Statistical models for evaluating formulations.
Purpose: To develop ampicillin-loaded chitosan nanoparticles by modified ionic gelation method for evaluating their antimicrobial activity onto Escherichia coli.
- chitosan nanoparticles
- drug delivery
- Escherichia coli
- ionic gelation
Drug delivery systems are effective for implementing sustained release of many kinds of drugs. Chitosan (CHT), in particular chitosan nanoparticles (CHT-NPs), have been frequently used in drug delivery applications . CHT is the generic name for a family of strongly polycationic derivatives of poly-N-acetyl-D-glucosamine (chitin) found in the exoskeletons of crustaceans such as crabs and shrimps. It can also be found in the cell wall of fungi and bacteria. Structurally, it is a linear polymer of cationic character formed by units of 2-amino-2-deoxy-D-glucose and 2-acetamido-2-deoxy-D-glucose linked by 1–4 bonds . Having a positive charge, CHT is ideal for many drug delivery applications [3, 4]. CHT is biodegradable, non-toxic, non-immunogenic and biocompatible as well as the only naturally occurring polycationic polymer. Along with its derivatives, CHT has received a great deal of attention from the pharmaceutical industry as antimicrobial and antifungal agent [3, 4]. An extensive review of biocompatibility, hydrophilicity, biodegradability and broad spectrum gram-negative/positive antibacterial and anti-fungal effects of chitosan can be found .
CHT-NPs exhibit great drug encapsulation efficiency (EE) and diverse release profiles, adequate for transporting different types of drugs in many environments . By controlling synthesis parameters conveniently (i.e. reagent concentrations, stirring speed), their diameter, surface charge and other properties can be easily modified thereby leading to a versatile delivery vehicle. CHT-NPs can be synthesized using the ion gelation
method with tripolyphosphate (TPP) as cross-linking agent [6, 7]. Ionic cross-linking of CHT is a typical non-covalent interaction, which can be realized by association with negatively charged multivalent ions of TPP. There is a considerable body of literature on the production of CHT-TPP using this method, with many variations concerning the concentrations and ratios of the initial components, of which [1, 7] are seminal works in these matter. Ion gelation allows encapsulation of various compounds, including β-lactams.
Ampicillin is a beta-lactam antibiotic with amino-penicillin skeleton that easily penetrates outer membrane gram-positive/negative bacteria and irreversibly inhibits the transpeptidase enzyme—needed for synthesis of the bacterial cell wall —in the third and final stage of synthesis in binary fission leading cell lysis (bacteriolytic) . Bacterial resistance to antibiotics, including ampicillin, has been observed and investigated for three decades as a serious threat and health crisis . Several mechanisms have been found liking transpeptidase synthesis and activation pathways to bacterial resistance to ampicillin in E. coli. Given that ampicillin drug acts as a broad-spectrum penicillin/antibiotic in treatment of infections caused by gram-positive/negative bacteria, and the broad-spectrum β-lactamases among Enterobacteriaceae  that is evolutionary conserved, finding novel synergistic ways to curb such general and versatile resistance mechanisms is essential for ensuring continued effectiveness of antimicrobial agents .
In this work, ampicillin-loaded CHT-NPs were synthesized and the effectiveness of their antimicrobial activity was evaluated against ampicillin and CHT-NPs on E. coli. A set of synthesis conditions was investigated in terms of their effect upon particle size, morphology and surface charge. After selection of the most viable delivery system from the later analysis, the β-lactamic antibiotic (ampicillin) was encapsulated in order to evaluate the encapsulation efficiency of the CHT-NPs. A release profile was obtained in order to assess their applicability as antimicrobial agents against gram-negative bacteria. The aim of the present study is the development of a simple method of synthesis of nano-carriers capable of transporting and releasing a drug such as ampicillin, ultimately inhibiting growth on E. coli.
2. Materials and methods
CHT with molecular weight between 100,000 and 300,000 (ACROS Organics™), sodium tripolyphosphate and glacial acetic acid were used. Ultra-pure water was obtained using the Milli-Q A10 system (Millipore). Ampicillin sodium salt (Fisher BioReagents) was used as the antimicrobial agent.
