New Chloramphenicol Derivatives from the Viewpoint of Anticancer and Antimicrobial Activity

Over the last years, we have been focused on chloramphenicol conjugates that combine in their structure chloramphenicol base with natural polyamines, spermine, spermidine and putrescine, and their modifications. Conjugate 3, with spermidine (SPD) as a natural polyamine linked to chloramphenicol base, showed the best antibacterial and anticancer properties. Using 3 as a prototype, we here explored the influence of the antibacterial and anticancer activity of additional benzyl groups on N1 amino moiety together with modifications of the alkyl length of the aminobutyl fragment of SPD. Our data demonstrate that the novel modifications did not further improve the antibacterial activity of the prototype. However, one of the novel conjugates (4) showed anticancer activity without affecting bacterial growth, thus emerging as a promising anticancer agent, with no adverse effects on bacterial microflora when taken orally.


Introduction
Chloramphenicol (CAM) is an effective antibiotic with bacteriostatic action against Gram-positive and Gram-negative bacteria and was the first broad-spectrum antibiotic to be used clinically. However, the undesired toxicity resulting from CAM administration restricts its clinical use. Case reports indicate that CAM is responsible for hematological disorders such as bone marrow depression and aplastic anemia. While the first is a rare complication, the second is more frequent and was ascribed as a mitochondrial protein synthesis disorder [1][2][3][4]. Despite its side effects, CAM remains one of the best-studied protein synthesis inhibitors, and efforts towards its derivatization have continued in order to provide a new generation of compounds with better antibacterial activity and minimal side effects [5]. Numerous functional and structural studies have concluded that CAM binds to the large ribosomal subunit occupying the A-site of the peptidyltransferase center (PTC) [6][7][8]. Its binding allows the delivery and initial binding of the A-site aminoacyl-tRNA but prevents its full accommodation at

Synthesis of PA-CAM Conjugates
The synthesis of the new PA-CAM conjugates 4-7 is depicted in Scheme 1. It involves the onepot acylation of the commercially available chloramphenicol base (CLB) with succinic anhydride followed by coupling with the appropriate N 1 ,N 1 -dibenzylated linear diamines 10a-c or the N 1 ,N 1 ,N 8 ,N 8 -tetrabenzylspermidine (12) in the presence of the coupling agent O-(benzotriazol-1-yl)-N,N,N',N'-tetramethyluronium hexafluorophosphate (HBTU) and Et3N or ethyldiisopropylamine (DIEA) in dimethylformamide (DMF). The required diamine derivatives 10a-c were obtained in the form of their bistrifluoroacetate salts from the corresponding linear, commercially available amino acids β-alanine (βAla, 8a), γ-aminobutyric acid (γΑba, 8b) and ε-aminocaproic acid (εAca, 8c) through a four-step sequence, also depicted in Scheme 1. This involved the protection of the amino function with the triphenylmethyl (trityl, Trt) group [18,19] followed by coupling with dibenzylamine in the presence of HBTU and Et3N to give the amides 9a-c. These were reduced with lithium aluminium hydride (LAH) in tetrahydrofuran (THF), and then the trityl group was removed by trifluoroacetic acid (TFA)-mediated acidolysis to provide compounds 10a-c. On the other hand,

