Chemokine production induced by Corynebacterium pseudotuberculosis in a murine model

Abstract

Corynebacterium pseudotuberculosis is the etiological agent of caseous lymphadenitis. The main clinical sign of this disease is the development of granulomas, especially in small ruminants; however, the pathways that are involved in the formation and maintenance of these granulomas are unknown. Cytokines and chemokines are responsible for the migration of immune cells to specific sites and tissues; therefore, it is possible that chemokines participate in abscess formation. This study aimed to evaluate the induction of chemokine production by two C. pseudotuberculosis strains in a murine model. A highly pathogenic (VD57) and an attenuated (T1) strain of C. pseudotuberculosis, as well as somatic and secreted antigens derived from these strains, was used to stimulate murine splenocytes. Then, the concentrations of the chemokines CCL-2, CCL-3, CCL-4, and CCL-5 and the cytokines IL-1 and TNF were measured in the culture supernatants. The VD57 strain had a higher ability to stimulate the production of chemokines when compared to T1 strain, especially in the early stages of stimulation, which can have an impact on granuloma formation. The T1 lysate antigen was able to stimulate most of the chemokines studied herein when compared to the other antigenic fractions of both strains. These results indicate that C. pseudotuberculosis is a chemokine production inducer, and the bacterial strains differ in their induction pattern, a situation that can be related to the specific behavior of each strain.

Introduction

Corynebacterium pseudotuberculosis is a Gram-positive facultative intracellular bacterium and the etiologic agent of caseous lymphadenitis (CLA). This disease is a chronic infectious zoonosis that affects small ruminants. These animals, when infected by this bacterium, develop granulomatous lesions in external and/or internal lymph nodes and in some organs, such as the spleen, liver, and lungs. The disease causes economic losses due to leather depreciation, condemnation of viscera and carcasses, weight loss, and decreased milk production [1, 2].

Granulomatous inflammation is a marked histological characteristic of the infections by some intracellular agents, as a consequence of a chronic antigenic stimulation caused by bacterial persistence in phagocytes [3]. In Mycobacterium tuberculosis infection, it has already been verified that chemokines are essential in the formation and maintenance of granulomas, which may isolate the bacteria and prevent its spread, creating an environment that allows an immune response characterized by the activation of macrophages and T CD8 + lymphocytes [4].

Some studies concerning chemokines have been carried out involving M. tuberculosis, which is phylogenetic related to C. pseudotuberculosis and triggers a similar immune response. M. tuberculosis is a strong inducer of chemokine expression, and its cell wall components have a crucial role in this process; cytokines act in a coordinated manner in the formation of the granuloma and in the regulation of β-chemokine production, such as CCL-2, CCL-3, CCL-4, and CCL-5 [5, 6]. This situation indicates that these molecules can modulate the immune response against these bacterial infections [7, 8].

Until now, there are no studies based on the evaluation on the production of chemokines in C. pseudotuberculosis infections, and this knowledge is important because it can base further studies on the development of vaccines against caseous lymphadenitis, since it was already described that a more efficient cellular immune response-enhanced innate immune response against C. pseudotuberculosis can lead to a better prognosis of the disease [9]. In this way, the present work aimed to evaluate the induction of chemokine production by an attenuated and a highly pathogenic strain of C. pseudotuberculosis and its associated antigenic extracts in a murine model.

Materials and methods

Bacterial strains

Two C. pseudotuberculosis strains were selected for this study. The T1 strain was isolated from a naturally infected goat from Bahia state and maintained in the Microbiology Laboratory of the Federal University of Bahia (UFBA). After several passages, the strain was considered to present a low virulence as a consequence of a low intensity of synergistic hemolysis with Rhodococcus equi and an absence of granuloma development when inoculated into goats [10]. The second strain, named VD57, presented high virulence for mice and goats, inducing the development of severe caseous lymphadenitis clinical signs [10, 11]. These two strains belong to the biovar ovis and were chosen because they are C. pseudotuberculosis strains that have their virulence profile well characterized both in mice and in small ruminants, as well as due to the fact that they were molecularly identified and have their genome already sequenced (GenBank identification numbers CP015100.2 for the T1 strain and CP009927.1 for the VD57 strain) [12, 13].

