Introduction

Most fungal infections are common, localized to skin or mucosal surfaces and can be treated effectively with topical agents. While invasive fungal infections (IFIs) are more uncommon, they are very difficult to control medically, and are associated with high mortality rates, especially in immunosuppressed patients [1]. The two most common fungal species associated with IFIs are Aspergillus fumigatus and Aspergillus flavus. Importantly, advances in aggressive clinical practices in cancer therapy, and solid organ and hematological transplants have created growing patient populations subjected to extended timeframes of immunosuppression which in turn increases their susceptibilities to life-threatening invasive infections [2,3,4,5]. The most common IFI is invasive candidiasis, and is the fourth leading healthcare associated blood stream infection in the US, reaching close to 50% mortality rates [6, 7]. Invasive aspergillosis is also a cause of high mortality in immunocompromised patients with only one-year survival rates of 59% and 25% among solid organ [8] and stem cell transplant recipients [4], respectively. Invasive pulmonary aspergillosis leads to acute respiratory distress syndrome and high mortality in severe influenza patients [9] and a similar clinical progression has been reported recently in intensive care COVID-19 patients [10, 11].

There are only three classes of antifungal agents widely used to treat IFIs; polyenes (e.g. amphotericin B), azoles (e.g. fluconazole, itraconazole, isavuconazole, posaconazole, and voriconazole) and echinocandins (e.g. caspofungin, micafungin, and anidulafungin). All classes target key components of the fungal cell membrane or cell wall, namely ergosterol (polyenes and azoles) or β-(1,3,)-D-glucan (echinocandins), respectively [12]. The limited spectrum of antifungal agents coupled with important shortcomings in their use, including modest success rates in reducing high mortality rates, emerging resistance, especially to the azole and echinocandin classes, detrimental drug interactions, and toxicity liabilities for amphotericin B and several azoles, make the current antifungal armamentarium very vulnerable. Therefore, there is an urgent need for the development of new antifungal agents that are mechanistically distinct from current antifungals to effectively treat these dangerous infections and reduce the potential for appearance of cross resistance.

We have developed a series of non-peptide analogs of the host defense proteins (HDP mimics, smHDPs) as antimicrobial agents that have distinct advantages over peptides for pharmaceutical use [13, 14]. Our approach is to recapitulate the biological and structural properties of HDPs into oligomeric backbones without trying to duplicate the dimensional structures of the peptides, but rather by creating small structurally constrained cationic compounds. The smaller size and improved metabolic stability of smHDPs optimizes their pharmacokinetic and tissue distribution properties over peptidic HDPs and, with the selection of appropriate building blocks, smHDPs can be produced economically. Furthermore, synthetic chemistry provides considerably greater chemical space beyond amino acids to fine-tune structures for enhanced antimicrobial and safety properties.

A number of structurally distinct smHDPs have shown potent activity against Candida spp in vitro and in vivo. Minimal inhibitory concentrations (MICs) against Candida albicans and non-albicans Candida species [15, 16] ranged between 0.5 and 8 µg/mL and selectivity indices were >100-fold over mammalian cells. Several of the compounds were also rapidly fungicidal versus C. albicans at 2 – 4x over the MIC showing > 2log10 reductions in colony forming units (cfus) within 2–5 h of exposure [16]. Selected compounds were highly efficacious in mouse models of oral candidiasis after a single dose applied topically to the infected tongue [17] and in a disseminated candidiasis model measuring kidney burdens 24 h after a single subcutaneous dose [18]. These animal results indicate that suitable safety and absorption, distribution, metabolism, and elimination (ADME) properties can be incorporated into smHDPs for in vivo administration and activity. Many of the smHDPs described above had limited activity against Aspergillus spp. and other molds, so we have embarked on an optimization program to improve the breadth of activities against fungal pathogens.

Recently, we described bis-guanidine smHDPs with a range of substitutions on heteroaryl central rings between bis-aryl rings and terminal guanidines [19]. Our investigations revealed that a 1,2,3,6-tetrahydropyridine was a superior central linker between the aryl rings and the guanidines. In particular, compound 1, Fig. 1, was found to be non-cytotoxic, as well as highly potent and broadly active against Candida, Aspergillus and Fusarium spp. [19]. These compounds were significantly more potent against multiple fungal pathogens than other guanidine-containing antifungal agents including bis-guanidine pyrrole derivatives [17], bis-guanidine peptidomimetics of lactoferrin [20], macrocyclic analogs of guazatine [21, 22], the arylguanidine abafungin [23] and mono-aminoguanidine analogs described by Miyazaki et al. [24]. More generally, a wide range of guanidine-containing agents have been described as having antifungal [25] and antimicrobial [26] effects, indicative of the potential of guanidine-containing derivatives to serve as potential drugs.

