Introduction

Teicoplanin is a member of the glycopeptide antibiotics family, such as vancomycin, as the ‘last resort of defence’ drug to treat severe infections of Gram-positive bacteria including methicillin-resistant Staphylococcus aureus (MRSA) and enterococci1. Its chemical structure was first determined in 19842,3. Figure 1 shows that the core aglycone structure of teicoplanin is a linear heptapeptide, combined with three monosaccharide residues: α-D-mannose, N-acetyl-β-D-glucosamine, and one of five subtypes of N-acyl-β-D-glucosamines depending on teicoplanin A2-1 through A2-54. The mixture of five subtypes of teicoplanin is produced by Actinoplanes teicomyceticus5 and is generally administered as a single product in clinical practice6. The prominent difference with vancomycin is the presence of a long fatty acid chain attached to the (N-acyl-) β-D-glucosamine residue, compared with vancomycin bearing a non-acylated disaccharide residue7. Both glycopeptide antibiotics show antibacterial activity in the same manner. Namely, they bind to bacterial membranous protein (Lipid II) preventing this peptidoglycan precursor from insertion into the bacterial cell wall8. However, to increase their affinity for the peptidyl D-Ala-D-Ala motif of Lipid II, vancomycin cooperatively dimerises in a back-to-back manner, while teicoplanin shows a much greater tendency to self-associate: in the concentrations 0–10 mg/mL teicoplanin A2 was found to self-associate plateauing > 1 mg/mL to give a molar mass of (35,400 ± 1000) g/mol corresponding to ~ (19 ± 1) mers, with a sedimentation coefficient9 s20, w =  ~ 4.65 S.

Figure 1
figure 1

Structure of teicoplanin (adapted from the National Institute for Health/ National Centre of Biotechnology Information10 based on an original structure given by F. Parenti11) and teicoplanin lipoform A2-2 (M1 = 1879.7 g/mol) with the other major types of A2 subtypes with different acyl chains are shown in the inset.

Full size image

Bacterial infection of the eyes ranges from relatively easily treated conditions, such as conjunctivitis and blepharitis12, to more serious ones including keratitis13, and notoriously endophthalmitis14. The significance of MRSA strains for ocular infections is on the increase. For example, Harford et al.15 reported that in the U.K. the percentage of MRSA-positive cases from eye swabs was 2% on average in 2013–2019. Those cases are rising in the U.S.16 and South India17.

Therefore, the eradication of MRSA strains on ocular surfaces is routinely performed for proven positive patients in pre-operative screening. Those patients receive chloramphenicol drops before eye surgery and intracameral vancomycin intraoperatively15. Vancomycin eye drops (50 mg/mL) are also used for the treatment of MRSA-positive keratitis18. On the other hand, teicoplanin drops are reported to have no or little corneal penetration in rabbits19 and in patients at 10 mg/mL20. For orbital cellulitis, intravenous teicoplanin can be applied to patients with penicillin allergy, along with oral ciprofloxacin and metronidazole21.

However, the emergence of further resistant species against these glycopeptide antibiotics becomes a burning issue for antibiotic selection in ocular infections. Vancomycin- and teicoplanin-resistant enterococcal species were reported in the U.S.22 and Europe23,24 before the 1990s. Resistance gene clusters are transferred between bacterial cells through plasmids6. Glycan antibiotics attach to the D-Ala-D-Ala motif of Lipid II through five hydrogen bonds, though these resistance operons modify that targeted motif into D-Ala-D-Lac for vanA, vanB, vanD, vanF, and vanM, and D-Ala-D-Ser for vanC, vanE, vanG, vanL, and vanN25. This modification reduces the number of hydrogen bonds and Lipid II affinity. Overuse and misuse of antibiotics are considered major causes of antimicrobial resistance26. This is due to excessive exposure of bacterial populations to antibiotics. These human activities lead to selection pressure, increasing resistance genes in microbes27.

