Synthesis of New Proteomimetic Quinazolinone Alkaloids and Evaluation of Their Neuroprotective and Antitumor Effects

New quinazolinone derivatives of the marine-derived alkaloids fiscalin B (3) and fumiquinazoline G (1), with neuroprotective and antitumor effects, were synthesized. Eleven quinazolinone-containing indole alkaloids were synthesized, proceeding the anti analogs via a one-pot method, and the syn analogs by the Mazurkiewicz-Ganesan approach. The neuroprotection capacity of these compounds on the rotenone-damage human neuroblastoma cell SH-SY5y was evaluated using the MTT assay. Compounds 1, 3, 5, and 7 showed more than 25% protection. The antitumor activity was investigated using the sulforhodamine B assay and some compounds were tested on the non-malignant MCF-12A cells. Fumiquinazoline G (1) was the most potent compound, with GI50 values lower than 20 µM. Compounds 5, 7, and 11 were more active in all tumor cell lines when compared to their enantiomers. Compounds 5, 7, 10, and 11 had very little effect in the viability of the non-malignant cells. Differences between enantiomeric pairs were also noted as being essential for these activities the S-configuration at C-4. These results reinforce the previously described activities of the fiscalin B (3) as substance P inhibitor and fumiquinazoline G (1) as antitumor agent showing potential as lead compounds for the development of drugs for treatment of neurodegenerative disorders and cancer, respectively.


Introduction
The pathophysiology of neurodegenerative diseases is poorly understood, and there are few therapeutic options, making neuroprotective drug discovery appealing for medicinal chemists.

Chemistry
Two synthetic approaches were used to prepare the syn and anti enantiomers of quinazolinone alkaloids. The syn enantiomers 1 (fumiquinazoline G) and 2 were synthesized by the Mazurkiewicz-Ganesan procedure [17] (Scheme 1.A) by coupling anthranilic acid (i) with D-tryptophan methyl ester (ii) for 1 or with L-tryptophan methyl ester (vi) for 2, using 1,1,3,3-tetramethylaminium tetrafluoroborate (TBTU) in alkaline condition to obtain the dipeptide iii or vii. Then, the coupling of iii or vii with N-protected α-amino acid chloride in a two-phase Schotten-Baumann condition yielded a tripeptide (dehydrate β-keto amides) v or ix. The oxazole intermediates were obtained by adding the dehydrating agent, triphenylphosphine (Ph3P), and I2 to dehydrate β-keto amide v or ix, and N-deprotection by 20% piperidine afforded 1 and 2. On the other hand, a highly effective and environmentally friendly approach using a microwave-assisted multicomponent polycondensation of amino acids was used to prepare a series of the anti Marine-derived indolylmethylpyrazinoquinazoline alkaloids, with a pyrazino[2,1-b]quinazoline-3,6-dione linked to an indole moiety (Figure 1), have attracted our attention due to their promising antitumor activities [3], with epi-fiscalin A [4], fumiquinazoline A [5][6][7][8][9], fumiquinazoline G [7], and versiquinazolines [10] as the most active analogs. Moreover, the response of fiscalins A-C [11] and (−)-spiroquinazoline [12] ( Figure 1A) as substance P inhibitors was also reported as a novel neuroprotective therapy in the intrastriatal 6-hydroxydopamine model of early stage of Parkinson's disease (PD) [13]. It is well known that among compounds implicated in neurodegeneration, non-proteinogenic amino acids may cause significant collateral neurodegenerative damage [14]. Rodgers et al. [15] reported that proteomimetic L-tyrosine of L-DOPA is cytotoxic in vitro and capable of generating protein aggregation, whereas non-protein amino acid β-methylamino-L-alanine (BMAA) has been linked to neurological diseases such as amyotrophic lateral sclerosis (ALS) and PD since BMAA was detected in brain protein of LAS and PD patients. In the previous work, we have described syntheses of a series of fiscalin B derivatives, which showed weak to moderate antitumor activity against non-small cell lung cancer (NCI-H460) and colorectal adenocarcinoma (HCT-15) cell lines [16]. These findings led us to develop a small library of proteomimetic quinazolinone-derived compounds ( Figure 1B) with different configurations at C-1 and C-4 to investigate their action on neurodegenerative disorders as well as to further explore their potential as tumor cell growth inhibitors, putting in evidence the influence of the stereochemistry of the derivatives.

