Palladium-mediated synthesis and biological evaluation of C-10b substituted Dihydropyrrolo[1,2-b]isoquinolines as antileishmanial agents

Palladium-mediated synthesis and biological evaluation of C-10b substituted Dihydropyrrolo[1,2-b]isoquinolines as antileishmanial agents Iratxe Barbolla a , Leidi Hern andez-Su arez a , Viviana Quevedo-Tumailli a , b , c , Deyani Nocedo-Mena a , Sonia Arrasate a , María Auxiliadora Dea-Ayuela d , Humberto Gonz alez-Díaz a , e , f , *** , Nuria Sotomayor a , ** , Esther Lete a , * a Departamento de Química Org anica e Inorg anica, Facultad de Ciencia y Tecnología, Universidad Del País Vasco / Euskal Herriko Unibertsitatea UPV/EHU, Apdo. 644, 48080, Bilbao, Spain b RNASA-IMEDIR, Computer Science Faculty, University of A Coru~ na, 15071, A Coru~ na, Spain c Universidad Estatal Amaz onica UEA, Puyo, 160150, Pastaza, Ecuador d Departamento de Farmacia, Facultad de Ciencias de La Salud, Universidad CEU Cardenal Herrera, Edificio Seminario S/n, 46113, Moncada, Valencia, Spain e Basque Center for Biophysics CSIC-UPV/EHU, University of the Basque Country UPV/EHU, 48940, Bilbao, Spain f IKERBASQUE, Basque Foundation for Science, 48011, Bilbao, Spain article info Article history: Received 15 February 2021 Received in revised form 12 March 2021 Accepted 5 April 2021 Available online 16 April 2021 Keywords: Leishmaniasis Pyrroloisoquinoline Palladium Cascade reactions Machine learning Cheminformatics abstract The development of new molecules for the treatment of leishmaniasis is, a neglected parasitic disease, is urgent as current anti-leishmanial therapeutics are hampered by drug toxicity and resistance. The pyrrolo[1,2-b]isoquinoline core was selected as starting point, and palladium-catalyzed Heck-initiated cascade reactions were developed for the synthesis of a series of C-10 substituted derivatives. Their in vitro leishmanicidal activity against visceral (L. donovani) and cutaneous (L. amazonensis) leishmaniasis was evaluated. The best activity was found, in general, for the 10-arylmethyl substituted pyrroloisoquinolines. In particular, 2ad (IC 50 ¼3.30 m M, SI >77.01) and 2bb (IC 50 ¼3.93 m M, SI >58.77) were approximately 10-fold more potent and selective than the drug of reference (miltefosine), against L. amazonensis on in vitro promastigote assays, while 2ae was the more active compound in the in vitro amastigote assays (IC 50 ¼33.59 m M, SI >8.93). Notably, almost all compounds showed low cytotoxicity, CC 50 >100 m g/mL in J774 cells, highest tested dose. In addition, we have developed the first Perturbation Theory Machine Learning (PTML) algorithm able to predict simultaneously multiple biological activity parameters (IC 50 ,K i ,etc.) vs. any Leishmania species and target protein, with high values of specificity (>98%) and sensitivity (>90%) in both training and validation series. Therefore, this model may be useful to reduce time and assay costs (material and human resources) in the drug discovery process. ©2021 The Authors. Published by Elsevier Masson SAS. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). 1. Introduction Leishmaniasis is a neglected parasitic disease, endemic in about 100 countries, with morbidity and mortality increasing daily. The disease is caused by protozoan pathogens of the Leishmania genus that are transmitted by sandflies. Leishmania spp exists in two morphologically distinct forms: a motile flagellated form (promastigotes) and an intracellular non-flagellated form (amastigotes). There are four main forms of the disease: visceral leishmaniasis (VL, also known as kala-azar); post-kala-azar dermal leishmaniasis (PKDL); cutaneous leishmaniasis (CL); and mucocutaneous leishmaniasis (MCL). While CL is the most common form of the disease, VL is the most serious and can be fatal if untreated [1]. Immunosuppressed patients related to HIV co-infection or solid organ transplantation are prone to infection by Leishmania, which can also promote cancer development [2]. *Corresponding author. ** Corresponding author. *** Corresponding author. Departamento de Química Org anica e Inorg anica, Facultad de Ciencia y Tecnología, Universidad del País Vasco / Euskal Herriko Unibertsitatea UPV/EHU, Apdo. 644, 48080, Bilbao, Spain. E-mail addresses: humberto.g[email protected] (H. Gonz alez-Díaz), nuria. sotomay[email protected] (N. Sotomayor), esther[email protected] (E. Lete). Contents lists available at ScienceDirect European Journal of Medicinal Chemistry journal homepage: http://www.elsevier.com/locate/ejmech https://doi.org/10.1016/j.ejmech.2021.113458 0223-5234/©2021 The Authors. Published by Elsevier Masson SAS. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). European Journal of Medicinal Chemistry 220 (2021) 113458 Current antileishmanial therapeutics are hampered by drug toxicity, high cost, need for parenteral administration, increasing treatment failure rates, and emergence of Multi-Drug Resistant (MDR) Leishmania species or strains. Treatment depends not only on the etiological species and the infection type (CL, VL, PKDL, or MCL), but also on the geographical location where the disease was acquired [3]. Besides, there are few well-validated molecular drug targets in Leishmania, but the molecular targets of the current clinical molecules are unknown. VL is treated with different multidrug therapy, different combinations of pentavalent antimonials, paromomycin, liposomal amphotericin B, and miltefosine. Its use is further limited by associated life-threatening toxicities like cardiac arrhythmias, prolonged QT interval (QTc), ventricular premature beats, tachycardia or fibrillation in pentavalent antimonials or nephrotoxicity, hypokalemia and myocarditis in amphotericin B therapy [4]. The toxicity and the narrow therapeutic margin is one of the main limitations of the compounds currently used against leishmaniasis, therefore it is important to search for new therapeutic alternatives that show a reduced toxicity. Moreover, these drugs can have significant side effects, e.g. miltefosine can cause birth defects if taken within three months of getting pregnant [5]. On the other hand, treatment of CL should be decided by the clinical lesions, etiological species, the different response to drugs of species and strains, and the possibility to develop into mucosal leishmaniasis [6]. But proven treatments of CL are scarce, being pentavalent antimonials, paromomycin, pentamidine, and the triazol derivative fluconazole the most effective drugs [7](Fig. 1). Among the latest advances in this field stand out novel antileishmanial drug-like chemical series based on nitrogen heterocyclic scaffolds, as pyrazolopyrimidine GSK3186899/DDD853651 [8] and benzabenzoxazole