Synthesis of PA-CAM Conjugates
The synthesis of the new PA-CAM conjugates 4-7 is depicted in Scheme 1. It involves the one-pot acylation of the commercially available chloramphenicol base (CLB) with succinic anhydride followed by coupling with the appropriate N 1 ,N 1 -dibenzylated linear diamines 10a-c or the N 1 ,N 1 ,N 8 ,N 8 -tetrabenzylspermidine (12) in the presence of the coupling agent O-(benzotriazol-1-yl)-N,N,N ,N -tetramethyluronium hexafluorophosphate (HBTU) and Et 3 N or ethyldiisopropylamine (DIEA) in dimethylformamide (DMF). The required diamine derivatives 10a-c were obtained in the form of their bistrifluoroacetate salts from the corresponding linear, commercially available amino acids β-alanine (βAla, 8a), γ-aminobutyric acid (γAba, 8b) and ε-aminocaproic acid (εAca, 8c) through a four-step sequence, also depicted in Scheme 1. This involved the protection of the amino function with the triphenylmethyl (trityl, Trt) group [18,19] followed by coupling with dibenzylamine in the presence of HBTU and Et 3 N to give the amides 9a-c. These were reduced with lithium aluminium hydride (LAH) in tetrahydrofuran (THF), and then the trityl group was removed by trifluoroacetic acid (TFA)-mediated acidolysis to provide compounds 10a-c. On the other hand, Antibiotics 2019, 8, 9 4 of 16 the spermidine derivative 12 was synthesized from amide 9a as follows. Compound 9a was routinely detritylated with TFA and the thus-liberated primary amino function was acylated with succinic anhydride, in the presence of DIEA. Without isolation, the obtained intermediate was further coupled in a one-pot fashion with dibenzylamine in the presence of HBTU and DIEA and the thus-obtained trisamide 11 was finally reduced with LAH to give the tetrabenzylated SPD 12. Experimental details for the syntheses described above and the complete characterization of intermediates and final products are provided in the Supplementary Material (Section S1).
Antibiotics 2018, 7, x 4 of 16 the spermidine derivative 12 was synthesized from amide 9a as follows. Compound 9a was routinely detritylated with TFA and the thus-liberated primary amino function was acylated with succinic anhydride, in the presence of DIEA. Without isolation, the obtained intermediate was further coupled in a one-pot fashion with dibenzylamine in the presence of HBTU and DIEA and the thus-obtained trisamide 11 was finally reduced with LAH to give the tetrabenzylated SPD 12. Experimental details for the syntheses described above and the complete characterization of intermediates and final products are provided in the Supplementary Material (Section S1).

Bacterial Strains and Cell Lines
The in vivo antibacterial activity of the conjugates was determined using the following three bacterial strains: wild-type Escherichia coli K12 (E. coli K12), Escherichia coli ΔTolC mutant strain (E.

Bacterial Strains and Cell Lines
The in vivo antibacterial activity of the conjugates was determined using the following three bacterial strains: wild-type Escherichia coli K12 (E. coli K12), Escherichia coli ∆TolC mutant strain (E. coli ∆TolC) lacking the TolC protein, which is involved in the efflux pumps operation, and wild-type Staphylococcus aureus (S. aureus). The anti-tumor activity of the conjugates was assessed using the human ZL34 and MeT5A cell lines, which were kindly offered by Professor G. Stathopoulos (University of Patras). ZL34 are cancer cells, derived from malignant pleural mesothelioma, while MeT5A are immortalized cells, which originate from healthy mesothelial tissue [20].

In Vivo Antibacterial Activity
For the experimental procedures, the bacteria were cultured in luria broth (LB) medium (tryptone 10 g/L, yeast extract 5 g/L, NaCl 10 g/L) as previously described [10]. The bacteria cultures were grown in the presence of CAM or each conjugate at increasing concentration (1-100 µM). The cultures were incubated at 37 • C until the absorbance (OD 560 ) of the reference culture in the absence of antibiotic reached the value of 0.7. At this point, the growth of the cultures was terminated by placing them in an ice bath, and the measurement of the optical density followed [10]. These measurements were used to calculate the half-maximal effective concentration (EC 50 ) values [21].

Affinity Measurement of PA-CAM Conjugates for the E. coli 70S Ribosome
Reassociated 70S ribosomes were prepared from E. coli K12 cells as described previously [22]. 70S ribosomes were incubated in buffer A (100 mM Tris-HCl pH 7.2, 100 mM ammonium acetate, 10 mM magnesium acetate, 6 mM β-mercaptoethanol) with 10 µM [ 14 C]-chloramphenicol (150 dpm/pmol) at a final concentration of 0.20 µM [23]. After incubation for 10 min at 37 • C, the mixture was diluted with 3 mL of cold buffer A and filtered through a 25-mm diameter cellulose nitrate membrane filter (Millipore 0.45 µm pore size). The filter was washed three times with 3 mL of cold buffer A and the radioactivity which remained bound on the filter was measured. The binding of [ 14 C]-chloramphenicol was studied in competition with CAM or PA-CAM conjugates by keeping the concentration of [ 14 C]-CAM constant (10 µM) and increasing the concentration of non-radioactive conjugates [23].

