Insights into interspecies protein binding variability using clindamycin as an example

Abstract

Background

In the preclinical development of new drugs, animal models are often employed to predict their efficacy in humans, relying on translational pharmacokinetic/pharmacodynamic (PK/PD) studies. We performed in vitro experiments focusing on the comparison of plasma protein binding (PPB) and bacterial growth dynamics of clindamycin, a commonly used antimicrobial agent, across a range of drug concentrations and plasma environments.

Methods

Human, bovine and rat plasma were used for determining PPB of clindamycin at various antibiotic concentrations in buffer and media containing 20% to 70% plasma or pure plasma using ultrafiltration (UF) and equilibrium dialysis (ED). Also bacterial growth and time–kill assays were performed in Mueller–Hinton broth (MHB) containing various percentages of plasma.

Results

Protein binding of clindamycin correlated well between UF and ED. Notably, clindamycin exhibited substantially lower protein binding to rat plasma compared with human and bovine plasma. Staphylococcus aureus growth was significantly reduced in 70% human, bovine, and rat plasma after 4, 8 and 24 h compared with standard MHB. Time–kill data demonstrated that bacterial counts at both 20% and 70% plasmas were less when compared with MHB at drug concentrations lower than MIC after 4 and 8 h of incubation. For rat plasma, the difference was maintained over 24 h of incubation. Furthermore, a complete bacterial killing at 16 mg/L was observed after 24 h in 20% and 70% human and bovine plasma, but not for rat plasma.

Conclusions

Recognizing interspecies differences in PB might be essential for optimizing the translational relevance of preclinical studies.

Introduction

During the preclinical development of new drugs, animal models are used to forecast their effectiveness in humans, relying heavily on translational pharmacokinetic/pharmacodynamic (PK/PD) studies.¹ Animal models play a pivotal role in the development of new drugs for several reasons as they provide a valuable platform for predicting how humans may respond to new drugs.² As crucial components of the drug development pipeline they complement other research methods such as in vitro studies and computational modelling.² However, differences in drug PK/PD among species can significantly affect potency and safety, leading to differences in therapeutic outcomes.³ To effectively translate preclinical findings into clinical practice, it is crucial to understand and account for interspecies variations in physiological processes, since differences may impact drug metabolism, distribution and absorption.

It has been demonstrated that protein binding (PB) can impact antimicrobial activity.^4–7 Albumin, alpha-1-acid glycoprotein (AAG) and lipoproteins are considered the most relevant plasma proteins in terms of drug binding.⁸ Generally, AAG values in rodents (0.1–0.3 g/L) are lower compared with human (0.5 to 1.0 g/L).^9–12 The saturation of AAG binding sites can contribute to an increase in unbound concentration with rising total concentration.¹³ Albumin serves as the primary binding site for β-lactam antibiotics and fluoroquinolones, though they may also bind to other serum proteins.¹⁴ Therefore, accurately measuring unbound antibiotic concentrations is essential. A previous study focused on the interspecies PK/PD investigation of cefazolin, a first-generation β-lactam antibiotic used for the treatment of several bacterial infections, where we examined the drug’s plasma protein binding (PPB), its in vitro susceptibility against Escherichia coli, bacterial growth in different media, and time–kill curves under various plasma conditions.¹⁵ In view of the importance of the subject, in this study we considered a drug with different antimicrobial spectrum and chemical properties.

Clindamycin is employed in the treatment of infections with Gram-positive pathogens, including Streptococcus and Staphylococcus species, as well as anaerobic bacteria.^16,17 It is approved for treating severe bacterial infections, such as intra-abdominal infections, septicaemia, lower respiratory tract infections, bone and joint infections, and skin and soft tissue infections, but is also under investigation for its effectiveness in combination with other agents to combat resistant pathogens including MRSA.¹⁸ The unique mechanism of action of clindamycin involves inhibiting bacterial protein synthesis by binding to the 50S subunit of the bacterial ribosome.¹⁹ Clindamycin, like many drugs, binds to plasma proteins such as albumin and AAG.²⁰ Interspecies variation in protein binding may arise due to several factors including differences and concentration of individual plasma proteins.^15,21 Variations in the amino acid sequences of albumin and AAG between species can affect drug binding affinities.²² Furthermore, glycosylation of proteins like AAG can differ between species, altering the binding properties of drugs.²³ As a consequence of species differences in the quality and levels of albumin and AAG the fraction of the drug that is bound can vary, thereby impacting PK.²¹ Several other factors present in plasma can potentially impact bacterial growth and the results of experiments. Plasma contains various ions such as sodium, potassium, calcium, magnesium, chloride and bicarbonate. These ions can affect bacterial growth by influencing osmotic balance, enzyme activity and other physiological processes. Moreover, plasma contains nutrients such as glucose, amino acids, lipids and vitamins that bacteria can utilize for growth. Variations in these nutrients between experimental conditions can affect bacterial metabolism and growth rates. There are also antimicrobial proteins and peptides (AMPs) such as defensins, cathelicidins and lysozyme in the plasma that can have direct bactericidal or bacteriostatic effects on various bacterial species.^5,24

