Introduction

Leptospirosis, a neglected re-emerging enzootic spirochetal disease, affects millions of people worldwide causing an overall mortality rate of 65000 per year (1). In addition, it causes serious health problems in animals of agricultural interest which leads to substantial economic losses mostly in tropical and subtropical countries. Assessing the true severity of leptospirosis can be incredibly challenging, especially when early diagnosis is difficult due to nonspecific symptoms overlapping with other illnesses (2). Recent outbreaks of both human and canine leptospirosis in New York and California in 2020 to 2022 (3, 4) underscores the need for development of effective strategies to control this disease. Although serovar-specific vaccines are available for animals and at least one is available for humans, no broadly effective vaccine is available for either (5). The absence of an effective cross-protective vaccine candidate increases the risk of disease re-emergence on a global scale. Efforts to use leptospiral surface antigens in various vaccine formulations have shown limited success in conferring protection against leptospiral dissemination and shedding, as well as severe disease. Leptospira’ immune evasion strategies contribute to the complexities of finding good vaccine candidates.

The genus Leptospira is broadly categorized into 2 major clades P and S (Pathogens and Saprophytes) and further categorized into 4 subclades, P1, P2, S1 and S2 based on their virulence properties, growth conditions and genetic make-up (6, 7). Subclade P1 is further divided in two phylogenetically related groups named P1+ (high-virulence pathogens, established pathogenic species, e.g. L. interrogans) and P1-(low-virulence pathogens, phenotypically not well-characterized) (8). Leptospira survives in moist conditions and are free-living organisms naturally found in soil and water bodies (9, 10). The spread of infection occurs through contaminated water contact with breached skin or mucosal surfaces (11). Saprophytic strains of Leptospira, such as L. biflexa (S1), are unable to establish disease due to the lack of certain virulence factors (7) and have been found in natural environments around the world alongside pathogenic serovars (6, 11, 12). Moreover, L. biflexa exhibits certain niche-specific adaptations that allow them to persist in both environmental and host setting (13, 14).

Our previous studies (15, 16) demonstrated that L. biflexa triggers a robust innate immune response in mice during the acute phase of infection. This raised the question of whether saprophytic Leptospira induced immune responses could confer any degree of resistance or immune memory that could suppress a subsequent challenge with a pathogenic serovar of Leptospira. Answering that question was the main goal of the current study.

Materials and Methods

Animals

Male C3H/HeJ mice (n=6-8/group) were purchased from The Jackson Laboratory (Bar Harbor, ME) and were maintained in a pathogen-free environment in the Laboratory Animal Care Unit of the University of Tennessee Health Science Center (UTHSC). All experiments were performed in compliance with the UTHSC Institutional Animal Care and Use Committee (IACUC), Protocol no. 19-0062.

Bacteria

Non-pathogenic Leptospira biflexa serovar Patoc (LB) belonging to subclade S1 was purchased from ATCC and grown in EMJH media. Pathogenic Leptospira interrogans serovar Copenhageni strain Fiocruz L1-130 (LIC) belonging to subclade P1+ (high-virulence pathogens) (6, 8) was grown in EMJH media and subsequently passaged in hamster to maintain virulence. EMJH culture passage 2 was used to inoculate mice (10⁸) after counting Leptospira under a dark-field microscope (Zeiss USA, Hawthorne, NY) using a Petroff-Hausser chamber.