Ocular mucins may also contribute to antimicrobial resistance (AMR). There is no doubt that the mucin layer of tear films protects against bacterial adherence14. It is reported that ocular mucus, as well as bovine submaxillary mucin (BSM), inhibited the adherence of Pseudomonas aeruginosa on the rabbit corneal epithelium28. However, it is also reported that antibiotics bind to mucins of intestinal29 and respiratory30 tracts, showing a substantial reduction of their microbial activity31. Samad et al.32 suggested that mucin glycoproteins (MUC5AC, MUC2, and MUC5B) interacted with two antibiotics against gram-positive P. aeruginosa (polymyxin and fluoroquinolone), increasing its growth after exposure to antibiotics. Dinu et al.33 reported that gastrointestinal mucins and BSM induced aggregation with vancomycin. Therefore, mucin components of tear films might also exacerbate AMR through their interactions with antibiotics.

In this study, we use commercially available BSM as a popularly used model ocular mucin to explore potential interaction and aggregation behaviour with teicoplanin. BSM – which in common with ocular mucin has a lower degree of glycosylation compared with respiratory, reproductive and alimentary tract mucins—has been widely used as a tear film model for the evaluation of contact lenses34,35,36 and for the interactions with polysaccharides37,38 and proteins39. Setälä et al.39 reported the validity of BSM as an ocular mucin model for in vitro interaction experiments, pointing out that both ocular mucins and BSM interacted with phospholipid transfer protein. In contrast, it should be noted that BSM may not necessarily behave like ocular mucins, such as MUC5AC, the most abundant and gel-forming mucin since commercially available BSM is not a gel-forming mucin40. Furthermore, Rivera and Tessarollo41 warned against dependence on a single animal model to extrapolate its findings to human pathophysiology, such as human carcinogenesis42 and inflammatory diseases43,44. However, it is worth conducting in vitro interaction experiments with BSM as a substitute for whole mucin components of tear films before sufficiently collecting and purifying human ocular mucins.

This interaction study is based on a relatively novel combination of hydrodynamic and microscopic methods. Sedimentation velocity in the analytical ultracentrifuge (SV-AUC) is the gold standard method used to evaluate the integrity of various macromolecules, such as glycoproteins45. SV-AUC provides sedimentation coefficients and distributions relating to molecular sizes46. Dynamic light scattering (DLS) also provides macromolecular sizes complementary to SV-AUC data47. The hydrodynamic results can be confirmed by visualising macromolecular aggregates from samples. Atomic force microscopy (AFM) can detect morphological dynamic changes on a nanometre scale48. AFM is a powerful tool for visualising nanostructure and has been used in studies with ocular mucin aggregation49 and gastric mucin-chitosan interactions50.

The combination of hydrodynamic and microscopic methods was first used by Dinu et al.33 to assess the aggregation of vancomycin with mucins, such as BSM. Many studies have been designed to focus on how glycan antibiotics bind to Lipid II51 and the genomic profiles of resistant microbes32, though the study on environmental factors including mucin-antibiotic binding is of great importance. The hydrodynamic and microscopic methods will also evaluate teicoplanin regarding the degree of interactions with BSM as an outer mucus model.

Methods

Teicoplanin and Bovine submaxillary mucin, BSM

Teicoplanin A2 powder (the mixture of teicoplanin A2-1 with monomer molar mass M1 = 1877.6 g/mol, teicoplanin A2-2 and A2-3 with M1 = 1879.7 g/mol, and teicoplanin A2-4 and A2-5 with M1 = 1893.7 g/mol) was obtained from Sigma-Aldrich, the United Kingdom. A refractive increment dn/dc of 0.188 mL/g was used52. The stock solution concentration was then measured with a differential refractometer (Atago DD7, Tokyo, Japan). The final concentrations (0.125 mg/mL, 1.25 mg/mL, and 12.5 mg/mL) of teicoplanin A2 were prepared in a phosphate-chloride buffered saline solution (PBS or “Paley buffer”) at pH ~ 6.8, ionic strength of I = 0.1 mol/L53.

