Polytrimethylenimines: Highly Potent Antibacterial Agents with Activity and Toxicity Modulated by the Polymer Molecular Weight

Cationic polymers have been extensively investigated as a potential replacement for traditional antibiotics. Here, we examined the effect of molecular weight (MW) on the antimicrobial, cytotoxic, and hemolytic activity of linear polytrimethylenimine (L-PTMI). The results indicate that the biological activity of the polymer sharply increases as MW increases. Thanks to a different position of the antibacterial activity and toxicity thresholds, tuning the MW of PTMI allows one to achieve a therapeutic window between antimicrobial activity and toxicity concentrations. L-PTMI presents significantly higher antimicrobial activity against model microorganisms than linear polyethylenimine (L-PEI) when polymers with a similar number of repeating units are compared. For the derivatives of L-PTMI and L-PEI, obtained through N-monomethylation and partial N,N-dimethylation of linear polyamines, the antimicrobial activity and toxicity were both reduced; however, resulting selectivity indices were higher. Selected materials were tested against clinical isolates of pathogens from the ESKAPE group and Mycobacteria, revealing good antibacterial properties of L-PTMI against antibiotic-resistant strains of Gram-positive and Gram-negative bacteria but limited antibacterial properties against Mycobacteria.


■ INTRODUCTION
Despite an intense development of antibiotics and other methods of treatment, our densely populated and aging society is exposed to growing threats from microbial diseases. Currently, antibiotic-resistant bacterial infections are responsible for approximately 700,000 deaths annually, and according to predictions, this value will reach 10 million by 2050. 1 The increasing number of antibiotic-resistant pathogenic strains drives studies toward new antimicrobial agents with a novel mechanism of activity.
One promising area of research is the use of naturally occurring antimicrobial peptides (AMPs) 2 and their synthetic mimics (SMAMPs). 3−6 Their mode of action is based on electrostatic attraction to the negatively charged bacterial cell surface and subsequent disruption of its integrity. Such mechanism, based on multiple nonspecific interactions, is potentially less susceptible to development of resistance in comparison to antibiotics in current clinical use. 7,8 Among many investigated SMAMPs, 3,9,10 serious research efforts have been devoted to polycations with amine groups allocated within the main polymeric chain, such as polyethylenimine (PEI) and its derivatives. 11,12 PEI is a close analogue of naturally occurring low-molecular-weight polyamines, like spermine, spermidine, norspermidine, and putrescine. Such short amines are essential compounds in cellular processes of prokaryotic and eukaryotic organisms, acting as, e.g., growth factors and intracellular pH regulators. Importantly, in physiological pH, they are fully protonated, which allows them to bind nucleic acids and other anionic cellular structures via electrostatic interactions. Spermine and spermidine, for example, condense DNA in mammal semen 13 but also stabilize RNA of viruses. These biogenic polyamines display poor antimicrobial activity, 14 but they are considered as potent substrates for the synthesis of derivatives with anticancer or antibacterial properties due to the probability of higher biocompatibility compared to fully artificial compounds. 13 Studies also show that they are able to improve bacterial susceptibility for β-lactam antibiotics. 15 Derivatives of PEI have been extensively studied as antimicrobial agents, 16,17 as nonviral transfection vectors, 18,19 and for other biomedical applications. 19,20 The results, however, have shown that the polymer exhibits relatively high cytotoxicity, which limits its applications to nonmedical uses such as antibacterial coatings or paints. 20,21 On the other hand, polytrimethylenimine (PTMI), # an analogue of PEI with a three-methylene spacer between amine groups, has received substantially lower attention. Its dendrimeric form has gained popularity as a nonviral transfection vector and in drug delivery systems, 22,23 although there are very limited reports on a linear form of PTMI. 24−28 Different N-methylated PTMI derivatives were synthesized; 24,29,30 however, they have not been the subject of antimicrobial studies. Both PEI and PTMI may be obtained in the branched, 20 linear, 27,31 and dendrimeric forms. 23 Linear PEI (L-PEI) and linear PTMI (L-PTMI) are usually obtained by hydrolysis of poly(alkyl-2-oxazoline)s 32,33 and poly(alkyl-2-oxazine)s, 28,34 respectively. Whereas precursors of L-PEI, poly(alkyl-2-oxazoline)s, have already been widely studied for various biomedical applications, poly(alkyl-2-oxazine)s, the precursors of L-PTMI, have been just gathering interest in terms of drug delivery systems 35, 36 and polymer brushes. 