Despite the clinical success of clindamycin, its PK/PD intricacies, especially regarding PB, remain subjects of ongoing investigation. Considering the importance of understanding these factors in optimizing dosing strategies and improving treatment outcomes, our study investigated cross-species dynamics of clindamycin PB and bacterial killing of clindamycin in multispecies plasma in vitro.

Materials and methods

Antibiotic and plasma

Clindamycin was procured from Sigma-Aldrich and prepared/preserved throughout the experiments in accordance with the manufacturer’s instructions. Human plasma was commercially purchased from Octapharma under the Octaplas product name. Octaplas^® is a pharmaceutical-grade human plasma for infusion, offering standardized quality and exceptional safety profiles. Rat plasma was obtained from Sigma-Aldrich (Germany). Bovine plasma was supplied by the Veterinary Department of the Medical University of Vienna.

Quantification of PPB of clindamycin through equilibrium dialysis and ultrafiltration

We conducted equilibrium dialysis (ED) and ultrafiltration (UF) to elucidate the PPB of clindamycin in human, rat and bovine plasmas. Our experiments used 20%, 70% and 100% concentrations of human, rat and bovine plasmas for PPB assessments. Clindamycin concentrations of 20, 5 and 2 mg/L were chosen to mirror typical serum levels following standard human doses. Plasma samples from human, rat and bovine sources were stored at −20°C, thawed and centrifuged at approximately 2000 rpm for 5 min to remove any precipitates. Subsequently, the pH was adjusted to 7.4 by carefully adding NaH₂PO₄. A 1 mL sample of clindamycin-spiked plasma was prepared by directly diluting the clindamycin stock solution into the plasma. This mixture was introduced into one compartment of an ED chamber [membrane material: regenerated cellulose 8000 MWCO (molecular weight cutoff), membrane width: 10 mm, insert dimensions: 33.8 mm × 16.6 mm × 7.6 mm], with the other compartment containing 1 mL of 0.01 M PBS. The ED apparatus was immersed in a water bath at 37°C and rotated at 20 rpm, demonstrating equilibrium attainment within 5 h. Upon achieving equilibrium, the contents of the cell chambers were extracted and analysed, with samples taken in triplicate. Non-specific binding, in the absence of plasma, was also determined. The mean recovery of clindamycin was assessed by sampling both dialysis chambers at each clindamycin concentration.

In the case of UF, the studied combinations were subjected to a 30 min incubation at 37°C to facilitate PPB. Subsequently, 250 μL aliquots of each combination were dispensed into a centrifugal filter unit containing a low-binding regenerated cellulose membrane (Centrifree; Millipore ref. 4041, USA) UF unit (MWCO: 30 kD). The ultrafiltration process was carried out through centrifugation at 3500 rpm for 40 min at room temperature. Each UF procedure was conducted in triplicate. The determination of clindamycin concentrations in human, bovine and rat plasma samples, as well as in ultrafiltrate, was conducted as described before.¹⁵

In vitro susceptibility tests

S. aureus (ATCC 29213) was obtained from the ATCC. Columbia agar (Columbia + 5% sheep blood) and MHB broth (Oxoid Ltd, UK) were for bacterial growth. MIC values were determined using the serial broth microdilution method, following CLSI (formerly NCCLS) guidelines.²⁵ Overnight, S. aureus was precultured on Columbia agar plates and then inoculated into MHB at an initial inoculum of ∼5 × 10⁵ cfu/mL. The growth media contained defined concentrations of clindamycin. The lowest concentration of antibiotic that prevented visible bacterial growth after 20 h of incubation at 37°C was defined as the respective MIC.

