Introduction

Nowadays, a large number of hospital-acquired infections are caused by antibiotic-resistant opportunistic pathogens. Furthermore, several clinically relevant organisms are rapidly evolving towards multidrug and even pan-drug-resistant phenotypes, whose global spread represents a significant Public Health issue. Whereas tackling the problem of antimicrobial resistance (AMR) relies on the implementation of several approaches (e. g. antibiotic stewardship), the discovery or development of new antibiotics, especially those active on carbapenem-resistant Enterobacteriaceae, Pseudomonas spp., and Acinetobacter spp., remains a critical priority [1].

The current COVID-19 pandemic increased awareness of the fragility of health systems worldwide [2, 3] and further highlighted the urgent need for novel pharmacologic treatments to tackle drug-resistant pathogens. Indeed, although the actual impact of the SARS-CoV-2 pandemic on AMR is still debated [4], the incidence of bacterial coinfections was recently reported to be as high as 30% [5]. Furthermore, as antibiotic prophylaxis is also part of the COVID-19 management [6], the related overuse of such valuable therapeutics [4] could have worsened this already dramatic scenario.

In the last years, we have been involved in the development of AlkylGuanidino Ureas (AGUs) endowed with remarkable antibacterial properties [7,8,9]. Among them, compound 1a (Fig. 1) showed a broad-spectrum bactericidal activity, including against antibiotic-resistant Gram-negative species (Minimal Inhibitory Concentration (MIC) values as low as 1 µg/mL); in-depth biological characterization and in vitro preliminary Absorption, Distribution, Metabolism, and Excretion (ADME) profile were also recently reported [9, 10].

Fig. 1
figure 1

Structure of AGU 1 as TF-Acetate (a) and HCl (b) salts or free base (c) and the Boc-intermediate 2

Full size image

However, compound 1a and the whole AGUs library were obtained and tested as TF-Acetate salts since the final reaction step involved the treatment of the intermediates like 2 (Fig. 1) with TFA. Briefly, all the reported synthetic routes were based on the tert-butoxycarbonyl (Boc)-protecting group strategy to soften the high polarity of free guanidino functions and to enhance their synthetic accessibility and handling. Thus, after the Boc-cleavage in TFA solution, the final compounds are endowed with the guanidino moieties positively charged and TF-Acetate counterions.

Although TFA is widely employed in protecting groups-based strategies and peptide synthesis, especially in solid-supported techniques [11] in relation to its strong acidity (pKa = 0.52) and low boiling point, the presence of residual amounts (up to 10–7 M) of TFA or TF-Acetate counteranion in the final samples subjected to biological assays represents a thoroughly debated issue for the scientific community [12]. In fact, it was proved to affect the accuracy and reproducibility of cellular assays by inhibiting cell proliferation [12] or favoring the viability of several cell types unspecifically [12,13,14], limiting further in vivo experiments [15] and potential medical uses. This is especially true for synthetic antimicrobial peptides (AMPs) or compounds bearing positively charged residues or moieties and strong bases. The TF-Acetate content should thus be appropriately considered, even if studies on the potential effects of counterions on cell-based assays are sparse [16] and a straightforward correlation between counterion type and in vitro biological activity still has to be established [17]. In general, no significant differences emerged when testing a compound as a free base or as salts from different organic or inorganic acids [18, 19]. However, especially in the case of antimicrobial compounds, a modulation of their activity was noticed [20,21,22], resulting in MIC values differing by at 2 log2 dilutions (fourfold). This could be in part explained by the different molecular weights of the test compounds, since MIC values are commonly reported in µg/mL (or mg/L), rather than being a direct influence on antimicrobial action [20], but it is unclear whether the presence of counterions could further impact on the antibacterial activity. Furthermore, contaminant TFA or TF-Acetate counterion is highly detrimental to biological material also in preclinical and in vivo Pharmacokinetics (PK) studies [17]. In fact, in vivo TFA is reported to trifluoroacetylating proteins, causing hepatitis [23], and breaking down the intermolecular structure of water, while TF-Acetate anions can interfere with or disrupt membrane function, enzymatic catalysis, secondary structures of proteins, and protein stability [24] and lead to immune response-inductions [25,26,27]. Also, compounds as TF-Acetate salts were reported to exert a low pharmacological efficacy compared to the same compounds with other counterions [28]. Hence, each case should be considered individually and more than one counterion should be investigated to achieve the optimal biological profile [17].