Fig. 1
figure 1

The lead compound from the heteroaryl-linked series

Full size image

Results and discussion

In seeking alternate scaffolds for our antifungal agents, we were cognizant that there was a high degree of aromaticity in our heteroaryl-linked analogs such as 1, Fig. 1 [19]. With this in mind, we initially chose to replace the central aromatic ring with an amide bond, and embarked upon the synthesis of novel compounds. The first compound made was the bis-piperazine derived compound 2, Table 1. This compound was found to show good potency against four of the six mold strains tested against, and suggested that this would be a productive series to pursue further. We have separately described the use of the 1,2,3,6-tetrahydropyridine as a linker of value in the production of antifungal agents [19], and incorporated this moiety in the amide series, generating derivative 3, which showed excellent activity against all six of the mold strains, but was found to have higher than acceptable cytotoxicity in NIH3T3 cells (41 μM) (we tested our compounds for cytotoxicity in both NIH3T3 and HepG2 cells). Compared to amphotericin B, 3 was more potent against all six of the strains tested, and showed superior or equivalent potency as posaconazole against the same six strains tested. Hybridization of the 1,2,3,6-tetrahydropyridine and the piperazine resulted in analog 4, which was reasonably potent but not as active as 3. Wishing to vary the distance between the two guanidine moieties, combinations of the 1,2,3,6-tetrahydropyridine on the left-hand side with alternate alkyl linkers on the right-hand side were assembled, and two of these, 5 and 6, were found to possess good potency. Overall, it was clear to us that the optimal template for antifungal activity was with the 1,2,3,6-tetrahydropyridine as the linker on each side, and we focused our efforts in this area.

Table 1 N-phenyl benzamide containing analogs
Full size table

Focusing on the 1,2,3,6-tetrahydropyridine linked analogs, we next investigated substitutions on the aryl rings. We determined that a 3-fluoro-substitution on the left-hand aryl ring 7, was well tolerated, Table 2, and chose to utilize this scaffold for investigations in to the substitution preferences for the right-hand aryl ring. Mono-methylation gave 8 and 9, for which we felt that overall the 3-position substitution, 9, was preferable. Focusing on this position, we incorporated F (10), Cl (11), and OMe (12), with the methoxy compound (12) being particularly interesting to us due to its improved aqueous solubility (1210 μg/mL).

Table 2 SAR around the N-phenyl-3-fluorobenzamide series
Full size table

Based on the excellent potencies achieved with a number of the phenyl analogs, we next chose to investigate heteroaryl ring substitution in this system. On the left-hand side, 2,5-disubstituted furan 13 showed good activity, Table 3, while the 2,4-disubstituted furan derivative 14 showed exceptional potency against all six strains. The equivalent thiophene compounds were also synthesized, 15 and 16, and in particular the 2,5-disubstituted analog, 15, like 14, displayed excellent potency. On the right-hand side, two pyridyl derivatives were synthesized, with 2-pyridyl 17 showing high potency against all six strains, Table 4, while 3-pyridyl 18 was slightly less active, but still highly potent.

Table 3 Left-hand side aryl derivatives
Full size table
Table 4 Right-hand side pyridyl analogs
Full size table

We were also interested in the possibility of replacing one or both of the guanidine moieties with alternate groups so as to increase the functional group diversity of our derivatives, and with this in mind a select number of analogs were made as derivatives of 7, Table 5. The guanidine replacements were selected to include hydrogen-bond donors and hydrogen-bond acceptors. Unfortunately, none of the derivatives 19, 20, or 21 were found to possess any antifungal activity, indicating that two guanidine functional groups were necessary for antifungal activity.

Table 5 Right-hand side guanidine replacements
Full size table

Having found that amide-linked analogs have shown good antifungal activity, vide supra, we sought to expand the scope of our derivatives further, so as to discover a new series with different properties, and embarked upon the synthesis of novel bis-amide containing compounds. We initially synthesized symmetrical alkyl linked guanidines, with either a methylene or ethylene linker, 22 and 23, Table 6, as the bis-amide containing analogs were perceived to provide a greater distance between the two guanidine moieties, which we believed to be an important parameter. Of these, compound 22 displayed superior antifungal activity, though far lower than we desired. We have separately described the use of the 1,2,3,6-tetrahydropyridine as a linker of value in the production of antifungal agents, [19], and incorporated this moiety in to the bis-amide linked series, generating derivative 24, which showed improved activity against four of the mold strains. An alternate connector between the guanidine and aryl ring that was investigated was the piperazine, which gave 25, and this compound was less active than 24. Hybridization of the methylene linker from 22 and the 1,2,3,6-tetrahydropyridine linker from 24 resulted in the unsymmetrical 26, which was highly potent against all six strains. Compound 26 was found to possess higher cytotoxicity than we would prefer (152–170 μM), has high mouse plasma protein binding (>99%) and modest aqueous solubility at pH 7.4 (171 μg/mL). With this information in hand, we moved forward seeking to improve the in vitro properties and cytotoxicity, while maintaining the high potency of 26.