derivatives GNF6702 and LXE408 [9](Fig.1). Besides, sitamaquine, an 8-aminoquinoline, is a drug candidate for the treatment of VL by oral route, although phase II clinical trials point out some adverse effects, such as methemoglobinemia and nephrotoxicity, which have to be considered for a further development decision [10](Fig. 1). More recently, 4aminostyrylquinolines, quinolone-metronidazoles, and ferrocenylquinoline-based derivatives have also shown promising antileishmanial profiles [11]. Currently, identification of new effective and safe antileishmanial drugs is crucial to advance in obtaining new lead compounds and disease control [12]. Our group has experience in the development of synthetic methodologies for the preparation of different benzo(hetero)fused six-membered heterocycles [13], in particular quinoline, isoquinoline and their dihydro counterparts, whose structural cores are among many biologically active natural products and pharmaceuticals [14]. In particular, Lycorane-type of Amaryllidaceae alkaloids may present biological activity against tropical diseases caused by protozoan parasites [15]. On the other hand, Bringmann and Moll have reported that the presence of an aromatic ring linked to the nitrogen atom of the isoquinoline nucleus is crucial for the leishmanicidal activity of these heterocycles [16]. Therefore, we envisioned that the change of the aryl group for an heteroaromatic ring, the pyrrole, but fused to the isoquinoline nucleus by the bside, would lead to the pyrrolo[1,2-b]isoquinoline core, thus combining structural features of both types of heterocycles. We recently reported the synthesis of C-10 (hetero)arylmethyl substituted pyrrolo[1,2-b]isoquinolines 2[17] via a Heck/Suzuki cascade reaction (Scheme 1)[18]. Therefore, we decided to explore a palladium-catalyzed intramolecular Heck/anion capture cascade reaction, which would allow to vary the nature of the substituent at Fig. 1. Antileishmanial drugs with heterocyclic motifs and miltefosine. Scheme 1. Synthesis of C-10 substituted pyrrolo[1,2-b]isoquinolines 2and 3for antileishmanial assays: (a) previously reported Heck/Suzuki cascade; (b) Heck/anion capture cascade. I. Barbolla, L. Hern andez-Su arez, V. Quevedo-Tumailli et al. European Journal of Medicinal Chemistry 220 (2021) 113458 2 C-10b for the study of structure-activity relationships. Thus, intramolecular Heck-type reaction of 2-alkenyl N-(o-iodobenzyl)pyrroles 1proceeds via an initial 6-exo cyclization process to give a s - alkylpalladium(II) intermediate that would be trapped by a cyanide ion. The use of 1,1-disubstituted alkenes directs the carbopalladation to the most substituted position avoiding the syn b -hydride elimination. Thus, the sequential formation of two CeC bonds can be achieved to give pyrrolo[1,2-b]isoquinolines 3with a cyanomethyl substituent at the quaternary stereocenter (Scheme 1). In this context, Cheminformatics models can be useful to carry out high-throughput computational in silico pre-screening of large libraries of compounds. These studies allow to prioritize some families of compounds (potential lead compounds) in the pharmacological assays in order to reduce the time and costs (material and human resources) of the drug discovery process. Thus, assay of new compounds by trial-and-error tests are avoided. Nevertheless, to our best knowledge, there are no reports of computational models for antileishmanial compounds that include data for multiple species and types of assays. The main pitfalls of classic Cheminformatics models are the impossibility to predict simultaneously multiple biological activity parameters of drugs against different target proteins, cell lines, organisms of assay, etc. They fail to perform multi-label and multi-output classification of new compounds. Our group has introduced the PTML ¼Perturbation Theory (PT) þMachine Learning (ML) algorithm to solve similar problems (multiple structures vs. multiple species and conditions of assay) in the drug discovery process [19]. PTML models start with a known value (reference) of expected biological activity for a group of compounds or property and add the effect of perturbations (deviations) of the structure and/or conditions of assay of the new case (query compound) with respect to the reference. The PTML approach uses PT Operators (PTOs) such as deviations, moving averages, etc. to quantify the effect of these deviations or perturbations over the final biological activity. The simpler PTML models are linear additive models, but more complicated and general PTML models can be constructed. PTML models have been used successfully for predicting different parameters of biological activity and/or toxicity (K i ,IC 50 ,LD 50 , Km,% inhibition, etc.) for the interaction of different compounds with different biological targets (proteins, tissues, cell lines, pathogenic organisms, etc.) [20]. Therefore, PTML models are useful to select compounds out of the series to be sent to pharmacological assays. We describe herein the application of a carbopalladation initiated domino reaction to the synthesis of a series of pyrrolo[1,2-b] isoquinoline scaffolds, and their in vitro biological evaluation against two species of Leishmania,L. amazonensis and L. donovani, which cause CL and VL, respectively. The effect of varying substitution pattern of the aromatic ring, as well as the nature of the C10b substituent will be explored. In addition, we have developed the first general-purpose PTML model to predict the antileishmanial activity of these series of pyrroloisoquinolines in different biological assays against different Leishmania species. 2. Chemistry The pyrrolo[1,2-b]isoquinolines 3were synthetized through a intramolecular domino Heck/nucleophile capture anion reaction on 2-alkenyl N-(o-iodobenzyl)pyrroles 1, which were prepared following previously reported procedures by us [17]. These Heck/ nucleophile capture anion cascade reactions have been mainly applied to the construction of five-membered rings, so our aim will also be to expand the scope of the procedure to the synthesis of sixmembered rings. The main challenge was to control the chemoselectivity of the process to favor the initial 6-exo carbopalladation over the early coupling of aromatic ring with cyanide ion, so all experimental conditions should be carefully chosen to control the nature of the intermediate palladium(II) species that determine the selectivity. In these cascade reactions it is crucial to choose, not only the adequate palladium catalyst, but also the cyanide source. Since Grigg’s seminal work on palladium catalyzed alkene arylcyanation using KCN as cyanide source [21], the method has been