BSM (Sigma-Aldrich, U.K., catalogue no. M3895, type I-S) was purchased and dissolved in the previously described PBS buffer. A refractive increment dn/dc of 0.181 mL/g for BSM54 was used. The stock solution concentration was then determined with the differential refractometer (Atago DD7, Tokyo, Japan).

Sedimentation velocity in the analytical ultracentrifuge (SV-AUC)

SV-AUC experiments were conducted at 20.0 °C with the Optimal XL-I analytical ultracentrifuge (Beckman, Palo Alto, U.S.A.) coupled to Rayleigh interference optics. Reference solvent (PBS) of 420 μL and sample solutions (teicoplanin A2, BSM, and mixtures of teicoplanin A2 and BSM) of 400 μL were injected into the channels of 12 mm double-sectored cells with sapphire windows and rotated at 47,500 rpm for a run time of ~ 24 h until the specimen completely sedimented. The data was then obtained by the interference system to monitor changes in the concentration of samples in fringe units over radial displacement. This involves the acquisition of multiple (> 50) radial scans as a function of time from which accurate sedimentation coefficient distributions are obtained with the SEDFIT algorithm55.

This algorithm produces by the least squares ls-g*(s) method the sedimentation coefficient distribution, g(s) versus sT,b, where s is the sedimentation coefficient at temperature T and in buffer b. The value of sT,b, given by the unit of Svedberg (S) = 10–13 s, was normalised to standard conditions with viscosity and density of water solvent at 20.0 °C, \({s}_{20,w}\), using the equation 56:

$${s}_{20,w}=\frac{1 - \overline{\upsilon }\cdot {\rho }_{20,w}}{1 - \overline{\upsilon }\cdot {\rho }_{T,b}}\cdot \frac{{\eta }_{T,b}}{{\eta }_{20,w}}\cdot {s}_{T,b}$$
(1)

where \({\rho }_{T,b}\) and \({\eta }_{T,b}\) are the density and the viscosity of buffer b at temperature T, respectively. The following sets of samples were used: 1 mg/mL BSM control (high enough to give a good signal but still clear of molecular overlap), teicoplanin A2 control (0.125 mg/mL, 1.25 mg/mL, and 12.5 mg/mL), and the mixtures (1 mg/mL BSM + 0.125 mg/mL teicoplanin A2, 1 mg/mL BSM + 1.25 mg/mL teicoplanin A2, and 1 mg/mL BSM + 12.5 mg/mL teicoplanin A2).

Dynamic light scattering (DLS)

DLS experiments were performed using a fixed scattering angle Zetasizer Nano-S system (Malvern Panalytical Ltd., Malvern, U.K.) coupled to a 4 mV He–Ne laser at 632.8 nm57,58. Sample measurements were performed in a 1 cm square cuvette at 20.0 °C using a scattering angle of 173°. The data were obtained and analysed with the “Zetasizer Software (Version 7.1)” (Malvern Panalytical Ltd, Malvern, U.K.). For each sample analysed multiple autocorrelation profiles were obtained for 10 scans per 3 replicates averaged over 10 min. The CONTIN programme59 gives volume distributions of translational diffusion coefficients, Dz. The z-average hydrodynamic radii, rz, (nm) were then determined from Dz using the Stokes–Einstein equation57:

$${r}_{z}=\frac{{k}_{B}T}{6\pi \eta {D}_{z}}$$
(2)

where kB is the Boltzmann constant. In these experiments, non-ideality effects were assumed to be insignificant because the sample solutions are sufficiently dilute and sample sizes are small. Therefore, an extrapolation to zero concentration is unnecessary. Additionally, for translational diffusion, non-ideality is related to the two major factors in the hydrodynamic and thermodynamic terms, though these factors can compensate for and thus cancel each other60,61. Moreover, the teicoplanin samples–whether monomeric or multimeric forms in solution–were assumed not to be asymmetric. Therefore, the measured values of Dz were independent of the angle and extrapolation to zero angle to removed rotational diffusion/ anisotropic effects was unnecessary57.