Evaluation of the Anticancer Activity
The antitumor activity of the conjugates was assessed using the human ZL34 and MeT5A cell lines as previously described [10]. ZL34 and Met5A cells were plated in sterile 6-well microtiter plates and grown in Dulbecco's modified Eagle's medium (DMEM), supplemented with 10% heat-inactivated fetal bovine serum and 1% penicillin/streptomycin. Cultures were maintained in a humidified atmosphere with 5% CO 2 at 37 • C. Compounds 3 and 4 were added at final concentrations of 30 and 60 µM, and then cells were grown for 24, 72 and 96 h. In parallel, solutions of conjugates combined with a ten-fold concentration of SPD were incubated under the same conditions. After treatment, the drug was removed by washing the cells twice with phosphate-buffered saline (PBS). The cells were then trypsinized (0.5 mL trypsin-EDTA×1 solution/well, 5 min at 37 • C), mixed with 1 mL DMEM and collected by centrifugation at 3000 × g for 5 min. Cell viabilities were determined by the trypan blue exclusion assay, using a TC10 automated cell counter (BIORAD) [10]. Viable cells were expressed as a percentage of total cells.

Immunoblotting
Cell lysates were prepared after treatment with 60 µM of conjugates for 48 h. Total cellular proteins were isolated at 4 • C using RIPA buffer (ABCAM) and mixed with 4% SDS-containing loading buffer. Protein concentrations were determined using Bradford assay [24]. Lysates (100 mg protein/well) were reduced in Laemmli buffer containing 1 mM dithiothreitol, resolved by 10% SDS-PAGE and electrotransfered to Immobilon-P PVDF membranes (Millipore, Massachusetts, USA) according to standard procedures [25]. Blots were incubated at 4 • C overnight with rabbit monoclonal antibody against mitochondrial cytochrome c oxidase subunit II (COX2) (anti-mtCO2 antibody, ABCAM) at a 1:1000 dilution in blocking buffer. After washing with TBS-Tween and incubation with secondary antibody coupled to horse radish peroxidase, protein bands were visualized with Antibiotics 2019, 8, 9 6 of 16 ECL TM chemiluminescence reagent (Amersham) and detected by autoradiography. The intensity of bands was quantified by Image Analysis, using the Image-Pro Plus 7 software (Media Cybernetics). Blots were re-probed with a mouse anti-actin antibody (Santa Cruz Biotechnology) to verify equal protein loading.

Quantification of the Intracellular Levels of Conjugates 3 and 4 Plus Polyamines
PAs and conjugates 3 and 4 in ZL34 and Met5A cells were determined by reverse phase high-performance liquid chromatography (RP-HPLC-DAD), as described previously [11]. Briefly, 100 mg cells (wet weight), either ZL34 or Met5A, which were grown in the presence of 60 µM of conjugates, were washed and then homogenized in 0.6 M HClO 4 . Cultures grown in the absence of the conjugates were used as reference. In the isolates, hexamethylenediamine dihydrochloride was added as an internal standard, followed by dansylation and subsequent extraction by toluene. The extracts were dried, reconstituted in acetonitrile/10 mM phosphate buffer pH 4.4 (45:55 v/v) and filtered through a 0.2-µm membrane (Spartan 13/0.2 RC Whatman GmbH, Dassel, Germany) prior to HPLC analysis. The HPLC system included a Varian Star 9010 solvent delivery system with a 250 × 4.6 mm Luna 5 µm C18(2) 100 A column (Phenomenex Co., USA) and a photodiode array detector (Shimadzu SPDM20A).
The gradient elution profile is presented in Table 1. The flow rate was set at 1.0 mL/min. Polyamines were quantified at 258 nm and the conjugates at 275 nm. The detection limits for PUT, SPD, and SPM were 52 pmol, 18 pmol, and 36 pmol at 258 nm, respectively. The detection limits for compounds 3 and 4 were 25 pmol and 30 pmol at 275 nm, respectively. All experiments were run in triplicate. The coefficients of variation (C.V.) of the intra-assays were 5-7%, while those of the inter-assays ranged from 13 to 17%. LC Solution Software (Shimadzu) was used for data processing and the results were expressed as nmol/mg protein. The binding modes of conjugates 3 and 4 in a region near the catalytic center (A2938, C2939) of the human mitochondrial large ribosomal subunit (hmLRS) have been modeled using molecular dynamics simulations. Three-dimensional models of the compounds were produced with Arguslab 4.0.1 (Planaria Software LLC, Seattle, WA, USA, http://www.arguslab.com), starting with the 3D structure of CAM derived from the 50S ribosomal subunit structure of E. coli in complex with CAM (PDB:3OFC). CHARMM force field parameters and topology files were generated by the SwissParam Tool [26]. The 3D structure of the hmLRS (PDB:3J7Y) with a resolution of 3.4 Å was used to model a~28 Å radius environment around each analog near the catalytic center. The analogs were initially positioned with their CAM moieties within the drug crystallographic pocket mapped into the hmLRS according to the intersystem correspondence of large subunit relevant nucleotides shown in Table S1 of Supplementary Material S3. The systems were then solvated with~10,000 TIP3 water molecules, and then neutralized with Mg 2+ ions using the VMD program [27]. All systems derived as above were energy minimized and then subjected to short NPT molecular dynamics (MD) simulations to stabilize their volume at T = 300 K and P = 1 Atm. Subsequently, NVT MD simulations for 30 ns at 300 K, with the particle mesh Ewald (PME) algorithm, and the rigid bonds assignments were run using the NAMD software [28] and the CHARMM27 force field for proteins and nucleic acids. During MD simulations, harmonic constraints have been imposed to all nucleic acid backbone atoms. Finally, an average structure over the last 1 ns of each simulation trajectory was energy minimized and used