Bacterial growth curves

The S. aureus strain was cultured overnight on Columbia agar plates and then inoculated into three different media: MHB, MHB containing 20% plasma, and MHB containing 70% plasma. The initial inoculum was set at approximately 5 × 10⁵ cfu/mL. The culture tubes containing 3 mL aliquots were then incubated in a water bath at 37°C. Bacteria counts were performed at 0, 4, 8 and 24 h of incubation. After vortexing each culture tube, two 50 µL samples were removed and diluted serially in 0.9% sodium chloride. After each dilution step, 20 µL was spotted to Columbia agar plates and incubated for 24 h at 37°C. The colonies were then counted and back-extrapolated to determine cfu/mL. Each experiment was repeated at least three times.

Time–kill curves

The impact of various clindamycin concentrations on S. aureus viability was investigated by determining bacterial killing curves. S. aureus was inoculated into MHB broth containing antibiotic concentrations of 0.015, 0.031, 0.062, 0.125, 0.25, 1, 4 and 16 mg/L. The same procedure was repeated with MHB broth containing 20% and 70% bovine plasma to assess the influence of plasma on bacterial sensitivity to clindamycin. For time–kill experiments, 20% and 70% plasma were used as 100% plasma is more viscous than MHB, which can make it more difficult to handle and mix uniformly in experiments. High viscosity can affect the diffusion of oxygen and other molecules, potentially altering bacterial growth rates. Diluting plasma with MHB reduces viscosity and makes the experimental conditions more consistent and manageable. Besides, 100% plasma has a strong impact on bacterial growth. Moreover, by using 20% or 70% plasma, we can explore a range of conditions that mimic different in vivo environments while still maintaining the experimental rigour and control necessary for reproducible and interpretable results.

To establish the bacterial killing curves, S. aureus was inoculated into 3 mL aliquots of MHB or MHB containing 20% or 70% plasma and incubated in a water bath at 37°C for 30 min to allow for plasma binding. After PB, the inoculum of S. aureus was adjusted to approximately 5 × 10⁵ cfu/mL. Samples were collected at 0, 4, 8 and 24 h of incubation, and the bacterial concentration was determined by plating serial dilutions onto Columbia agar plates. Each experiment was repeated at least three times to ensure reproducibility. To eliminate the confounding effect of antibiotic-free growth, medium growth controls were also performed without the addition of clindamycin. The bacterial killing was quantified as the log₁₀ difference in bacterial cfu/mL between the control and clindamycin-treated samples.

Statistical calculations

Statistical analyses were conducted using commercially available software (GraphPad Prism, San Diego, CA, USA; www.graphpad.com). Paired t-test analysis of the means from the human, bovine and rat PPB data at three different concentrations of clindamycin (20, 5 and 2 mg/L) was performed (P value <0.05). Correlation analysis was performed on PPB data, and Spearman’s correlation coefficient, r, was calculated. The Mann–Whitney test was used to compare differences in bacterial growth. A P value <0.05 was considered statistically significant.

Results

Protein binding of clindamycin

Mean protein-binding (±SD) values for clindamycin using UF and ED are depicted in Figure 1(a and b), respectively. Our data demonstrate that PPB values for clindamycin in human and bovine were similar, whereas the binding in rat plasma was lower compared with that in human and bovine (Table S1, available as Supplementary data at JAC Online), independent of the method used. Statistical analysis of the human PPB data at three different concentrations of clindamycin (20, 5 and 2 mg/L) indicated that it was significantly higher than that observed for the rat data when applying either ED or UF (P value <0.05). Similarly, PPB in bovine was found to be significantly higher compared with rat (P value <0.05), whereas no significant difference in PPB was found between human and bovine plasma (20%, 70% and 100%), when applying either ED or UF (P value >0.05). The percentage of clindamycin binding was higher when performed in 70% plasmas compared with that in the 20% plasmas, whereas there was no significant change in PB in 100% plasmas compared with that in 70% plasmas (Table S1).

Figure 1.

Clindamycin plasma protein binding in human (red), bovine (blue), and rat (green) plasmas using (a) ultrafiltration and (b) equilibrium dialysis. The experiments were conducted in triplicate (mean ± SD).

