Pyrrolo[2,3-e]indazole as a novel chemotype for both influenza A virus and pneumococcal neuraminidase inhibitors

Influenza infections are often exacerbated by secondary bacterial infections, primarily caused by Streptococcus pneumoniae. Both respiratory pathogens have neuraminidases that support infection. Therefore, we hypothesized that dual inhibitors of viral and bacterial neuraminidases might be an advantageous strategy for treating seasonal and pandemic influenza pneumonia complicated by bacterial infections. By screening our in-house chemical library, we discovered a new chemotype that may be of interest for a further campaign to find small molecules against influenza. Our exploration of the pyrrolo[2,3-e]indazole space led to the identification of two hit compounds, 6h and 12. These molecules were well-tolerated by MDCK cells and inhibited the replication of H3N2 and H1N1 influenza A virus strains. Moreover, both compounds suppress viral and pneumococcal neuraminidases indicating their dual activity. Given its antiviral activity, pyrrolo[2,3-e]indazole has been identified as a promising scaffold for the development of novel neuraminidase inhibitors that are active against influenza A virus and S. pneumoniae.

Inuenza A and B viruses cause seasonal epidemics of acute respiratory diseases with an annual death mortality of 290 000-650 000 people. 1 Moreover, there have been at least ve inuenza A virus pandemics in the past 150 years, such as the Spanish u (1918)(1919)(1920), Asian u (1957)(1958), Hong Kong u (1968)(1969), Russian u (1977Russian u ( -1979 and swine u (2009-2010), claiming millions of lives. [2][3][4] These viral infections are oen exacerbated by co/secondary bacterial infections, most commonly caused by Streptococcus pneumoniae, Staphylococcus aureus, Streptococcus pyogenes, or Haemophilus inuenzae. 5,6 Indeed, such a secondary infection may cause high mortality from severe pneumonia. Historical records and several systematic reviews indicate that a high number of deaths during the inuenza pandemics were the result of secondary bacterial pneumonia rather than primary inuenza infection. [7][8][9][10][11][12][13][14][15][16] Co/ secondary bacterial infection was identied in 25-95% of cases during the 1918, 1968, and 2009 pandemics, with S. pneumoniae being the most common pathogen. [14][15][16] Murine studies have shown that inuenza A virus promotes S. pneumoniae transmission and infection, 17 demonstrating a lethal synergism between the virus and the bacterium. 18,19 Hence, therapeutic approaches targeting both pathogens would enable the improvement of global healthcare systems against future respiratory epidemics and pandemics.
Neuraminidase (NA) is an important target in modern anti-inuenza drug discovery campaigns. [20][21][22] Neuraminidase is a key enzyme in an inuenza virus lifecycle: it facilitates viral release and spread. Its function is to catalyze the hydrolysis of sialic acid residues from host-cell glycoprotein receptors. 23,24 Inuenza virus neuraminidase supports S. pneumoniae infection, released sialic acids can be catabolized by these bacteria. 25,26 Moreover, the resulting de-sialylated glycoproteins can be used by S. pneumoniae to bind to lung epithelial cells. 5,27-29 S. pneumoniae also express neuraminidases -NanA, NanB (both found in most strains), and NanC (rarely identi-ed). 30 NanA and NanB play important roles in host colonization. 31,32 The active sites of inuenza virus and pneumococcal neuraminidases show structural similarity, suggesting that S. pneumoniae could be targeted by neuraminidase inhibitors. 33,34 Indeed, oseltamivir is active against both viral and bacterial neuraminidases in vitro and improves survival in a mouse model of inuenza A virus-S. pneumoniae synergism. 35,36 Another NA inhibitor, peramivir, also showed efficacy in inuenza virus-infected mice co-infected with S. pneumoniae, demonstrating prolonged survival rate and reduced bacterial burden and virus titre. 37 However, no studies aimed to determine the effect of these drugs on the treatment or prevention of secondary bacterial complications following inuenza infection in humans. Although resistance to these drugs occurs infrequently, resistant inuenza viruses can emerge and become a major problem as seen in case of oseltamivir in the 2008 inuenza season. [38][39][40] Therefore, the search for novel compounds with activity against both inuenza virus and pneumococcal neuraminidases may be an attractive way for the development of dual-acting anti-infective agents.