2.2. Bacterial strain
E. coli ATCC 25922 was used for this study.
2.3. Synthesis of chitosan nanoparticles
CHT-NPs were produced using a modified ion gelation method. Briefly, CHT was dissolved at 0.5% w/v in 2% v/v acid acetic solution. Sodium tripolyphosphate was dissolved in ultra-pure water to obtain a 0.1% w/v concentration. An ampicillin volume of 1 ml was then added and later 1 ml of the TPP solution was added drop-wise to 1.5 ml of the CHT solution and magnetically stirred for 1 h. The final suspension was centrifuged at 11,000 rpm for 20 min [6, 13]. Three concentrations of CHT (0.1, 0.3 and 0.5% w/v) and five concentrations of TPP (0.005, 0.01, 0.05, 0.1 and 0.3% w/v) were employed in order to determine their optimal ratio in terms of particle size and surface charge.
2.4. Statistical analysis
In order to evaluate the role of synthesis parameters on particle size and surface charge, including random variations in CHT and then TPP concentrations, selection of linear mixed effects models was performed using standard the Akaike information criterion (AIC), the Bayes information criterion (BIC) and the negative 2 log likelihood criterion (−2LL) . The generalized form () of the usual coefficient of determination (R2) was used . A total of five models (including interactions) were evaluated for each response variable, that is, particle size and ζ potential. Calculations were performed using the R Statistical Computing package  with the lmer4 package for linear mixed effects models . The Staterthwaite approximation of p values was computed for the corresponding observable (Table 1) .
|Model||Response variable||Fixed effects|
|4||Size||CHT + TPP concentrations|
|5||Size||CHT + TPP + CHT*TPP concentrations|
|7||ζ potential||CHT concentration|
|8||ζ potential||TPP concentration|
|9||ζ potential||CHT + TPP concentrations|
|10||ζ potential||CHT + TPP + CHT*TPP concentrations|
2.5. Determination of particle size and ζ potential
The determination of particle size (apparent hydrodynamic diameter) was performed by dynamic light scattering (DLS) and surface electric charge using a Zetasizer Malvern Nano SZ-90 particle analyser, reported as either ζ potential or electrophoretic mobility .
2.6. Morphology analysis of CHT-NPs
In order to study the morphology of nanoparticles, topographic images of CHT-NPs were taken on a multi- mode atomic force microscope (AFM) Asylum Research MFP-3D. The AFM probes used for this study were rectangular silicon probes with a nominal spring constant of 40 nN/nm. Similarly, image visualization was carried out in a scanning electron microscope (SEM) Hitachi S-3700 with a 15 nm gold coating on the diluted samples (1/10) using an aluminum base at an acceleration voltage of 15 kV .
2.7. Determination of encapsulation efficiency of loaded CHT-NPs
The encapsulation efficiency (EE) of the CHT-NPs was determined according to the method described in the previous studies . In brief, the nanoparticle suspensions were separated by centrifuging at 11,000 rpm for 20 min and the contents of ampicillin in the supernatants were measured by HPLC-DAD Perkim Elmer. A blank sample of CHT-NPs without ampicillin was obtained but treated similarly as the drug-loaded CHT-NPs. All samples were measured in triplicate. The EE were calculated using.
where F is the free amount of ampicillin in the supernatant, T is total amount of ampicillin.
2.8. Release profile of loaded CHT-NP
Release studies were carried out in PBS (pH 7.4) as follows: 1.5 ml ampicillin-loaded CHT-NPs and 1.5 ml PBS were incubated at 37°C and shaken at 200 rpm. Triplicate samples were analyzed at each time step, between 0 and 24 h. The samples were centrifuged and the concentrations of ampicillin released in the supernatant were determined by HPLC-DAD.
2.9. Determination of antimicrobial activity of loaded CHT-NPs
The spectrophotometrically adjusted inoculum (100 μl) of 104 bacterial cells was added to each well in the sterile flat-bottomed microtiter plate containing the test CHT-NPs. The design of experiments includes duplicated wells of ampicillin-loaded nanoparticles with three different concentrations of ampicillin (5, 10, 20 mg/ml), two wells with ampicillin as growth inhibition control, two wells containing bacterial suspension with CHT-NPs (growth control) and two wells containing only media (background control) were included in this plate. For the case of wells with ampicillin and CHT-NPs, dilutions were halved at each consecutive level in the gradient. Optical densities were measured for 24 h at 37°C using a multi-detection microplate reader Biotek Synergy HT at 600 nm and automatically recorded for each well every 30 min. Turbidimetric growth curves were obtained depending on the changes in the optical density of bacterial growth for each CHT NP sample and the drug-free growth control.
3. Results and discussion
3.1. Particle size and surface charge of loaded CHT-NPs
Figure 1 corresponds to measures of nanoparticle size (a) and ζ potential (b) according to the concentrations indicated above for both CHT and TPP. As TPP concentration increases, diameter decreases at 0.3 and 0.5% w/v CHT concentrations (a). The opposite occurs at a CHT concentration of 0.1% w/v. In the case of ζ potential (b), all CHT concentrations yield decreasing values as TPP concentration increases.