Comparative analysis of ED and UF for assessing bound plasma concentrations of clindamycin

We conducted correlation analysis on the PB data of clindamycin obtained using UF and ED. The correlation was strongest for human plasma (r = 0.9261) and bovine plasma (r = 0.8913) and moderate for rat plasma (r = 0.8565) (Figure 2).

Figure 2.

Comparative analysis of ultrafiltration and equilibrium dialysis using correlation analysis of plasma protein binding percentage data determined in (a) human, (b) bovine and (c) rat.

In vitro susceptibility tests

The MIC value for S. aureus was 0.125 mg/L for clindamycin, which was within the CLSI range.²⁵

Bacterial growth in different media

Growth profiles of S. aureus in MHB and MHB containing 20% and 70% human, rat and bovine plasmas are depicted in Figure S1. Our growth assays in 20% human and bovine plasmas showed significant growth inhibition compared with that in MHB after 4 and 8 h of growth. When we performed growth curve analysis in 70% human, rat and bovine plasmas, we observed a significant decrease in the count of S. aureus compared with MHB after 4, 8 and 24 h of growth (Figure S1).

Time–kill curves

Figure 3 presents bacterial time–kill profiles of S. aureus at exposure to clindamycin at concentrations equal to the 0.015, 0.031, 0.062, 0.125, 0.25, 1, 4 and 16 mg/L in MHB and MHB containing 20% and 70% human, bovine and rat plasmas (Figure 3). Mean differences in bacterial log₁₀ cfu/mL between the initial inoculum and 4, 8 and 24 h after exposure are shown (Table 1). The time–kill data reveal that bacterial counts in both 20% and 70% plasma were reduced compared with those in MHB at drug concentrations below the MIC following 4 and 8 h of incubation. However, this difference disappeared after 24 h for human and bovine plasmas. On the other hand, for rat plasma, the difference persisted beyond the 24 h incubation period. Moreover, complete bacterial elimination at 16 mg/L was noted after 24 h in 20% and 70% human and bovine plasma, but not in rat plasma.

Figure 3.

Time–kill curves of S. aureus ATCC 29213 in pure MHB and in 20% and 70% human, bovine and rat plasmas at 0, 0.015, 0.031, 0.062, 0.125, 0.25, 1, 4 and 16 mg/L clindamycin. Viability was enumerated at the indicated timepoints by serial dilution plating. Black lines indicate cfu data in MHB, whereas red, blue and green represent data in human, bovine and rat plasmas, respectively. All data points shown are mean ± SEM.

Table 1.

Drug response of clindamycin to S. aureus ATCC 29213 grown in pure MHB with 20% human plasma, 20% bovine plasma, 20% rat plasma, 70% human plasma, 70% bovine plasma and 70% rat plasma

Medium	Time, h	Clindamycin concentration, mg/L
		16	4	1	0.25	0.125	0.061	0.031	0.0156	0
		Log10-transformed Δcfu/mL
MHB
	4	−1.48	−0.34	−0.27	−0.20	−0.17	0.02	0.20	1.30	2.19
	8	−1.66	−1.53	−0.44	−0.45	−0.48	0.44	1.14	2.33	2.35
	24	−3.83	−2.64	−1.78	−1.74	−1.85	1.04	1.84	2.20	2.33
20% Human
	4	−1.15	−0.44	−0.06	−0.10	0	0.11	−0.052	−0.06	0.67
	8	−2.14	−1.29	−0.44	−0.40	−0.14	−0.13	−0.18	1.52	2.36
	24	−6.52	−2.37	−1.22	−1.34	−0.86	1.07	2.67	2.67	2.70
20% Bovine
	4	−1.23	−0.35	−0.25	−0.27	−0.22	−0.19	−0.09	−0.18	−0.27
	8	−1.45	−1.07	−0.54	−0.34	−0.33	−0.25	−0.23	0.66	0.68
	24	−6.61	−2.46	−1.55	−1.39	−1.28	−1.23	1.70	2.91	2.92
20% Rat
	4	−1.08	−0.11	0.04	−0.02	−0.09	0.04	0.08	0.06	0.08
	8	−1.15	−0.14	−0.07	−0.96	−1.04	0.04	0.10	1.52	1.91
	24	−2.74	−1.35	−1.10	−0.96	−1.04	0.00	1.77	3.00	3.02
70% Human
	4	−1.36	−1.14	−0.08	0.04	0.03	−0.28	−0.29	−0.27	−0.27
	8	−2.65	−2.16	−0.17	−0.04	−0.55	−0.43	0.48	1.04	1.10
	24	−6.51	−2.69	−1.20	−0.28	−0.16	1.61	2.15	2.37	2.28
70% Bovine
	4	−1.16	−0.35	−0.24	−0.19	−0.11	−0.10	−0.11	−0.16	−0.04
	8	−1.59	−0.68	−0.36	−0.24	−0.11	−0.14	−0.07	1.27	1.25
	24	−6.62	−1.56	−1.20	−0.31	0.72	0.69	1.81	1.85	1.85
70% Rat
	4	0.01	−0.05	0.06	0.06	0.06	0.09	0.08	0.11	0.19
	8	−0.16	−0.21	−0.21	−0.22	−0.01	−0.06	−0.08	−0.25	−0.28
	24	−1.20	−1.27	−1.03	−1.12	−1.03	−1.20	−2.05	−2.50	−2.62