Secreted antigen concentrated by the three-phase partition method (TPP)

With the objective to produce the secreted proteins of both strains, 4 L of culture from each of the strains were used, obtained from cultivation in BHI medium (Sigma-Aldrich, Saint Louis, MO) at 37 °C for 48 h, followed by a protocol for concentration of secreted proteins established by Paule et al. [14]. The samples were stored at − 20 °C until use.

Somatic antigen

The bacterial strains were cultivated in BHI medium at 37 °C for 48 h. Then, 50 mL of the cultures were centrifuged at 3,000 × g for 30 min at 4 °C. The bacteria masses were washed twice with phosphate buffered saline (PBS—137 mM NaCl, 2.7 mM KCl, 8 mM Na₂HPO₄, and 2 mM KH₂PO₄) pH 7.4 (5 parts of PBS for 1 part of the pellet). A sonication step was performed at 60 Hz, using five cycles of 60 s each (Sonifier 450, Branson, Brookfield, CT). The samples were centrifuged at 10,000 × g for 30 min at 4 °C and stored at − 20 °C until use.

Animals and survival curve

Forty BALB/c mice, aged between 4 and 6 weeks, were used in this study. Twenty animals were used to determine the survival curve after a challenge with the bacterial strains. Ten Balb/c mice were inoculated with 1 × 10⁶ CFU/mL of T1 strain, and another ten mice were inoculated with the same amount of the VD57 strain. All inoculations were performed intraperitoneally. The survival rate was monitored during the period of 30 days, with daily inspection with the objective to record deaths.

Splenocyte obtaining

Twenty mice were euthanized by cervical dislocation, washed with 70% ethyl alcohol and then dried with paper towels. They were kept in the supine position in a laminar flow hood and sectioned with sterile dissection materials with the objective to remove the spleen. The spleens were placed in a Petri dish containing 20 mL of RPMI medium (Sigma-Aldrich) and kept at 4 °C. For the obtaining of the splenocytes, the organs were homogenized using a tissue grinder (cell dissociation sieve—tissue grinder kit CD1-1KT, Sigma-Aldrich), and the whole product of the maceration was placed in 50mL centrifuge tubes and centrifugated at 4 °C for 10 min at 500 × g. The supernatant was discarded, and the pellet was resuspended in 1 mL of RPMI. Splenocytes were counted in a Neubauer chamber, the viability was assessed by the Trypan Blue method (Roche, Basel, Switzerland), the cell suspension was diluted to a 1 × 10⁶ cells/mL concentration, and 1 mL of this final cell suspension was added to each well of a 24 well culture plate and cultivated in a CO₂ incubator at 37 °C and 5% CO₂ supplementation.

Stimulation of splenocytes

Each 1 × 10⁶ cells/mL splenocyte suspension (each suspension obtained from one mouse, from a total of 10 mice) was stimulated as described below: negative control (NC), with the addition of sterile PBS only; positive control (PC), stimulated with a solution of Pokeweed mitogen (PWM—Sigma Aldrich) in a concentration of 5 µg/mL; T1 strain, stimulated using concentrations of 10², 10⁴, and 10⁶ CFU/mL of the bacteria; and VD57 strain, stimulated using concentrations of 10², 10⁴, and 10⁶ CFU/mL of the bacteria. All experiments were performed in triplicate, and the supernatant was removed at 24, 48, and 72 h after the stimulation.

Splenocytes obtained from other 10 animals (each splenocyte suspension obtained from one mouse) were removed and stimulated as follows: negative (NC) and positive (PC) controls, as described above; 250 µL of the antigenic solutions: T1 strain secreted antigen; T1 strain somatic antigen; VD57 strain secreted antigen; and VD57 strain somatic antigen. All antigenic solutions had a protein concentration of 80 µg/mL, as determined by the Bicinchoninic Acid method (Thermo Fisher, Waltham, MA). These experiments were also performed in triplicate, and the supernatant was collected at 24, 48, and 72 h after the stimulation. The antigenic solution concentrations were used as described by Rebouças et al. [15].

Determination of chemokine and cytokine concentration in culture supernatants

The supernatants from the splenocyte cultivations with different stimuli were subjected to the quantification of the cytokines IL-1 and TNF, and the chemokines CCL-2, CCL-3, CCL-4, and CCL-5, using commercially available ELISA kits (R&D Systems, Minneapolis, MN), and following the manufacturer’s instructions.