Hence, starting from a TF-Acetate AGU (1a), we synthesized the corresponding compound with a different counterion as a hydrochloride salt (1b) or as a free base (1c) (Fig. 1), in order to compare their antibacterial activity and to gain insights on the influence of counterions in such preparations.

Results and discussion

Although several protocols avoiding the use of TFA were reported [29,30,31,32], such synthetic approaches were barely used for compounds with several guanidino functions on an aliphatic skeleton. As an example, preparation of the antimicrobial polyhexamethylene guanidine (PHMG) polymer involved the use of non-protected guanylating agents, yielding the guanidinium hydrochloride without additional reactions involving strong acids [33]. It would not be unlikely that the presence of several guanidines and aliphatic chains could hamper the reaction completion or promote the degradation of the compound.

Surprisingly, the use of HCl on the Boc-intermediate 2 (Scheme 1) was found unsuccessful to yield the desired product, although HCl-based protocols are generally widely used in Boc cleavage due to their faster kinetics compared to that using TFA. It is common knowledge that, when the milder TFA is employed, higher reaction concentrations and amounts of acid are required to achieve reasonable reaction times [34]. However, the synthesis of 1b was very challenging and passed across TF-Acetate derivative 1a (ii, Scheme 1), as shown in Scheme 1.

Scheme 1
scheme 1

Synthetic routes explored to obtain 1b and 1c. Successful (iii, vi) and failed (i, iv, v) reaction steps. In green and blue are presented the approaches for 1b and 1c, respectively. Reagents and reaction conditions: ii. freshly dist. TFA (20% v/v), dry CH2Cl2, sealed flask, r.t., 5 h [10]; iii. Amberlite IRA400 chloride form, CH3OH, r.t., 72 h; vi. NaOEt in situ dry EtOH, N2, 0 °C-r.t., 30 min. Unsuccessful reactions are reported in Tables 1, S1, and S2

Full size image
Table 1 Attempts to synthesize compound 1b
Full size table

Different synthetic strategies were explored and some attempts are reported in Table 1.

The first strategies to obtain 1b were based on the Boc-cleavage of intermediate 2 through the direct employment of hydrochloric acid in 1,4-dioxane saturated solution [35] (Entry 1, Table 1) or in situ-generated from acetyl chloride in methanol [36] (Entry 2). However, chromatographic purification in alumina or reverse-phase silica failed to provide an acceptable purity of the sample for biological evaluation (> 95%), as determined by HPLC–MS analysis. Isolation of such compound, using semi-preparative HPLC, could not be investigated further due to the low intensity of UV absorbance and an unsatisfactory resolution of compounds 1a and 1b when neither acids nor bases were added to the mobile phase in analytical HPLC.

Then, a different strategy starting from the synthesis of the TF-Acetate salt 1a [10, 37] (ii., Table 1) and the counterion exchange with the chloride anion was implemented. To confirm the completion of the anion exchange, we resorted to the detection of the presence/disappearance of the TF-Acetate in the samples by 19F NMR since its fluorine atoms show a characteristic singlet signal at 77 ppm in CD3OD [17, 38]. In fact, in our case, the traditionally employed IR spectroscopy technique is not effective in monitoring the anion exchange since the absorption bands of urea and guanidines belonging to AGUs overlap to that of TF-Acetate, complicating the interpretation of the resulting spectra [15].

We also excluded RP-HPLC technique since it shows limited efficacy in anion exchange [15, 17] and not possible in our case. In fact, the optimized analytical protocol set up for the characterization of AGU compounds relied on the use of ammonium acetate as a mobile phase additive, while the use of acids resulted in very broad and low-resolved signals in the UV-chromatogram. Thus, several in-batch synthetic approaches were tried (Entries 3–7, Table 1).

Initially, 1a was treated with HCl 5 N in methanol [39] (Entry 3) but no conversion was observed. Then we tried a protonation/reprotonation approach consisting of the in situ generation of the free base guanidino intermediate 1c, followed by the addition of concentrated HCl (37% w/w). Thus, we stirred 1a with different organic bases, such as triethylamine (TEA) (Entry 4) and 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU) (Entry 5). On one hand, the low basicity of TEA (pKb = 3.2) was not sufficient to deprotonate guanidinium moieties, as confirmed by 19F NMR spectra, while treatment with DBU, which possesses a lower pKb (1.1), caused the compound degradation. In Entry 6, an attempt with a caustic alkali in an aqueous medium [40] is reported but, also in this case, the isolation of 1b failed.

