Abstract
Tick-borne encephalitis virus (TBEV) is a leading cause of vector-borne viral encephalitis with expanding endemic regions across Europe. Although currently used inactivated whole virus vaccines are effective, vaccination breakthroughs have been reported for which the reasons are unclear. In this study we tested in mice the efficacy of pre-infection with a closely related low-virulent flavivirus, Langat virus (LGTV strain TP21), or a naturally avirulent TBEV strain (TBEV-280) in providing protection against lethal infection with the highly virulent TBEV strain TBEV-Hypr (referred to as TBEV-Hypr). LGTV has been evaluated as an experimental live vaccine against TBE, but further development was abandoned due to too high residual pathogenicity of a LGTV-based vaccine. Here we show that prior infection with TP21 or TBEV-280 is efficient in protecting mice from lethal TBEV-Hypr challenge. Histopathological analysis of brains from non-immunized control mice revealed neuronal TBEV infection and necrosis. Neuroinflammation, gliosis and neuronal necrosis was however also observed in some of the TP21 and TBEV-280 pre-infected mice although at reduced frequency as compared to the non-immunized TBEV-Hypr infected control mice. Interestingly, qPCR detected the presence of viral RNA in the brains and spinal cord of both TP21 and TBEV-280 immunized mice after TBEV-Hypr challenge, but significantly reduced compared to mock-immunized mice. Our results indicate that although TBEV-Hypr infection is effectively controlled in the periphery upon immunization with low-virulent LGTV or naturally avirulent TBEV-280, it may still enter the CNS of these animals. These findings improve our understanding of potential causes for vaccine failure in individuals vaccinated with TBE vaccines.
Introduction
Tick-borne encephalitis virus (TBEV) is an enveloped virus belonging to the family Flaviviridae which contains several other vector-transmitted viruses such as West Nile virus, yellow fever virus, dengue virus, Japanese encephalitis virus and Zika virus (ZIKV), which all may cause life threatening conditions in humans (1). TBEV is a major cause of concern in Europe and Asia with currently expanding endemic areas (2). TBEV is primarily transmitted by a tick-bite of Ixodes species, which can lead to strain-dependent outcome of illness (3,4). The clinical course of infections caused by TBEV strains circulating in Europe often shows a biphasic pattern. Approximately 7-14 days after infection, first flu-like symptoms may appear. After an asymptomatic phase, 20-30% of the patients develop a second phase with headache, high fever, and neurological symptoms as a consequence of severe meningitis and meningoenephalomyelitis. Of these patients, 2% exhibit long-lasting neurological sequelae (5,6). Preventive vaccination is the only available specific intervention method for TBE to date. Current TBEV vaccines are based on formalin inactivated whole virus preparations, which have proven to be effective in conferring temporary protection. Therefore repeated vaccinations are needed to induce satisfactory immunity with booster doses being required every 5-10 years (6–9). Live attenuated vaccines have been proven to confer better protection, nevertheless associated safety risks have discouraged their usage (10,11). LGTV, a naturally occurring low-virulent flavivirus, that has high amino acid sequence identity with TBEV (∼80%), has shown potential as a live attenuated vaccine candidate. In Russia LGTV was used in 1970s to vaccinate against TBE, which was regarded highly successful until among vaccinees cases displaying neurological symptoms (about 1:20.000) were observed (2,12–15). Apparently, this LGTV-based vaccine was under-attenuated. Protection against TBEV and other flaviviruses in the periphery is largely mediated by virus-specific antibodies. In contrast, virus control within the central nervous system (CNS) is dependent on the interplay between infiltrating virus specific T cell and CNS cells. It has been shown that this interaction may have both beneficial and detrimental effects (16–18). Investigations into the pathogenesis of TBE in patients revealed that granzyme B releasing T cells and microglia cells / macrophages contribute to the tissue damage after infection, resulting in neuronal death and astrogliosis (19). It was recently shown that even low dosage LGTV infection in mice may result in astrogliosis and microglia activation in the hippocampus, suggesting that infection with non-lethal doses of flaviviruses can indeed lead to histopathological changes in the brain (20).
To gain more insights in live vaccine induced protective immunity against TBE, we studied the efficacy of LGTV and naturally avirulent member of the TBEV serogroup (TBEV-280; a strain closely related to another naturally avirulent and well-characterized strain TBEV-263 (21)) in conferring protection against challenge with highly pathogenic TBEV strain in mice. Our findings show the potential of TP21 and TBEV-280 in providing protection against infection with the highly pathogenic TBEV-Hypr strain and advance our understanding of mechanisms within the CNS that may be involved in vaccination breakthroughs.