37,38 Polymers have been investigated as antimicrobial agents in various strategies as stand-alone active molecules or conjugates with known active ingredients. 39−41 The antimicrobial potency and cytotoxicity of polycations depend on many structural parameters, e.g., type of repeating unit (r.u.), type of positively charged group, hydrophilic−lipophilic balance, molecular architecture, and average molecular weight (MW). 42 The impact of MW on the antimicrobial activity of polymers has been studied for several types of polycations. Reported results indicate that such correlation is highly dependent on the type of polymer and that an increase of MW may lead to an increase, 43−46 a decrease, 47−50 or no significant change 51−55 of antimicrobial activity. Molecular weight also affects the hemolytic and cytotoxic activity of polymers, often leading to the lowered selectivity of materials with a higher polymerization degree. 53,56,57 The antimicrobial activity of L-PEI has been the subject of several studies; 11,58 however, the antimicrobial activity of L-PTMI has never been tested. In this work, a series of L-PTMI with different MWs, and L-PEI as a benchmark, were synthesized and characterized. The obtained L-PEI and L-PTMI were additionally subjected to exhaustive N-monomethylation and subsequent partial N,N-dimethylation. Such library of novel L-PTMI and L-PEI derivatives allowed us to study the influence of MW, type of cationic group, and positive charge density on their antimicrobial activity, including antibiotic resistance species and Mycobacterium, and toxicity toward mammalian cells. Presented results reveal the key role of MW in fine-tuning the selectivity of L-PTMI and unprecedentedly high antimicrobial activity of this polycation. ■ EXPERIMENTAL SECTION Materials. Reagents were purchased from Sigma-Aldrich, Acros Organics, Alfa Aesar, or Fluorochem. The AR grade solvents were purchased from Chempur or POCH. Deuterated solvents were purchased from Euroisotop. 2-Methyl-2-oxazoline 98% (MeOx, Sigma-Aldrich) and methyl tosylate 98% (MeOTs, Sigma-Aldrich) were additionally purified by distillation over CaH 2 and stored under an argon atmosphere. Acetonitrile used for polymerization was dried over molecular sieves (13×) and stored under an argon atmosphere. All other reagents were used without further purification. Water was purified using the Milli-Q system (Millipore).
The freeze-drying process was performed using the Labconco FreeZone 2.5 L Benchtop Freeze Dry System. Broths for antimicrobial assays were prepared using the commercially available Mueller−Hinton broth powder (Biocorp), Sabouraud broth (SAB) Instruments. Microwave-assisted reactions were performed in a single-mode microwave reactor (CEM Discover LabMate) equipped with an infrared (IR) temperature sensor. GC analysis was carried out using an Agilent 7820A GC System equipped with an FID detector. A 30 m × 0.32 mm i.d. × 0.25 μm film thickness HP-5 (Agilent 19091J-413) column was used. Helium (99.999%) with a constant flow rate of 6 mL·min −1 was used as carrier gas. The injector and detector temperature was set at 250°C (injection volume 1 μL). The split flow rate was 360 mL·min −1 , and the split ratio was 60:1. The column oven was kept at 40°C for 1.0 min, and then the temperature was increased up to 80°C at a rate of 35°C·min −1 and subsequently to 250°C at a rate of 75°C·min −1 and held at this temperature for 3 min. After this time, the column oven temperature was increased to 300°C at a rate of 75°C·min −1 and held for 1 min. The total running time was 9.08 min. 1 H NMR spectra were recorded using a Varian 400 MHz spectrometer and D 2 O, CD 3 OD, CD 3 CN, or CDCl 3 as solvents. 1 H NMR chemical shifts were referenced to the residual signal of the protonated solvent (δ 7.26 for CDCl 3 , δ 3.31 for CD 3 OD, δ 1.94 CD 3 CN, and δ 4.79 for D 2 O). SEC analysis was performed using an Agilent 1260 Infinity liquid chromatograph equipped with an RID detector, the PSS NOVEMA Max 5 μm analytical 300 × 8 mm column with a precolumn (PSS GmbH), and a mobile phase containing 54/23/23 (v/v/v%) water/ methanol/acetic acid and 0.5 M sodium acetate. All chemicals were HPLC grade. To calibrate the method, a series of poly(2vinylpyridine) standards (PSS GmbH) in the range of molar masses 620 Da−540 kDa were used. Samples, dissolved in the eluent at 5 mg· mL −1 concentrations (injection volume 20 μL), were analyzed at 50°C with a flow rate of 0.4 mL·min −1 . Molar masses M w and M n and dispersity Đ M were calculated using the Agilent GPC Addon Rev. B.01.02 software.
Material Synthesis. 1 H NMR spectra of studied polymers and starting materials are presented in Supporting Information (SI) Figures S1−S12. Representative SEC traces are presented in SI Figure  S14.