improved. Thus, Neuville and Zhu used K 4 Fe(CN) 6 in the intramolecular palladium-catalyzed domino Heck/cyanation sequence for the synthesis of a 3,3 0 -disubstituted oxindole [22]. Later, Lautens reported the diastereoselective synthesis of dihydroisoquinolinones starting from enantiopure N-allylcarboxamide with Zn(CN) 2 as cyanide source [23]. The same reagent was used in a related Pdcatalyzed dearomative indole bisfunctionalization reaction for the preparation of indolines with congested stereocenters [24]. To test the viability of the method outlined above, we carried out an optimization of the Heck/cyanation reaction conditions on N-(oiodobenzyl)pyrrole 1a.Wefirst explored the reaction with phosphane-free catalytic systems for economical and environmental reasons, testing several cyanation agents. Thus, after examining various pre-catalysts [Pd(OAc) 2 , Pd(PPh 3 ) 4 , Pd(dba) 2 , Pd 2 (dba) 3 $CHCl 3 ,etc.], organic (NEt 3 , PMP) and inorganic bases (K 2 CO 3 ,Cs 2 CO 3 , NaHCO 3 ), polar and non-polar solvents, additives and reaction temperatures, the optimal results were obtained using Pd(OAc) 2 (10 mol%) as catalyst, K 4 Fe(CN) 6 $3H 2 O (0.22 equiv.) as cyanide source, Na 2 CO 3 (1.3 equiv.) as inorganic base and nBu 4 NCl (1 equiv.) as additive, using a mixture of DMF:H 2 O (8:2) as solvent at 120  C (See Supplementary Material for the details of the optimization of the reaction conditions). It is noteworthy that K 4 Fe(CN) 6 $3H 2 O is the best cyanation agent, as it is non-toxic compared with traditional cyanation reagents [KCN, NaCN, Zn(CN) 2 , TMSCN] and it can easily be handled without special precautions [25]. In addition, substoichiometric amounts of reagent can be used, as all cyanide ions bound to the iron (II) center can be released in the cyanation reaction. In fact, the reaction was slowed (24 h vs. 1 h) when using higher amounts of K 4 Fe(CN) 6 $3H 2 O (0.44 vs. 0.22 equiv.), probably because higher concentration of free cyanide ions in the reaction mixture can form catalytically inactive palladium(II)-cyano complexes that lead to partial catalyst deactivation [26]. Notably, the presence of nBu 4 NCl as phase-transfer catalyst was essential to shorten the reaction time and to enhance the formation of pyrroloisoquinoline 3a. In fact, it is known that the addition of tetrabutylammonium halides increases the rates of some steps of the catalytic cycle of the Heck reaction, favoring the 6-exo-trig cyclization step [27]. However, the yield is moderate (65%), as the competitive direct cyanation coupling reaction to give 4a (26%) could not be completely suppressed. The use of a series of phosphane ligands [PPh 3 ,PtBu 3 , P(2-furyl) 3 ] did not improve the efficiency of the cascade cyclization of 1a. Selected examples are shown on Table 1 (See Supplementary Material for more details). Therefore, either using conditions that favor cationic or neutral mechanisms for the syn insertion of the ArePd(II)-X complex to the alkene, where the selectivity of the Heck reaction is controlled [28], we could achieve the selective synthesis of the pyrroloisoquinolines 3. With an optimized set of conditions in hand, the scope of the Heck/cyanation cascade was examined. Generally, different substitution patterns on the aromatic ring are tolerated, affording the pyrroloisoquinolines 3in moderate yields, as the direct cyanation process was always competitive (Table 2). Specifically, strongly electron-donating groups as methoxy and methylenedioxy groups (Table 2, entries 3e4 and 7e9) and electron-withdrawing groups (F) (entry 6), as well as unsubstituted derivatives (Table 2, entry 2) were viable under the optimized reaction conditions. Finally, the cyano group of the pyrrolo[1,2-b]isoquinolines 3 I. Barbolla, L. Hern andez-Su arez, V. Quevedo-Tumailli et al. European Journal of Medicinal Chemistry 220 (2021) 113458 3 obtained through the domino Heck/cyanation reaction has been efficiently derivatized to different functional groups, such as aldehyde, amide and amine, showing the versatility of the procedure (see Supplementary Material). 3. Antileishmanial assays The new synthesized 10-cyanomethyl substituted 5,10dihydropyrrolo[1,2-b]isoquinoline derivatives 3, together with the previously obtained C-10 (hetero)arylmethyl substituted derivatives 2[17], were screened against two species of Leishmania, L. amazonensis and L. donovani, which cause CL and VL, respectively (Fig. 2). In vitro promastigote susceptibility assays and in vitro intracellular amastigote susceptibility assays have been carried out, as well as cytotoxicity assay on J774 cell line of macrophages, a line of macrophages used to test cytotoxicity of drugs in vitro prior to animal tests (see Experimental Section). Miltefosine was the drug of reference, as it can be used for the treatment of different forms of the disease. The initial screening on the in vitro promastigote assays revealed that some 5,10-dihydropyrrolo[1,2-b]isoquinolines favorably compare to miltefosine in terms of activity and selectivity against L. amazonensis. In general, best activity was found for the 10-arylmethyl substituted derivatives 2aa,2ab, 2ad,2ae,2ag,2ah, 2bb, and 2db. In particular, the most active and selective compounds 2ad (IC 50 ¼3.30 ±2.80 m M, SI >77.01) and 2bb (IC 50 ¼3.93 ±0.23 m M, SI >58.77) were approximately 10-fold more potent and selective than the drug of reference (miltefosine), followed by compound 2db (IC 50 ¼8.00 ±0.28 m M, SI >34.4). In contrast, the presence of a cyanomethyl group at C-10 resulted in compounds with weaker antileishmanial activity against L. amazonensis (Table 3, entries 12e19). Only compounds 3h and 3i, which have an O-benzyl group at the 7-position of the pyrroloisoquinoline core, have an activity against L. amazonensis comparable to that of miltefosine (Table 3, entries 18e19 vs. entry 20). Therefore, the presence of a benzyl group in the pyrroloisoquinoline core seems to be crucial for the antileishmanial activity. The aromatic substitution pattern of the pyrroloisoquinoline core of this series 2also plays an important role, having the electron-donating substituents a positive impact on antileishmanial activity against L. amazonensis. Thus, compound 2cb, with an unsubstituted aromatic ring, is 4-47-fold less potent than other compounds of series 2(Table 3, entry 10 vs. entries 1 and 11). Besides, as shown in Table 3, replacement of the methyl group at the 10-position of the pyrroloisoquinoline ring system by a trifluoromethyl group resulted in more active compounds, being 2bb over 4-fold more active against L. amazonensis than 2ab (Table 3, entry 2 vs. entry 9). Regarding the substitution pattern of the benzyl moiety, the presence of