Infection of mice and study design

We carried out two experiments set apart by a single or double exposure to a saprophytic serovar of Leptospira (L. biflexa) before challenge with a pathogenic serovar (L. interrogans). Groups of mice were inoculated with 10⁸ Leptospira intraperitoneally (IP) both for exposure to L. biflexa and for challenge with L. interrogans. Each experiment was reproduced once. In the single exposure study (Fig 1A), Group 1 (n=3) was the naive control which received PBS (PBS¹), Group 2 (n=4) was inoculated with 10⁸ L. biflexa once at 6 weeks (LB¹), Group 3 (n=4) received PBS for 2 weeks followed by challenge with 10⁸ L. interrogans at 8 weeks (PBS¹LIC¹), and Group 4 (n=4) was inoculated with 10⁸ L. biflexa at 6 weeks and challenged with 10⁸ L. interrogans at 8 weeks (LB¹LIC¹). In the double exposure study (Fig 2A), Group 1 (n=3) was the naive control which received PBS (PBS²), Group 2 (n=4) received 10⁸ L. biflexa IP at 6 and 8 weeks (LB²), Group 3 (n=4) received PBS for 2 weeks followed by challenge with 10⁸ L. interrogans at 10 weeks (PBS²LIC²), and Group 4 (n=4) was inoculated with 10⁸ L. biflexa at 6 and 8 weeks and challenged with 10⁸ L. interrogans at 10 weeks (LB²LIC²). Weight was monitored daily. Mice were euthanized 15 days after L. interrogans challenge or when they reached the endpoint criteria (20% body weight loss post-infection). Blood and kidney were collected at euthanasia: blood was used for quantification of anti-Leptospira antibody; kidney was used for quantification of Leptospira load (16S rRNA) and it was cultured in EMJH media for evaluation of bacterial viability. Furthermore, kidney samples were stored in 10% formalin for Hematoxylin and Eosin (H&E) staining. Spleen for flow cytometric analysis was collected from mice after euthanasia only in the double exposure study given that all mice consistently survived challenge.

Figure 1:

Figure 2:

Leptospira detection through qPCR

Isolation of genomic DNA from blood, urine and kidney were carried out using NucleoSpin tissue kit (Clontech, Mountain View, CA) according to the manufacturers’ instructions. Leptospira 16S rRNA primers (Forward-CCCGCGTCCGATTAG and Reverse-TCCATTGTGGCCGAACAC) and TAMRA probe (CTCACCAAGGCGACGATCGGTAGC) were used for detection of Leptospira genus using qPCR with a standard curve of 10⁵ to 1 L. interrogans (17, 18). Similarly, qPCR was performed with kidney tissues placed in EMJH to grow live Leptospira after culturing for 5 days and visually quantified under a dark field microscope (20X, Zeiss USA, Hawthorne) on d3 and d5 post culture inoculation.

RNA isolation and RT-PCR

Kidneys were stored in RNA later after euthanasia. RNeasy mini kit (Qiagen) was used to extract total RNA from kidney tissue according to the manufacturer’s specifications. RNA purity was measured using a Nanodrop instrument (Thermo Scientific) at A260/280 ratio. cDNA was prepared using cDNA reverse transcription kit (Applied Biosystems). ColA1 primers (Forward-TAAGGGTACCGCTGGAGAAC, Reverse-GTTCACCTCTCTCACCAGCA), TAMRA probe (AGAGCGAGGCCTTCCCGGAC) and β-actin primers (Forward-CCACAGCTGAGAGGGAAATC, Reverse-CCAATAGTGATGACCTGGCCG), TAMRA probe (GGAGATGGCCACTGCCGCATC) were purchased from Eurofins Genomics.

Histopathology by H&E staining

Kidney tissues were fixed in formalin buffer. Histopathology was performed at the Histology Department, UT Methodist University Hospital, Memphis, TN. Digital scanning of inflammatory cell infiltration was measured by taking images of ∼ 5 fields per sample under 20x magnification. Images were captured after digitally scanning the H&E slides using Panoramic 350 Flash III (3D Histech, Hungary) and CaseViewer software.

ELISA

Leptospiral extract for Leptospira biflexa and Leptospira interrogans were prepared as described previously (19). Briefly, Leptospira was cultured in EMJH media and once confluency was observed, cells were centrifuged to obtain a pellet. This pellet was then incubated with BugBuster® solution (1mL) at RT in a shaker incubator (100 rpm) for 20 min and homogenized by vortexing. Stocks were stored at -20°C. This whole-cell extract of Leptospira was then diluted in 1X sodium carbonate coating buffer. Nunc MaxiSorp flat-bottom 96 well plates (eBioscience, San Diego, CA) were coated with extracts prepared from 10⁷-10⁸ bacteria per well and incubated at 4℃ overnight. Cells were washed using 1X PBST the following day and blocked for 1 h using 1% BSA solution, followed by another wash with 1X PBST. Serum samples (1:100) were added to the antigen coated wells and incubated at 37°C for 1 h, washed twice with 1X PBST, followed by HRP conjugate secondary anti-mouse-IgG1, IgG2a and IgG3 (1:10000) which was incubated for 30 mins. After washing the plate 3 times with 1X PBST the color was developed using TMB Sureblue followed by Stop solution before the absorbance was measured at OD 450 nm using an ELISA plate reader (Molecular Devices Spetramax).

