁵⁵ According to the Ames test, compound 3f was identified as a mutagenic agent because the number of colonies was increased by more than 2-fold in two consecutive doses compared with the negative control, 9.5–10.5 colonies upon compound 3f (at the four consecutive highest doses) vs. 4.0 colonies for the negative control. Meanwhile, derivative 3b was identified as a non-mutagenic agent because it could not duplicate the number of spontaneous revertant colonies (0.0 μg/plate of compound) for at least two consecutive dose levels in all assayed doses,⁵⁵ whereas revertant colonies were obtained with the mutagenic positive control (4-nitro-o-phenylenediamine) (Table 5). Thus, the nitro-derivative 3b was identified as a safe, efficacious, and selective compound for further evaluations against trypanosomatids.

In silico physicochemical, pharmacokinetic and drug-likeness parameters of the key 2-arylquinazolin-4-(furfurylidene)hydrazones 3a, 3b, and 3f.

Type of parameter	Parameter	3a	3b	3f
Physico-chemical properties	M.W. (g mol−1)a	359.34	377.33	393.78
	No rotatable bonds	5	5	5
	No H-bond acceptors	6	7	6
	No H-bond donors	1	1	1
	Molar refractivity	103.54	103.50	108.55
	TPSA (Å2)b	109.13	109.13	109.13
Lipophilicity	Log Po/w (iLOGP)	2.06	2.23	2.36
	Log Po/w (XLOGP3)	4.71	4.81	5.34
	Log Po/w (WLOGP)	4.05	4.61	4.71
	Log Po/w (MLOGP)	3.36	3.75	3.86
	Log Po/w (SILICOS-IT)	1.75	2.17	2.39
	Consensus log Po/w	3.19	3.51	3.73
Water solubility	Log S (ESOL)	−5.28	−5.43	−5.87
	Solubility	5.28 μM (soluble)	3.68 μM (moderately soluble)	1.35 μM (moderately soluble)
	Log S (Ali)	−6.73	−4.74	−7.38
	Solubility	0.19 μM (poorly soluble)	0.15 μM (poorly soluble)	0.04 μM (poorly soluble)
Pharmakinetic properties	GI absorptionc	High	High	High
	BBB permeantd	No	No	No
	P-gp substratee	No	No	No
	CYP1A2 inhibitor	Yes	Yes	Yes
	CYP2C19 inhibitor	Yes	Yes	Yes
	CYP2C9 inhibitor	Yes	Yes	Yes
	CYP2D6 inhibitor	No	No	No
	CYP3A4 inhibitor	No	No	No
	Log Kp (skin permeation)	−5.15 cm s−1	−5.19 cm s−1	−4.91 cm s−1
Druglikeness	Lipinski	Fit; 0 violation	Fit; 0 violation	Fit; 0 violation
	Ghose	Fit	Fit	Fit
	Veber	Fit	Fit	Fit
	Egan	Fit	Fit	Fit
	Muegge	Fit	Fit	No fit; 1 violation: XLOGP3 > 5
	Bioavailability score	0.55	0.55	0.55
Medicinal chemistry	PAINSf	0 alert	0 alert	0 alert
	Brenk filter	2 alerts: imine_1, nitro_group	2 alerts: imine_1, nitro_group	2 alerts: imine_1, nitro_group
	Leadlikeness	No; 2 violations: MW > 350, XLOGP3 > 3.5	No; 2 violations: MW > 350, XLOGP3 > 3.5	No; 2 violations: MW > 350, XLOGP3 > 3.5

^aMW: molecular weight.

^bTPSA: topological polar surface area.

^cG.I.; gastrointestinal.

^dBBB: blood brain barrier.

^eP-gp: P-glycoprotein.

^fPAINS: pan assay interference structure. General drug-likeness map can be found in Fig. S7.†

Ames results in S. typhimurium TA 98 strain.

Comp.	Doses (μM)	Revertants number	Conclusion
3b	0.0	4.0 ± 0.0	Non-mutagenic
	100.0	2.0 ± 0.0
	300.0	2.5 ± 0.1
	1000.0	3.5 ± 0.2
	3000.0	3.0 ± 0.0
	9000.0	3.0 ± 0.2
3f	0.0	4.0 ± 0.0	Mutagenic
	90.0	3.0 ± 0.0
	270.0	10.0 ± 1.0
	830.0	9.5 ± 0.5
	2500.0	10.5 ± 1.0
	7500.0	9.5 ± 0.5
Positive control (NPD)	20.0	365.0 ± 10	Mutagenic

Conclusion

The chemical system, 2-arylquinazoline, seems to possess an intrinsic antifolate response that promotes specific antitrypanosomal activity. Chemical functionalization of 2-arylquinazoline at the 4-position using key pharmacophores (e.g. hydrazinyl or 5-nitrofurfurylidene-hydrazinyl) promoted the stimulation of other specific mechanisms (e.g. promotion of NO or ROS production, respectively), which implied a significant increase the in vitro antitrypanosomal response against L. infantum and T. cruzi parasites. 5-Nitrofurfurylidene-hydrazinyl functionality provided the most active compounds with the lowest IC₅₀ values (in sub-micromolar ranges) against the promastigote and amastigote of L. infantum and epimastigote of T. cruzi, and the highest selectivity index (>17). This strategy allowed identification of a promising candidate, compound 3b, for further preclinical studies. It showed a convenient medicinal-chemistry profile, including appropriate cytotoxicity and selectivity index with respect to both parasites (≥17). This strategy was focused on promotion/modulation of the mechanism of action through chemical functionalization using recognized key pharmacophores involved in specific mechanistic pathways. This represents a rational alternative to assess active and potent antitrypanosomal compounds. The 2-arylquinazoline system represents an attractive framework for the design of new active multitarget compounds against trypanosomatids.

Author contributions

A. H. R. carried out synthetic experiments, supervised biological assays and mechanistic studies, organized the investigation, analyzed the experimental and theoretical data, and prepared and revised the manuscript. E. A. carried out biological experiments related to L. infantum and T. cruzi as well as mechanistic biological studies. B. D. carried out the Ames Test. L. G. undertook theoretical experiments. G. C. and H. C. provided financial resources. H. C. supervised investigation, analyzed the experimental data, and revised the manuscript. All authors approved the final version of the manuscript.

Conflicts of interest

The authors declare no conflicts of interest.