Chemistry
Two synthetic approaches were used to prepare the syn and anti enantiomers of quinazolinone alkaloids.
The syn enantiomers 1 (fumiquinazoline G) and 2 were synthesized by the Mazurkiewicz-Ganesan procedure [17] (Scheme 1A) by coupling anthranilic acid (i) with D-tryptophan methyl ester (ii) for 1 or with L-tryptophan methyl ester (vi) for 2, using 1,1,3,3-tetramethylaminium tetrafluoroborate (TBTU) in alkaline condition to obtain the dipeptide iii or vii. Then, the coupling of iii or vii with N-protected α-amino acid chloride in a two-phase Schotten-Baumann condition yielded a tripeptide (dehydrate β-keto amides) v or ix. The oxazole intermediates were obtained by adding the dehydrating agent, triphenylphosphine (Ph 3 P), and I 2 to dehydrate β-keto amide v or ix, and N-deprotection by 20% piperidine afforded 1 and 2. On the other hand, a highly effective and environmentally friendly approach using a microwave-assisted multicomponent polycondensation of amino acids was used to prepare a series of the anti enantiomers of pyrazinoquinazoline alkaloids [18], as described in our previous work [16]. This methodology was used to synthesize new derivatives of fiscalin B (3) and fumiquinazoline G (1), 4, 5, 6, 7, and 8 (Scheme 1B). The syn isomer 9 was obtained along with 8, and both were isolated by preparative thin layer chromatography (TLC). Diastereoisomers of 10 and 11 were obtained after deprotection of O-benzyl group from 8 and 9, respectively, using boron trichloride, according to Okaya et al. [19] with a slight modification. Compound 12 was also synthesized using microwave irradiation from 3,5-dichloroanthranilic acid (xiii). The purity of the compounds was determined by a reversed-phase liquid chromatography (LC, C18, MeOH:H 2 O; 60:40 or CH 3 CN:H 2 O; 50:50) and was found to be higher than 90%. The enantiomeric ratio (er) was determined by a chiral LC equipped with amylose tris-3,5-dimethylphenylcarbamate column, using hexane:EtOH (80:20) or (70:30) as a mobile phase.
The reaction carried out using microwave with high temperature resulted not only in low yields of the products in the range of 2.2 to 21.7%, but also with a high degree of epimerization (Scheme 1). Contrary to what has been found in our previous study [16] that the reaction under a microwave irradiation was regioselective and yielded only anti isomers, the synthesis of 8, by a microwave irradiation, produced also its syn epimer, 9 [4-(benzyloxy)-1-methylbenzyl at C-1], with a 22% yield. This study suggested that microwave irradiation is beneficial for the synthesis of quinazolinone alkaloids with bulky substituents at C-1 which was previously reported as unsuccessful by Mazurkiewicz-Ganesan method [17]. However, this methodology failed for the synthesis of syn enantiomers as described in the experimental section for 4 and 6. The syn enantiomers of 1 and 2 were synthesized by Mazurkiewicz-Ganesan approach [17] and gave moderate yields (37 and 26%, respectively). Compounds 10 and 11 were obtained by deprotection in good yields (30 and 69%, respectively).
Moreover, the methodology involving microwave irradiation was characterized by producing partial epimerizations. Surprisingly, 4 and 5, with three stereogenic centers, gave a higher enantiomeric ratio (er) of 99%. Similar to the previous report for fiscalin B analogs [16], the multi-step approach gave a better yield and, in most cases, higher enantiomeric ratios due to milder conditions; nonetheless, the one-pot reaction is a faster alternative to provide anti enantiomers with diversity of substituents at C-1. h ; Boc = tert-butyloxycarbonyl, R'= H when no mentioned; er = enantiomeric ratio calculated from the peak area from chiral LC experiments (by using equation X × 100/Xn in which X is the peak area of each peak and Xn is the total peak area. * referred to previous work [16].