Atomic force microscopy (AFM) analysis

A Dimension ICON (Bruker Nano, Santa Barbara, CA, USA) using dedicated software (Nanoscope 9.4) was used to image 3 independent 500 nm × 500 nm areas of teicoplanin (1.25 mg/mL), BSM (1 mg/mL), and teicoplanin-BSM at increasing teicoplanin concentrations of 0.125 mg/mL, 1.25 mg/mL, and 12.5 mg/mL. All specimens were prepared by depositing 10 µL on cleaved mica and left to air dry at room temperature for 24 h before imaging. RTESPA-150 (Bruker Nano, Santa Barbara, CA) cantilevers were used across all samples, imaging in Tapping mode™ operating at a resonance frequency of 150 kHz in air at room temperature. Particle size analysis was performed using Nanoscope Analysis Version 1.9 monitoring height (nm) and diameter (nm) across all analysed samples. Statistical analysis between teicoplanin and BSM-teicoplanin at increasing teicoplanin concentrations was performed using a series of Kruskal–Wallis ANOVAs.

Results

Analytical Ultracentrifugation (AUC) of teicoplanin-BSM solutions

Figure 2 shows the sedimentation coefficient distributions for the interactions of teicoplanin with BSM 1 mg/mL. The values of \({s}_{20,w}\) for the BSM control was ~4.6S, and teicoplanin controls of 0.125 mg/mL (Fig. 2a), 1.25 mg/mL (Fig 2b), and 12.5 mg/mL (Fig. 2c) are ~0.9S (0.125 mg/mL teicoplanin), ~4.5S (1.25 mg/mL teicoplanin) and ~3.3S (12.5 mg/mL teicoplanin). For the mixture of BSM and 0.125 mg/mL teicoplanin (Fig. 2a), there was no clear interaction between each component (\({s}_{20,w}\) of the mixture was ~ 5.4S).

Figure 2
figure 2

Sedimentation coefficient distributions of mixtures of 1 mg/mL BSM with (a) 0.125 mg/mL teicoplanin (b) 1.25 mg/mL teicoplanin, and (c) 12.5 mg/mL teicoplanin (c). Solutions in phosphate-chloride buffer pH 6.8, I = 0.10. Rotor speed 47,500 rpm, at 20.0 °C.

Full size image

On the other hand, there were clearer shifts for BSM with 1.25 mg/mL teicoplanin (\({s}_{20,w}\) of the mixture ~5.9S), and for BSM with 12.5 mg/mL teicoplanin (\({s}_{20,w}\) of the mixture was ~6.3S). Furthermore, some aggregates emerged over far higher values (20–40S) under the presence of the mucin. For Fig. 2b, there was a loss of the teicoplanin peak at ~0.9S (teicoplanin unimer) in the mixture, suggesting that all teicoplanin unimers were exhausted by complexation with BSM.

Dynamic light scattering (DLS)

Further evidence of BSM-teicoplanin aggregation is shown in Fig. 3 providing the distributions of apparent hydrodynamic radii (after transformation from the apparent translational diffusion coefficients through the Stokes–Einstein equation, Eq. (2)). Distributions were obtained from software based on the CONTIN algorithm of Provencher59.

Figure 3
figure 3

The distributions of z-average apparent hydrodynamic radii, rz from DLS measurements on 1 mg/mL BSM with teicoplanin 0.125 mg/mL, 1.25 mg/mL, and 12.5 mg/mL. The different colours denote each sample: teicoplanin controls (0.125 mg/mL for the black line, 1.25 mg/mL for the red line, and 12.5 mg/mL for the green line), BSM control (violet line), the mixture of BSM and 0.125 mg/mL (sky blue line), the mixture of BSM and 1.25 mg/mL (purple line), and the mixture of BSM and 12.5 mg/mL (dark yellow line).