Statistical analysis

The statistical analyses were made using the SPSS v. 20. software (IBM, Armonk, NY). Statistical analysis was performed using the Kruskal–Wallis method. The comparisons were made between the negative and positive control, as well as between the negative control and the stimulations with each concentration of the live strains, with the antigenic fractions and between each equal antigenic preparations of the different strains at different times. The analysis of the correlation between cytokine and chemokine production was performed using the Spearman method. The survival curves of mice experimentally infected with C. pseudotuberculosis strains were compared using the log-rank statistical test.

Results

Survival curve of experimentally infected mice

With the objective to demonstrate the pathogenicity of each strain, a survival curve was performed using twenty Balb/c mice, half of them inoculated with T1 strain and the other half with VD57 strain, and the mice were monitored for 30 days (Fig. 1). Among the animals challenged with the T1 strain, 90% remained alive during the studied period, while only 50% of the mice inoculated with VD57 strain survived, which proves the pathogenicity previously described for each strain [10, 12, 13, 16].

Fig. 1.

Survival curve of C. pseudotuberculosis experimentally infected Balb/c mice. The graph shows the rate of survival animals, using ten mice inoculated with the VD57 strain and ten with the T1 strain, which were monitored for 30 days after the challenge. Survival curves were compared using the log-rank statistical analysis, with a statistical difference set at p < 0.05

Chemokine production

Regarding the production of chemokines by murine splenocytes, the positive control of all cultures showed a statistically significant difference in relation to the negative control (NC), guaranteeing the quality control of the cultivation and stimulation processes (data not shown). The NC occasionally showed slight production of cytokines and chemokines, a situation that is expected as a consequence of cultivation characteristics and is already described in the scientific literature [17].

Considering the production of CCL-2 (Fig. 2), there was a significant production of this chemokine when the cells were stimulated with the VD57 strain within 24 h of incubation (p < 0.05), and the stimulation with the VD57 strain at 10⁴ CFU/mL showed a statistical difference when compared to the production of CCL-2 induced by the T1 strain at the same concentration (Fig. 2a). When analyzing the stimulation period of 48 and 72 h, only the lysed antigen of the T1 strain showed a difference when compared to the NC and showed a significant difference when compared to the stimulation with the VD57 lysate antigen (p < 0.05) (Fig. 2b and c ).

Fig. 2.

Production of CCL-2 after antigenic stimulus of Balb/c mice splenocytes. The data shows the concentrations of CCL-2 after a 24 h, b 48 h, and c 72 h of stimulation. PBS buffer was used as negative control (NC). The (*) within the graph stands for statistical difference between the antigen and NC; (**) for difference between equal antigenic preparations of different strains. The analysis was performed using the Kruskal–Wallis test with p < 0.05

Considering the chemokine CCL-3 (Fig. 3), only the different concentrations of the live VD57 strain induced significant production of this chemokine after 24 h of stimulation (p < 0.05), but the stimulus with the concentration of 10⁴ CFU/mL of this strain did not differ when compared to the T1 strain stimulus (Fig. 3a). When we analyzed the production of CCL-3 48 h after stimulation, there was no significant production after the stimulus with the VD57 and T1 secreted antigens and with the lysate antigen of the T1 strain (p < 0.05) (Fig. 3b). The production of this chemokine 72 h after stimulation was only significant when the cells were stimulated with the live strains, except for the T1 strain at 10⁴ CFU/mL (Fig. 3c).

Fig. 3.

Production of CCL-3 after antigenic stimulus of Balb/c mice splenocytes. The data shows the concentrations of CCL-3 after a 24 h, b 48 h, and c 72 h of stimulation. PBS buffer was used as negative control (NC). The (*) within the graph stands for statistical difference between the antigen and NC; (**) for difference between equal antigenic preparations of different strains. The analysis was performed using the Kruskal–Wallis test with p < 0.05

Figure 4 shows the production of CCL-4 at the three analyzed incubation times. This chemokine was only significantly produced when the cells were stimulated with the live strains, being only produced by the stimulus with the VD57 strain within 24 and 72 h of incubation (Fig. 4a and c ) and by both strains, except for T1 strain at 10⁴ CFU/mL, at 48 h (Fig. 4b) (p < 0.05). A difference between the stimulation with the same antigen concentration for both strains was observed at 24 h for all the live strains (Fig. 4a) and at 72 h only for the highest concentration (Fig. 4c) (p < 0.05).