Flow cytometry

Spleens were chopped into small pieces and macerated to prepare single cell suspensions on the same day of euthanasia to avoid loss of cell viability. RBC lysis was performed using ACK lysis buffer (Gibco). AO/PI dual staining was used to count live/dead cells on a Luna counter (Logos Biosystems, South Korea). 10⁶ cells were seeded in a 96 well microtiter plate after washing with 1X PBS twice. Blocking was performed with anti-mouse CD16/32 antibody (1:100), followed by 20 minutes incubation on ice. Fluorochrome-conjugated antibodies (Table S1) were used to stain specific cell surface markers after 30 mins incubation in the dark at 4°C. Freshly prepared flow staining buffer was used for washing stained cells. Cells were fixed using 4% Paraformaldehyde for 10-15 mins at room temperature. Beads were stained using specific fluorochrome conjugated antibodies and used for compensation, while FMO prepared with spleen cells simultaneously were used for gating controls. Cells were resuspended in flow staining buffer and the BioRad ZE5 cell analyzer was used for data acquisition. Data analysis was done using Flow Jo software. Gating strategy used for immunophenotyping of spleen cells is provided in supplementary Fig S1.

Statistical analysis

One-way ANOVA with Tukey’s multiple comparison test and unpaired t-test with Welch’s correction were used to analyze differences between experimental groups. GraphPad prism software was used to plot graphs; a value of p<0.05 is considered significant.

Results

Exposure to saprophytic Leptospira before infection with a pathogenic serovar prevents disease and increases survival of C3H-HeJ mice

We inoculated adult C3H-HeJ male mice with a single dose of L. biflexa two weeks before challenge with L. interrogans (LB¹LIC¹) at 8 weeks (Fig 1A) and measured a significant rescue of weight loss over a period of 15 days as compared to mice infected at 8 weeks that did not receive L. biflexa (PBS¹LIC¹) (p<0.0001); unchallenged control mice that received L. biflexa (LB¹) or PBS (PBS¹) gained weight throughout the corresponding 15 days (Fig 1B). Survival curves were generated after the mice reached the following endpoint criteria: 20% weight loss or 15 days post challenge with L. interrogans or 15 days post inoculation with L. biflexa /PBS for the controls (Fig 1C). All mice infected with L. interrogans (PBS¹LIC¹) reached the 20% weight loss endpoint criteria between d9 and d12 post infection. In contrast, 3/4 (75%) of the mice that received one dose of L. biflexa before challenge with L. interrogans (LB¹LIC¹) survived and gained significant body weight (Fig 1B) which was similar to the naïve control that received only PBS. Analysis of bacterial dissemination was done by qPCR of the Leptospira 16S gene in genomic DNA purified from blood, kidney tissue and urine. Of note, although 16S rRNA primers can amplify L. biflexa 6h post infection (20), we processed the tissue samples 30 days or 45 days post L. biflexa exposure. We also found that a single exposure to L. biflexa before challenge did not prevent dissemination of pathogenic L. interrogans in blood (Fig S2A) or shedding in urine (Fig 1D), or kidney colonization (Fig 1E). Culture of kidney in EMJH media showed presence of ∼ 2500 motile, morphologically intact L. interrogans under dark field microscopy which was confirmed by 16S qPCR (Fig 1F) both on d3 and d5 post culture of kidney collected from LB¹LIC¹ mice; kidney from PBS¹, LB¹ and PBS¹LIC¹ did not produce positive cultures by dark field microscopy or 16S qPCR.

In the double exposure study, mice were inoculated with 2 bi-weekly doses of L. biflexa two weeks before challenge with L. interrogans (LB²LIC²) at 10 weeks in comparison with the respective controls (Fig 2A). As expected, mice infected at 10 weeks with LIC that did not receive L. biflexa (PBS²LIC²) lost ∼ 11 % of weight on d11 post challenge and did not recover (Fig 2B). In contrast, mice that received a double dose of L. biflexa two weeks before challenge at 10 weeks with L. interrogans (LB²LIC²) lost a maximum of 5% weight on d10 and recovered fully by d15 post infection; unchallenged control mice that received L. biflexa (LB²) or PBS (PBS²) gained weight throughout the 15 days (Fig 2B). Survival curves generated after the mice reached endpoint criteria (Fig 2C) show that all experimental groups survived LIC infection. Analysis of bacterial dissemination showed that a double exposure to L. biflexa before challenge did not prevent dissemination of pathogenic L. interrogans in blood (Fig S2B), or shedding in urine (Fig 2D), or kidney colonization (Fig 2E). Culture of kidney in EMJH media showed presence of 5,000-10,000 motile, morphologically intact L. interrogans on d3 and d5 post culture of kidney collected from PBS²LIC² mice in contrast to 1,000 to 2,500 live L. interrogans observed in culture from kidney collected from LB²LIC² mice which was confirmed by 16S qPCR (Fig 2F); kidney from PBS² and LB² mice did not produce positive cultures by dark field microscopy or 16S qPCR.