Structure Elucidation
A series of 1D and 2D NMR experiments and HRMS were used to confirm the structures of all the new compounds (Supplementary information, Figures S1-S57). The amide proton (H-2) appeared as a broad singlet (brs) in the anti isomers, i.e., 4, 5, 6, 7, 8, 10, and 12, but as a doublet (d) in the syn isomers, i.e., 1, 2, 9, and 11 (Supplementary information, Figures S1-S21). Each stereoisomer exhibited different chemical shift values for H-2, H-4, and H-1. For example, for isomers 2, 4, and 5, with the isobutyl group at C-1, the syn and the anti isomers were distinguished by the chemical shift values of H-1 and H-1'; which were ca. δ H 4.03 and 0.93, respectively for the syn isomer (2), and at ca. δ H 2.80 and 2.3, respectively, for the anti isomers (4 and 5), due to the absence of the shielding effect by the aromatic ring of the indole moiety [20].
This assignment was also confirmed by HMBC correlations which distinguished between the isomers of 2, 4, and 5 by the presence or absence of correlations from H-2 to C-3, C-4, and C-14. In 2, H-2 showed no correlation to C-3, C-4, and C-14 whereas in 4 and 5, H-2 showed correlations to those three carbons. H-4 also showed correlations to different carbons among isomers. H-4 showed correlations to C-3 and C-4' in 2, but to C-3, C-14, C-4', and C-5' in 4, while there were no such correlations observed in 5. H-1 of the syn isomer of 2 showed correlation only to C-14 while in the anti isomer of 4 and 5, it displayed correlations to C-14, C-1' and C-3" (Figure 2). In contrast, in 1, H-2 and H-4 showed no correlations to carbons that are two or three bonds away ( 2 J or 3 J), and H-1 showed cross peaks to C-3 and C-14. Furthermore, H-8 and H-10 of 12 (with Cl at C-9 and C-11) appeared as two doublets (δ H 8.  Figure S56). The NOESY spectrum for the syn configuration compound 2 (1S, 4S) exhibited correlations between H-1 and H-1', H-2', H-3", and H-2 as well as from H-4 to H-4'. Also, correlations between H-4' and H-1' in the syn configuration compound 2 could be noted while these was absent in the anti configuration compound 6 (Supplementary information, Figure S57). These observations were similar to the previously described for fiscalin B isomers [16]. This assignment was also confirmed by HMBC correlations which distinguished between the isomers of 2, 4, and 5 by the presence or absence of correlations from H-2 to C-3, C-4, and C-14. In 2, H-2 showed no correlation to C-3, C-4, and C-14 whereas in 4 and 5, H-2 showed correlations to those three carbons. H-4 also showed correlations to different carbons among isomers. H-4 showed correlations to C-3 and C-4' in 2, but to C-3, C-14, C-4', and C-5' in 4, while there were no such correlations observed in 5. H-1 of the syn isomer of 2 showed correlation only to C-14 while in the anti isomer of 4 and 5, it displayed correlations to C-14, C-1' and C-3" (Figure 2). In contrast, in 1, H-2 and H-4 showed no correlations to carbons that are two or three bonds away ( 2 J or 3 J), and H-1 showed cross peaks to C-3 and C-14. Furthermore, H-8 and H-10 of 12 (with Cl at C-9 and C-11)