Results
TP21 and TBEV-280 immunization protects mice from lethal TBEV-Hypr infection
First, we tested the efficacy of immunity induced by pre-immunization with TP21 or TBEV-280 to confer protection against lethal TBEV-Hypr infection. As demonstrated in Figure 1, all mock-treated mice challenged with TBEV-Hypr started showing clinical signs from 7 dpi onward with body weight loss accompanied by signs of weakness, reduced activity, pilo-erection, kyphosis, and lethargy, often combined with neurological indicators of ataxia and paresis of hind legs. Nearly 50% of mice attained clinical scores corresponding to humane endpoint at 8 dpi and the rest had to be euthanized due to severe disease by 10 dpi. Notably, all mice that were immunized with TP21 or TBEV-280 were protected from developing clinical signs and survived lethal TBEV infection until study endpoint (14 dpi) with no notable changes in the body weights. These results show that pre-immunization with TP21 or TBEV-280 is effective in inducing immunity in mice, which conferred protection from clinical and lethal TBEV infection.
Effect of TP21 and TBEV-280 immunization on TBEV-Hypr replication in organs
To gain insight into the virus control in immunized mice, we compared distribution of TBEV-Hypr in peripheral organs and in the CNS of the mock-treated, TP21 or TBEV-280 immunized mice after subsequently infected with TBEV-Hypr. As assessed by qPCR analysis designed to detect TBEV RNA, high copy numbers of viral RNA were found in the spleen and brain of mock-immunized mice that were sacrificed between 8-10dpi due to high clinical scores (Figure 2A). Furthermore, neuro-tropism and ability of TBEV-Hypr to predominantly replicate in neural tissue was evident from highest virus load observed in brains as compared to spleens of control mice (Figure 2A). At the study endpoint (14dpi), viral RNA was not detected in the spleens of any of the mice that experienced a prior immunization with TP21 or TBEV-280, which shows effective protection of peripheral organs from virus replication. In contrast, significant numbers of viral RNA copies were detected in the brains of some of the TP21 and all of the TBEV-280 pre-immunized mice (Figure 2A). However, viral copy numbers detected in these mice were significantly lower than those of mock control mice. Similar observations were made by qPCR analysis of the spinal cords (Supplementary Figure S1). TBEV primers used in this assay do not detect LGTV RNA at numbers less than 105 copies(data not shown). Since viral copy numbers detected in LGTV immunized mice are below detection limit we can confirm that only TBEV RNA is detected in these mice. Immunostaining of brain sections for TBEV antigen confirmed presence of TBEV-Hypr infected cells across different regions of brains of mock control mice. Cerebral cortex, thalamus, hypothalamus and pons appeared to be the regions with highest viral loads (Figure 2B-C). Interestingly, no TBEV antigen was detected in TP21 or TBEV-280 immunized mouse brains (Figure 2D-E). Taken together these results confirm that TP21 and TBEV-280 immunization qPCR positive induces protective immune response against TBEV-Hypr and effectively keeps virus replication and spread under control. Apparently, virus escaping from peripheral immunity may enter the CNS and replicate there before being cleared by local inflammatory or immune mechanisms.
TBEV infects neurons
Immunofluorescence staining for TBEV antigen of brain sections from control, TBEV-Hypr-infected mice revealed high numbers of infected cells across different brain regions. Based on location and morphology, most of the infected cells appeared to be neurons (Figure 3A-B), as has been reported previously (22,23). However, to further investigate whether major glial cell types such as microglia and astrocytes could also constitute a target of TBEV, double labeling of brain sections using antibodies targeting TBEV antigen and microglia/macrophages (Iba-1+) or astrocytes (GFAP+) was performed. Interestingly, none of the regions that were screened displayed astrocytes (GFAP+ cells) that co-localized with TBEV antigen (Figure 3C-D). Largely, microglia/macrophage (Iba-1+cells) also failed to show co-localized TBEV antigen. However, in some areas of cerebral cortex and midbrain of single TBEV-Hypr infected control mice we detected Iba-1+ cells co-stained with TBEV antigen (Figure 3E-F). Collectively these results confirm that neurons constitute the primary target of TBEV infection.