2-n-Propyl-2-oxazine (PrOzi). The process was adapted from the synthetic procedure of 2-substituted cyclic imino ethers. 59,60 Butyronitrile (278 g, 4.02 mol) was heated to 100°C with a catalytic amount of zinc acetate (14.7 g, 80.1 mmol), and 3-aminopropan-1-ol (452 g, 6.02 mol) was added dropwise to the stirred mixture. The reaction solution was refluxed for 72 h at 135°C. Upon cooling to room temperature, the mixture was diluted using 450 mL of dichloromethane and subsequently extracted with 450 mL of water and 450 mL of brine. The product was purified twice by vacuum distillation and subsequent vacuum distillation over CaH 2 , yielding 232 g of a colorless liquid (46%). 1  General Procedure for the Preparation of Poly(2-n-propyl-2-oxazine) (PPrOzi). Poly(2-n-propyl-2-oxazine)s were synthesized by a typical procedure for CROP of 2-alkyl-2-oxazines. 60−62 A representative example is described below, and differences in parameters of individual reactions are summarized in Table 1.
Poly(2-n-propyl-2-oxazine)_3k (PPrOzi_3k). A total of 3.3 mL of PrOzi (3.17 g, 24.9 mmol, 1 equiv) and 2.5 mL of dried acetonitrile were added to a dry, magnetic-stirring-equipped microwave reaction vessel under the flow of inert gas. Subsequently, the mixture was stirred, and 1.0 mL of 0.25 M MeOTs (0.01 equiv) stock solution in dried acetonitrile was added. The reaction vessel was placed in the microwave reactor and heated to 120°C for t = 40 min. After microwave heating, the polymerization mixture was cooled to 50°C and quenched by the addition of methanol. After evaporation of the solvent on a rotatory evaporator, the residue was dissolved in EtOH and precipitated with Et 2 O. The obtained yellow oil was dried in a vacuum, yielding 0.96 g (48%) of the polymer. 1  General Procedure for the Preparation of Polytrimethylenimines (L-PTMIs). Linear polytrimethylenimines were synthesized by adapting procedures of the acidic hydrolysis of poly(2-alkyl-2oxazines). 63,64 A representative example is described below, and yields and differences in parameters of individual reactions are summarized in Table 2.
Linear Polytrimethylenimine_0.8k (L-PTMI_0.8k). PPrOzi_0.8k (1.20 g, 9.45 mmol of r.u.) was dissolved in hydrochloric acid (6 mL, 6 M) in a microwave reaction vessel and heated to 140°C for 2.5 h. The volatiles were removed in vacuo, and the residue was dissolved in distilled water, frozen, and freeze-dried for 72 h, yielding 1.03 g (87%). 1 Linear N-Methyltrimethylenimine (Me-L-PTMI_7k). L-PTMI Nmonomethylation was performed according to the Eschweiler−Clarke procedure adapted to postpolymerization modification. 24 L-PTMI_7k (1.00 g, 17.6 mmol of r.u.), water (5.6 mL), formic acid (2.8 mL), and 35% formaldehyde (2.8 mL) were added to a 25 mL round-bottom flask, and the mixture was refluxed overnight. Upon cooling to room temperature, NaOH (20% water solution) was added to reach pH = 12, and the product was extracted using chloroform. The organic phase was dried using anhydrous MgSO 4 , and the solvent was removed using the vacuum evaporation process. The obtained paleyellow residue did not require further purification (yield 1.06 g, 85%). 1 To a solution of Me-L-PTMI_7k (0.35 g, 4.86 mmol of r.u.) in methanol (14 mL) in a two-neck-round bottom flask under inert gas, methyl iodide solution in methanol (14 mL) was dosed, and the mixture was refluxed overnight. Then, the solvent was removed using a rotatory evaporator, and the pale-yellow solid was dried overnight in a vacuum. The degree of quaternization (DQ) was determined using 1 H NMR spectra. Amounts of MeI, yields, and DQ are presented in Table 3.