Table 6 Initial bis-amide containing analogs
Full size table

One method we have sought to utilize for the improvement in aqueous solubility is via the incorporation of an additional basic, saltable, nitrogen into the series. With this in mind, we synthesized two pyridyl derivatives, as variations on the central aromatic ring, Table 7. Both compound 27 and 28 were highly potent, and it was found that while 27 had higher cytotoxicity (CC50: 3T3 > 117 μM, HepG2 > 117 μM), and apparently lower solubility, compound 28 displayed appreciably lower cytotoxicity (CC50: 3T3 > 1171 μM, HepG2 739 μM). Unfortunately, it was shown that 28 had low aqueous solubility at pH 7.4, of only 4.5 μg/mL, and as a result of this information the series was deprioritized.

Table 7 Central pyridyl bis-amide analogs
Full size table

Heteroaryl left-hand ring analogs of 26 were also synthesized, Table 8. Pyridyl derivative 29, pyridazine 30, and pyrazine 31 were all highly potent. The cytotoxicity of all three analogs were superior to 26 [29 (CC50: 3T3 544 μM, HepG2 353 μM), 30 (CC50: 3T3 231 μM, HepG2 326 μM), and 31 (CC50: 3T3 > 585 μM, HepG2 585 μM)], however the aqueous solubilities of 29, 30, and 31, were all noticeably lower than for 26, although 30 and 31 were two and four times respectively more soluble than 29.

Table 8 Central pyridyl bis-amide analogs
Full size table

Seeking to improve the aqueous solubility of compound 26, we investigated methoxy substitution on the left-hand aryl ring, Table 9. Methoxy was chosen as a group capable of disrupting planarity in the system, and hence improving the solubility [27], and also improving the aqueous solubility of the analogs by virtue of its polarity. Methoxy substitution adjacent to the 1,2,3,6-tetrahydropyridine provided 32, which was reasonably potent, while substitution next to the amine, 33, gave a derivative that was less potent. Due to the relatively lowered antifungal activity compared to 26, neither 32 nor 33 was considered worthwhile to pursue further.

Table 9 Central pyridyl bis-amide analogs
Full size table

An alternate way that we sought to improve the physical properties of the derivatives was via incorporation of saturation of the central ring, so as to make the compounds less aromatic, more three-dimensional, and more drug-like [28]. We have previously found in the symmetrical bis-amide linked series, that the central phenyl ring could be replaced by a trans-substituted cyclohexane while maintaining potency- data not shown. For the new analogs we chose to use the bicyclco[2,2,2]octane and bicyclo[1,1,1]pentane rings- rings that contain no chiral center, and are also known to be phenyl isosteres [29]. Bicyclo[2,2,2]octane derivative 34 displayed excellent potency against five of the six strains, and bicyclo[1,1,1]pentane analog 35 was also highly potent, Table 10. We were delighted to find that both compounds possessed far lower mouse plasma protein binding than 26 (89 and 83% respectively), and 35 also has five times higher aqueous solubility than 26 (770 μg/mL).

Table 10 Saturated central ring analogs
Full size table

Comparing the different amide-linked analogs, we were most interested in compound 14, and chose to investigate this compound further. Susceptibility to 14 was tested against a clinical isolate of Candida albicans, and the MIC (50% growth inhibition) was 0.78 µg/mL, indicative of broad antifungal activity against both mold and yeast pathogens. Cytotoxicity assays measuring 50% inhibition of metabolic activity (CC50) following 24-h exposures in mouse NIH3T3 fibroblasts and human HepG2 liver cells showed 14 had low cytotoxic CC50 values of 68 and 129 μM, respectively, for selectivity indices (CC50/MIC) that ranged over 150-fold. With this data in hand, we chose to move 14 forward into in vitro ADME and off-target studies (Table 11). Compound 14 was stable in mouse liver microsomes (>60 min half-life), had moderately high protein binding in mouse plasma (98.3%), showed little inhibition of three cytochrome P-450 (CYP) enzymes at 10 µM (0 to 23%), and had no meaningful hERG activity at 10 µM in an automated patch clamp assay (15.2%). The aqueous solubility of 14 at pH 7.4 was found to be 58 μg/mL. Unfortunately, when 14 was tested in a three-day mouse maximum tolerated dose (MTD) study, the MTD was determined to be only 5 mg/Kg.