Hence, in the end, we resorted to a resin-based procedure (iii., Table 1), which resulted to be successful in yielding the pure chloride salt 1b. Briefly, we converted TF-Acetate 1a into 1b through a long-period stirring in presence of Amberlite IRA 400 chloride form resin [41,42,43]. The completion of conversion was checked by 19F NMR spectroscopy. 1H and 19F NMR spectra of 1a and 1b are reported in Fig. 2 and in the Supplementary Information.

Fig. 2
figure 2

Comparison of 1H and 19F NMR spectra of AGUs 1a–c. A 1H NMR spectra of 1a (purple) and 1b (green) were recorded at Bruker 400 MHz. 1H NMR spectrum of 1c (blue) was recorded at Bruker 600 MHz. B 19F NMR spectra of 1a (purple) and 1b (green) were recorded at Bruker 600 MHz. All the experiments were performed in CD3OD. Spectra were visualized and analyzed through MestreNova 14.2 [44]

Full size image

The expected small changes in chemical shifts were observed (A, Fig. 2), confirming the absence of fluorine atom after anion exchange (B).

The stability issue of the free base compounds remains unsolved, due to their high hygroscopicity and tendency to capture CO2 after air exposure to form carbonate salts in aqueous solution. Although basic molecules are generally available on the market in the form of acid salts [45], several AMPs and guanidino compounds were prepared and tested as free base derivatives. Indeed, bacterial culture media are characterized by pH values above 6.0 [19] and free base guanidines will fast pick up protons from the environment to increase their chemical stability.

On that basis, we focused our chemical efforts toward the preparation of 1c. However, as expected, great synthetic difficulties were encountered and several approaches were tried by exploring both in batch, microwave (MW), and salt-exchange procedures, starting from salts 1a and 1b or Boc-intermediate 2, as shown in Scheme 1. Attempts involving 2 were unsuccessful, probably due to the high complexity rate of polyguanidines cleavage and are described in Table S1 in the Supplementary Information. To the best of our knowledge, the acid and base-free cleavage of highly guanylated substrates is sparsely reported in literature and during these experiments, degradation phenomena or by-products formation often occurred. Further experiments were performed aimed at obtaining 1c from the TF-Acetate 1a and the reaction conditions are reported in Table S2 in the Supplementary Information. Unfortunately, no product was obtained since the TF-Acetate counterion strongly interacts with the positively charged guanidines, resulting in a challenging displacement [15] that requires first a replacement with a stronger acid-derived counteranion, such as the chloride ion from hydrochloric acid (pKa = − 5.9) [46].

Hence, we treated the hydrochloride derivative 1b with sodium ethoxide in absolute ethanol [47], based on the poor solubility of the resulting sodium chloride in the reaction solvent [48] and the possibility to be easily filtered off. To assess the total absence of hydrochloride counterion, a precipitation assay-based analysis with silver nitrate was performed. The pictures of an ethanolic solution of 1b and 1c after filtration are reported in Figure S1 in Supplementary Information.

Finally, derivatives 1b and 1c were tested against a panel of representative bacterial species and data are shown in Table 2.

Table 2 Antibacterial activity of compounds 1a–c against representative Gram-positive and Gram-negative bacterial species, including multi-drug resistant clinical isolates
Full size table

Table 2 reports the antibacterial activity of the same AGU compound (1), in its salt (a-b) and free base (c) forms, in both μg/mL and μM, the latter unit limiting the influence of variations of molecular weights. The Clinical & Laboratory Standards Institute (CLSI) and the European Committee on Antimicrobial Susceptibility Testing (EUCAST) do not provide guidelines on the unit system for reporting MICs. Indeed, both weight/volume (µg/mL) and molarity (mol/L) unit systems can be used. In general, in clinical practice, a relevant parameter to understand the potency of an antibiotic is the ratio between MIC value and blood concentration which is conventionally expressed as weight/volume. Also, microbiologists usually resort to the latter system although molar concentrations can also be used when the tested compounds are pure and fully characterized antimicrobials.