L. biflexa exposure before challenge with L. interrogans mitigates renal histopathological changes

As expected, H&E staining of histological slices of all kidneys from mice challenged with L. interrogans in both single and double exposure experiments (PBS¹LIC¹ and PBS²LIC²) showed signs of inflammation with increased immune cell infiltration (Fig 3A and 3C). In contrast, H&E staining of kidney slices from the groups of mice exposed to L. biflexa before L. interrogans challenge (LB¹LIC¹ and LB²LIC²) showed reduced immune cell infiltration. We also measured expression of a marker (ColA1) for kidney fibrosis. In both experiments, kidneys from mice challenged with L. interrogans (PBS¹LIC¹ and PBS²LIC²) had significantly higher expression of ColA1 as compared to the controls; in contrast, kidneys from mice challenged with L. interrogans after exposure to L. biflexa (LB¹LIC¹ and LB²LIC²) were not different than the controls (Figs 3B and 3D).

Figure 3:

In addition, we were able to collect kidneys from experimental mice subjected to the double exposure of L. biflexa because they all survived subsequent challenge with L. interrogans. As such, we did a comparative gross morphological analysis between the 4 groups (Fig S3). We found that kidney from LB²LIC² mice maintained their normal gross anatomy and coloration as did kidneys from control mice (PBS² and LB²), in contrast to kidneys from mice challenged with L. interrogans that were not previously exposed to L. biflexa (PBS²LIC²).

Serologic IgG2a responses to L. interrogans were significantly higher in mice pre-exposed to L. biflexa before challenge with L. interrogans

In both single and double L. biflexa exposure experiments we measured anti-L. interrogans total IgM, total IgG, and IgG subtypes IgG1, IgG2a and IgG3 in serum collected 2 weeks after challenge with L. interrogans (IgG subtypes shown in Figs 4A and 4B). In both experiments, total IgM and IgG were significantly increased in PBS-LIC and LB-LIC when compared to the respective controls, but not between PBS-LIC and LB-LIC groups. Regarding IgG isotypes, IgG1 was generally low and IgG2a as well as IgG3 were generally high in groups infected with L. interrogans. Although differences between groups (PBS²v PBS²LIC² and LB^1/2 v LB^1/2LIC^1/2) were significant, differences between the LIC infected groups (PBS^1/2LIC^1/2 v LB^1/2LIC^1/2) were not significant for IgG1 and for IgG3 in contrast to IgG2a (p=0.001 for single exposure and p=0.0095 for double exposure).

Figure 4:

Exposure to non-pathogenic L. biflexa before pathogenic L. interrogans challenge induced increased frequencies of effector helper T cells in spleen

We immunophenotyped the spleen cells from the mice subjected to double L. biflexa exposure because all animals survived to the term of the experiment. Mice in the single exposure experiment met endpoint criteria before the term of the experiment and thus we were not able to process spleen for immunophenotyping. In the L. biflexa double exposure experiment, we measured increased frequencies in B cells when LIC infected mice were compared to the respective controls, but not between PBS²LIC² and LB²LIC² mice (Fig 5A). We measured decreased frequencies in T cells when LIC infected mice were compared to the respective controls but not between PBS²LIC² and LB²LIC² mice (Fig 5B). No differences were observed in NK cells between any of the groups (Fig 5C). We measured increased frequencies in helper T cells between all groups; of note, PBS²LIC² vs LB²LIC² p=0.006 (Fig 5D). We also measured decreased frequencies in cytotoxic T cells between all groups; of note PBS²LIC² vs LB²LIC² p=0.0056 (Fig 5E). Furthermore, T cell subset typing (Fig 6) showed that frequency of early effector CD4+ T helper cells (Fig 6B, CD44-CD62L-) and effector T helper cells (Fig 6C, CD44+CD62L-) were significantly increased when compared between the LIC challenged groups and the respective controls (except PBS² v PBS²LIC² early effectors) and that frequency of early effector and effector T helper cells were higher in the LB²LIC² group than PBS²LIC². No major changes were measured in memory CD4+ T helper cells (Fig 6D, CD44+CD62L+). In the CD8+ cytotoxic T cell subsets, we measured significant decreases in frequency of naïve T cells between LIC infected groups and the respective controls (Fig 6E, CD62L+CD44-) and this was replicated in the CD8+ cytotoxic memory except that differences with the LIC infected groups were not significant (Fig 6H, CD44+CD62L+).