two trifluoromethyl groups was detrimental for activity and selectivity, being 2ac the less active and more toxic derivative of both series (Table 3, entry 3). However, it should be pointed out that the introduction of this CF 3 group at C10 has less effect (Table 3, entry 2 vs. entry 9). Conversely, all pyrroloisoquinolines tested are notably less active than miltefosine for the treatment of L. donovani, being again compound 2bb one of the more active and selective of both series with IC 50 ¼16.41 ±4.90 m M and SI >14.09. In addition, and very notably, almost all pyrroloisoquinolines are much less toxic than the drug of reference with values of concentration of the compound that produces 50% reduction of cell viability (Cytotoxic Concentration, CC 50 ) in the range 195e416 m M in J774 cells. In fact, the Selectivity Index (SI ¼CC 50 /IC 50 ) is higher for almost all pyrroloisoquinolines than for miltefosine whose SI is only 4.43. It also should be pointed out that although better IC 50 values have been reported for different types of molecules against different Leishmania species [29], they usually have lower CC 50 values than most Table 1 Optimization of the carbopalladation/cyanation cascade reaction of 1a. Entry [Pd] Ligand Time (h) Yield 3a (%) a Yield 4a (%) a 1 Pd(OAc) 2b e154 32 2 Pd(OAc) 2c e165 26 3 Pd(TFA) 2b e446 30 4 Pd(dba) 2b e48 37 d 48 5Pd 2 (dba) 3 $CHCl 3b e48 53 e 27 6 Pd(OAc) 2c L1 644 40 7 Pd(OAc) 2c L2 152 35 8 Pd(OAc) 2c L3 24 24 56 9 Pd(TFA) 2c L2 24 33 31 10 Pd(dba) 2c L2 642 33 11 Pd 2 (dba) 3 $CHCl 3c L2 136 39 a Isolated yield. b Na 2 CO 3 (1 equiv.). c Na 2 CO 3 (1.3 equiv.). d Conversion 88%. e Conversion 86%. Table 2 Carbopalladation/cyanation cascade reaction. Synthesis of 3a-i. Entry Ar Time (h) Product Yield (%) a Product Yield (%) a 1 1 3a 65 4a 26 243b 25 4b 41 313c 37 4c 38 423d 39 4d 29 548 3e 21 b 4e - c 6 48 3f 16 4f 18 763g 23 d 4g 28 813h 23 4h 15 924 3i 16 4i 30 a Yield (%) of isolated pure compound. b Conversion: 63%. c 4e was detected by 1 H NMR in reaction crude, but it could not be isolated. d Conversion: 94%. I. Barbolla, L. Hern andez-Su arez, V. Quevedo-Tumailli et al. European Journal of Medicinal Chemistry 220 (2021) 113458 4 of our pyrroloisoquinolines, which are not toxic at the highest tested dose (CC 50 >100 m g/mL). Therefore, this is a very interesting result, considering that currently available chemotherapy for leishmaniasis is hampered by toxicity and drug resistance. Then, the most active compounds of the promastigote assays were screened against amastigotes of L. amazonensis and L. donovani (Table 4). All pyrroloisoquinolines tested were less active than miltefosine when tested against L. donovani amastigotes.However, these compounds have IC 50 values in the range 33.59e77.12 m M, which are similar or even better than miltefosine (IC 50 ¼47.60 ±7.04 m M) in the test against L. amazonensis. All of them have SI values between 3.58 and 8.93, which are 2to 4-fold higher than miltefosine (SI ¼2.85). In this case, pyrroloisoquinoline 2ae showed the best activity with IC 50 ¼33.59 ±2.64 m M and higher selectivity with SI >8.93. We can observe that 10-arylmethyl substituted dihydropyrroloisoquinoline series seems to be more selective and presents similar or even better activity than the drug of reference for the treatment of L. amazonensis amastigotes.The result is promising because in the present model of experimental assay the amastigotes are inside the human macrophage cells. Therefore, it seems that these pyrroloisoquinolines could be able to cross/attached host and parasite barriers to exert their activity without damaging the host membrane (macrophage membrane). First, they need to cross the membrane of the host cells (macrophages), next they need to cross Parasitophorous Vacuole Membrane (PVM). Last, if the activity is not over PVM directly, the compounds should reach/cross the parasite membrane to reach their molecular target in the membrane or inside the parasite. The PVM prevents the acidification of the media by lysosomes of the host cell to destroy an Fig. 2. C-10 substituted 5,10-dihydropyrrolo[1,2-b]isoquinolines 2and 3screened against L. amazonensis and L. donovani. Table 3 IC 50 leishmanicidal and cytotoxic effects from pyrroloisoquinoline derivatives (expressed as m M) on in vitro promastigote assay. Entry Comp. L. amazonensis L. donovani Macrophages J774 IC 50 ±SD ( m M) a SI b IC 50 ±SD ( m M) a SI b CC 50 ±SD ( m M) c 12aa 26.41 ±2.20 >10.40 47.40 ±1.79 >5.80 275.13 e 22ab 18.00 ±0.79 >14.68 79.93 ±4.36 >3.31 264.24 e 32ac 39.62 ±2.13 1.18 111.34 ±10.8 0.42 46.80 ±4.00 42ad 3.30 ±2.80 >77.01 38.30 ±3.36 >6.63 254.14 e 52ae 17.09 ±0.60 >17.54 30.29 ±2.01 >10.09 299.91 e 62af 36.71 ±0.44 >5.95 86.01 ±1.60 >2.54 218.54 e 72ag 23.29 ±1.41 >15.15 124.43 ±14.70 >2.84 352.89 e 82ah 12.24 ±0.56 >23.00 33.66 ±2.06 >8.27 278.20 e 92bb 3.93 ±0.23 >58.77 16.41 ±4.90 >14.09 231.30 e 10 2cb 157.35 ±6.28 >2.00 159.40 ±26.20 >1.97 314.09 e 11 2db 8.00 ±0.28 >34.49 30.35 ±3.92 >9.09 275.95 e 12 3a 309.20 ±41.10 >1.14 226.21 ±11.60 >1.57 354.18 e 13 3c 77.35 ±1.13 >4.85 120.12 ±12.01 >3.13 375.51 e 14 3d 171.07 ±6.38 >2.07 nd d nd d 354.18 e 15 3e 93.85 ±6.02 >3.77 nd d nd d 354.20 e 16 3f 96.55 ±1.66 >4.31 192.19 ±7.66 >2.17 416.18 e 17 3g 123.25 ±10.70 >3.21 359.99 ±34.44 >1.10 396.32 e 18 3h 15.83 ±0.06 14.04 60.25 ±0.34 3.69 222.30 ±42.70 19 3i 24.82 ±0.28 7.86 34.37 ±5.30 5.68 195.30 ±27.30 20 miltefosine 30.70 ±0.98 4.43 0.15 ±0.02 906.00 135.90 ±10.30 a IC 50 : Concentration of the compound that produced a 50% reduction in parasites; SD: Standard Deviation. b SI: Selectivity Index, SI ¼CC 50 /IC 50 . c CC 50 : Concentration of the compound that produced a 50% reduction of cell viability in treated culture cells with respect to untreated ones. d nd: not determined. e CC 50 values, expressed as m M, correspond to 100 m g/mL, which was the higher doses tested. I. Barbolla, L. Hern andez-Su arez, V. Quevedo-Tumailli et al. European Journal of Medicinal Chemistry 220 (2021) 113458 5 invading parasite. PVM is shaped by the parasite using parts of the membrane of the host cell. The PVM surrounds the intracellular parasite, creating a separate bubble of cytoplasm-filled plasma membrane within the host cell [30]. 