Halogenation as a tool to tune antimicrobial activity of peptoids

Antimicrobial peptides have attracted considerable interest as potential new class of antibiotics against multi-drug resistant bacteria. However, their therapeutic potential is limited, in part due to susceptibility towards enzymatic degradation and low bioavailability. Peptoids (oligomers of N-substituted glycines) demonstrate proteolytic stability and better bioavailability than corresponding peptides while in many cases retaining antibacterial activity. In this study, we synthesized a library of 36 peptoids containing fluorine, chlorine, bromine and iodine atoms, which vary by length and level of halogen substitution in position 4 of the phenyl rings. As we observed a clear correlation between halogenation of an inactive model peptoid and its increased antimicrobial activity, we designed chlorinated and brominated analogues of a known peptoid and its shorter counterpart. Short brominated analogues displayed up to 32-fold increase of the activity against S. aureus and 16- to 64-fold against E. coli and P. aeruginosa alongside reduced cytotoxicity. The biological effect of halogens seems to be linked to the relative hydrophobicity and self-assembly properties of the compounds. By small angle X-ray scattering (SAXS) we have demontrated how the self-assembled structures are dependent on the size of the halogen, degree of substitution and length of the peptoid, and correlated these features to their activity.


Results and discussion
Probing the link between halogenation and antimicrobial activity. We synthesized 36 peptoids using a scaffold containing alternating NLys and Npm units which vary by length (6-, 8-, 10-, 12-mers), and the level of halogen substitution (full or alternate). Halogen atoms (fluorine, chlorine, bromine or iodine) were introduced via phenyl rings in position 4 and synthesized using submonomer approach (Fig. 1).
Halogenation had no effect on the activity of the peptoids against either E. coli and P. aeruginosa (Table 1). However, a clear correlation was observed between antimicrobial activity against Gram-positive strains, the level of substitution, and the nature of a halogen, among all sets. The fully halogenated peptoids demonstrated drastically enhanced activity against wild type and resistant strains of both S. aureus and S. epidermidis. Interestingly, for 6-(2-5) and 8-mers (11)(12)(13)(14) the activity rose from fluorine to iodine, where the latter was most potent. For the 6-mers, addition of just three iodine atoms led to up to > 64-fold increase against S. aureus (MIC for 1 = > 512 μg/mL; for 5 = 8 μg/mL) and 256-fold increase against MRSE (MIC for 1 = 512 μg/mL; for 5 = 2 μg/ mL), while 8-mer 14 exhibited the same activity against S. aureus, and 128-fold increase against MRSE (MIC for 10 = 256 μg/mL; for 14 = 2 μg/mL). However, moving to fully substituted 10-mer (20-23) and 12-mer (29)(30)(31)(32) sets, the antimicrobial trend is lost as compounds bearing iodine atoms displayed lower potency against both S. aureus and multidrug resistant S. epidermidis compared to their chlorinated and brominated analogues. Bromination of the 10-mer led to > 256-fold and 128-fold increase of activity against S. aureus and MRSE compared to the unsubstituted control peptoid (S. aureus: MIC for 19 = > 512 μg/mL; for 22 = 2 μg/mL; MRSE: MIC for The "half-substituted" sets fell under a similar trend, though generally displaying similar or lower potency. Interestingly, the half-substituted peptoids bearing bromine exhibited comparable activity to their fully substituted analogues. For example, compound 26 showed MICs between 1-8 μg/mL versus 1-4 μg/mL for the fully substituted analogue 22. As expected, the peptoids' hydrophobicity increased with the addition of halogen atoms, where fluorine displayed a less pronounced effect while incorporation of iodine led to noticeably higher hydrophobicity profiles. In parallel, the peptoids' antimicrobial activity fell into a well-established correlation, where increase in hydrophobicity led to an increased antimicrobial activity. However, when a certain hydrophobicity threshold was met (e.g. for compounds 23 and 32), the activity was lost, the phenomena that has been previously observed in other peptide/peptoid studies 36,37 . All in all, we could see that introduction of halogen atoms led to an increase of hydrophobicity that accompanies an increase of antimicrobial activity against Gram-positive bacteria. However, when a certain level of hydrophobicity was met, the activity of peptoids started decreasing.