Poly(2-methyl-2-oxazoline)_4k (PMeOx_4k). The polymer was synthesized by a typical procedure for CROP of 2-alkyl-2-oxazolines. 65 Under an argon atmosphere, 6.3 mL of MeOx (6.33 g, 74.4 mmol) and 9.6 mL of dried acetonitrile were added to a dry microwave reaction vessel equipped with a magnetic stirring bar. The mixture was stirred, and 3 mL of 0.25 M MeOTs stock solution in dry acetonitrile was added. The reaction vessel was placed in a microwave reactor and heated to 120°C for 4 min. After microwave heating, the polymerization mixture was cooled to 50°C and quenched by the addition of methanol. After evaporation of the solvent and residual monomer on a rotatory evaporator, the polymer was dried under a vacuum and did not require further purification. Yellow oil was .00 g of PPrOzi_7k (31.5 mmol of r.u.) and 20 mL of 6 M HCl were used. b After removing volatiles in vacuo, the residue was dissolved in hot water and alkalinized with 8 M sodium hydroxide. After cooling, precipitated L-PTMI was centrifuged and washed with aqueous NH 3 (12.5% water solution) until the supernatant conductivity was in the order of magnitude of NH 3 solution conductivity. Then, the white solid was dispersed in distilled water, frozen, and freeze-dried for 72 h. Subsequently, the polymer was dissolved in 6 M HCl (1.1 equiv). The solution was concentrated and dried in vacuo, and the residue was dissolved in distilled water, frozen, and freeze-dried for 72 h. Linear Polyethylenimine (L-PEI_4k). The procedure was adapted from procedures of acidic hydrolysis of poly-2-alkyl-2-oxazolines. 63,64 PMeOx_4k (2.95 g, 34.7 mmol of r.u.) was dissolved in hydrochloric acid (25 mL, 6 M) in a microwave reaction vessel and heated to 140°C for 1 h. The volatiles were removed in vacuo, and the residue was dissolved in hot water (100 mL) and alkalized with 8 M sodium hydroxide. After cooling, precipitated L-PEI was centrifuged and washed with aqueous NH 3 (12.5% water solution) until the supernatant conductivity was in the order of magnitude of NH 3 solution conductivity. Then, the white solid was dispersed in distilled water, frozen, and freeze-dried for 72 h (1.17 g, 78%). 1 Linear Poly(N-methylethylenimine) (Me-L-PEI_4k). L-PEI Nmonomethylation was performed according to the Eschweiler−Clarke procedure adapted to postpolymerization modification. 24 L-PEI_4k (0.91 g, 21.1 mmol of r.u.), water (7.0 mL), formic acid (3.5 mL), and 35% formaldehyde (3.5 mL) were added to a 25 mL round-bottom flask, and the mixture was refluxed overnight. After cooling to room temperature, NaOH (20% water solution) was added to pH = 12, and the solution was extracted with chloroform. The organic phase was dried with anhydrous MgSO 4 , and the solvent was removed in a vacuum. The obtained pale-yellow oil did not require further purification (0.97 g, 82%). 1 To a solution of Me-L-PEI_4k (0.35 g, 6.14 mmol of r.u.) in methanol (14 mL) in a two-neck round-bottom flask under inert gas, methyl iodide solution in methanol (14 mL) was dosed, and the reaction was stirred at room temperature for 48 h. Subsequently, the solvent was removed using a rotatory evaporator, and the paleyellow oil was dried overnight in a vacuum. The degree of quaternization was determined based on 1 H NMR spectra. Amounts of MeI, yields, and DQ are presented in Table 4.

Minimum Inhibitory Concentration (MIC) Determination.
The broth microdilution method was applied following CLSI M07-A9 Vol. 32 No. 2 (for bacteria) and CLSI M27-A2 Vol. 22 No. 15 (for yeast) protocols. Single colonies of bacteria or yeast were used to inoculate 5 mL of the Mueller−Hinton broth (MHB) or Sabouraud broth, respectively, and the cultures were grown overnight at 37°C with shaking (240 rpm). The polymer stock solutions (5120 μg· mL −1 ) prepared in Milli-Q water were diluted with the appropriate broth (MHB or SAB) to a concentration of 1024 μg·mL −1 and used to prepare a series of different polymer concentrations in the broth (from 512 to 0.25 μg·mL −1 ; 100 μL each) in 96-well plates by the twofold dilution method. Tested concentrations of oligomeric polyamines were within the range from 4096 to 16 μg·mL −1 . Subsequently, 100 μL of the microbial suspension (2 × 10 5 CFU· mL −1 for bacteria and 2 × 10 3 CFU·mL −1 for yeast) was added to each well. Uninoculated broth, uninoculated broth with polymer solutions, and inoculated broth without any antimicrobial agent were used as controls. Four replicates were performed for each concentration of polymer and the control. The plates were incubated for 20 h at 37°C. The optical density at 600 nm (OD 600 ) was measured using the Synergy H4 Hybrid Microplate Reader (Biotech, Winooski, VT, USA). The recorded MIC value was the lowest concentration of the polymer at which no microbial growth was observed with the microplate reader.
MIC against Mycobacterium. The assay was performed by the twofold serial microdilution method (in 96-well microtiter plates) using a Middlebrook 7H9 Broth medium (Beckton Dickinson) containing 10% of OADC (Beckton Dickinson). The inoculum was prepared from fresh LJ culture in the Middlebrook 7H9 Broth medium with OADC, adjusted to a no. 0.5 McFarland tube, and diluted 1:20. The stock solution of a tested molecule was prepared in water and diluted in the Middlebrook 7H9 Broth medium with OADC by fourfold the final highest concentration to be tested. Compounds were diluted serially in sterile 96-well microtiter plates using 100 μL of the Middlebrook 7H9 Broth medium with OADC. Concentrations of tested agents ranged from 0.25 to 512 μg·mL −1 . A growth control containing no antibiotic and a sterile control without inoculation were also prepared on each plate. After incubation at 37°C for 21 days, the MICs were visually assessed as the lowest concentration showing complete growth inhibition of the reference microbial strains. Isoniazid (INH) and rifampicin (RMP) were used as reference drugs.