4. Computational model As described above, we have measured the experimental IC 50 ( m M) values for the series of pyrroloisoquinolines 2and 3from the tests against two species of Leishmania,L. amazonensis and L. donovani, at two different stages (S) of development of the parasite: amastigotes (A) and promastigotes (P). We could observe a certain tendency in the behavior of these pyrroloisoquinolines against the promastigotes of the two different species of Leishmania studied. In fact, we found a regression coefficient of R ¼0.67 for the IC 50 of pyrroloisoquinoline derivatives in L. amazonensis vs. L. donovani. However, we have not yet clues about the possible target proteins of these compounds. In addition, there are many other species of Leishmania (including MDR strains) that could not be assayed and may present different susceptibilities to the same drugs. In order to put our results into context, a ChEMBL dataset of >145,000 preclinical assays of putative antileishmanial compounds was downloaded and explored, as detailed in Experimental Section and the Supplementary Material. The experiments include up to 10 different conditions of assay c j ¼[c 0 ,c 1 ,c 2 ,…c 10 ], which were split into two partitions or subsets of experimental conditions c I and c II . The principal conditions c I are those that characterize the biological experiment per se (parameter measured, target protein, parasite stage, organism of assay, etc.). The dataset includes values for n(c 0 )>50 different biological parameters measured for compounds vs. at least one out of n(c 1 )¼32 target proteins, n(c 2 )¼29 cell lines, n(c 3 )¼40 organisms of assay (not all are parasites), and n(c 4 )¼37 Leishmania parasite species or strains, etc. Instead of outcomes for all possible experiments, the dataset includes n(c I )>240 combinations of these principal conditions c I (different experiments) for numerous compounds. In addition, the dataset includes n(c II )¼80 combinations of secondary c II conditions related to the biological nature and/or accuracy of data (target type, target mapping, value exactitude, etc.). This study revealed that above 70 assays involving 20 species/strains of Leishmania are most commonly used. Unfortunately, even limiting the analysis to these assays it may be costly in terms of resources and time. Testing a short series of only 10 compounds would require n assay ¼10$70 ¼700 experimental assays. Therefore, we decided to carry out a preliminary computational study of the susceptibility of other species to this series of compounds. As stated above, there are no reports on a computational model to perform such study. Therefore, we first developed a new PTML model to carry out this type of predictions. The equation of the best PTML model found is the following: fð n ijÞcalc ¼59:918599 $fð n ijÞref þ1:631537$ D D1ðcIÞ þ0:041494$ D D2ðcIÞ2:675709 $ D D3ðcIÞ 1:562187$ D D1ðcIIÞe0:041886$ D D2ðcIIÞ þ2:649511$ D D3ðcIIÞ25:182671 n¼109389 c 2¼135169:7p<0:05 The idea of this PTML model is to start using as input a function of reference f(v ij ) ref , which is obtained from the experimental assays of a set of compounds of reference carried out under specific conditions c j . This function quantifies the expected prior probability to give a positive result in the specific assays for a compound selected at random. Next, the values of the PT Operators (PTOs) with the general form D D k (c j ) were added to the f(v ij ) ref . These PTOs quantify the deviation ( D ) of the molecular descriptors D k (structure) of our query compound with respect to the group of reference compounds (see Experimental Section). The descriptors D k are D 1 ¼ALOGP, the n-octanol/water partition coefficient, D 2 ¼TPSA, the Topological Polar Surface Area, and D 3 ¼NRV, the Number of Violations to Rule of V (Lipinski’sorPfizer’s rule) were used to identify each compound in the equation. The descriptors ALOGP and TPSA quantify the molecular structure of the drug by means of a weigthed sum of different molecular fragments in the molecules. NRV rule quantifies the likeness/similarity of one compound with respect to known drugs based on molecular weight, hydrogen bonds, etc.[31]. The values of D k were extracted from ChEMBL dataset and/or calculated with the software DRAGON for new compounds. Finally, the output of the model f(v ij ) calc is a scoring function used to calculate the probability of activity p(f(v ij ) pred ¼1) of different compounds. In order to train this model, we selected at random a large training series of n ¼109,389 preclinical assays downloaded from ChEMBL database. The values of Specificity (Sp) and Sensitivity (Sn) are in the range z90e98% for training series (see Supplementary Material), which are excellent values for this type of ML classification models. Moreover, the p-level <0.05 for the Chi-square test with c 2 ¼135,169.7 points to a statistically significant discrimination Table 4 IC 50 Leishmanicidal and cytotoxic effects from pyrroloisoquinoline derivatives (expressed as m M) on in vitro amastigote assay. Entry Compound L. amazonensis L. donovani Macrophages J774 IC 50 ±SD ( m M) a SI b IC 50 ±SD ( m M) a SI b CC 50 ±SD ( m M) c 12aa 51.56 ±12.00 >5.34 45.00 ±7.51 >6.12 275.13 e 22ab 56.12 ±13.50 >4.70 29.62 ±6.87 >8.92 264.24 e 32ad 60.79 ±5.74 >4.18 46.30 ±0.33 >5.49 254.14 e 42ae 33.59 ±2.64 >8.93 55.03 ±3.96 >5.45 299.91 e 52ag 43.54 ±9.53 >8.10 255.88 ±42.10 >1.38 352.89 e 62ah 69.26 ±6.96 >4.02 16.74 ±0.14 >16.61 278.20 e 72bb 56.72 ±4.58 >4.08 22.17 ±4.16 >10.43 231.30 e 82db 77.12 ±19.10 >3.58 68.65 ±9.63 >4.02 275.95 e 93h nd d e44.75 ±8.53 4.97 222.30 ±42.70 10 miltefosine 47.60 ±7.04 2.85 0.37 ±0.05 369.30 135.90 ±10.30 a IC 50 : Concentration of the compound that produced a 50% reduction in parasites; SD: Standard Deviation. b SI: Selectivity Index, SI ¼CC 50 /IC 50 . c CC 50 : Concentration of the compound that produced a 50% reduction of cell viability in treated culture cells with respect to untreated ones. d nd: not determined. e CC 50 values, expressed as m M, correspond to 100 m g/mL, which was the higher doses tested. I. Barbolla, L. Hern andez-Su arez, V. Quevedo-Tumailli et al. European Journal of Medicinal Chemistry 220 (2021) 113458 6 between active (f(v ij ) obs ¼1) and non-active compounds (f(v ij ) obs ¼0) in all these assays. This model may be used to predict new compounds not included in training series. First, the values of the f(v ij ) ref and PTOs (containing D k of drug and <D k (c j )>of assay) were substituted in the equation to calculate the output function f(v ij ) calc . Next, the values of f(v ij ) calc were transformed into posterior probabilities of success p(f(v ij ) pred ¼1) for each compound in different assays using a sigmoid function. Once the values of probabilities were calculated, the compounds could be classified. Thus, those outcomes in the range of probability p(f(v ij ) pred ¼1) >0.5 are considered interesting for assay f(v ij ) pred ¼1. Then, the present PTML model was tested with a very large validation series, obtaining values of Sp and Sn also in the range z90e98% for the external validation series. In conclusion, this simple but powerful PTML model predicts very well (overall Accuracy ¼97.8%) a large dataset (training þvalidation) of antileishmanial activity preclinical assays (n assay >145,000) involving 96,800 unique compounds. 