Studying the effects of halogen substitutions on peptoid self-assembly in solution.
To further understand the impacts of variation in length, size of halogen groups and degree of substitution, the nanostructures of these compounds were studied in detail using Small Angle X-ray Scattering (SAXS).
SAXS allows for the determination of whether these peptoids self-assemble into nanostructures or instead exist as single molecules in aqueous solution [38][39][40][41] . Furthermore, through detailed theoretical modelling, the techniques allow for an accurate estimation of molecular weight, shape and the overall physical structures of peptoid assemblies. The results revealed that the observed structures depend on the length and hydrophobicity of the various peptoids; and self-assembly into defined nanostructures was observed for a few of them. Scattering intensity is plotted as a function of the modulus of the scattering vector, q = 4πsin(θ/2)/λ, where λ is the wavelength of the X-rays and θ is the scattering angle (Fig. 2). It should be noted that 1/q has the dimension of length and the quantity represents a sort of 'measuring stick'; at low q, large structures are probed, while at high q, SAXS is sensitive to more local structures.
For the 10-mers (compounds 19-27), the scattering curve for the fully brominated 22 and fully iodinated 23 peptoids exhibited significantly higher intensity and a different shape as compared to the rest of the 10-mers ( Fig. 2A,B). The latter exhibited typical polymer-like scattering pattern for random (Gaussian) chains although an upturn at low q revealed a small fraction of larger structures or aggregates. The upturn follows a power law of ~ q −2 indicating plate-like fibers, and no larger aggregates that would typically follow the power law of ~ q −442 . Through fit analysis of these data using a model with a combination of free chains and rectangular fibers, we obtained a radius of gyration (Rg) of the free chains of 7-9 Å, and a small mole fraction of only 0.001-0.0005% fiber-like sheets (see supporting information Table S1 for full list of fit parameters).
The scattering for the fully brominated 22 and all iodinated 23 did not exhibit the same upturn at low q and overall shape at intermediate and high q could therefore not be explained with the fit model described above. Instead, the scattering intensity exhibited a flatter q-dependence at low q indicating discrete, smaller nanostructures. The data from both compounds could be analyzed using a bundle model where cylinders representing folded/helical units are assembled into trimeric/tetrameric bundles 39 . The scattering of a concentration range www.nature.com/scientificreports/ from 5-0.6 mg/mL of both systems was analyzed simultaneously and the obtained fit parameters are listed in Table S1. The fit analysis also revealed the critical aggregation concentration (CAC) for the self-assembled structures and indicated a CAC value of 2.3 mg/mL and 0.4 mg/mL for the fully brominated 22 and iodinated 23, respectively (see supporting information Figure S1). The lower CAC value of 23 is a result of the high hydrophobicity as reported in Table 1. The increased hydrophobicity might provide an explanation of the observed loss of antimicrobial activity for the fully iodinated 10-and 12-mer (c.f. Table 1). These results suggest that a clear correlation between self-assembly and activity as seen by Chu-Kung and co-workers in the past 32 , cannot be drawn for these compounds as the estimated lowered CAC from SAXS is higher than the MIC values. To further investigate if the self-assembly properties of the iodinated peptoids, we proceeded with the 6-, 8-and 12-mer of the fully iodinated peptoids (5,14,23,32). The results are shown in Fig. 3 (see Table S2 for fit parameters). From the fit analysis, we found a CAC of 2.8 mg/mL for the fully iodinated 6-mer compound 5, 1.4 mg/mL for the 8-mer 14 and 0.4 mg/mL for the 12-mer 32.
These results show that even at very short lengths, these peptoids self-assemble in the probed concentration range due to their high hydrophobicity, which is evident from the retention times in Table 1. The detected CAC is highly correlated with the length of the peptoid. However, as seen from the MIC values these shorter peptoids are highly active compared to the 10-mer and 12-mer, indicating a potential threshold of hydrophobicity. Furthermore, the detected CAC is still highly correlated with the length of the peptoid, but a link between CAC and MIC could not be drawn. This is in contrast with the fatty acid conjugated peptides studied by Chu-Kung and co-workers, who found a correlation 32 . The reduction of activity seen for the 10-mer and 12-mer is therefore likely related to the increased hydrophobicity, indicating that there is a threshold to stay within than the self-assembly properties themselves. However, further studies into the consequences of hydrophobicity and self-assembly with the activity and toxicity for peptoids are needed to fully explain the observed trends.