Discussion

Our study focused on the interspecies PK of clindamycin, employing two distinct methodologies, namely, in vitro UF and ED. Comparative analysis between UF and ED techniques for assessing free drug concentrations revealed consistent PB results, with a slightly lower PB observed at 20% plasma concentrations using UF (Figure 1). We further provide a comprehensive examination of clindamycin, focusing on its in vitro susceptibility against S. aureus, bacterial growth in different media, and time–kill curves under various plasma conditions.

PB plays a significant role in the PK/PD of various anti-infective agents, such as clindamycin.^4–6 Previous in vitro experiments have revealed a high degree of PB for clindamycin, with a clear association between this binding and a decrease in antimicrobial activity.²⁶ This reduction in activity is attributed to the hindered uptake of clindamycin by bacteria in the presence of serum. Our results revealed PPB of clindamycin ranging from 47.2% to 93.9% depending on the concentration of human plasma used and the amount of drug. These findings are consistent with the results of Kays et al.²⁷ in 1992, when they reported that PPB of clindamycin was 61.3%–92.4% depending on the concentration of drugs and human serum used. Besides, similar human PPB values were observed in vitro when studies were performed in MHB with different concentrations of clindamycin and albumin.²⁶ The similarity in PB between human and bovine plasma in our experiments, with a lower binding in rat plasma, implies species-specific variations that may influence the drug's distribution and pharmacological activity. Rat plasma contains various components that can contribute to antibacterial activity. These components include a complement system, which is a part of the immune system that enhances the ability of antibodies and phagocytic cells to clear pathogens. Besides, there are antimicrobial peptides, which are small proteins that can disrupt bacterial cell membranes and have been shown to exhibit antibacterial activity.^28–30 Also, proteins like transferrin and lactoferrin may sequester iron, limiting bacterial growth, whereas lysozyme can break down bacterial cell walls.³¹ Although rat plasma contains these general immune components, the specific antibacterial potency may differ depending on the bacterial species and the environment. Furthermore, the higher percentage of clindamycin binding in 70% plasma compared with 20% plasma underscores the importance of considering different plasma compositions in PK studies. Moreover, research on other antibiotics, such fluoroquinolones, has also highlighted the impact of different plasma compositions on drug binding.³²

In the 20% plasma condition, the lower plasma protein concentration results in less drug being bound to proteins, leading to a higher free (unbound) drug concentration. This theoretically should enhance the antimicrobial activity, as more of the drug is available to act on the bacteria. Conversely, in the 70% plasma condition, the higher plasma protein concentration increases drug binding, reducing the free drug concentration. This would typically result in reduced bacterial killing since less of the drug is available in its active form. The 20% plasma condition, being closer to the nutrient-rich MHB, likely supports faster bacterial growth. This could potentially offset the higher free drug concentration by giving bacteria more opportunities to replicate before being killed by the antibiotic. The 70% plasma condition, due to higher protein content and possibly lower nutrient availability, likely slows bacterial growth. Slower bacterial growth often makes bacteria less susceptible to many antibiotics, which target rapidly dividing cells, thus balancing the reduced free drug concentration with a slower growth rate. The observed results suggest a complex interaction between drug binding, bacterial growth rates and the antimicrobial action of the drug. The findings highlight the importance of considering both drug PK (free drug concentration) and bacterial physiology (growth rates) when interpreting in vitro experiments.