When the MIC values were converted from μg/mL to μM, no significant differences were observed between the hydrochloride AGU 1b and the free base 1c, whose molecular weight was approximately 24 and 35% lower than that of the corresponding TF-Acetate 1a, resulting in a no negligible difference. However, despite compounds 1a and 1b showing similar antibacterial activities, the free base AGU (1c) showed a low potency on Gram-negative species, both considering data in weight/volume or molar systems, although a good profile is still retained on Gram-positive strains. This indicates that the counterions may affect the intrinsic antibacterial activity of AGU derivatives. The most significant drop in activity was observed on Gram-negative pathogens, particularly towards P. aeruginosa, pointing out that the presence of the counterion may somehow contribute to enhancing the antibacterial activity of these compounds (e.g. by altering the physicochemical conditions of the medium or modifying the permeability of biological membranes).

Compounds 1b and 1c were also tested on resistant clinical isolates and the panel was also extended for the reference compound 1a. HCl salt form 1b exhibited an enhanced antibacterial profile on C. meningosepticum, A. hydrophila, B. cepacia, and S. maltophilia. However, a detrimental decrease of the antibacterial properties was observed for 1c.

Among the different bacterial resistance mechanisms, the changes in the lipid composition and distribution in the membranes are very common and, along with the counterion effect, may have led to activity fluctuations depending on the strain and the counterion type or presence.

Minimal bactericidal concentrations (MBC) values were also measured on a panel of multi-drug resistant bacterial strains, as shown in Table 3. Data outlined a good bactericidal profile, as the obtained MBC values are almost identical to those of MIC, as already observed for different AGUs derivatives [10]. Moreover, the counterion does not seem to affect the bactericidal behavior of compounds 1a-c proving that the salt form does not have an impact on the effect of the compounds on the bacterial cultures.

Table 3 MBCs of 1a–c against representative drug-resistant bacterial species
Full size table

Conclusion

We prepared the hydrochloride 1b and the free base 1c forms of a previously characterized AGU compound 1a, showing potent and broad-spectrum antibacterial activity, including on antibiotic-resistant clinical isolates. Compounds 1b and 1c were obtained using anion-exchange and in batch protocols, respectively. Several synthetic strategies, which were reported in literature to be successful for monoguanidino compounds, were attempted. However, these approaches were found unable to produce the desired compounds, being less efficient when applied to the preparation of polyguanidino compounds.

Quite strikingly, the antibacterial activity, determined using a panel of relevant bacterial species, of TF-Acetate (1a) and HCl (1b) salts were comparable, besides an improvement of activity highlighted by 1b especially towards resistant clinical isolates. However, the free base 1c retained the activity on a few Gram-positive bacterial species, exhibiting a somewhat decreased potency on Gram-negative strains and clinical isolates with resistant phenotypes. While it remains unclear whether this could be more relevant when testing polycationic substances, these data highlight the need to carefully consider the potential impact of counterions when testing antimicrobial compounds, as their presence may modulate the intrinsic activity of such compounds.

Material and methods

Chemistry

All commercially available chemicals and solvents were used as purchased. Anhydrous reactions were performed into flame-dried glassware after three cycles of vacuum/dry nitrogen and were run under a positive pressure of dry nitrogen. Anhydrous solvents were prepared prior to use: CH2Cl2 was dried over calcium hydride and alcohols over I2/Mg. TLCs were visualized under UV light and stained with ninhydrin or basic permanganate stains. 1H NMR spectra were recorded on Bruker Avance 400 or 600 spectrometers at 400 or 600 MHz, while 13C NMR spectra on Bruker Avance 400 at 100 MHz. 1H NMR spectra were reported in parts per million (δ scale) and internally referenced to CD3OD signal at δ 3.31 ppm. Chemical shifts for carbon are reported in parts per million (δ scale) and referenced to the carbon resonances of CD3OD at δ 49.00 ppm. Data are shown as following: chemical shift, multiplicity (s = singlet, d = doublet, t = triplet, q = quartet, qi = quintet, m = multiplet and/or multiplet resonances, br = broad signal), integration and coupling constants (J) in Hertz (Hz). 19F NMR spectra were recorded on a Bruker 600 spectrometer at 564 MHz. Mass spectra (LCMS) were acquired using an Agilent 1100 LC-MSD VL system (G1946C) by direct injection with a 0.4 mL/min flow rate using a binary solvent system of 95/5 CH3OH/H2O. UV detection was monitored at 221 or 254 nm. Mass spectra were acquired in positive or negative mode scanning over the mass range 100–1500 m/z, using a variable fragmentor voltage of 0–70 V.

The purity of final products was assessed by HPLC–UV-MS analysis as reported [10].