Neuroprotection Activity
The neuroprotection assay was performed on human neuroblastoma cell SH-SY5y treated with rotenone, a toxin that acts by interfering with the electron transport chain in mitochondria, inhibiting the transfer of electron from iron-sulfur centers in complex I to ubiquinone. This in turn interferes with NADH, therefore creating reactive oxygen species (ROS), which can damage DNA and other components, leading to cell death [21,22]. In animal experimentation, rotenone reproduces features of PD, including selective nigrostriatal dopaminergic degeneration and alpha-synuclein-positive cytoplasmic inclusions [23]. Furthermore, rotenone triggers mitochondrial impairment, oxidative damage, and cell death in neuronal culture, phenomena that are common in neurodegenerative diseases [24].
In this assay, the SH-SY5y cells were treated with 2 µ M of rotenone for 24 h. The MTT assay, which assesses cell metabolic activity through the activity of NAD(P)H-dependent cellular oxidoreductase enzyme that reflect the number of viable cells [24], was used for quantifying the cell death. The cellular protection of 1, 2, 4, 5, 6, 7, 8, 9, 10, 11, and 12 against the toxin was determined by MTT assay and expressed as percentage referred to the cell treated with rotenone at 10 different concentrations to produce the dose-dependent curve. Synthetic fiscalin B (3), previously obtained by some of us [16] and was reported as a substance P antagonist [12], was also tested in this assay for comparison. Compounds considered as neuroprotective must have (i) more than 25% of protection, (ii) statistically significant difference, and (iii) protection is more than one dose [23,24].
Most of the compounds, at the highest concentration tested (100 µ M), were found to increase toxicity, which ensures that the compounds were assayed at their maximal tolerated dose (MTD). Compounds 1, 2, 3, 4, 5, 10, 11, and 12 showed high toxicity at 100 µ M and some of them also at 50

Neuroprotection Activity
The neuroprotection assay was performed on human neuroblastoma cell SH-SY5y treated with rotenone, a toxin that acts by interfering with the electron transport chain in mitochondria, inhibiting the transfer of electron from iron-sulfur centers in complex I to ubiquinone. This in turn interferes with NADH, therefore creating reactive oxygen species (ROS), which can damage DNA and other components, leading to cell death [21,22]. In animal experimentation, rotenone reproduces features of PD, including selective nigrostriatal dopaminergic degeneration and alpha-synuclein-positive cytoplasmic inclusions [23]. Furthermore, rotenone triggers mitochondrial impairment, oxidative damage, and cell death in neuronal culture, phenomena that are common in neurodegenerative diseases [24].
In this assay, the SH-SY5y cells were treated with 2 µM of rotenone for 24 h. The MTT assay, which assesses cell metabolic activity through the activity of NAD(P)H-dependent cellular oxidoreductase enzyme that reflect the number of viable cells [24], was used for quantifying the cell death. The cellular protection of 1, 2, 4, 5, 6, 7, 8, 9, 10, 11, and 12 against the toxin was determined by MTT assay and expressed as percentage referred to the cell treated with rotenone at 10 different concentrations to produce the dose-dependent curve. Synthetic fiscalin B (3), previously obtained by some of us [16] and was reported as a substance P antagonist [12], was also tested in this assay for comparison. Compounds considered as neuroprotective must have (i) more than 25% of protection, (ii) statistically significant difference, and (iii) protection is more than one dose [23,24].