Figure 5:

Figure 6:

Discussion

Understanding host immune responses to Leptospira infection is crucial for advancing our ability to develop new control measures for leptospirosis (19, 21–25). Given their widespread presence in nature, the likelihood of humans or animals getting exposed to non-pathogenic serovars of Leptospira is likely high (10, 11). Our previous studies showed innate immunity engagement during saprophytic L. biflexa infection in mice (15, 16). In addition, L. biflexa extracts can be used to detect Leptospira-specific antibody in up to 67% of serum from patients with clinically confirmed leptospirosis (19) which points to a high degree of immunodominant cross-reactive epitopes between L. biflexa and pathogenic Leptospira. Further, the current hypothesis regarding evolution of Leptospira species is that symbiosis of Leptospira with eukaryotes emerged from free-living ancestral species (26); in other words, pathogenic Leptospira may have evolved from an environmental ancestor (6). Thus, we hypothesized that these highly cross-reactive immunodominant epitopes may also induce cross-protective immune responses. The objective of the current study was to assess whether exposure to a live saprophytic serovar of Leptospira provides any heterologous cross-species protection against a subsequent challenge with a pathogenic serovar in a mammalian host (mouse).

In the initial analysis of pathogenesis (Fig 1, Fig 2, Fig S2) we observed that prior exposure to one or two doses of saprophytic L. biflexa rescues weight loss in mice challenged with pathogenic L. interrogans at 8 weeks (LB¹LIC¹) and at 10 weeks (LB²LIC²), respectively. Weight gain correlated with survival (75% survival in LB¹LIC¹ versus 0% survival of the PBS¹LIC¹ group) in the single exposure experiment, where we expected all mice infected at 8 weeks with pathogenic L. interrogans (PBS¹LIC¹) to irreversibly lose weight and meet endpoint criteria for euthanasia before the 2 week term of the experiment (23, 27). Loss of mice due to irreversible weight loss is not expected if mice are infected with LIC at 10 weeks of age (17), as observed in the L. interrogans control group in the double exposure experiment (PBS²LIC²). In both experiments, mice exposed to L. biflexa before challenge with LIC, produced evidence of L. interrogans dissemination in blood, shedding in urine and kidney colonization of the kidney.

Histological inspection of kidney slices (Fig 3) showed that exposure to a saprophytic Leptospira before challenge supported normal structural morphology and prevented infiltration of immune cells in both single and double exposure experiments; in addition, it significantly reduced a fibrosis marker (ColA1) in the single exposure experiment. In the double exposure experiment, differences in ColA1 fibrosis marker are not significant between the two LIC infected groups because 10-week old C3H-HeJ infected with LIC are more resistant to pathology resultant from infection. Our findings are intriguing as they suggest that while prior live saprophytic exposure did not prevent infection or leptospiral dissemination, it may confer protection against kidney fibrosis.

Our data also shows that prior exposure to non-pathogenic Leptospira before pathogenic challenge, induced higher antibody titer in the serum, specifically IgG2a antibodies against L. interrogans in both single and double exposure experiments (Fig 4). Increased IgG2a response in serum is associated with induction of a Th1 biased immune response. Others have recently found that saprophytic L. biflexa induced Th1 responses, higher T cell proliferation and IFN-g producing CD4+ T cells (28). Persistent IgM and strain specific IgG responses was observed during a homologous leptospiral challenge in C57BL6/J mice (22). In our study, exposure to a saprophytic Leptospira induced antibody responses that may provide heterologous protection against the pathogenic strain of Leptospira. This supports a promising broad-spectrum efficacy. Thus, live vaccines derived from a saprophytic strain of Leptospira could offer broader protection and overcome the limitation of serovar specificity often observed with killed whole-cell vaccines based on pathogenic strains.