Fig. 4.

Production of CCL-4 after antigenic stimulus of Balb/c mice splenocytes. The data shows the concentrations of CCL-4 after a 24 h, b 48 h, and c 72 h of stimulation. PBS buffer was used as negative control (NC). The (*) within the graph stands for statistical difference between the antigen and NC; (**) for difference between equal antigenic preparations of different strains. The analysis was performed using the Kruskal–Wallis test with p < 0.05

The results for the last chemokine herein studied are shown in Fig. 5. The production of CCL-5 was mainly stimulated by the three concentrations of the live VD57 strain, but also showed significant production after the stimulation with the higher concentration of the live T1 strain and its lysate antigen (Fig. 5a). After 48 h of the antigenic stimulation, the live T1 strain at the concentration of 10² CFU/mL induced a significant production of CCL-5, but its lysate antigen did not present significant results (Fig. 5b). Considering the last stimulation time, all concentrations of live strains showed significantly higher results than the NC, and again the T1 lysate antigen presented significant results (Fig. 5c) (p < 0.05). Differences between the stimulations with the same antigen concentration for both strains were observed at 24 h of stimulation for the live strains at 10⁴ and 10⁶ CFU/mL (Fig. 5a) and at 72 h only for the T1 lysate antigen (Fig. 5c) (p < 0.05).

Fig. 5.

Production of CCL-5 after antigenic stimulus of Balb/c mice splenocytes. The data shows the concentrations of CCL-5 after a 24 h, b 48 h, and c 72 h of stimulation. PBS buffer was used as negative control (NC). The (*) within the graph stands for statistical difference between the antigen and NC; (**) for difference between equal antigenic preparations of different strains. The analysis was performed using the Kruskal–Wallis test with p < 0.05

Cytokine production

The analysis of the production of cytokines showed that IL-1 was mostly stimulated by the live VD57 strain at all incubation times (Fig. 6), by the VD57 lysate antigen at 24 h (Fig. 6a) and by the T1 lysate antigen at 72 h (Fig. 6c) of stimulation (p < 0.05). Differences between the stimulations with the antigenic preparations of both strains were also observed for IL-1 production, and it is shown in Fig. 6.

Fig. 6.

Production of IL-1 after antigenic stimulus of Balb/c mice splenocytes. The data shows the concentrations of IL-1 after a 24 h, b 48 h, and c 72 h of stimulation. PBS buffer was used as negative control (NC). The (*) within the graph stands for statistical difference between the antigen and NC; (**) for difference between equal antigenic preparations of different strains. The analysis was performed using the Kruskal–Wallis test with p < 0.05

Regarding the TNF production, the stimulus with the live strains only showed significant results at 72 h and only when the splenocytes were stimulated with the T1 strain at 10² and 10⁶ CFU/mL (Fig. 7c). The T1 lysate antigen was able to stimulate TNF production, presenting statistical differences when compared to the stimulus with the VD57 lysate antigen (Fig. 7). The VD57 TPP antigen stimulates the production of TNF at 24 h (Fig. 7a), as well as the T1 TPP antigen at 48 h (Fig. 7b), but these two antigenic preparations could not significantly stimulate the TNF production at other incubation periods (p < 0.05).

Fig. 7.

Production of TNF after antigenic stimulus of Balb/c mice splenocytes. The data shows the concentrations of TNF after a 24 h, b 48 h, and c 72 h of stimulation. PBS buffer was used as negative control (NC). The (*) within the graph stands for statistical difference between the antigen and NC; (**) for difference between equal antigenic preparations of different strains. The analysis was performed using the Kruskal–Wallis test with p < 0.05

Correlation between chemokine and cytokine production

The Spearman’s statistical method was performed with the objective to demonstrate the correlation between chemokine and cytokine production stimulated by the antigenic preparations that presented statistical differences when compared to the NC (Table 1). The production of TNF stimulated by the live T1 strain correlated with all chemokine studied herein, except CCL-2. In the other hand, the stimulus of IL-1 production by the live VD57 strain only correlated with the production of CCL-3. Markedly, the IL-1 and TNF production induced by the T1 lysate antigen correlated with the production of all chemokines, but the levels of CCL-4 were not found to be significantly induced by this antigen.