Statistical models revealed that the best alternative for explaining nanoparticle size in terms of CHT and TPP concentrations is model 5, which includes both factors as well as their interaction (Table 2). Both AIC and −2LL have a larger distance from the baseline than in model 3 (size only depending on TPP concentration), being also more significant and with a better fit than it. Figure 1(a) shows that nanoparticle size decreases as TPP concentration increases and, simultaneously, CHT concentration decreases. In that sense, model 5 also indicates that while TPP is the largest driver of nanoparticles diameter (χ2 = 3.536, df = 4, p(χ2) < 0.001), the interaction between CHT and TPP is responsible for the observed change in slope (dotted lines for each line). CHT concentration seems not to have a large role by itself.
A similar analysis was carried out for ζ potential (Figure 1(b)). By the same criteria, model 10 was the best criteria for all AIC, BIC and −2LL simultaneously. While all factors are statistically significant, TPP concentration.
central (χ2 = 6604.98, df = 4, p(χ2) < 0.001): as it increases, ζ potential decreases thanks to a lower amount of available binding sites in the CHT matrix (Table 3).
Finally, the most significant levels for both size (diameter in nm) and ζ potential that include CHT and TPP concentrations were identified (p < 0.001) as candidates (Table 4). In agreement with existing literature, nanoparticles with a large ζ potential (ζ ≥ 40 mV) are desirable due to their good stability [20, 21] having sizes in between 200 and 580 nm [5, 22] for the particular case of CHT-NPs. From Figure 2, it is clear that levels 1 and 2 reported in Table 4 (matching Samples M and N in Table 5) comply with these requirements. The choice of sample N maximizes the value for the ζ potential while preserving a small nanoparticle size within the range mentioned above.
|1||CHT 0.5%w/v, TPP 0.05%w/v|
|2||CHT 0.5%w/v, TPP 0.1%w/v|
|3||CHT 0.5%w/v, TPP 0.3%w/v|
|A||0.1||0.005||323.57 ± 186.84||60.23 ± 2.85|
|B||0.1||0.01||565.60 ± 97.49||59.67 ± 0.92|
|C||0.1||0.05||133.60 ± 18.34||48.37 ± 0.31|
|D||0.1||0.1||347.43 ± 17.60||10.73 ± 0.59|
|E||0.1||0.3||492.63 ± 16.35||2.92 ± 0.16|
|F||0.3||0.005||331.73 ± 168.06||57.17 ± 2.28|
|G||0.3||0.01||562.73 ± 155.08||57.33 ± 2.06|
|H||0.3||0.05||241.90 ± 34.51||47.30 ± 0.53|
|I||0.3||0.1||239.30 ± 27.14||51.73 ± 1.25|
|J||0.3||0.3||214.93 ± 11.40||9.53 ± 0.40|
|K||0.5||0.005||906.97 ± 264.60||54.63 ± 2.26|
|L||0.5||0.01||498.30 ± 49.73||56.27 ± 0.95|
|M||0.5||0.05||90.88 ± 30.90||52.63 ± 2.53|
|N||0.5||0.1||213.50 ± 29.76||53.73 ± 3.33|
|O||0.5||0.3||130.56 ± 15.65||20.03 ± 1.46|
3.2. Morphology analysis of loaded CHT-NPs
A batch of CHT-NPs without cargo was synthesized and visualized using AFM imaging according to sample preparation N (Figure 3). Scan areas are (A) 5 μm × 5 μm and (B) 1 μm × 1 μm respectively for 3A and 3B. The distribution of nanoparticle diameters is reported in Figure 4 after post-processing of the AFM image.
Additionally, SEM images were performed upon samples with and without ampicillin cargo (Figure 5). Images 5A, 5B and 5C are from a CHT-NPs sample. It verifies that CHT-NPs have a dispersed, corrugated and spherical morphology with a diameter between 100 and 200 nm. Complementary, images 5D, 5E and 5F belong to 10 mg/ml ampicillin-loaded CHT-NPs sample with an EE more than 40% (Figure 6) these samples have a smooth spherical morphology with a diameter between 500 and 1000 nm, showing an aggregation effect between nanoparticles.
3.3. Encapsulation efficiency of loaded CHT-NPs
For encapsulation efficiency (Figure 6) the initial ampicillin concentration was compared against final encapsulated concentration (Figure 6(a)) and later transformed into EE (Figure 6(b)). EE lies within a range of 15–41% with a peak value at a concentration of ampicillin of 4 mg/ml in the final CHT-NPs solution volume.