Differences in antibody titer among the L. interrogans infected group pre-exposed to saprophytic L. biflexa can be attributed to the robust trafficking and differentiation of B and helper T cells (CD4+) measured in spleen (Fig 5 and Fig 6). Presence of effector helper T (CD4+) cells in the spleen indicate a robust cellular immune response as these cells produce cytokines that play a pivotal role in activating other immune cells, including antibody-producing B cells. Moreover, our findings align with another observation which further reinforces the potential immunostimulatory properties of components (polar lipids) derived from saprophytic L. biflexa, indicating that these components could play a crucial role in inducing robust B cell responses (29). Induction of helper T cell responses along with dynamic transition from naïve to early effector and effector without T helper memory reflects an orchestrated immune response upon pathogenic challenge in the saprophytic pre-exposed group that is typical of effective responses to vaccines. Previous studies have highlighted the significance of activated CD4+ T cells during Leptospira infection in providing protective immunity to the host and mitigating the severity of leptospirosis by releasing cytokines (30). Correlating induction of chemo-cytokines by saprophytic Leptospira with subsequent adaptive immune responses, such as the activation of CD4+ T cells or the production of specific antibodies, provides insights into how innate immune signals drive the adaptive immune response against a pathogenic threat. It may also aid in identifying key signaling molecules or pathways that could be targeted for therapeutic interventions or vaccine design.

While other researchers have explored vaccination strategies using live attenuated or mutant strains of pathogenic serovars, our approach was to utilize a live saprophytic bacterial strain which is unique in the field (21, 23, 31, 32). We previously showed that oral delivery of a probiotic strain, Lactobacillus plantarum, reduces the severity of leptospirosis by recruiting myeloid cells (21) which suggests that a general phenomenon of trained immunity may be involved. Current vaccines based in inactivated pathogenic species provide equivalent protection to the one achieved in this study (5, 33) with the caveat of being serovar specific (32–34). Although our current study conclusively shows protection from severe leptospirosis after heterologous challenge, it remains to be shown if protection extends to multiple pathogenic serovars of Leptospira. Using a live saprophytic strain of Leptospira as control strategy could pave the way for development of novel broadly effective vaccines against leptospirosis. Such a vaccine could have a substantial economic impact if applied to animals of agricultural interest.

Another interesting aspect of our current study is that it shows that exposure to a live saprophytic strain of Leptospira provides protection against a pathogenic serovar. Thus, in the real-life scenario where individuals or animals may naturally encounter a saprophytic Leptospira species, they may develop immune responses that mitigate severe disease outcomes if the host later encounters a pathogenic strain of Leptospira. By exploring the immune dynamics during the co-exposure to different Leptospira serovars, this study could open avenues of research on strategies that leverage natural exposure to saprophytic species to devise safe control measures against leptospirosis. This concept is important for understanding the epidemiological risk factors of leptospirosis and it should be applicable to other infectious diseases caused by direct contact between the pathogen and mucosal membranes or abraded host skin.

Importantly, we found that in mice pre-exposed to live saprophytic Leptospira there was a correlation between kidney health after LIC infection (less infiltration of immune cells in kidney and less fibrosis marker ColA1) and higher shedding of live LIC in urine. This suggest that a status of homeostasis was reached after kidney colonization that helps the spirochete complete its enzootic cycle. Additional research is needed to fully understand the mechanisms involved in kidney homeostasis after LIC infection.

Data availability

All data generated or analyzed during this study are included in this manuscript. All relevant data are available from the authors.

Acknowledgements

This work was supported by the National Institute of Allergy and Infectious Diseases (NIAID), United States National Institutes of Health (NIH), grant numbers R01 AI139267 (MGS), R21 AI 142129 (MGS). The content of this manuscript is totally the responsibility of the authors and does not involve the official views of NIAID or NIH.

Competing Interests

MGS and SK have a relevant patent that might pose a conflict of interest.

Author Contributions

MGS is responsible for the overall concept. SK, AS conducted the experiments. SK, MGS designed the experiments and analyzed the data. MGS secured funding, administered the project and supervised personnel. SK and MGS wrote the manuscript.