Statistics
All data presented in the figures and tables represent the mean value and standard deviation, obtained from at least two independently performed experiments. Statistical tests (data variability, t-tests, one-way ANOVA) were performed using the program IBM Statistics 24. The statistical significance threshold was set at p = 0.05.

Antibacterial Activity of Novel PA-CAM Conjugates
All cells were cultivated in the presence of CAM or PA-CAM conjugates in concentrations reaching up to 100 µM. An E. coli strain was used as a model of Gram-negative bacteria, and an S. aureus strain was used as a model of Gram-positive bacteria. The antibacterial activity of the conjugates is expressed by the EC 50 value ( Table 2). With the exception of compound 3, none of the new compounds exhibited inhibitory activity against the growth of bacterial cultures. Thus, either new derivatives failed to cross the bacterial cell membrane or they are non-active within the bacterial cells. The latter could be attributed either to new derivatives not binding on the bacterial ribosome, or binding without effect [29]. An alternative explanation would be that the conjugates may enter the bacteria but are exported by efflux pumps, the specific bacterial transporters which represent a resistance mechanism. However, since E. coli ∆TolC growth was not inhibited, such a possibility does not seem feasible. 11.0 ± 0.9 8.9 ± 0.7 6.8 ± 0.5 4 >200 >200 >200 5 >200 >200 >200 6 >200 >200 >200 7 >200 >200 >200

The New PA-CAM Conjugates Compete with CAM to Bind on the Bacterial Ribosome
To clarify whether the new PA-CAM conjugates maintain their binding site on the 70S bacterial ribosome, we carried out competition experiments. These data ( Figure S1, Supplementary Material) allowed us to measure the concentration required for the 50% displacement of the radioactive CAM (IC 50 ), in the presence of increasing concentrations of all PA-CAM conjugates. The affinity constants, K i , were calculated using the following Cheng-Prusoff relationship (1) [30] and their values are presented in Table 3. According to the values of Table 3, the decrease of the affinity for 70S ribosome follows the order 3>7>4>6>CAM>5.
where [I] is the concentration of [ 14 C]-CAM (constant at 10 µM) and K D is the dissociation constant of ribosome-chloramphenicol complex formation. 1.3 ± 0.2 7 1.0 ± 0.1

Antiproliferative Activity of Compound 4
Based on the established antitumor activity of compound 3, we were expecting that the addition of the benzyl groups on the spermidine domain of the compound could also affect this feature. Compound 4 demonstrated cytotoxicity for ZL34 cancer cells, which appeared to be inhibited when the cells were cultured in the presence of exogenous SPD in the medium ( Figure 2C). Furthermore, conjugate 4 exhibited milder action against immortalized mesenchymal Met5A cells relative to conjugate 3, as presented in Figure 2D. The other compounds 5, 6 and 7 did not show any important antiproliferative activity.

Antiproliferative Activity of Compound 4
Based on the established antitumor activity of compound 3, we were expecting that the addition of the benzyl groups on the spermidine domain of the compound could also affect this feature. Compound 4 demonstrated cytotoxicity for ZL34 cancer cells, which appeared to be inhibited when the cells were cultured in the presence of exogenous SPD in the medium ( Figure 2C). Furthermore, conjugate 4 exhibited milder action against immortalized mesenchymal Met5A cells relative to conjugate 3, as presented in Figure 2D. The other compounds 5, 6 and 7 did not show any important antiproliferative activity.