Full size image

There was a partial loss of mucin components (observed at ~ 30 nm and 3000 nm in the control) in the mixture of BSM and 1.25 mg/mL teicoplanin. There was also a complete loss of these two components in the mixture of BSM with 12.5 mg/mL teicoplanin, and new aggregates emerge at 300 nm. These results from DLS experiments appear to reinforce the observations of SV-AUC in terms of the interaction/ aggregation behaviour of the mixtures.

The apparently higher proportion of aggregate appearing in DLS is caused by the disproportionate higher scattering of larger particles, compared to smaller ones: this does not affect the sedimentation velocity in the same way, where the optical detection system is based on interference from refractive index differences between solvent and solution, and it gives a much more effective separation of particles of different size.

AFM Imaging of teicoplanin-BSM aggregates

Further evidence comes from atomic force microscopy. Figure 4A–C (and Supplementary Information) shows example 500 nm × 500 nm AFM images of teicoplanin A2 samples at increasing concentrations of 0.125 mg/mL, 1.25 mg/mL, and 12.5 mg/mL, with BSM 1 mg/ml shown in Fig. 4D. Figure 4E–G shows BSM and teicoplanin mixtures at increasing concentrations of teicoplanin at 0.125 mg/mL, 1.25 mg/mL and 12.5 mg/mL respectively. It is clear from the AFM imaging an observable increase in particle size is obtained when increasing the concentration of Teicoplanin and Teicoplanin in the BSM mixtures shown in Fig. 4A–G: Table 1 shows the average particle height (nm) and diameter (nm) for each samples, from 3 independent areas across the sample cohort.

Figure 4
figure 4

Atomic Force Microscopy (AFM) images (500 nm x 500 nm) of (A) teicoplanin 0.125 mg/mL, (B) teicoplanin 1.25 mg/ml (C) teicoplanin 12.5 mg/mL, (D) BSM 1 mg/mL, (E) teicoplanin-BSM 0.125 mg/mL, (F) teicoplanin-BSM 1.25 mg/mL and (G) teicoplanin-BSM 12.5 mg/mL.

Full size image
Table 1 Average particle height (nm) and diameter (nm) for all analysed samples and concentrations. *p < 0.05 for comparison between TP (teicoplanin) 0.125 mg/mL, TP 1.25 mg/mL. TP 12.5 mg/mL, BSM-TP 0.125 mg/mL and BSM-TP 1.25 mg/mL.
Full size table

Teicoplanin alone at increasing concentrations of 0.125 mg/mL, 1.25 mg/mL and 12.5 mg/mL exhibited significant increases in particle height from (3.0 ± 0.6) nm to (8.3 ± 1.6) nm (p < 0.05) and (16.2 ± 4.5) nm (p < 0.05) respectively. This was also evident in particle diameter analysis although the only significant increase in diameter for teicoplanin controls was increasing the teicoplanin concentration from 1.25 mg/mL to 12.5 mg/mL (p < 0.05) as shown in Table 1. BSM-teicoplanin mixtures at increasing teicoplanin concentrations of 0.125 mg/mL and 1.25 mg/mL show significant increasing particle heights of (6.6 ± 1.3) nm and (19.5 ± 1.9) nm (p < 0.05) shown Table 1. This was also evident in particle diameter when increasing the concentration of teicoplanin in the BSM-teicoplanin mixtures from 0.125 mg/mL to 1.25 mg/mL, exhibiting (44.1 ± 3.3) nm and (150.5 ± 30.3) nm respectively (p < 0.05). At the greatest teicoplanin-BSM concentration, particle sizes could not be calculated as they exceed the size of the analysed image area (more than 500 nm). From the AFM imaging shown in Fig. 4 and particle size analysis in Table 1, it is evident aggregation is occurring with increasing teicoplanin concentration both alone and in the presence of BSM, consequently increasing particle size properties.