Tumor Cell Growth Inhibitory Activity
Compounds 1-2, 4-8, 10 and 11 were tested for tumor cell growth inhibitory activity on three human tumor cell lines: NCI-H460 (non-small cell lung cancer), BxPC3 (human pancreatic adenocarcinoma), and PANC1 (human pancreatic adenocarcinoma) using the sulforhodamine B (SRB) colorimetric assay [16,25]. Cells were exposed to five concentrations of each compound (at a maximum concentration of 25, 150 or 200 µM, depending on the compound) for 48 h. Doxorubicin was used as a positive control for NCI-H460 cell line, and gemcitabine was used as a positive control for the BxPC3 and PANC1 cell lines. The antitumor activity was reported as GI 50 concentration (drug concentration that inhibits the growth of cancer cells by 50%).
Compounds 2, 4-8, 10, and 11 showed weak to moderate growth inhibitory, with GI 50 ranking from 27.93 ± 0.8 to 151.07 ± 2.9 µM (Table 1). In general, compounds with the indolylmethyl substituent on C-4 whose configuration of C-4 is R showed better antitumor activity in all cell lines when compared to those with 4S configuration. This was evidenced by stronger antitumor activities of 1, 5, and 7 than those of 2, 4, and 6. Only fumiquinazoline G (1) with R-configuration for both C-1 and C-4 showed strong growth inhibitory effect in all cancer cell lines tested (GI 50 ranging from 7.62 ± 0.7 to 17.34 ± 1.7 µM). In addition, these results agree with those reported in our previous publication in which enantiomers with R-configuration at C-1 and C-4 showed better antitumor activity than enantiomers with S-configuration [16]. Unfortunately, 9 and 12 tested in this study could not be evaluated regarding their tumor cell growth inhibitory activity due to contamination of the compounds (data not shown). The GI 50 concentrations (µM) were determined by the SRB assay and results are the mean of ± SEM of three independent experiments. Gemcitabine was use as a positive control for the BxPC3 and PANC1 cell lines, and doxorubicin as a positive control for the NCI-H460 cell line. (-) indicates not-determined.

Activity in Non-Tumor Cells
Compounds presenting the best neuroprotection and/or antitumor effects, namely 5, 7, 10, and 11, were also evaluated against the non-malignant MCF-12A human breast epithelial cells. For that, one concentration of each compound (corresponding to approximately the highest GI 50 value obtained in the cell growth inhibitory activity assay) was tested and the percentage of cell growth inhibition was determined by the SRB assay. In this assay, the duration of the SRB assay had to be longer since non-malignant cells have a much slower growth rate than tumor cells. This longer duration of the assay also allowed to evaluate possible delayed effects of the compounds in these non-malignant cells. Therefore, the assay in the MCF-12A cells was performed following 7 days of treatment with compounds (48 h with compound incubation plus 5 days without the compounds). As shown in Table 2, all the tested compounds caused a small effect in the growth of these non-malignant cells, meaning that the cell growth inhibition detected in MCF-12A cells after 7 days of treatment was much lower than that detected in the tumor cell lines after 2 days of treatment (when tested at the same concentration).

Structural-Activity Relationship (SAR)
Structure-activity relationship analysis showed that the obtained results were consistent with data previously reported for the natural product fumiquinazoline G [26] and fiscalin B [11,[27][28][29] and their derivatives [16] (Figure 4). Moreover, it was found that the configurations of C-1 and C-4 have strong influence on antitumor activity since fumiquinazoline G (1), with R-configurations at C-1 and C-4, showed the strongest antitumor effect against NCI-H460, BxPC3, and PANC1 cell lines. Comparing the enantiomeric pairs 4 and 5, compound 5 with also R-configuration at C-4 and S at C-1 showed stronger antitumor effect. In contrast, their isomer, compound 2, with S-configurations both at C-1 and C-4 were S exhibited the weakest inhibitory effect on all cell lines. In addition, the substituent at C-1 also affected the antitumor activity. Alkyl residues (isoleucine residues) in 2, 4, and 5 showed better antitumor activity than that found in compounds with a sulfur atom (methionine residues) such as 6 and 7. Aromatic groups such as the tyrosine residue present in 10 and 11 produced good antitumor activity, with GI 50 values ranging from 27.93 ± 0.8 to 50.59 ± 2.3 µM, but the substitution by benzyl groups as in the case of 8 caused a 2-fold decrease in the antitumor activity. Regarding neuroprotection capacity, SAR suggests that the R-configuration at C-4 is also important (when comparing compounds 6 and 7, Figure 3); however, increasing the molecular weight of C-1 substituent has a negative effect in neuroprotection. In addition, this study confirmed that quinazolinone alkaloids which act as substance P inhibitors (i.e., fiscalin B, 3), showed a potential as neuroprotective agents. Therefore, the studied fumiquinazoline-derived alkaloids showed promising antitumor and neuroprotection effects and deserves to be further explored.
Molecules 2019, 24, x 9 of 18 longer since non-malignant cells have a much slower growth rate than tumor cells. This longer duration of the assay also allowed to evaluate possible delayed effects of the compounds in these non-malignant cells. Therefore, the assay in the MCF-12A cells was performed following 7 days of treatment with compounds (48 h with compound incubation plus 5 days without the compounds). As shown in Table 2, all the tested compounds caused a small effect in the growth of these non-malignant cells, meaning that the cell growth inhibition detected in MCF-12A cells after 7 days of treatment was much lower than that detected in the tumor cell lines after 2 days of treatment (when tested at the same concentration).