Hemolytic Activity Determination. Fresh blood was obtained from the Regional Center for Blood Donation and Blood Treatment in Warsaw. Samples were centrifuged (700g, 10 min), the supernatant plasma was rejected, and erythrocytes were washed with ice-cold phosphate-buffered saline (PBS) three times (by centrifuging, 700g, 10 min). After final centrifugation, erythrocytes were diluted 10 times with PBS. Polymer solutions were prepared in PBS buffer. Erythrocyte suspensions (500 μL) were added to polymer solutions (500 μL) at investigated concentrations. Samples were incubated for 1 h at 37°C and then centrifuged (10 min, 700g), and hemoglobin release was measured in supernatants using the spectrophotometric method (absorption at λ = 540 nm). PBS buffer and 0.2% Triton X-100 served as negative and as positive controls, respectively. Experiments were performed at the Faculty of Chemical and Process Engineering WUT with OSH approval for research with human blood.
Cytotoxicity Assay (XTT). The L929 mouse fibroblast cell line was cultured in 75 cm 2 cell culture flasks and kept in an incubator at 37°C with 5% CO 2 . The culture was monitored under the microscope every 2 days, dissociated, and divided when the cells were near 100% confluent. The cell dissociation protocol was based on a 0.25% trypsin−EDTA solution procedure. The cell concentration was counted on a Thoma cell counting chamber (Marienfeld). Polymer samples were dissolved in DMEM (no supplementation with phenol red), sterilized using 0.22 μm PES membrane, and supplemented with fetal bovine serum (FBS) (10% (v/v)) and antibiotics (1% (v/v)). Final polymer concentrations were reached by dilution with DMEM containing FBS and antibiotics. A sterile solution of 0.1% Triton X-100 in DMEM with FBS was prepared as a positive cytotoxicity control; DMEM with FBS and antibiotics was used as a negative control. The L929 cell line was maintained in 96well plates for 24 h in 10 5 cells·mL −1 concentration and 100 μL of culture medium per each well. Subsequently, DMEM was replaced by the polymer solutions, and after 24 h of cultivation with polymer solutions, cells were rinsed two times by adding 100 μL of PBS.
For the XTT viability assay, 100 μL of DMEM, without phenol red and FBS, and 50 μL of XTT solution with coupling reagent (The Cell Proliferation Kit II, Roche) were added to each culture well and incubated for 4 h. After the XTT was reduced to formazan pigment by living cells, the assay medium, 100 μL from each well, was transferred to a new 96-well plate, and the absorbance at 475 nm was measured in a plate spectrophotometer. The relative cell viability was defined as the ratio between the absorbance from the sample and the absorbance measured for negative control. Experimental data were fitted to the Hill equation, and IC 50 (half-maximal inhibitory concentration) values ■ RESULTS AND DISCUSSION Polymer Synthesis and Characterization. The main aim of this work was to investigate the antimicrobial activity and toxicity of L-PTMI characterized by different molecular weights and compare them with the biological activity of extensively studied L-PEI. Second, we studied how antimicrobial properties are influenced by postpolymerization modifications of these linear polyamines using tertiary amines and quaternary ammonium salt derivatives. Series of L-PTMIs, with different MWs, were synthesized using a cationic ring-opening polymerization protocol 27,60,66 of n-propyl-2-oxazine and subsequent acidic hydrolysis of obtained poly(n-propyl-2oxazine) (PPrOzi) (Figure 1a). L-PEI was synthesized in a similar process, starting from methyl-2-oxazoline, and used as a reference material. 64 As reported in the literature, polymerization rates of n-alkyl-2-oxazines are approximately 4 times lower than n-alkyl-2-oxazolines. 60,67 Slower propagation leads to a higher contribution of side reactions, such as termination. As a result, the polymerization of oxazines is more challenging to control and more sensitive to traces of impurities. To optimize PTMI polymerization parameters and to obtain PPrOzi with welldefined molecular weights, monomer conversion over the reaction time was monitored with the use of GC. Results indicate that polymerization follows first-order kinetics with respect to the monomer ( Figure S13), allowing one to determine the propagation rate (k p = (5.3 ± 0.1) × 10 3 L· mol −1 ·s −1 ). In case of shorter PPrOzi, polymerization reactions were performed with complete conversion of the monomer. For larger MWs of PPrOzi, polymerizations were performed in a monomer conversion regime no higher than 70−80% to avoid side reactions and to achieve lower dispersity. MeOx polymerization also follows first-order kinetics ( Figure S13, k p = (41.2 ± 0.8) × 10 3 L·mol −1 ·s −1 ). Molecular weights (M n and M w ) and dispersity (Đ M ) of obtained materials, determined by SEC analysis, are presented in Table 5. Degrees of polymerization (DPs) based on 1 H NMR and conversion of the monomer via GC are presented in Table S1.