Discussion

Teicoplanin A2 demonstrates clear interactions with BSM at concentrations > 1.25 mg/mL, based on independent orthogonal hydrodynamic and microscopic assessments. Additionally, there are some teicoplanin-BSM aggregates at > 1.25 mg/mL, preventing teicoplanin unimer from maintaining its antibiotic activity in a mucin-rich environment. Interestingly, unlike vancomycin which has a net positive charge under physiological conditions, both teicoplanin A2 and BSM molecules are negatively charged54,62,63. The hydrophobic force might be employed rather than the electrostatic forces with its acyl tail in terms of complexation. The partial interactions and aggregations of teicoplanin with BSM have three consequences, as follows–(1) since teicoplanin A2 uses its hydrophobic tail to allocate itself near the Lipid II precursor as a monomer64, the multimeric formation with mucins inevitably reduces its antimicrobial activity. (2) Also, since aggregates are more likely to be removed and transferred into nasolacrimal ducts by blinking65, total concentrations of teicoplanin decrease on ocular surfaces and as a result more frequent application of eye drops. (3) Most importantly, there is a good possibility that aggregations of teicoplanin with mucins result in longer exposure to antibiotics, leading to more selection pressure towards antimicrobial resistance between bacterial populations. Therefore, teicoplanin eye drops may be preferred to be at least < 1.25 mg/mL or more preferably, around 0.125 mg/mL, where there were no clear interactions and aggregates with BSM.

Concluding remarks

The hydrodynamic data together with a high-resolution microscopic method–AFM, provide clear evidence of interactions between teicoplanin A2 with BSM resulting in the formation of undesirable aggregates, with a concentration-dependent growth in the size of these aggregates. We conclude that the appropriate concentrations of teicoplanin A2 eye drops based on this evidence using BSM as an ocular mucin model would be at least less than 1.25 mg/mL to avoid aggregational behaviours with mucins. The combination of the three independently orthogonal techniques–analytical ultracentrifugation (AUC), dynamic light scattering (DLS) and atomic force microscopy (AFM) has thus proven useful for understanding our model mucin system. AUC and DLS possess an advantage over many other methods for studying interactions of being matrix free and being pure solution techniques—i.e. not requiring separation columns or membranes or interaction with quartz interfaces or tethering to surfaces. AUC possesses the additional advantage of having an inherent separation and analysis facility. AFM permits the visualization of complexes, within the limitation of not being in a solution environment (even if a layer of liquid is placed on top). The three are very much complementary, hence the selection. Nonetheless the use of additional techniques in future work, such as surface plasmon resonance and quartz crystal microbalance measurements would help to elucidate the mechanism of the interaction (e.g. electrostatic interactions, etc.), quantify its kinetics and thus help to further corroborate our hypotheses.

Our experiments have been performed at 20.0 °C: further work at physiological temperatures may prove useful although the ocular surface is likely to be considerably less than 37 °C. More seriously, the validity of BSM as the ocular mucin model–although a good indicative first step–remains open to question for the study of antibiotics in ophthalmology. Many studies have used commercially available BSM as the ocular mucin model34,35,36,39, though there are some differences with ocular mucins, especially gel-forming mucins including MUC5AC. On the other hand, it is also reported that the mucus layer on the ocular surface is not the same as the truly gel-forming layer presented in respiratory and intestinal tracts, due to its relatively thin (1 μm) mucus layer66. Whatever the ocular mucus layer would be, it should be noted that differences with human ocular mucins also need to be explored in terms of the interaction study with teicoplanin. Ocular mucins have been shown to have virtually no surface activity67 and this may lead to reduced aggregation. This current study should therefore be regarded as providing the groundwork for further comparative studies—when evaluated in sufficient quantity—with ocular mucins.