5. Predictive study As mentioned above, some compounds of our series show interesting IC 50 values and can be considered active (f(v ij )¼1)) using the cutoff of IC 50 ¼10 m M in the range of more typical experimental studies. Then, we decided to use this new model to carry out a computational prediction of the outcomes of the 19 pyrroloisoquinolines of our series vs. different Leishmania species (>20) in more than 160 different preclinical assays. This PTML model is able to predict the posterior probability p(f(v ij ) pred ¼1) of getting the desired level for more than 50 different biological properties (IC 50 ,K i ,K m ,etc.). In this preliminary study, IC 50 was selected as unique property, as it was the first property experimentally measured in the early stages of screening, using a very low cut-off value of IC 50 ¼0.01 m M, in an effort to reduce the number of false positive cases from the first steps of screening. Consequently, the model was used to calculate the values of probability p(IC 50 ( m M) <0.01) calc for which a compound would show an IC 50 ( m M) <0.01 in different assays. The total number of calculations for in vitro assays that do not specify the target protein was n calc1 ¼n cmpd $n assay ¼19$162 ¼3078. The full results of this predictive study are compiled in Table S9 of the Supplementary Material. The values of p(IC 50 ( m M) <0.01) calc for assays with known protein targets were also predicted, being the number of calculations n calc2 ¼n cmpd $n prot ¼19$32 ¼608 for the 19 compounds vs. 32 target proteins of different species. It makes a total of n calc ¼n calc1 þn calc2 ¼3686 calculations for our 19 compounds in different assays. For the sake of simplicity, we only changed the conditions c I (assay per se) using a fix sub-set of conditions c II for the 3686 calculations. Consequently, the total number of predicted values of p(IC 50 ( m M) <0.01) calc was 3686. In order to summarize the results and withdraw conclusions, we calculated the average value of the probability p(IC 50 ( m M) <0.01) avg , which is the average probability that the compound is predicted to have an IC 50 ( m M) <0.01 in multiple assays vs. the same species. The PTML model predicts a coherent behavior for this series of compounds as a homologous series. It means that species resistant/susceptible to one compound are predicted to be resistant to the action of almost all compounds of the whole series. The model does not detect overall significant differences on the behavior of both subseries (2aa-2db vs.3a-3i subseries) of compounds, which could be coherent with the fact that both subseries of compounds have the pyrroloisoquinoline core. The average value of probability p(IC 50 ( m M) <0.01) avg for almost all compounds of the series are in the range 0.1e0.4, p(IC 50 ( m M) <0.01) <0.5, for many species of Leishmania, which is in agreement with the experimental findings. Table 5 shows selected results of the predictive study vs. different species for the two top lead compounds of each subseries (2ad,2bb, 3h, and 3i) experimentally tested in this work (see Supplementary Material for full details). Interestingly, the PTML model predicts values of p(IC 50 ( m M) <0.01) avg >0.8 for some species/strains not previously tested, despite of the demanding computational threshold value used in the computational study. See, for example, the values of p(IC 50 ( m M) <0.01) avg of the four top hit compounds for L. braziliensis strain M2904 (L. brm.), and L. major strain Friendlin (L. maf.), and L. mexicana (L. mex.), shown in Table 5.Regarding the possible target protein, the model predicts values of p(IC 50 ( m M) < 0.01) avg >0.7 for both subseries of pyrroloisoquinolines versus some proteins. The study points to the Trypanothione reductase of L. donovani (P39050) [32], L -isoaspartate( D -aspartate) O-methyltransferase (P22061) of L. donovani [33] and the Ornithine decarboxylase of L. donovani (P27116) [34] as plausible targets to be tested, although more candidates are also collected in Table 5.Itis interesting that some of these proteins are enzymes with amino acid derivatives (Trypanothione, D -aspartate, and Ornithine) as substrates. Therefore, this computational study opens the door to a further testing of these compounds or their derivatives vs. other species of Leishmania and specific tests vs. probable protein targets. 6. Conclusions In conclusion, the palladium-catalyzed Heck-initiated cascade reactions are effective procedures for the construction of the pyrrolo[1,2-b]isoquinoline scaffolds with a quaternary center at C-10b. The in vitro evaluation of their leishmanicidal activity against visceral (L. donovani) and cutaneous (L. amazonensis) leishmaniasis revelead that almost all compounds showed very low cytotoxicity, CC 50 >100 m g/mL in J774 cells (highest tested dose). This is an important feature, as drug toxicity is one of the main limitations of current chemotherapy for leishmaniasis. In general, 10-arylmethyl substituted pyrroloisoquinolines showed best activity against L. amazonensis on in vitro promastigote assays. In particular, 2ad (IC 50 ¼3.30 m M, SI >77.01) and 2bb (IC 50 ¼3.93 m M, SI >58.77) were approximately 10-fold more potent and selective than the drug of reference (miltefosine). On the other hand, 2ae was the more active compound in the in vitro amastigote assays (IC 50 ¼33.59 m M, SI >8.93). In addition, it has been demonstrated that Perturbation Theory Machine Learning (PTML) algorithms are useful to model large (>145,000 cases) ChEMBL datasets of antileishmanial preclinical assays. It is possible to use the developed PTML model to reduce assay costs by predicting the probability with which a query compound of this or other series of compounds reaches a desired level for multiple parameters (IC 50 ,K i ,etc.) vs. different Leishmania species and target proteins, with high values of specificity (>98%) and sensitivity (>90%) in both training and validation series. 