Albumin is highly conserved across species, but there are still important structural differences. For instance, human albumin has different binding sites compared with rodent or bovine albumin, leading to variations in drug binding capacity.⁵ Bovine albumin is used commonly in research due to its stability and similarity to human albumin, but it shows lower affinity for some drugs compared with the human variant.³³ Rat albumin may have different secondary and tertiary structural conformations,³⁴ which could affect the binding of specific drugs like clindamycin. AAG is involved in drug transport and immune modulation.³⁵ Human AAG has unique glycosylation patterns that affect its binding to drugs.³⁶ The human version is often more heavily glycosylated than that of other species.³⁷ Bovine AAG may have less complex glycosylation, and differences in the glycan structures can influence drug interactions.²² Rat AAG tends to have a lower molecular weight and a slightly different structure compared with human AAG, which can lead to altered drug-binding profiles.³⁸

Few studies have reported species-specific variations in PB.^15,21,39,40 The present data are widely comparable to our previous study involving in vitro experiments on cefazolin across species, where we investigated different PK/PD aspects including PPB, bacterial growth and time–kill assays. Both studies used UF and ED to determine PPB and conducted bacterial growth and time–kill assays in MHB supplemented with different percentages of plasma. Whereas the cefazolin study noted a significant decrement in drug binding to bovine plasma compared with human plasma, with rat plasma showing a pattern more consistent with human plasma, the current study found that clindamycin exhibits reduced binding affinity to rat plasma compared with human and bovine plasma. Interestingly, both studies observed significant growth inhibition of bacteria in plasma-supplemented media compared with pure MHB, with differences observed across species. Considerable growth inhibition of E. coli was observed in 70% bovine or rat plasma compared with human plasma or pure MHB, whereas there was a substantial growth reduction of S. aureus in 70% human, bovine and rat plasma compared with MHB.

At higher concentrations, clindamycin can exhibit a time-dependent bactericidal action against sensitive strains, inducing bacterial killing with a post-antibiotic effect.^41–43 This means that its effectiveness is related to the duration of time that its concentration remains above the MIC of the target bacteria.⁴⁴ The post-antibiotic effect refers to the persistent suppression of bacterial growth even after the antibiotic concentration falls below the MIC.⁴⁵ The PK/PD target for clindamycin is typically an AUC/MIC ratio.²⁰ This means that achieving a sufficiently high AUC (which is related to both the peak concentration and the duration of exposure) relative to the MIC for the bacteria is crucial for optimal bacterial killing. The frequency of administration and the dose are both important determinants of their antimicrobial activity. Time above the MIC has been found to be an important PK/PD parameter to correlate with efficacy of lincosamides.^42,43 Our results demonstrate that antimicrobial activity is maximum after 24 h of administration.

There are, however, certain limitations of our study. PB exhibits not only inter-individual variations within healthy subjects but also significant divergence based on the specific ailment or the severity of each patient's condition.⁴⁶ Moreover, in vitro studies may not perfectly mimic the physiological conditions in vivo. In addition, given the well-documented role of biofilm formation in S. aureus biology, infection and antimicrobial resistance,^47,48 biofilm studies for the S. aureus strain under different plasma and drug conditions could provide valuable insights. Besides, another limitation of this study was investigating only one drug regarding PB and its effect on bacterial killing. Expanding the scope of research to include additional drugs would offer further insights into the relationship between PB of clindamycin and its reduced antimicrobial activity.

Conclusion

This study highlights the importance of considering species-specific differences of PPB when assessing the PK/PD of antimicrobial agents in preclinical studies, which could have implications for drug development and translational research. Highlighting the differences in protein binding across species underscores the importance of selecting appropriate animal models during preclinical drug development. This emphasizes the need for more comprehensive preclinical studies that account for species-specific PK parameters. Moreover, the observed differences in bacterial growth dynamics in different plasma environments (human, bovine and rat) also hold importance. Although the interspecies differences in protein binding of clindamycin may not be surprising in principle, their specific implications for drug development and clinical translation are significant. Addressing these differences can enhance the accuracy of preclinical studies, improve the predictability of drug efficacy and safety in humans, and ultimately contribute to more effective therapeutic strategies.

Funding

Financial support was provided by the European Union’s Horizon 2020 research and innovation programme (grant agreement no. 861323).

Transparency declarations

None to declare concerning the present work.

Supplementary data

Figure S1 and Table S1 are available as Supplementary data at JAC Online.