L-PTMI_7k, containing 116 repeating units (r.u.) by average, and L-PEI_4k (91 r.u.) were used for further polymer modifications, namely, N-monomethylation and subsequent partial quaternization of amine groups (Figure 1a,b). The Nmonomethylation reaction was performed using the Eschweiler−Clarke reaction, 33,68 leading to fully N-methylated polymers (Me-L-PTMI_7k and Me-L-PEI_4k). The last stage of the postpolymerization modification was the partial transformation of N-methylamines into N,N-dimethylammonium groups with the use of MeI to reach the designed degree of quaternization (DQ). All polymers studied in this work had a linear structure. In the case of partially quaternized polymers, the letter "L" was omitted from the acronyms for simplicity. The arbitrarily chosen DQ levels for both polymers were motivated by the limited stability of PEI upon full quaternization. When more than 50% of r.u. of PEI is quaternized, the polymer undergoes degradation due to the proximity of adjacent charged groups within the polymeric chain. 69,70 Molecular Weight Influence on Biological Activity. Table 6 summarizes the basic biological activity of investigated molecules. Minimum inhibitory concentrations (MICs) were determined by the broth microdilution method against model   (Table 6). For L-PTMI, selectivity clearly changes with MW and displays a maximum for L-PTMI_1.2 k for all tested model organisms ( Figure S17). This trend provides the possibility of tuning antimicrobial and cytotoxic properties by adjusting MW. The impact of the molecular weight of polycations on their antimicrobial activity was reported previously in numerous papers; the effect is however strongly dependent on the polymeric structure. 44,45,48,52−55 The cytotoxicity of cationic polymers, similarly to its antibacterial activity, is a function of cationic charge density 56 and hence typically increases for longer polymers. For example, Tyagi and Mishra 57 reported that derivatives of polymethacrylamide, bearing a primary amine group in side chains, were highly cytotoxic to human cells for MWs higher than 5 kDa. In few reports, a U-shaped correlation between MW and biological activity is described for cationic amphiphilic polymers. 71−73 Short oligomers typically display limited antimicrobial and toxic activities, which increase along with MW. It can be explained by a higher overall cationic charge of the macromolecule and multivalent interactions of a single polymeric chain with the cell envelope. 72,74 For very high molecular weights, activity may be lowered, which is explained by the lower ability of large polymers to penetrate across the cell wall and membrane. 71,75,76 However, this may also be the effect of lower polymer solubility or aggregation, preventing interactions with the lipid bilayer. Ikeda et al. 43,77 reported that the activity of polymers bearing biguanide moieties in the main chain against S. aureus increases with increasing MWs up to 50 kDa and drops for weights above 120 kDa as a result of limited diffusion across the cell wall. Some reports indicate that the permeability of the cell wall of Gram-positive bacteria is not a concern for molecular weights up to 50−70 kDa. 51,73,78 In this context, our results for L-PTMI remain in good agreement with published data. a Literature values are given in parentheses. b Hemolytic yield for the polymer concentration of 2 mg·mL −1 is given in parentheses. c Uncertainties of IC 50 values, expressed as a standard deviation, are summarized in SI ( Figure S16 and Table S3). d Selectivity index = IC 50 /MIC. e Hemolysis assay was carried out for the polymer solution in 130 mM NaCl instead of PBS due to solubility issues.

Biomacromolecules pubs.acs.org/Biomac Article
The L-PTMI results were compared against naturally occurring, low-molecular-weight polyamines, namely, spermine, spermidine, putrescine, norspermidine, and 1,3-diaminopropane (Figure 1c), as similar structural analogues. 11,58,79 All of them display lower antimicrobial activity than L-PTMI, which is particularly well visible for direct structural L-PTMI analogues (norspermidine and 1,3-diaminopropane) ( Table 7, Figure 3a, Figure S15) against all tested microorganisms (ca. 4−8 times lower against E. coli and S. aureus compared to the least active synthesized L-PTMI and 16−64 times lower against C. albicans). This well supports the hypothesis that to obtain a bacteriostatic material, a certain minimum number of repeating units are required. 44,80,81 Additionally, it is worth to highlight that for L-PTMI, MW/ activity dependence remains in two regimes (Figure 3a, Figure  S15). MIC values decrease linearly in a log−log scale up to a certain polymer mass, and above this mass, MIC values remain constant at the low level. The linear negative correlation may indicate the presence of a single dominating biochemical mechanism responsible for the MW/activity effect in this area.