7. Experimental section 7.1. Chemistry General. All commercial chemicals were reagent grade and were used without further purification unless otherwise specified. Palladium catalysts were commercially available, and were used without further purification: Pd(OAc) 2 : 98% purity; Pd(TFA) 2 :97% purity; Pd(PPh 3 ) 4 : 99% purity.; Pd(dba) 2 : 99.9% purity; Pd 2 (dba) 3 $CHCl 3 : 97% purity. The 1-(o-iodobenzyl)-2alkenylpyrroles 1were synthetized according to the procedures previously reported by us [17]. All solvents used in reactions were anhydrous and purified according to standard procedures. All airor moisture-sensitive reactions were performed under argon; the I. Barbolla, L. Hern andez-Su arez, V. Quevedo-Tumailli et al. European Journal of Medicinal Chemistry 220 (2021) 113458 7 glassware was dried (130  C) and purged with argon. TLC was carried out with 0.2 mm-thick silica gel Merck F254 plates. Visualization was accomplished by UV light ( l ¼254 nm and 360 nm). Flash column chromatographic separations and purifications were performed on silica Flash P60 (Silicycle), 230e400 mesh ASTM. Final compounds were purified to 95% purity as assessed by 1 H NMR spectra and analytical liquid chromatography. Melting points were measured in a Büchi B-540 apparatus in unsealed capillary tubes. IR spectra were obtained using Attenuated Total Reflection (ATR) in a JASCO FT/IR 4100 in the interval between 4000 and 400 cm 1 with a 4 cm 1 resolution. Only characteristic bands are given in each case. 1 H and 13 C NMR spectra were recorded at 20e25  C on either a Bruker AC-300 spectrometer (300 MHz for 1 H and 75.5 MHz for 13 C) and on a Bruker AC-500 spectrometer (500 MHz for 1 H and 125.7 MHz for 13 C). Chemical shifts are reported in parts per million (ppm) relative to an internal solvent reference. Recorded peaks are listed in the order multiplicity (s, singlet; d, doublet; dd, doublet of doublets; m, multiplet), coupling constants, and number of protons. Assignments of individual 13 C and 1 H resonances are supported by DEPT experiments and 2D correlation experiments (COSY, HSQCed or HMBC) when necessary. High resolution mass spectra (HRMS) were performed by the Mass Spectrometry General Service at the University of the Basque Country using an ultra performance liquid chromatograph (Acquity UPLC, Waters Chromatography.), in tandem with a QTOF mass spectrometer (SYNAPT G2 HDMS, Waters Chromatography), with an electrospray ionization source in a positive mode. Table 5 PTML prediction of average value of probability p(IC 50 ( m M) <0.01) avg for pyrroloisoquinolines 2ad,2bb,3h, and 3i against >20 different Leishmania species. Leish. S b Target Compound Leish. S b Target Compound Species Protein 2ad 2bb 3h 3i Species a Protein 2ad 2bb 3h 3i L. aet. Ae0.02 0.01 0.01 0.01 L. maj. eP37268 0.41 0.34 0.30 0.32 L. aet. Pe0.02 0.01 0.01 0.01 L. maj. eQ01782 0.58 0.51 0.47 0.49 L. ama. Ae0.10 0.08 0.06 0.07 L. maj. eQ0GKD7 0.43 0.36 0.32 0.34 L. ama. Pe0.21 0.17 0.15 0.15 L. maj. eQ4Q5S8 0.04 0.03 0.03 0.03 L. ama. eO96394 0.64 0.58 0.53 0.55 L. maj. eQ4Q5W4 0.78 0.73 0.69 0.71 L. ari. ee 0.21 0.17 0.15 0.16 L. maj. eQ4QBL1 0.52 0.46 0.41 0.43 L. bra. Ae0.12 0.09 0.08 0.09 L. maj. eQ4QE15 0.02 0.02 0.01 0.02 L. bra. Pe0.20 0.16 0.14 0.15 L. maj. eQ6S996 0.17 0.13 0.11 0.12 L. brm. Ae0.93 0.91 0.89 0.90 L. maj. eQ9LM02 0.04 0.03 0.02 0.03 L. brm. Pe0.11 0.09 0.07 0.08 L. maj. Pe0.06 0.05 0.04 0.05 L. cha. Ae0.18 0.15 0.13 0.14 L. maj. P Q01782 0.41 0.35 0.31 0.33 L. cha. Pe0.25 0.21 0.18 0.19 L. maf. Pe0.94 0.92 0.90 0.91 L. don. Ae0.30 0.27 0.25 0.25 L. mex. Ae0.30 0.26 0.24 0.25 L. don. eP39050 0.83 0.79 0.76 0.77 L. mex. eP04406 0.73 0.67 0.63 0.65 L. don. eQ95WR6 0.16 0.13 0.11 0.12 L. mex. eP36400 0.35 0.29 0.26 0.27 L. don. eQ95Z89 0.29 0.24 0.21 0.22 L. mex. eQ01558 0.60 0.54 0.50 0.51 L. don. eQ9NJG8 0.07 0.05 0.04 0.05 L. mex. eQ27686 0.96 0.95 0.95 0.95 L. don. Pe0.25 0.23 0.22 0.23 L. mex. eQ4U254 0.41 0.35 0.31 0.32 L. don. P P22061 1.00 1.00 1.00 1.00 L. mex. eQ9U5N6 0.34 0.28 0.25 0.26 L. don. P P27116 1.00 0.99 0.99 0.99 L. mex. Pe0.27 0.22 0.20 0.21 L. dod. ee 0.21 0.17 0.15 0.16 L. mex. P P11166 0.09 0.07 0.06 0.06 L. enr. Pe0.13 0.10 0.09 0.09 L. mem. eQ27686 0.29 0.24 0.21 0.22 L. gar. ee 0.21 0.17 0.15 0.16 L. mem. Pe0.02 0.01 0.01 0.01 L. guy. Ae0.08 0.06 0.05 0.06 L. mev. ee 0.21 0.17 0.15 0.16 L. guy. Pe0.28 0.23 0.20 0.21 L. pan. Ae0.10 0.08 0.07 0.07 L. inf. Ae0.32 0.28 0.26 0.27 L. pan. Pe0.42 0.38 0.35 0.36 L. inf. eQ8I6E4 0.28 0.23 0.20 0.21 L. per. Pe0.23 0.19 0.16 0.17 L. inf. Pe0.20 0.17 0.15 0.15 L. pif. Ae0.11 0.09 0.07 0.08 L. maj. Ae0.23 0.20 0.18 0.18 L. pif. Pe0.21 0.17 0.14 0.15 L. maj. eO15826 0.76 0.70 0.67 0.68 L. pro. Pe0.11 0.09 0.07 0.08 L. maj. eO96526 0.07 0.06 0.05 0.05 L. tar. ee 0.10 0.08 0.07 0.07 L. maj. eP00374 0.21 0.17 0.15 0.16 L. tro. Pe0.18 0.14 0.12 0.13 L. maj. eP07382 0.22 0.18 0.16 0.16 L. tur. ee 0.08 0.06 0.05 0.05 a Leishmania species: L. aethiopica ¼L. aet., L. amazonensis ¼L. ama., L. aristidesi ¼L. ari.,L. braziliensis ¼L. bra., L. braziliensis M2904 ¼L. brm,L. chagasi ¼L. cha., L. donovani ¼ L. don.,L. donovani donovani ¼L. dod.,L. enriettii ¼L. enr.,L. garnhami ¼L. gar., L. guyanensis ¼L. guy,L. infantum ¼L. inf.,L. major ¼L. maj., L. Mexicana ¼L. mex.,L. mexicana mexicana ¼L. mem.,L. mexicana venezuelensis ¼L. mev.,L. panamensis ¼L. pan.,L. peruviana ¼L. per.,L. pifanoi ¼L. pif.,L. tarentolae ¼L. tar., L. tropica ¼L. tro., L. turanica ¼L. tur., L. major strain Friendlin ¼L. maf. b S¼Stage: A ¼Amastigotes, P ¼Promastigotes. I. Barbolla, L. Hern andez-Su arez, V. Quevedo-Tumailli et al. European Journal of Medicinal Chemistry 220 (2021) 113458 8 Intramolecular Heck/cyanide capture cascade reaction on 1. Synthesis of pyrrolo[1,2-b]isoquinolines 3. General procedure. Pd(OAc) 2 (0.1 mmol) was added to a mixture of N-(o-iodobenzyl) pyrrole 1(1 mmol), potassium hexacyanoferrate(II) trihydrate (0.22 mmol), sodium carbonate (1.3 mmol), and tetrabutylammonium chloride (1 mmol) in an 8/2 mixture of DMF/H 2 O (3 mL). The mixture was stirred at 120  C for the time indicated in each case. H 2 O (15 mL) was added and the resulting aqueous phase was extracted with EtOAc (3 20 mL). The combined organic extracts were washed with brine (3 20 mL), dried over anhydrous Na 2 SO 4 and concentrated in vacuo. Purification by column chromatography (silica gel) of the resulting residue afforded the corresponding pyrroloisoquinoline 3and direct aryl halide cyanation product 4 (see Supplementary Material). 2-(7,8-Dimethoxy-10-methyl-5,10-dihydropyrrolo[1,2-b]isoquinolin-10-yl)acetonitrile (3a). According to General Procedure, N-(o-iodobenzyl)pyrrole 1a (115 mg, 0.30 mmol) was treated with Pd(OAc) 2 (6.7 mg, 0.03 mmol), potassium hexacyanoferrate(II) trihydrate (27.9 mg, 0.07 mmol), sodium carbonate (41.3 mg, 0.39 mmol) and tetrabutylammonium chloride (83.4 mg, 0.30 mmol) in a 8/2 mixture of DMF/H 2 O (1 mL) for 1 h. After workup, purification by column chromatography (silica gel, petroleum ether/EtOAc 8/2) afforded 3a (54.9 mg, 65% yield) as a yellow solid: m. p. (petroleum ether/EtOAc): 132e134  C. IR (ATR): 2935, 2245, 1515 cm 1 . 