Table 11 In vitro ADME and off-target activities for 14
Full size table

From the different bis-amide linked analogs previously described, we felt that compound 34 was overall the most interesting, and elected to carry out additional studies on this derivative. Susceptibility to 34 was tested against a clinical isolate of Candida albicans, and the MIC (50% growth inhibition) was 0.2 µg/mL, showing that this compound has broad antifungal activity against both mold and yeast pathogens. Cytotoxicity assays measuring 50% inhibition of metabolic activity (CC50) following 24-h exposures in mouse NIH3T3 fibroblasts and human HepG2 liver cells showed 34 had low cytotoxic CC50 values of 439 and 253 μM, respectively, for selectivity indices (CC50/MIC) that ranged over 80-fold. With this data in hand, we chose to move 34 forward into in vitro ADME and off-target studies (Table 12). Compound 34 was stable in mouse liver microsomes (>60 min half-life), had moderately high protein binding in mouse plasma (89.4%), showed little inhibition of three CYP enzymes at 10 µM (0 to 29%), and had low hERG activity at 10 µM in an automated patch clamp assay (5.3%). The aqueous solubility of 34 at pH 7.4 was found to be 46 μg/mL. When 34 was tested in a three day mouse maximum tolerated dose study, we were delighted to find that the MTD for this compound was 40 mg/Kg.

Table 12 In vitro ADME and off-target activities for 34
Full size table

A representative scheme for the synthesis of the amide-linked analogs is exemplified by the synthesis of 14, Scheme 1, [30]. Starting from commercially available bromo-furan 36, Suzuki coupling with 37, followed by saponification of the ester gave the sodium salt 39. Separately, Suzuki coupling of bromoaniline 40 with 37 provided 41 in high yield. Amide coupling of fragment 39 with fragment 41, mediated by HATU, produced amide 42 in good yield. Bis-Boc removal on 42 using trifluoroacetic acid (TFA) gave a quantitative yield of bis-amine 43. Guanylation of 43 using the commercially available guanylating agent 44 provided 45 in good yield, and final step deprotection using TFA gave a quantitative yield of 14.

Scheme 1
scheme 1

(i) Pd(dppf)Cl2 • DCM, 2 M Na2CO3, 1,4-dioxane, 80 °C, 7 h, 61%; (ii) 1 N NaOH, MeOH - THF, 40 °C,16 h, 98%; (iii) Pd(dppf)Cl2 • DCM, 2 M Na2CO3, 1,4-dioxane, 80 °C, 16 h, 90%; (iv) HATU, EtNiPr2, DMF, 20 oC, 16 h, 58%; (v) TFA, DCM, 20 oC, 3 h, Quantitative; (vi) EtNiPr2, DMF, 20 oC, 16 h, 72%; (vii) TFA, DCM, 20 oC, 3 h, Quantitative Synthesis of 14

Full size image

For the bis-amide linked analogs, the synthesis of 34 is exemplary, and is shown in Scheme 2 [30]. Starting from commercially available bicyclooctane carboxylic acid 46, an amide coupling was carried out with 4-bromo-aniline 47 to give the mono-amide 48. Saponification of ester 48 with sodium hydroxide gave acid 49, and HATU-mediated amide coupling with the requisite aryl amine provided 51. Suzuki coupling of 51 with commercially available boronate ester 37 gave 52, and bis-Boc-removal using TFA provided bis-amine 53 in high yield. Reaction of 53 with guanylating agent 44 gave the bis-guanidine 54, and final step deprotection using TFA provided 34.

Scheme 2
scheme 2

(i) HATU, NEt3, DMF, 20 °C, 16 h, 85%; (ii) NaOH, 1,4-dioxane, water, 80 °C, 20 h, 90%; (iii) HATU, NEt3, DMF, 20 °C, 16 h, 53%; (iv) Pd(OAc)2, S-Phos, K2CO3, 1,4-dioxane, water, 95 °C, 16 h, 34%; (v) TFA, CH2Cl2, 20 °C, 16 h, 93%; (vi) NEt3, DMF, 35 °C, 16 h, 46%; (vii) TFA, CH2Cl2, 20 °C, 16 h, 30% Synthesis of 34

Full size image

Conclusions

We have shown that amide- and bis-amide-linked bis-guanidine-containing smHDPs are highly potent, and display broad-spectrum antifungal activities which are comparable or superior to current clinically utilized antifungal agents. For the amide-linked series, a range of substitutions have been shown to be tolerated on the amide-linked rings, and heteroaryl rings are also in general well tolerated in the system. Investigations into the linker between the guanidine and aryl rings showed that the 1,2,3,6-tetrahydropyridine was the optimal linker on both sides of the molecules. In the bis-amide-linked series, it has been shown that asymmetry, with the incorporation of one methylene-linked guanidine and one 1,2,3,6-tetrahydropyridine-linked guanidine was important for high antifungal activity. In this series, replacement of the central aromatic ring with saturated rings was well tolerated. In particular, compound 34 was found to be highly potent, broadly active, non-cytotoxic, and to have good in vitro ADME and off-target properties. As a result of the overall profile of the compound, 34 (FC12406) was selected as a pre-clinical development candidate.