Structural-Activity Relationship (SAR)
Structure-activity relationship analysis showed that the obtained results were consistent with data previously reported for the natural product fumiquinazoline G [26] and fiscalin B [11,[27][28][29] and their derivatives [16] (Figure 4). Moreover, it was found that the configurations of C-1 and C-4 have strong influence on antitumor activity since fumiquinazoline G (1), with R-configurations at C-1 and C-4, showed the strongest antitumor effect against NCI-H460, BxPC3, and PANC1 cell lines. Comparing the enantiomeric pairs 4 and 5, compound 5 with also R-configuration at C-4 and S at C-1 showed stronger antitumor effect. In contrast, their isomer, compound 2, with S-configurations both at C-1 and C-4 were S exhibited the weakest inhibitory effect on all cell lines. In addition, the substituent at C-1 also affected the antitumor activity. Alkyl residues (isoleucine residues) in 2, 4, and 5 showed better antitumor activity than that found in compounds with a sulfur atom (methionine residues) such as 6 and 7. Aromatic groups such as the tyrosine residue present in 10 and 11 produced good antitumor activity, with GI50 values ranging from 27.93 ± 0.8 to 50.59 ± 2.3 µ M, but the substitution by benzyl groups as in the case of 8 caused a 2-fold decrease in the antitumor activity. Regarding neuroprotection capacity, SAR suggests that the R-configuration at C-4 is also important (when comparing compounds 6 and 7, Figure 3); however, increasing the molecular weight of C-1 substituent has a negative effect in neuroprotection. In addition, this study confirmed that quinazolinone alkaloids which act as substance P inhibitors (i.e., fiscalin B, 3), showed a potential as neuroprotective agents. Therefore, the studied fumiquinazoline-derived alkaloids showed promising antitumor and neuroprotection effects and deserves to be further explored.