1 H NMR (300 MHz, CDCl 3 ): d 7.05 (s, 1H), 6.75e6.74 (m, 2H), 6.26 (dd, J¼3.6, 2.7 Hz, 1H), 6.20 (dd, J¼3.6, 1.7 Hz, 1H), 5.18 (d, J¼15.9 Hz, 1H), 5.07 (d, J¼15.9 Hz, 1H), 3.95 (s, 3H), 3.90 (s, 3H), 2.72 (s, 2H), 1.88 (s, 3H). 13 C{ 1 H} NMR (75.5 MHz, CDCl 3 ): d 148.6, 148.4, 132.5, 129.8, 124.0, 119.3, 117.8, 109.2, 108.8, 108.4, 103.5, 56.2, 56.0, 47.1, 37.5, 34.2, 26.4. MS (ESI) m/z (rel intensity): 283 (MH þ , 100), 242 (46). HRMS (ESI-TOF): calcd for C 17 H 19 N 2 O 2 [MH þ ] 283.1447; found, 283.1450. 2-(10-Methyl-5,10-dihydropyrrolo[1,2-b]isoquinolin-10-yl) acetonitrile (3b). According to General Procedure, N-(o-iodobenzyl)pyrrole 1b (101 mg, 0.30 mmol) was treated with Pd(OAc) 2 (6.7 mg, 0.03 mmol), potassium hexacyanoferrate(II) trihydrate (27.9 mg, 0.07 mmol), sodium carbonate (41.3 mg, 0.39 mmol) and tetrabutylammonium chloride (83.4 mg, 0.30 mmol) in an 8/2 mixture of DMF/H 2 O (1 mL) for 4 h. After workup, purification by column chromatography (silica gel, petroleum ether/EtOAc 95/5) afforded 3b (17.4 mg, 23% yield) as a colorless oil: IR (ATR): 2925, 2250, 1515 cm 1 . 1 H NMR (300 MHz, CDCl 3 ): d 7.59e7.56 (m, 1H), 7.40e7.28 (m, 3H), 6.76 (dd, J¼2.7, 1.7 Hz, 1H), 6.26 (dd, J¼3.6, 2.7 Hz,1H), 6.21 (dd, J¼3.6, 1.7 Hz,1H), 5.24 (d, J¼16.1 Hz,1H), 5.14 (d, J¼16.1 Hz,1H), 2.74 (s, 2H),1.89 (s, 3H). 13 C{ 1 H} NMR (75.5 MHz, CDCl 3 ): d 137.9, 132.4, 131.8, 128.1, 127.5, 126.6, 125.2, 119.4, 117.5, 108.8, 103.7, 47.5, 37.8, 33.8, 26.0. MS (ESI) m/z (rel intensity): 223 (MH þ , 65), 182 (100). HRMS (ESI-TOF): calcd for C 15 H 15 N 2 [MH þ ] 223.1235; found, 223.1237. 2-(10-Methyl-5,10-dihydro- [1,3]dioxolo[4,5-g]pyrrolo[1,2-b] isoquinolin-10-yl)acetonitrile (3c). According to General Procedure, N-(o-iodobenzyl)pyrrole 1c (110 mg, 0.30 mmol) was treated with Pd(OAc) 2 (6.7 mg, 0.03 mmol), potassium hexacyanoferrate(II) trihydrate (27.9 mg, 0.07 mmol), sodium carbonate (41.3 mg, 0.39 mmol) and tetrabutylammonium chloride (83.4 mg, 0.30 mmol) in a 8/2 mixture of DMF/H 2 O (1 mL) for 1 h. After workup, purification by column chromatography (silica gel, petroleum ether/EtOAc 8/2) afforded 3c (29.5 mg, 37% yield) as a yellow oil: IR (ATR): 2915, 2250, 1485 cm 1 . 1 H NMR (300 MHz, CDCl 3 ): d 7.01 (s,1H), 6.74e6.70 (m, 2H), 6.26e6.25 (m,1H), 6.18 (dd, J¼3.7, 1.7 Hz, 1H), 6.00 (s, 2H), 5.15 (d, J¼15.9 Hz,1H), 5.03 (d, J¼15.9 Hz, 1H), 2.73 (d, J¼16.4 Hz, 1H), 2.66 (d, J¼16.4 Hz, 1H), 1.84 (s, 3H). 13 C{ 1 H} NMR (75.5 MHz, CDCl 3 ): d 147.6, 147.0, 132.3, 131.3, 125.3, 119.2, 117.5, 108.8,106.4, 105.4, 103.6,101.5, 47.5, 37.8, 34.0, 26.3. MS (ESI) m/z (rel intensity): 267 (MH þ , 100), 226 (76). HRMS (ESI-TOF): calcd for C 16 H 15 N 2 O 2 [MH þ ] 267.1134; found, 267.1137. 2-(6,7-Dimethoxy-10-methyl-5,10-dihydropyrrolo[1,2-b]isoquinolin-10-yl)acetonitrile (3d). According to General Procedure, N-(o-iodobenzyl)pyrrole 1d (113 mg, 0.30 mmol) was treated with Pd(OAc) 2 (6.7 mg, 0.03 mmol), potassium hexacyanoferrate(II) trihydrate (27.9 mg, 0.07 mmol), sodium carbonate (41.3 mg, 0.39 mmol) and tetrabutylammonium chloride (83.4 mg, 0.30 mmol) in a 8/2 mixture of DMF/H 2 O (1 mL) for 2 h. After workup, purification by column chromatography (silica gel, petroleum ether/EtOAc 9/1) afforded 3d (32.4 mg, 39% yield) as a brown oil: IR (ATR): 2935, 2250, 1495 cm 1 . 1 H NMR (300 MHz, CDCl 3 ): d 7.26 (d, J¼8.7 Hz, 1H), 6.96 (d, J¼8.7 Hz, 1H), 6.79e6.78 (m, 1H), 6.26e6.24 (m, 1H), 6.17 (dd, J¼3.6, 1.7 Hz, 1H), 5.35 (d, J¼16.9 Hz, 1H), 5.05 (d, J¼16.9 Hz, 1H), 3.91 (s, 3H), 3.90 (s, 3H), 2.70 (s, 2H), 1.86 (s, 3H). 13 C{ 1 H} NMR (75.5 MHz, CDCl 3 ): d 151.2, 144.8, 132.5, 130.8, 126.1, 120.6, 119.7, 117.6, 111.7, 108.7, 103.4, 60.5, 55.8, 42.7, 37.4, 34.2, 26.1. MS (ESI) m/z (rel intensity): 283 (MH þ , 100), 242 (30). HRMS (ESI-TOF): calcd for C 17 H 19 N 2 O 2 [MH þ ] 283.1447; found, 283.1453. 2-(7,9-Dimethoxy-10-methyl-5,10-dihydropyrrolo[1,2-b]isoquinolin-10-yl)acetonitrile (3e). According to General Procedure, N-(o-iodobenzyl)pyrrole 1e (115 mg, 0.30 mmol) was treated with Pd(OAc) 2 (6.7 mg, 0.03 mmol), potassium hexacyanoferrate(II) trihydrate (27.9 mg, 0.07 mmol), sodium carbonate (41.3 mg, 0.39 mmol) and tetrabutylammonium chloride (83.4 mg, 0.30 mmol) in a 8/2 mixture of DMF/H 2 O (1 mL) for 48 h. After workup, purification by column chromatography (silica gel, petroleum ether/EtOAc 8/2) afforded 3e (18.1 mg, 21% yield) (63% conversion) as a yellow solid: m. p. (petroleum ether/EtOAc): 167e169  C. IR (ATR): 2935, 2250, 1460 cm 1 . 1 H NMR (300 MHz, CDCl 3 ): d 6.68 (dd, J¼2.7,1.7 Hz,1H), 6.47 (d, J¼2.5 Hz, 1H), 6.35 (d, J¼2.5 Hz, 1H), 6.33 (dd, J¼3.7, 2.7 Hz, 1H), 6.20 (dd, J¼3.7, 1.7 Hz, 1H), 5.25 (d, J¼16.3 Hz, 1H), 5.13 (d, J¼16.3 Hz, 1H), 3.90 (s, 3H), 3.83 (s, 3H), 3.54 (d, J¼16.3 Hz,1H), 2.97 (d, J¼16.3 Hz,1H),1.80 (s, 3H). 13 C{ 1 H} NMR (75.5 MHz, CDCl 3 ): d 159.7, 159.2, 134.6, 133.6, 118.8,117.9, 117.2, 109.3,103.2, 102.2, 98.8, 55.4, 55.3, 47.2, 37.8, 31.5, 28.9. MS (ESI) m/z (rel intensity): 283 (MH þ , 100), 242 (33). HRMS (ESI-TOF): calcd for C 17 H 19 N 2 O 2 [MH þ ] 283.1447; found, 283.1440. 2-(7-Fluoro-10-methyl-5,10-dihydropyrrolo[1,2-b]isoquinolin-10-yl)acetonitrile (3f). According to General Procedure, N-(o-iodobenzyl)pyrrole 1f (102 mg, 0.30 mmol) was treated with Pd(OAc) 2 (6.7 mg, 0.03 mmol), potassium hexacyanoferrate(II) trihydrate (27.9 mg, 0.07 mmol), sodium carbonate (41.3 mg, 0.39 mmol) and tetrabutylammonium chloride (83.4 mg, 0.30 mmol) in a 8/2 mixture of DMF/H 2 O (1 mL) for 48 h. After workup, purification by column chromatography (silica gel, petroleum ether/EtOAc 9/1) afforded 3f (11.8 mg, 16% yield) as a yellow oil: IR (ATR): 2925, 2250, 1500 cm 1 . 1 H NMR (300 MHz, CDCl 3 ): d 7.57 (dd, J¼8.8, 5.3 Hz, 1H), 7.11 (td, J¼8.8, 2.7 Hz, 1H), 7.01 (dd, J¼8.8, 2.7 Hz, 1H), 6.78e6.77 (m, 1H), 6.29e6.21 (m, 2H), 5.25 (d, J¼16.3 Hz, 1H), 5.14 (d, J¼16.3 Hz, 1H), 2.80e2.68 (m, 2H), 1.90 (s, 3H). 13 C{ 1 H} NMR (75.5 MHz, CDCl 3 ): d 161.6 (d, J¼247.5 Hz), 134.1 (d, J¼7.7 Hz), 133.7 (d, J¼3.3 Hz), 132.2, 127.2 (d, J¼8.3 Hz), 119.4, 117.4,115.1 (d, J¼21.4 Hz), 113.3 (d, J¼22.4 Hz),109.1,103.9, 47.4 (d, J¼2.2 Hz), 37.6, 34.0, 26.2. MS (ESI) m/z (rel intensity): 241 (MH þ , 59), 200 (100). HRMS (ESI-TOF): calcd for C 15 H 14 FN 2 [MH þ ] 241.1141; found, 241.1145. 2-(7-Methoxy-10-methyl-5,10-dihydropyrrolo[1,2-b]isoquinolin-10-yl)acetonitrile (3g). According to General Procedure, N-(o-iodobenzyl)pyrrole 1g (106 mg, 0.30 mmol) was treated with Pd(OAc) 2 (6.7 mg, 0.03 mmol), potassium hexacyanoferrate(II) trihydrate (27.9 mg, 0.07 mmol), sodium carbonate (41.3 mg, 0.39 mmol) and tetrabutylammonium chloride (83.4 mg, I. Barbolla, L. Hern andez-Su arez, V. Quevedo-Tumailli et al. European Journal of Medicinal Chemistry 220 (2021) 113458 9