Table 1.

Correlation between chemokine and cytokine production significantly induced by live strains of C. pseudotuberculosis and T1 lysate antigen, as detected by Spearman’s correlation test. ( +) positive correlation; ( −) absence of correlation; (¢) no significant production of chemokine

Antigen	Cytokine	Chemokine
		CCL-2	CCL-3	CCL-4	CCL-5
Live VD57 strain	IL-1	−	+	−	−
Live T1 strain	TNF	−	+	+	+
T1 lysate	IL-1	+	+	¢	+
	TNF	+	+	¢	+

Discussion

This study was conducted with the objective to better elucidate the dynamics of chemokine production during the infection by C. pseudotuberculosis in susceptible Balb/c mice and after stimulation with different antigenic extracts of this bacterium. This animal model was chosen due to numerous advantages, such as its easy handling and the existence of isogenic strains, in addition to the fact that it demonstrates a pattern of response to the infection that resembles the one observed in goats and sheep [10]. The inflammatory cytokines TNF and IL-1 were chosen to be included in this study because they are described in the literature as inducers of chemokine production in M. tuberculosis infection and to the fact that the absence of TNF production results in a granuloma malformation, which can impair the infection control in tuberculosis and caseous lymphadenitis [4, 18–20].

Chemokines are responsible for recruiting host defense cells to infection sites, regulating the traffic of lymphocytes and other leukocytes through peripheral lymphoid tissues. The expression of β-chemokines, including CCL-2, CCL-3, CCL-4, and CCL-5, has already been demonstrated in chronic processes, such as psoriasis [21], gingival inflammation [22], and in the formation of granulomas induced by Schistosoma mansoni [23]. These chemokines play fundamental roles in the formation of granuloma in tuberculosis [5, 8, 24]. Therefore, these chemokines were evaluated herein because they are described as important molecules that are produced during the infection with M. tuberculosis, a bacterial model that is widely studied and elicits an immune response that is similar to the one induced by C. pseudotuberculosis [4–6].

It is well documented that CCL-2 is a potent monocyte activator through the attraction of CD4⁺ T lymphocytes and dendritic cells, and it plays a key role in the control of tuberculosis in rats, although high levels of this molecule in humans can cause increased susceptibility to tuberculosis [3, 7]. This fact can corroborate our findings for CCL-2 production after splenocyte stimulation with the live strains of C. pseudotuberculosis, and we could notice that, within 24 h, this chemokine was significantly produced, which leads us to infer that there was an activation of the innate immune response against this bacterium.

Specifically, CCL-3, CCL-4, and CCL-5 in M. tuberculosis infection induce activation and proliferation of T cells and macrophages, and CCL-3 promotes Th1 cell differentiation [25]. In the present study, VD57 strain stimulated a significant production of CCL-3 at all periods of stimulation and in all concentrations of the live bacteria; the T1 strain stimulated CCL-3 production 48 h after the stimulus. It was noted that the innate immune response against the pathogenic strain was induced in a previous moment than the attenuated strain.

In a same way than CCL-2, CCL-4 and CCL-5 are also initially induced by the live pathogenic strain, proving to be important molecules in an initial signaling that will probably result in the formation of granuloma, the main clinical symptom of the infection generated by C. pseudotuberculosis. Besides the fact that the production of these chemokines is an attempt of the organism to control the infection, the high pathogenic VD57 strain can escape this mechanism probably due to the different composition of cell wall glycolipids, which may contribute to its high proliferation profile in target tissues because of a lesser recognition by IgM-specific antibodies [16].

The inflammatory chemokines and pro-inflammatory cytokines studied herein, mainly TNF and IL-1 [26, 27], are produced by leukocytes and other cell types (e.g., endothelial cells and fibroblasts) in response to a stimulus induced by some molecules of microorganisms, such as cell wall molecules [28]. It is known that the induction of the production of inflammatory cytokines by alveolar macrophages in tuberculosis stimulates the expression of chemokines in an autocrine way [29, 30]. Brieland and collaborators [31] found that TNF and IL-1 stimulate the production of CCL-2 in murine alveolar macrophages, although Streiter and collaborators [32] did not observe the same in human alveolar macrophages. These differences can be related to different culture conditions, and it is possible that the expression of inflammatory cytokines in infections by M. tuberculosis [18] and by C. pseudotuberculosis contribute to the increase in chemokine levels.