In the ongoing anti-inuenza virus screening for novel dualactive neuraminidase inhibitors, we found that compounds based on a previously unexplored chemotype, pyrrolo [2,3-e] indazole, prevent virus replication by inhibiting the activity of viral neuraminidase. We then obtained a series of pyrrolo [2,3-e] indazole-core compounds with various substituents at positions R 1 -R 4 to study structure-activity relationship. Pyrrolo [2,3-e] indazole is a heterocyclic system rarely mentioned in the scientic literature. We synthesized these compounds using the aza-Nenitzescu reaction as previously described by Lyubchanskaya and colleagues. 41,42 According to Scheme 1, commercially available 1,4-benzoquinone 1 is treated with the corresponding benzaldehyde phenylhydrazones 2a-f to afford hydroquinone adducts 3a-f. Subsequent oxidation of 3a-f with an aqueous solution of potassium ferrocyanide and potassium carbonate leads to indazolequinones 4a-f. 42 These intermediates are then reacted with the corresponding commercially available aminocrotonic esters 5a-d in the presence of acetic acid and acetic anhydride to cyclize into the nal pyrrolo[2,3-e]indazoles 6a-k. 41 2-Aminoalkyl-pyrrolo[2,3-e]indazoles 9a-c and 10a,b were synthesized according to Scheme 2 from pyrroloindazoles 6b,c. The protection of the 5-hydroxy groups of 6b,c with an acetyl group followed by bromination with bromosuccinimide (NBS) in the presence of benzoyl peroxide provides the corresponding 2-bromomethyl derivatives 8a,b. Further nucleophilic substitution of the bromine atom with piperidine, diethylamine, or morpholine results in the 2-substituted derivatives 9a-c. Using 4-nitrophenyl-pyrroloindazole 6e as a starting point, we synthesized a number of derivatives according to Scheme 3. O-Acylation with acetic anhydride of compound 6e yields 5acetyloxy-4-nitrophenyl-pyrroloindazole 11e. Mono-(5-C) and disubstituted (1,5-C) derivatives 11a-d were synthesized via alkylation of 6e with the corresponding alkyl iodides in the presence of potassium carbonate in NMP medium. Further, we reduced the nitro group of compound 6e by catalytic hydrogenation with Pd/C to obtain 4-aminophenyl-pyrroloindazole 13. As for 6e, we performed an O-acylation with acetic anhydride for compound 12 to form 5-acetyloxy-4-aminophenylpyrroloindazole 13. The pyrrole derivative 14 was prepared by condensation between primary amine of 12 and 2,5-dimethoxytetrahydrofuran in the presence of acetic acid as a catalyst.
To get a detailed insight into the structural information, Xray diffraction analysis for compounds 6a and 12 was performed. The crystallographic data were deposited with the Cambridge Crystallographic Data Centre (CCDC ID 2246679 for compound 6a and 2249815 for compound 12). † X-ray crystal structure of 6a established its identity as a 6,8-diphenyl-pyrrolo [2,3-e]indazole ( Fig. 1A and Table S1 †). All bond lengths and angles in compound 12 correspond well to those observed in compound 6a and related compounds from the Cambridge Structural Database (CSD) (Fig. 1B and Table S1 †). The triclinic syngony and proximity of unit cell dimensions in both compounds may indicate their crystal packing similarity. Surprisingly, both compounds form similar centrosymmetric hydrogen-bonded dimers despite the different numbers of active sites in their structures. Indeed, the hydroxy group in compound 12 participates in the formation of these dimers similar to the same group in 6a with close geometries (Fig. 1C and D). Specically, the O1/O3(1 − x, 2 − y, 1 − z) distance in 6a is 2.705(4) Å, while in 12 the O1/O3(1 − x, −y, 1 − z) distance is 2.711(9) Å. Interestingly, the amino group of compound 12 does not involve in any hydrogen bonding formation. Instead, it participates in weak intermolecular N/ H/p interactions that bind centrosymmetric dimers into ribbon-like stretches (Fig. S2 †).