Halogenation as a tool to improve antimicrobial activity. In order to investigate whether halogens can be used as a tool to improve the antimicrobial potency of a known peptoid, we chose the well-studied Peptoid 1 and synthesized two small libraries of chlorinated and brominated analogues as these halogens demonstrated overall higher potency compared to the fluorinated ones, and iodine variants raised the concern of aggregation and loss of activity. Both halogen atoms were introduced via Nspe units in position 4 on the phenyl rings. The level of substitution varied between full substitution (all phenyl rings bear a halogen atom) and "half " substitution (every other phenyl ring bears a halogen atom). This strategy yielded six chloro-and six bromomodified versions of Peptoid 1 (Fig. 4). www.nature.com/scientificreports/ The library of 12 compounds (37-48) was tested against the same panel of bacterial strains as the first generation of peptoids. However, none of the modifications led to increased potency, while most caused a loss of activity (Table 2). Judging from long HPLC retention times, this could be explained by the fact that the critical hydrophobicity level was reached, which causes aggregation and loss of activity similar to what we have seen for compounds 23 and 32, as the highest activity was observed for the compounds with lower retention times. Interestingly, compounds 38, 40-42 have the same number of chlorine atoms, but their hydrophobicity differ. For the brominated derivatives 44, 46-48, this effect was even more pronounced. It shows that even distribution of halogen atoms across the peptoid chain led to higher hydrophobicity compared to positioning chlorine or bromine atoms at the ends or the middle of the peptoid sequence.
Short brominated Peptoid 1 analogues. As seen from the increased values of retention times, we have hit the hydrophobicity ceiling during the introduction of halogens into the sequence of Peptoid 1 (Table 2), hence we decided to shorten the length of Peptoid 1 from twelve to six residues, cutting its length in half, and synthesized a small library of four analogues with different levels of halogenation (Fig. 4). As seen from Tables 1 and 2, introduction of fluorine didn't result in the desired increase of activity, while bromination had a more pronounced effect on the hydrophobicity than chlorination and similar to one displayed by the iodine-containing peptoids. Hence, three brominated short Peptoid 1 analogues (49-51) and one non-halogenated peptoid Pep1-6 mer were synthesized and tested against the same bacterial strains as the previous peptoids. Data are shown in Table 3 where the new data set is compared with the original Peptoid 1 data 27 .
Compound 49 has only one terminal bromine, 50 has two bromine atoms and for 51-all four phenyl rings are substituted with a bromine atom. The addition of just two bromine atoms was enough to improve the activity of a non-active short analogue, the Pep1-6 mer, 16-32-fold against both S. aureus and S. epidermidis. Incorporation of two extra bromines yielding four in total failed to significantly affect the activity. Addition of one terminal halogen atom, on the other hand, was already enough to see a several-fold increase in antimicrobial activity. Despite the previous data, where introduction of halogens did not improve the activity against Gram-negative  www.nature.com/scientificreports/ bacteria, in this case addition of two bromines was enough to reach sufficient hydrophobicity in order to obtain an MIC against P. aeruginosa that is close to the original Peptoid 1 and reach 32-fold increase of activity against E. coli compared to the Pep1-6 mer .
As we have shown that the incorporation of halogens can improve the activity of inactive peptoids, we decided to examine their cytotoxicity profiles. Peptoid 1, Pep1-6 mer and the three brominated analogues were tested towards a HaCaT cell line for 1 h. The results were obtained using MTS/PMS assay (Fig. 5, Table 3).
We have observed a clear trend between the number of halogen atoms and corresponding increased cytotoxicity, however while compound 50 demonstrated the same antimicrobial activity profile to Peptoid 1 it was less cytotoxic with IC 50 146.9 μg/mL versus only 35.0 μg/mL for the latter. All in all, initial results indicate reduced cytotoxicity of the brominated analogues when compared to Peptoid 1, though compounds 50 and 51 displayed similar activity and hydrophobicity profiles.