3.4. Release profile of loaded CHT-NP
A release profile was obtained for the ampicillin-loaded CHT-NPs (Figure 7). Released percentage was calculated in relation to the encapsulated ampicillin concentration. Release percentage oscillates between 5 and 20% across 24 h. The burst effect is clearly observable (between 0 and 2 h), to be later succeeded by a more stable behavior (from 2 to 18 h) and rising finally in the last stage (from 18 to 24 h). The observed pattern suggests that swelling of the first layer in the polymeric matrix releases a large amount of ampicillin in the medium (0–1 h), and becomes more stable until the innermost layers are reached, where the remaining contents are finally released. The latter is consistent with what is described by Carbinato et al.  for cross-linked pectin/high-amylose starch matrices.
3.5. Growth inhibition assays for E. coli ATCC 25922
Figure 8 shows the growth inhibition assays for E. coli ATCC 25922, an ampicillin−susceptible reference strain. Under the conditions of this assay, this strain shows a minimal inhibitory concentration (MIC) of 2 μg/ml, which is in agreement with the range values (2–8 μg/ml) previously reported . A clinical E. coli isolate is considered as non-susceptible when its MIC value is >8 μg/ml. The highest ampicillin concentration used as control was 500 μg/ml (dilution 1, Figure 8), generating a concentration gradient by double dilutions through 3.91 μg/ml (dilution 8, Figure 8). The three ampicillin-loaded CHT-NPs systems (C, D and E) show a similar growth inhibition pattern with respect to the positive control (ampicillin), indicating that under these conditions ampicillin is released from CHT − NPs to concentrations high enough to inhibit bacterial growth. Moreover, both CHT − NPs systems D and E, loaded with 10 mg/ml and 20 mg/ml of ampicillin respectively, maintain growth inhibition with more dilutions with respect to ampicillin. It is worth noting that inhibition in samples A and C lower their effectiveness rapidly after the fifth dilution, but samples D and E have a much wider range of inhibition at lower concentrations for all remaining dilutions.
CHT-NPs synthesis was successfully achieved and the manipulation of CHT and TPP concentrations towards optimization for drug delivery resulted in smaller particles and increased surface charge. Besides, a detailed physicochemical characterization of CHT-NPs was obtained. CHT-NPs were able to encapsulate and release ampicillin. The encapsulation efficiency of ampicillin CHT-NPs exceeds 30% in most cases while antibiotic release maintains a relatively stable profile, ranging between 5 and 20% in a 24-hour period. A microbiological assay was used as proof of principle in order to verify the release of the antibiotic ampicillin from the CHT-NPs systems. Growth inhibition of the ampicillin-susceptible E. coli ATCC 25922 reference strain was achieved in a similar pattern compared with ampicillin alone.
In terms of the obtained release profile in a period of 24 h, several comparisons can be drawn in terms of other alternatives. The cumulative release profile, computed from data shown in Figure 7, indicates that less than 40% of the cargo was released after 24 h. Ampicillin-loaded electrospun poly(ε-caprolactone) nanofiber yarns have a much faster release profile, releasing more than 90% of its cargo in the same period . By contrast, our formulation has a release profile faster than that reported for ampicillin-conjugated gum Arabic microspheres . In the later cases, no additional antibiotic or bacteriostatic effect is present in the polymeric matrix. CHT-NPs loaded with ampicillin reported here are comparable to chitosan microgranules loaded with diclofenac sodium in contrast to chitosan beads loaded with diclofenac sodium in terms of the order of magnitude of released antibiotic . Compared against ampicillin loaded methylpyrrolidinone chitosan and chitosan microspheres, our nanoparticles have a significantly lower encapsulation efficiency , yet the former are not suitable for sustained release. CHT-NPs are similar in size compared to existing literature on ampicillin-loaded nanoparticles and liposomes , with a release profile closer to that of liposomes. In the later study, liposomes had a larger inhibition halo at lower dilutions than nanoparticles of higher density and longer release time.
Finally, the synthesis protocol of CHT-NPs elaborated in this study constitutes a platform for the analysis of the encapsulation of other antibiotics with different structures as well as for assays on other bacteria.
The authors wish to acknowledge funding and support from the National Center for Advanced Technology Studies (CeNAT) and the National Council of Rectors for the period comprehended between 2012 and 2015.
Conflict of interest
The authors have no conflict of interest to declare.