As a part of an anti-inuenza screening, we evaluated synthesized compounds against inuenza A virus A/HK/1/68 (subtype H3N2) and inuenza A virus A/Jena/8178/09 (subtype A(H1N1)pdm09). Cytotoxicity and antiviral activity of the target compounds on MDCK cells were shown in Table 1. All pyrrolo [2,3-e]indazole-core compounds, regardless of structural fragments, showed no cytotoxicity at the maximum tested concentration, 100 mM, for MDCK cells supporting their further study for antiviral activity. The results of the structure-activity relationship study indicate a viral subtype-specic activity of the scaffold. For example, a number of compounds containing different substituents at positions R 2 -R 4 exclusively inhibited the cytopathic effect induced by inuenza virus A/Jena/8178/09 of subtype A (H1N1)pdm09 (compounds 6b-d, 7a,b, 9c, 10a  11a-e, 14). In contrast, pyrroloindazoles with a small group at the R 2 position (compound 6f) or without substituents at the R 2 -R 4 positions (compounds 6j,k) exhibited activity only against inuenza A virus A/HK/1/68 of subtype H3N2. At the same time, inhibition of both inuenza A virus subtypes was shown for compounds without any groups at positions R 2 -R 4 (exception for compound 13, R 4 = Ac) and with a R 1 -phenyl ring containing para-nitro, amino or triuoro groups (compounds 6e, 6h, 12). The control compounds oseltamivir and zanamivir acted as expected.
To investigate whether viral hemagglutinin or neuraminidase represent a target for pyrrolo [2,3-e]indazole-core compounds, we performed a human erythrocyte-based assay with inuenza A virus A/Jena/8178/09. Viral hemagglutinin causes hemagglutination of these erythrocytes at 4°C, and viral neuraminidase abrogates hemagglutination of human erythrocytes aer incubation of the assays at 37°C. Inhibitors targeting the viral hemagglutinin or neuraminidase might block  hemagglutination and/or its abrogation, respectively. With the exception of compound 13 (hemagglutination at 31.6 mM), our compounds inhibited hemagglutination weakly or not at all (results not shown). At the same time, all small molecules were found to inhibit the viral neuraminidase in the assay (Table 2). Among them, compounds 6h, 11a, 11e, 12, and 14 were the most active in the series with MICs in the range of 14.20-24.40 mM. The exception is compound 9a, which does not inhibit either viral neuraminidase activity (Table 2) or viral replication in the CPE reduction assay (Table 1). Structural similarities between inuenza and pneumococcal neuraminidases in their active sites assume that both neuraminidases can be targeted simultaneously by one neuraminidase small-molecule inhibitor. 33,34 For example, oseltamivir, but not zanamivir, exerts such dual activity (Table 2). 35,43 Inspired by the idea of developing small-molecule dual-acting neuraminidase inhibitors, we evaluated whether our pyrroloindazoles would also act on pneumococcal neuraminidase. To do this, we performed a hemagglutination-based NA inhibition assays with the neuraminidase NanA of S. pneumoniae strain DSM20566. With the exception of compound 13 all of the compounds tested were found to inhibit the enzyme at moderate concentrations ( Table 2). Among them, pyrroloindazoles 6h, 10b, 11e, 12 and 14 have the highest NanA-inhibitory activity with a MIC of 10.0 mM. Taking into account the above results, we represent compounds 6h and 12 as interesting molecules with dual neuraminidase inhibitory activity. Preliminary results of the structure-activity relationship for pyrrolo [2,3-e]indazole-core compounds are shown in Fig. 2A.