conclusions
In this study, we have investigated the effect of fluorine, chlorine, bromine and iodine on the antimicrobial activity of peptoids. First, using an inactive model (Nlys-Npm) n peptoid scaffold, we have identified that incorporation of chlorine or bromine led to an improvement of antimicrobial activity against Gram-positive bacteria, while fluorination did not display any pronounced effect. Introduction of iodine in 6-and 8-mers analogues dramatically increased the activity, but led to loss of activity due to aggregation in 10-and 12-mers. Afterwards, we attempted to improve the antibacterial potency of Peptoid 1 by incorporating chlorine or bromine atoms via Nspe units, which led to overall loss of activity. We interpret this oberservation as sign of the hydrophobicity limits being reached. However, bromination of a shorter inactive 6-mer analogue of Peptoid 1 resulted in the same activity as the 12-mer Peptoid 1 against some bacteria, while noticeably improving its cytotoxic profile. Therefore, here we have shown that halogenation, and particularly bromination, can be used to readily modify and alter the physicochemical and antibacterial properties of peptoids but the effect strongly depends on the choice of the halogen. In addition, the effect is quite sequence-and length-specific, and inclusion of halogens can also lower antimicrobial activity.

Methods
General information. Solvents, amines, and resins were purchased from Iris Biotech, Sigma-Aldrich, and Merck and used without further purification. For compounds 1-48: purity was determined as described previously 49   Small angle X-ray scattering. SAXS experiments of the 10-mer peptoids were performed at the automated BM29 bioSAXS beamline at the European Synchrotron Radiation Facility (ESRF) in Grenoble, France 45 . The data was obtained using an energy of 12.5 keV and a detector distance of 2.87 m, covering a q range ( q = 4πsin(θ/2)/ ), where θ is the scattering angle and is the X-ray wavelength) of about 0.0047 to 0.5 Å −1 .
The data set was calibrated to an absolute intensity scale using water as a primary standard. 40 µL samples were run through a capillary using the flow mode of the automated sample changer 46 . SAXS data was collected in ten successive frames of 0.5 s each to monitor radiation damage and the data reduction was done using the standard tool at BM29 47 .
The SAXS experiments on the fully iodinated 6-, 8-, and 12-mers to determine the CAC of the compounds were performed using a Bruker NANOSTAR equipped with a microfocus X-ray source (IμS Cu, Incoatec, Germany) and a VÅNTEC-2000 detector. Raw scattering data was calibrated to absolute intensity scale using water as a primary standard and radially averaged in order to obtain the 1D scattered intensity profile as a function of the scattering vector, with a wavelength of 1.54 Å. Two concentrations of compound 23 and 19 was also run on the NANOSTAR to verify that the results were comparable with the results from synchrotron SAXS at ESRF.
The modelling fit analysis of the scattering data is explained in detail in the supplementary information.
Cell viability assay. Cell viability assessment was performed on cell monolayers grown to ∼90% confluence in 96-well plates Corning Costar (Sigma-Aldrich, Brøndby, Denmark) by using the MTS/PMS assay as previously described 48 . Briefly, the adhered cells were washed with 37 °C PBS solution (ThermoFisher Scientific, Roskilde) and exposed for 1 h at 37 °C to 100 μL of peptoids dissolved in the medium also used for culturing of the cell line (at concentrations in the range 0-1,000 μg/mL). The precise exposure-time were selected to enable comparison with related peptoid compounds 21,35 . Then the cells were washed twice with 37 °C PBS and then 100 μL of an MTS/PMS solution in media, consisting of 240 μg/mL MTS (Promega, Madison, WI, United States) and 2.4 mg/mL PMS (Promega, Madison, WI, United States), were added to the cells, which then were incubated for 1 h at 37 °C protected from light. A plate reader (SpectraMax i3X; Molecular devices, San Jose, CA) was used to measure the absorbance at 492 nm. The relative viability was calculated by using 0.2% (w/v) sodium dodecyl sulfate (SDS) as the positive control, while cells exposed to medium without test compound were used as the negative control. Data were obtained in three independent biological replicates performed on separate passages of cells and on separate days with a total number of six replicates.