Conjugates 3 and 4 Are Inhibitors of Mitochondrial Protein Synthesis
The cytotoxic effect of conjugates 3 and 4 on human cell lines triggered the investigation of the underlying mechanism. Taking into account the mitochondrial toxicity of CAM [31], we examined the possible disruption of mitochondrial protein synthesis. For this purpose, a mitochondrially encoded protein compartment of OXPHOS Complex IV, COX2 was measured. Indeed, as presented in Figure 3A, a remarkable decrease in COX2 levels was observed in ZL34 malignant cells. Simultaneously, actin levels, which were used as a reference of cytoplasmic protein synthesis level, did not demonstrate any decrease. In Figure 3C, data were normalized as to actin levels. COX2 levels were affected to a lesser extent in Met5A cells compared to ZL34 ( Figure 3B,D). The cytotoxic effect of conjugates 3 and 4 on human cell lines triggered the investigation of the underlying mechanism. Taking into account the mitochondrial toxicity of CAM [31], we examined the possible disruption of mitochondrial protein synthesis. For this purpose, a mitochondrially encoded protein compartment of OXPHOS Complex IV, COX2 was measured. Indeed, as presented in Figure 3A, a remarkable decrease in COX2 levels was observed in ZL34 malignant cells. Simultaneously, actin levels, which were used as a reference of cytoplasmic protein synthesis level, did not demonstrate any decrease. In Figure 3C, data were normalized as to actin levels. COX2 levels were affected to a lesser extent in Met5A cells compared to ZL34 ( Figure 3B and D).

Derivatives' and Polyamines' Intracellular Concentrations
The results of the identification and quantification of compounds 3 and 4 (for compound 4, see also Figure Table 4. Regarding the ZL34 cancer cell line ( Figure 4A), it appears that the presence of conjugates 3 and 4 resulted in the decrease of PUT, SPD and SPM levels, which was particularly pronounced in the case of SPM. In contrast, in immortalized Met5A cells ( Figure 4B), in the presence of conjugates 3 and 4, there was no remarkable change in PUT and SPD levels. SPM, however, showed a downward trend in immortalized cells, but to a lesser extent related to malignant cells. It is worth noting that conjugate 4 exhibited a remarkably high level of internalization in ZL34 cells relative to compound 3 ( Figure 4C). Still, both conjugates were identified in higher concentrations in tumor ZL34 cells relative to Met5A mesothelial cells. The elevated polyamine intake rates of cancer cells could probably explain this observation [17].

Derivatives' and Polyamines' Intracellular Concentrations
The results of the identification and quantification of compounds 3 and 4 (for compound 4, see also Figure S2, Supplementary Material) and PAs ( Figure S3, Supplementary Material) are presented in Table 4. Regarding the ZL34 cancer cell line ( Figure 4A), it appears that the presence of conjugates 3 and 4 resulted in the decrease of PUT, SPD and SPM levels, which was particularly pronounced in the case of SPM. In contrast, in immortalized Met5A cells ( Figure 4B), in the presence of conjugates 3 and 4, there was no remarkable change in PUT and SPD levels. SPM, however, showed a downward trend in immortalized cells, but to a lesser extent related to malignant cells. It is worth noting that conjugate 4 exhibited a remarkably high level of internalization in ZL34 cells relative to compound 3 ( Figure 4C). Still, both conjugates were identified in higher concentrations in tumor ZL34 cells relative to Met5A mesothelial cells. The elevated polyamine intake rates of cancer cells could probably explain this observation [17].

System Modeling and Molecular Dynamics Simulations
Based on the previous data, conjugates 3 and 4 seem to interact with mitochondrial protein synthesis, possibly by binding on the mitochondrial ribosome. Thus, we attempted to verify the