Even though L-PEI has been relatively widely explored in comparison to L-PTMI, no systematic studies on its antimicrobial activity as a function of molecular weight have been summarized. In several reports, MIC values against model bacterial strains have been determined; however, most of the publications discuss no more than two different polymeric masses (Figure 3b, Figure S15). 11,58,79,82−84 Richter et al. 85 tested the activity of four L-PEIs that differed in the number of r.u. and observed a trend of monotonically decreasing MIC values against E. coli and S. aureus with an increasing number of r.u.. Falco et al. 86 investigated the transfection efficacy of L-PEI; therefore, they fractionated a highly dispersed polymer and examined its toxicity dependence on molecular weight. The results showed that up to 4 kDa L-PEI seems to be of low toxicity, whereas above 20 kDa, the toxicity is significantly higher.
In our studies, L-PEI_4k (91 r.u.) displays 8-and 16-times lower inhibiting activity against E. coli and S. aureus, respectively, in comparison to L-PTMI_7k (116 r.u.) ( Table  6). Toxicity toward mouse fibroblasts as well as the hemolytic activity of L-PEI_4k was also lower; e.g., IC 50 is 2 orders of magnitude higher than the value for L-PTMI_7k. There is no simple explanation for such effect, but at least two factors may be considered. The distance between cationic groups, interacting with phospholipids, was hypothesized to influence the inhibitory effectiveness of the molecule as reported Gilbert and Moore 87 for biguanide polymers. Therefore, one of the possible explanations for the increased activity of L-PTMI is that the presence of an additional methylene group in the alkyl spacer leads to the optimization of the steric arrangement of amine groups in contact with the lipid layer. 11,49,53 The second possibility could be that the longer alkyl linker between the amine groups enhances the hydrophobic character of the polymer, providing amphiphilic properties to the material. This factor is reported for many SMAMPs to increase both the microbiological activity 88,89 and red blood cells lysis. 53,55,90 The influence of the polymer molecular mass on the activity of L-PTMI against C. albicans appears to be significantly different than in the case of model bacteria. All of the studied L-PTMIs showed similar moderate activity against C. albicans in the tested range of M n = 0.8−18 kDa (MIC: 16−32 μg· mL −1 ). This outcome may be puzzling in contrast to MIC values obtained for the same materials against E. coli and S. aureus (MIC: 1−256 μg·mL −1 ). The different activity of L-PTMI against C. albicans is likely related to differences in the structure of microbial cell walls and membranes. Compared to bacteria, the fungal cell wall is thicker and contains chitin, hindering polymer migration to the lipid membrane surface. 91 The zeta potential of the fungal cell surface is also less negative, 92,93 further lowering susceptibility to polycationic agents. It is not trivial to explain the lack of a molecular weight effect, but it may be hypothesized that our polymers display antifungal intracellular activity. Some cationic polymers 94,95 have been reported to enter fungal cells without disrupting the integrity of the fungal membrane.

Biomacromolecules pubs.acs.org/Biomac Article
Unlike in the case of E. coli and S. aureus, a direct comparison of the antifungal activity between L-PTMI_7k and L-PEI_4k shows a very low difference in MIC values (16 and 32 μg·mL −1 , respectively). It appears that changes in the length of the alkyl spacer between amine groups, from two to three carbons, do not significantly contribute to increased antifungal activity.
Biological Activity of L-PTMI and L-PEI Derivatives. The N-monomethylation of both L-PTMI and L-PEI leads to decreased antibacterial activity (Table 6, Figure 4). This effect is stronger for S. aureus than for E. coli and weakest for C. albicans. For example, the MIC values for Me-L-PTMI_7k against E. coli and S. aureus rise 8 and 16 times, respectively, compared to L-PTMI. The MIC value for Me-L-PEI_4k against S. aureus increases 8 times compared to unmodified L-PEI_4k, whereas the MIC against E. coli increases 2-fold (from 16 to 32 μg·mL −1 ). The N-monomethylation does not change the toxicity of L-PEI but significantly reduces the cytotoxicity and hemolytic activity of L-PTMI.
It is likely that N-monomethylation limits hydrogen bond formation and accessibility of the lone electron pair. The steric hindrance may be the reason for weakened interactions between the polymer and phospholipids in the cell membrane. However, other factors, such as an increased hydrophobicity of the molecule or a lack of direct chemical reactivity of tertiary amines (methylated polymers) toward carboxylic esters (amidation), could also play a role. 56 Partial N,N-dimethylation of Me-L-PTMI_7k and Me-L-PEI_4k was performed to introduce a permanent cationic charge on the amine groups. Secondary and tertiary amines contain a cationic charge that depends on the level of protonation and is strictly linked to the pH of an environment. Quaternization provides a permanent charge and can improve electrostatic attraction between the polymer and the bacterial surface.