General Procedure
All reagents were from analytical grade. Dried pyridine and triphenylphosphite were purchased from Sigma (Sigma-Aldrich Co. Ltd., Gillingham, UK). Anthranilic acid (i) and Protected amino acids (ii and vii) were purchased from TCI (Tokyo Chemical Industry Co. Ltd., Chuo-ku, Tokyo, Japan). Column chromatography purifications were performed using flash silica Merck 60, 230-400 mesh (EMD Millipore corporation, Billerica, MA, USA) and preparative TLC was carried out on precoated plates Merck Kieselgel 60 F 254 (EMD Millipore corporation, Billerica, MA, USA), spots were visualized with UV light (Vilber Lourmat, Marne-la-Vallée, France). Melting points were measured in a Köfler microscope and are uncorrected. Infrared spectra were recorded in a KBr microplate in a FTIR spectrometer Nicolet iS10 from Thermo Scientific (Waltham, MA, USA) with Smart OMNI-Transmission accessory (Software 188 OMNIC 8.3, Thermo Fisher Scientific Inc., Austin, TX, USA). 1 H and 13 C NMR spectra were recorded in CDCl 3 or DMSO-d 6 (Deutero GmbH, Kastellaun, Germany) at room temperature unless otherwise mentioned on Bruker AMC instrument (Bruker Biosciences Corporation, Billerica, MA, USA), operating at 300 MHz for 1 H and 75 MHz for 13 C). Carbons were assigned according to HSQC and or HMBC experiments. Optical rotation was measured at 25 • C using the ADP 410 polarimeter (Bellingham + Stanley Ltd., Royal Tunbridge Wells, Kent, UK), using the emission wavelength of sodium lamp, concentrations are given in g/100 mL. Qualitative GC-MS analyses were performed on a Trace GC 2000 Series ThermoQuest gas chromatography (Thermo Fisher Scientific Inc., Austin, TX, USA) equipped with ion-trap GCQ Plus ThermoQuest Finnigan mass detector (Thermo Fisher Scientific Inc.). Chromatographic separation was achieved using a capillary column (30 m × 0.25 mm × 0.25 µm, cross-linked 5% diphenyl and 95% dimethyl polysiloxane) from Thermo Scientific TM (Thermo Fisher Scientific Inc.) and high-purity helium C-60 as carrier gas. High resolution mass spectra (HRMS) were measured on a Bruker FTMS APEX III mass spectrometer (Bruker Corporation, Billerica, MA, USA) recorded as ESI (Electrospray) made in Centro de Apoio Cientifico e Tecnolóxico á Investigation (CACTI, University of Vigo, Pontevendra, Spain). The purity of synthesized compounds was determined by reversed-phase LC with diode array detector (DAD) using C18 column (Kimetex ® , 2.6 EV0 C18 100 Å, 150 × 4.6 mm), and the mobile phase was methanol:water (60:40) or acetonitrile:water (50:50). Enantiomeric ratio was determined by chiral LC (LCMS-2010EV, Shimadzu, Lisbon, Portugal), employing a system equipped with a chiral column (Lux ® 5 µm Amylose-1, 250 × 4.6 mm) and UV-detection at 254 nm, mobile phase was hexane:ethanol (80:20) and the flow rate was 0.5 mL/min. Compound 3 was obtained according to previous described method [17]. Neuroprotection studies were performed in Fundación Centro de Excelencia en Investigación de Medicamentos Innovadores en Andalucía, MEDINA (Granada, Spain).

Testing Effect of Compounds on Non-malignant Breast Cells
Compounds 5, 7, 10 and 11 (which presented simultaneously the best neuroprotection and antitumor effects) were tested against the non-malignant MCF-12A human breast epithelial cells. For that, cells were incubated with specific concentrations of each compound (corresponding to approximately the highest GI 50 concentration obtained in the antitumor activity screening) for 48 h, followed by removal of the compound, addition of new medium to the cells and then 5 more days in culture. At the end of the 7 days in total, the sulforhodamine B (SRB) assay was performed, as previously described [32].

Conclusions
New quinazolinone alkaloid derivatives with anti and syn stereochemistry were synthetized by combining both a one-pot microwave-assisted reaction and a multi-step approach. Interestingly, fumiquinazoline G (1) presented a better antitumor activity in all the tumor cell lines tested, with GI 50 values lower than 20 µM. The antitumor activity of the remaining compounds was not relevant, with GI 50 values higher than 20 µM in the tested cell lines. The effect of the synthesized compounds in the growth of the tested non-malignant cells was smaller than the effect on the studied tumor cells. It is worth noting that among the compounds tested, only 1, 3, 5, and 7 showed potential for neuroprotection in a PD in vitro model. This finding highlights new insights into marine natural products belonging to the proteomimetic quinazolinone alkaloids. Funding: This research was partially supported by the Strategic Funding UID/Multi/04423/2013 through national funds provided by FCT-Foundation for Science and Technology and European Regional Development Fund (ERDF), in the framework of the program PT2020. The authors thank to national funds provided by FCT-Foundation for Science and Technology and European Regional Development Fund (ERDF) and COMPETE under the Strategic Funding UID/Multi/04423/2013, the projects POCI-01-0145-FEDER-028736 and PTDC/MAR-BIO/4694/2014 (POCI-01-0145-FEDER-016790; 3599-PPCDT).