Classic neuraminidase inhibitors, zanamivir and oseltamivir, share the same structural feature, that is the carboxylate group trapping three guanidinium groups of the active site arginine residues. [44][45][46] This structural pattern is similar to the transition state, which is formed in the NA active site during the cleavage of sialic acid from glycoconjugates. 47 In contrast, our compounds do not have such carboxylate substituent in their structures. Moreover, other anionic groups are not presented either. Therefore, the search for a binding site seems challenging, especially as these compounds are active to both viral and bacterial NAs and, therefore, should form stable complexes with two proteins.
We initially performed molecular docking of pyrrolo[2,3-e] indazole 12 to the wide area around the active sites of both viral and bacterial neuraminidases. The set of complexes obtained was then studied using classical MD simulations. Most of the complexes dissociated aer a short simulation time (less than 10 ns). However, we obtained another binding site that was stable for both NAs for longer than 100 ns (Fig. S1 †). Despite the low amino acid sequence similarity between these proteins, we found that compound 12 binds similarly to both of them (Fig. 2). Compound 12 is anchored by two hydrogen bonds with a negatively charged carboxylate, Asp417 in NanA and Glu228 in viral NA, and a positively charged side chain of an arginine residue, Arg366 in NanA and Arg156 in viral NA. Importantly, this arginine residue is not among those that are responsible for substrate carboxylate binding. Compound 12 and its analogues have hydrophobic fragments complementary to the hydrophobic fragments of NA binding sites (Fig. 2).
Another important bacterial NA is NanB, therefore we checked whether this binding site is conserved for NanB. We reconstructed the binding mode similar to that in NanA and viral NA, and performed MD simulation. We found that the complex remains stable, and the structural patterns are the same as for NAs discussed above (Fig. S1 †). Thus, we suppose that compound 12 may inhibit NanB as well. However, additional in vitro enzyme studies are needed to prove this.
Some studies reported that pneumococcal neuraminidase NanA may impact biolm formation. 48,49 So, biolm formation of S. pneumoniae probably can be affected by neuraminidase inhibitors. We have recently observed that some neuraminidase inhibitors suppress bacterial planktonic growth and biolm formation. [50][51][52] In contrast, the inuenza drugs oseltamivir and zanamivir do not inhibit either planktonic growth or biolm formation (Table 3). 50,51 Similarly, we observed no inhibition of planktonic growth and/or biolm production of S. pneumoniae by most of the pyrrolo[2,3-e]indazoles studied (Table 3). Compounds 6c, 6h, and 10a active against bacterial NanA (Table  2) moderately affected S. pneumoniae planktonic growth as well as biolm formation (Table 3). However, there is not enough evidence to conclude whether the inhibition of pneumococcal growth and biolm formation by these molecules were the result of NanA inhibition.
As a result of our antiviral studies, we have discovered pyrrolo [2,3-e]indazoles as a novel class of small molecule inhibitors targeting inuenza A virus neuraminidase. According to the results of structure-activity studies, the R 1 -phenyl ring containing para-nitro, amino or triuoro groups of pyrrolo [2,3-e] indazoles is important for the inhibition of both circulating inuenza A virus subtypes H3N2 and H1N1. Therapeutic advantage of pyrrolo[2,3-e]indazoles might represent their dual activity against the viral neuraminidase as well as the structurally related neuraminidase NanA of S. pneumoniae. Molecular dynamic simulations demonstrate that the ethyl ester moiety, hydroxy groups, and the pyrrolo[2,3-e]indazole core system are responsible for the formation of stable complexes with active sites of both viral and bacterial neuraminidases. Some pyrrolo [2,3-e]indazoles also inhibited bacterial growth. Taken together these results led us to conclude that pyrrolo[2,3e]indazole represents novel hit compounds warranting further development.
All authors discussed the manuscript and approved the submitted version of the paper.

Conflicts of interest
The authors declare that they have no conict of interest.