The partial N,N-dimethylation of L-PTMI_7k (DQ = 10% and DQ = 20%) does not change antimicrobial activity in comparison to N-methylated L-PTMI_7k (Figure 4a). The antimicrobial activity of Me-L-PEI_4k tends to slightly increase with the level of N,N-dimethylation for all tested microorganisms (Figure 4b), and the increase is the most pronounced for C. albicans. Quaternization of the studied polymers was expected to lead to a higher affinity to the cell surface, stronger binding to its anionic components, and improved antimicrobial activity. However, we do not observe this effect for L-PTMI, which may be due to the DQ being too low to induce a noticeable change in activity or the initial level of polymer protonation being high enough to mask any additional charge contribution. The postmodified L-PTMI_7k and L-PEI_4k do not show hemolytic activity at tested concentrations (2 mg·mL −1 ) ( Table 6), which may be due to the increased hydrophilicity of the polymers. 96 Interestingly, although methylation and partial quaternization do not improve antimicrobial properties of L-PTMI derivatives, they strongly reduce the cytotoxicity of these polymers (Figure 4a). As a result, some of them display substantial selectivity (Table 6, Figure S17). For postmodified L-PEI, quaternization also results in lower cytotoxicity ( Figure  4b). This effect, combined with the increase in antimicrobial activity, results in a selectivity index for MePEI-co-Me 2 PEI 20% exceeding 100 in case of E. coli and C. albicans (Table 6, Figure  S17).
Activity against Clinical Isolates. Three polymers (namely, L-PTMI_1.2 k, L-PTMI_7k, and MePTMI-co-Me 2 PTMI 20% ) were selected to assay MIC values against clinical isolates of species belonging to the ESKAPE group 97 that pose a particular medical threat by evading commonly used antibiotics. All polymers display substantial activity against clinical isolates with MIC values as low as 4 μg· mL −1 , and it seems that investigated molecules present higher activity against Gram-positive than Gram-negative bacteria (Table 8). L-PTMI_7k displays lower MIC values than L-PTMI_1.2 k and partially quaternized L-PTMI_7k. These results are highly consistent with our observations for model microorganisms, where higher MWs and nonmodified polymers are more active. Importantly, L-PTMI_1.2 k and L-PTMI_7k retain good activity against clinically isolated S. aureus, both methicillin-sensitive or methicillin-resistant species. Similarly, no difference in MIC values can be observed between various strains of K. pneumoniae that are sensitive and resistant to β-lactam antibiotics. The same activity against βlactam resistant and susceptible strains also supports the Biomacromolecules pubs.acs.org/Biomac Article hypothesis that the mechanism of activity of the investigated polymers is linked to the disruption of the bacterial membrane. Additionally, the polymers obtained in this work were tested against five different strains of Mycobacterium tuberculosis (Table 9), including isoniazid (INH) and rifampicin (RMP) resistant. 98 The investigated polymers are substantially less active against Mycobacterium than against studied Gramnegative and Gram-positive strains. Some limited activity can be observed for polymers with low molecular weights. The large difference in activity between Mycobacteria and other bacteria could be explained by significant differences in the cell wall structure. It is likely that the Mycobacterium cell wall, which is rich in mycolic acids and arabinogalactan, limits penetration of the investigated polymers into the cell membrane surface and prevents interactions with cell membrane components.

■ CONCLUSIONS
In this work, we investigated a series of polymeric antimicrobial agents, namely, polyamines L-PEI and L-PTMI, and their derivatives. L-PTMI appears to be an interesting novel agent, active against Gram-positive and Gram-negative bacteria but with a moderate activity against C. albicans and no activity against M. tuberculosis.
For the first time, the impact of L-PTMI molecular weight on its antimicrobial activity has been demonstrated and compared with literature data for L-PEI. The polymer activity increases along with an increase of the polymer MW within the tested range (0.8−18 kDa). Importantly, MIC values of L-PTMI are as low as 1 μg·mL −1 against S. aureus and 4 μg·mL −1 against clinically isolated P. aeruginosa and methicillin-resistant S. aureus.
For both antimicrobial activity (MIC) and toxicity (IC 50 and HC 50 ) of L-PTMI, crossing a certain polymer weight substantially increases the activity of the macromolecule. Importantly, this MW threshold is located at lower masses for MIC values compared to HC 50 and IC 50 . This provides an opportunity to utilize MW as a potential source of selectivity for novel polymeric antimicrobial agents.
L-PTMI is significantly more active than its analogue, L-PEI, when polymers with a similar number of r.u. are compared. Increasing the distance between amines in the polymer main chain from two to three methylene groups is beneficial for the antimicrobial activity of studied materials. Further postpolymerization modification of the polymers, namely, N-monomethylation and subsequent partial quaternization, typically reduces their antibacterial activity but also toxicity. As a result, those molecules bear substantial selectivity.
The materials obtained could serve as a starting point for active ingredients in disinfectants, antimicrobial films, or topical medical agents. More invasive uses such as wound dressing or intravenous applications would require further optimization of the structure to reduce toxicity.