1,3‐Bis(8‐carbamimidamidooctyl)‐1,3‐bis({8‐[N'‐(cyclopropylmethyl)carbamimidamido]octyl})urea Hydrochloride Salt (1b) To a solution of 1a (45.7 mg, 0.035 mmol) in CH3OH (5.8 mL), Amberlite IRA-400 Chloride form (137.0 mg) was added and the reaction mixture was stirred through a rocker shaker at room temperature for 24 h. Then, the mixture was diluted with CH3OH and filtered through a 0.45 µm PTFE filter. The filtrate was evaporated under vacuum to furnish the desired product as a pale yellow oil. Yield: 90%.1H NMR (CD3OD, 400 MHz) δ (ppm): 3.24–3.16 (m, 16H), 3.10 (d, J = 7.0 Hz, 4H), 1.64–1.55 (m, 8H), 1.53–1.50 (m, 8H), 1.43–1.29 (m, 32H), 1.09 (p, J = 5.6 Hz, 2H), 0.62 (d, J = 7.6 Hz, 4H), 0.31 (d, J = 5.0 Hz, 4H). 13C NMR (CD3OD, 100 MHz) δ (ppm): 3.4, 9.5, 26.2, 26.5, 27.5, 28.4, 28.4, 28.8, 28.9, 29.2, 41.0, 41.1, 45.8, 46.9, 47.1, 47.3, 47.5, 47.7, 47.9, 48.1, 155.6, 165.4. LCMS(ES +) m/z = 845.8 [M + H]+, 423.1 [M + 2H]2+, 282.5 [M + 3H]3+, 212.1 [M + 4H]4+.

1,3‐Bis(8‐carbamimidamidooctyl)‐1,3‐bis({8‐[N'‐(cyclopropylmethyl)carbamimidamido]octyl})urea Free Base (1c). To Boc-protected 2 (28.0 mg, 0.017 mmol), a solution of freshly distilled acetyl chloride(800 µL) in dry CH3OH (7.2 mL) was added dropwise at 0 °C in an ice bath, under N2. The reaction mixture was stirred at rt for 30 min. The completion of the reaction was assessed by LCMS analysis. Then, toluene and hexane were added and the mixture was evaporated. The crude was washed with hexane and CH2Cl2 several times. Then, it was treated with a freshly prepared ethanolic solution of sodium ethoxide: sodium ethoxide was prepared in situ by adding sodium (3.2 mg, 0.14 mmol) to dry absolute ethanol (4.25 mL) at 0 °C. The reaction mixture was stirred at room temperature for 1 h till precipitation of a white powder (NaCl). Then, the mixture was diluted with dry ethanol (5 × 10.0 mL) and filtered several times through 0.45 µm PTFE filters. Aliquots of the sample were analyzed with the silver nitrate assay to ensure the absence of chlorine atoms. After evaporation, the yellowish oil was treated with CH2Cl2 (5 × 10.0 mL) and filtered again with a 0.45 µm PTFE filter. After evaporation, a colorless oil was obtained. An aliquot was subjected to the silver nitrate assay. In brief, a 0.1 M solution of silver nitrate in deion. H2O was freshly prepared and used to assess the complete conversion of 1b to 1c and the removal of residual NaCl. In presence of chloride ions, silver chloride is formed as a white precipitate. No evidence of precipitation was observed. Yield: 80%. 1H NMR (CD3OD, 600 MHz) δ (ppm): 3.24–3.15 (m, 16H), 3.09 (d, J = 7.0 Hz, 4H), 1.63–1.59 (m, 8H), 1.56–1.52 (m, 8H), 1.39–1.37 (m, 24H), 1.31 (m, 8H), 1.11– 1.08 (m, 1H), 0.62–0.60 (m, 4H), 0.31 (d, J = 6.0 Hz, 4H). 13C NMR and LCMS(ES +) spectra are consistent with those of 1b.

In vitro antibacterial activity testing

Bacterial strains, including representatives of both Gram-positive and Gram-negative bacterial species, were obtained from the ATCC or CCUG culture collections (except clinical isolates, which were already present in our collection). Compounds were resuspended in dimethylsulfoxide (DMSO) at a final concentration of 50 or 100 mg/mL and subsequently diluted in the culture medium. MIC and MBC values of the compounds were determined using the micro-dilution broth method using Mueller–Hinton broth as recommended by CLSI. Bacterial inoculum was 5 × 104 CFU/well. MICs and MBCs were recorded after 18 h of incubation at 35 °C.