Our results show that C. pseudotuberculosis is capable of inducing chemokines like in a study with M. tuberculosis [33, 34], and it can be observed that the production of chemokines can vary according to the studied strains. According to Saukkonen et al. [35], a virulent strain was able to induce human alveolar macrophages to produce CCL-2, CCL-3, CCL-4, and CCL-5, a result also found by other researchers [18], and a non-virulent strain was able to induce the same chemokines, but with greater emphasis on CCL-3. In this study, when using different C. pseudotuberculosis strains, the live VD57 strain (pathogenic) was able to stimulate in a more efficient way all the molecules studied herein, when compared to the T1 strain; this situation indicates that the high virulent strain induced a faster and more significant response that may impact on the formation of granuloma [35, 36].

When analyzing these results regarding the stimulation of spleen cells from Balb/c mice with bacterial antigenic fractions (secreted and lysate antigens) of the strains presenting high and low virulence, we realized that the lysate antigen from the T1 strain induced greater chemokine and cytokine levels when compared to the other antigens used in this study. In a previous work with the same murine model, a significant stimulation of cytokines by the T1 lysate antigen has been described [37]. The high rates of induction of these molecules by the T1 lysate antigen are probably associated with the concentration of innate immune-inducing antigens, that are not naturally exposed on live bacteria, being covered by other membrane antigens; in this context, it is noteworthy that the cell wall glycolipids of the T1 strain have already showed a stronger recognition by small ruminant IgMs than the VD57 strain glycolipids [16].

It must be cited that the VD57 strain has a more pronounced growth rate [10] and spread [11] in the host, being able to have a wider contact with the host immune cells than the attenuated strain, and in this way, the virulent strain can induce a higher activation of dendritic and macrophagic cells which are responsible for the production of several chemokines, in a similar way that was described for a hypervirulent M. tuberculosis strain [38]. Also, it was already observed that the C. pseudotuberculosis main virulence factor, the phospholipase D (PLD) exotoxin, was 4.5 times more expressed when in contact with the host cells than in bacteria maintained in culture media [39]. These characteristics may have influenced the situation where VD57 strain secreted and somatic antigens (produced by C. pseudotuberculosis cultivated in BHI medium) were not able to induce a significant chemokine production, but the live strain was a good inducer of these molecules.

With all the results obtained herein, it can be concluded that, depending on the strain and the antigenic extract of C. pseudotuberculosis used, there may be a production of the CCL-2, CCL-3, CCL-4, and CCL-5 chemokines and the TNF and IL-1 cytokines. Thus, the production of these chemokines and cytokines, induced by the C. pseudotuberculosis antigens, allows a chemotaxis action, mainly of leukocytes and macrophages, as previously observed in the infection by M. tuberculosis [4, 18–20].

The high pathogenic strain stimulated a more intense chemokine production than the low virulence strain, and the T1 lysate antigen was the antigenic extract that showed the greatest capacity for inducing chemokines, while the other antigenic preparations were not able to significantly induce the molecules. It was observed that different bacterial strains were able to induce different chemokine production patterns, and these different levels of chemokines can influence the pathogenicity of the C. pseudotuberculosis strains and their ability to induce the formation of granulomas.

Author contribution

TDB, MFR, DL, JTR, and TJS conducted the antigen production and splenocyte stimulation and wrote the manuscript. JTR, LFMC, and TJS performed the survival study of mice inoculated with different strains of C. pseudotuberculosis. TDB proofread the manuscript. RM, RWP, and VA designed the study, participated in the coordination of the project, and critically revised the manuscript. All authors have read and approved the manuscript.

Funding

This work was funded by the Fundação de Apoio à Pesquisa do Estado da Bahia (FAPESB), grant number PPP 022/2009. RWP is a Technological Development fellow from the National Council for Scientific and Technological Development (CNPq—Proc. 313350/2019–1).

Availability of data and material

The data that support the findings of this study are available from the corresponding author upon request.

Code availability

Not applicable.

Declarations

Ethics approval

This project was approved by the Ethics Committee of the Institute of Health Sciences of the Federal University of Bahia (001/2009).

Consent for publication

All authors consent for publication.

Conflict of interest

The authors declare no competing interests.