System Modeling and Molecular Dynamics Simulations
Based on the previous data, conjugates 3 and 4 seem to interact with mitochondrial protein synthesis, possibly by binding on the mitochondrial ribosome. Thus, we attempted to verify the possible stereochemical interactions of our compounds with the mitochondrial ribosome by applying molecular dynamics simulations. We also exploited our results from footprinting analysis on the 70S bacterial ribosome, taking into account the similarity of bacterial and mitochondrial ribosomal particles, as has been formulated by the theory of endosymbiosis [32]. Conjugates 3 and 4 demonstrated a similar footprinting pattern to CAM. Figure 5 illustrates the outcome of the simulations. Both compounds appear to form hydrogen bonds with G2724 of mitochondrial 16S rRNA, which corresponds to the G2061 of E. coli 23S rRNA. In addition, conjugate 3 interacts with A2725 and C2726 (A2062 and C2063, respectively), bases that lie within the CAM binding site on the bacterial ribosome. Furthermore, a hydrogen bond between conjugate 4 and U3072 (U2585 on 23S E. coli rRNA) suggests a contribution of the polyamine segment to the compound ribosomal binding affinity. Finally, the interactions of the conjugates with G2992 (conjugate 3) and A2938 (conjugate 4) indicate that they may enter the cleft of the peptidyl transferase center of the mitochondrial ribosome. Table 5. Hydrogen bond pairs $ between atoms of mitochondrial rRNA and of compounds 3 and 4 # . Corresponding bases of E. coli 23S rRNA are noted in parenthesis. The matching of RNA nucleotides was based on data from Brown et al. [33].  [34]. # Atom numbering in the models of compounds 3 and 4 is given by Arguslab and is shown in Figure S4, Supplementary Material.
Antibiotics 2018, 7, x 11 of 16 possible stereochemical interactions of our compounds with the mitochondrial ribosome by applying molecular dynamics simulations. We also exploited our results from footprinting analysis on the 70S bacterial ribosome, taking into account the similarity of bacterial and mitochondrial ribosomal particles, as has been formulated by the theory of endosymbiosis [32]. Conjugates 3 and 4 demonstrated a similar footprinting pattern to CAM. Figure 5 illustrates the outcome of the simulations. Both compounds appear to form hydrogen bonds with G2724 of mitochondrial 16S rRNA, which corresponds to the G2061 of E. coli 23S rRNA. In addition, conjugate 3 interacts with A2725 and C2726 (A2062 and C2063, respectively), bases that lie within the CAM binding site on the bacterial ribosome. Furthermore, a hydrogen bond between conjugate 4 and U3072 (U2585 on 23S E. coli rRNA) suggests a contribution of the polyamine segment to the compound ribosomal binding affinity. Finally, the interactions of the conjugates with G2992 (conjugate 3) and A2938 (conjugate 4) indicate that they may enter the cleft of the peptidyl transferase center of the mitochondrial ribosome.  [34]. # Atom numbering in the models of compounds 3 and 4 is given by Arguslab and is shown in Figure S4, Supplementary Material.  (Table 5) as well as two stacking interactions stabilize compound 4 in the region. Note the hydrogen bond between N1 nitrogen of adenine in A2938 with a hydroxyl oxygen of compound 4 nearest to the CAM ring.

Discussion
Bacterial resistance to antibiotics poses a serious risk to public health. Resistance can be developed by a variety of mechanisms, such as modifications on the antibiotic's binding site [35], reduced membrane permeability [36,37], modifications on the antibiotic moiety [38] or increased  (Table 5) as well as two stacking interactions stabilize compound 4 in the region. Note the hydrogen bond between N1 nitrogen of adenine in A2938 with a hydroxyl oxygen of compound 4 nearest to the CAM ring.

Discussion
Bacterial resistance to antibiotics poses a serious risk to public health. Resistance can be developed by a variety of mechanisms, such as modifications on the antibiotic's binding site [35], reduced membrane permeability [36,37], modifications on the antibiotic moiety [38] or increased efflux through specific transporters (efflux pumps) [39,40]. Despite ongoing research efforts, only a small number of new antibacterial compounds have been introduced into clinical practice over the last years, rendering resistance a serious therapeutic problem. An approach to overcome this issue is the chemical modification of already known effective antibiotics [41] and led us to focus on CAM derivatization [10,11,21,[42][43][44].
Our new PA-CAM conjugates were tested either in Gram-positive or in Gram-negative bacteria and additionally in the mutant E. coli strain ∆TolC, which is more sensitive to many antibiotics. According to our results, neither compound inhibited bacterial growth, although both can be bound on E. coli ribosomes with binding affinities similar to CAM (Table 3). From the data of Table 3, it is obvious that all compounds can bind to the E.coli ribosomes with almost identical affinities. Neither the addition of two more benzyl groups on the lead molecule, nor the deletion of the aminopropyl moiety nor PA carbon skeleton modification changed the order of magnitude of their binding affinity, presented by the K i value. Moreover, while their binding affinities for the ribosome are high enough, none of them exerted important antimicrobial activity. Taking also into account the fact that the ∆TolC strain was not inhibited in the presence of our conjugates, it remains to be further clarified whether the bound conjugates inhibit ribosome function in vitro. Such a result will allow us to clarify their ability or inability to pass through the cell membrane.
Multiple reports have been published concerning the undesirable side effects following administration of CAM, which are due to antibiotic activity on the mitochondrial ribosome [4]. This finding has prompted the investigation of the potential anticancer action of the antibiotic, as tumor cells are particularly dependent on mitochondrial activity due to their increased energy needs [3]. Along with increased energy requirements, the need for active metabolites is also elevated in tumor cells [45]. Among these metabolites are PAs, essential molecules for cell growth and survival. Indeed, cancer cells have to cope with this need by increasing both the biosynthesis and import of PAs into the cell [46]. The increased activity of PATs in cancer cells is a distinctive element, compared to normal cells, that may be useful in the specialized targeting of neoplastic cells [17].
These observations prompted us to investigate the antitumor activity of the PA-CAM conjugates. Interestingly, previous studies have established the antitumor potential of compound 3 [10]. Among the new compounds tested, only compound 4 exhibited a similar antiproliferative behavior to compound 3, decreasing the viability of cancer cells and having a milder effect on Met5A non-tumor cells.
Interestingly, compound 4 demonstrated elevated internalization in cancer cells compared to 3, possibly due to an increased lipophilicity, derived from the two additional benzyl groups. Nevertheless, both compounds were identified in higher concentrations in ZL34 mesothelioma cells, exploiting the presence of more active transporters on cancer cell membranes [17]. Both derivatives are considered to enter the human cells using the PATs, since they competed with SPD, when the latter was added exogenously in the culture medium (Figure 2A,C).
CAM has been shown to act on mitochondria, inhibiting mitochondrial protein synthesis [47]. A hypothesis that we examined is the possibility that compounds 3 and 4 affect the viability of human cells in the same way. For this purpose, COX2 protein was used as a marker. COX2 is encoded by mitochondrial DNA and is translated with mitochondrial ribosomes [48]. Indeed, conjugates 3 and 4 caused a significant decrease in COX2 expression in tumor cells, while actin level, which was used as an indicator of cytoplasmic protein synthesis, remained unaffected, hinting that the mitoribosome may be the main target in the human cancer cells. Molecular dynamics simulations ( Figure 5), indicate the placement of compounds on the mitochondrial ribosome in a way that could possibly inhibit the protein synthesis. Both compounds appear to form hydrogen bonds with nucleotides of mitochondrial 16S rRNA of the large ribosome subunit, which could justify the inhibition of mitochondrial protein synthesis.
PA levels in the tumor cells cultured in the presence of the two conjugates showed a remarkable decrease. A possible explanation for this is that the presence of PA-CAM conjugates, combined with the already elevated levels of cell PAs, triggers the activity of SSAT (spermidine/spermine N 1 -acetyltransferase), which is responsible for the first step of PA catabolism. SSAT, being unable to act on the conjugates, would, however, deplete only the endogenous PAs, thereby disrupting PA homeostasis [49]. This hypothesis is strengthened by the observation that PUT levels were not affected, since PUT is not used by SSAT as a substrate [50]. Another suggestion is that PA-CAM conjugates functioned in a competitive manner, lowering the intracellular levels of the PAs.

Conclusions
In conclusion, the novel PA-CAM conjugates, synthesized with the aim of improving the activity of a previously developed SPD-CAM conjugate (3), did not demonstrate improved antibacterial activity, possibly due to their inability to penetrate the bacterial cell membrane. However, one of them, namely compound 4, carrying two additional benzyl groups, proved to be effective against human mesothelioma ZL34 cells, while having a milder effect on Met5A cells. Therefore, this new compound could be considered as a promising antitumor agent. Moreover, the absence of antibacterial activity could render the compound as a therapeutic agent with no unintentional disturbance of the bacterial microflora of the patient, a necessary feature for the maintenance of homeostasis and normal function, which is often a side effect of orally-administrated drugs [51]. Finally, the deletion of the aminopropyl moiety of the lead molecule rendered it inactive either as an antibacterial or an antitumor agent, while the additional modification of the polyamine carbon skeleton did not further modify their limited biological activity.