Marked neuronal necrosis observed after TBEV-Hypr infection in mock control mice
The results so far indicate that TBEV infects and replicate within neurons upon entering the CNS. Histopathological analysis of HE stained brain sections revealed significant neuronal necrosis in mock immunized mice that were challenged with TBEV-Hypr (Figure 4A-B). The extent of neuronal damage was assessed on a scale of 3 based on the presence of single scattered necrotic neurons per HPF (score 1) or greater than 30% affected neurons per HPF (score 3) in the regions evaluated. Neuronal damage was more prominent in the olfactory bulb, cerebral cortex and midbrain regions of heavily infected brains of mock treated mice. Interestingly, neuronal necrosis was also detected in thalamus, hypothalamus and pons of some of the TP21 immunized mice and occasionally in the cerebral cortex of TBEV-280 immunized mice (Figure 4C and Supplementary Figure S2). This indicates that the TBEV specific immune response induced by TP21 and TBEV-280 immunization protects mice from clinical signs but does not fully protect all immunized mice from neuronal damage.
Inflammation and gliosis in the brain
Neuroinfection often results in extensive inflammation within the CNS which is usually characterized by perivascular infiltrates (PVI) of recruited peripheral inflammatory cells as well as hypertrophy and hyperplasia of local glial cells (gliosis). Such changes are not observed in brain of naïve mice. Examination of HE stained brain sections revealed significantly more PVI of mononuclear cells across different regions of brains of mock-immunized mice compared to TP21- and TBEV-280-immunized mice (Figure 5 & Supplementary Figure S3). Similarly, areas of gliosis were more prominent in mock immunized mice than in TP21 and TBEV-280 immunized mice (Figure 6 and Supplementary Figure S4), in line with the attenuated course of disease after immunization. Interestingly, inflammatory infiltrates as well as multifocal, mild accumulation of glial cells were also detected within the brains of TP21 and TBEV-280 immunized mice. This again shows that TBEV possibly gained access to the CNS despite the presence of peripheral protective immunity and triggers neuroinflammation.
Discussion
In the present study, we have shown that immunization of mice with LGTV, a closely related low-virulent flavivirus, or a naturally avirulent TBEV strain (TBEV-280) provided protective immunity in mice against symptomatic and lethal infection with TBEV-Hypr strain of TBEV. Interestingly, traces of viral RNA and signs of neuroinflammation were found in the brains of LGTV and TBEV-280 immunized mice that did not display any clinical signs upon lethal TBEV-Hypr challenge. This is of particular interest in the light of breakthrough TBE that have been reported in vaccinated individuals (24,25).
Neuronal cells are regarded as primary targets of TBEV (22,23). Accordingly, in the absence of vaccine-induced protective immunity, high numbers of infected neurons were detected by immunostaining in TBEV-Hypr infected mice. As we could not detect viral antigen within neurons of immunized mice, our initial assumption was that immune response induced by LGTV and TBEV-280 pre-immunization effectively cleared TBEV-Hypr from the periphery and prevented its spread into the CNS. However, detailed qPCR analyses of brain and spinal cord from these mice reflected a different outcome. Indeed, the average viral RNA copy number detected in TP21 or TBEV-280 immunized mice were several folds lower when compared to the numbers detected in mock immunized mice. The observed viral RNA copy numbers by qPCR in the brains of the TP21 or TBEV-280 immunized mice (to the order of 105 genome copies/g) were significantly higher than could be expected without active virus TBEV replication. Hence, the most likely explanation for the presence of high viral RNA copy numbers in brains and spinal cords of the immunized mice is that TBEV-Hypr is able to enter the CNS of immunized mice and probably replicate within the neural tissue. This hypothesis is furthermore supported by the presence of inflammatory (PVI) and reactive (gliosis) changes within the CNS of vaccinated mice, which both constitute frequent findings during and after neuroinfection (18). Besides neurons, some studies have demonstrated TBEV infection of astrocytes (23,26,27). Furthermore, it has been shown that TBEV can infect and survive in rat and human astrocytes for several days without affecting each other’s viability (26,27). However, co-staining of brain tissue sections did not provide evidence of TBEV antigen within astrocytes either in mock-, LGTV- or TBEV-280-immunized mice after challenge. Similarly, only isolated microglia cells in certain brain regions in mock immunized mice showed co-labeling of Iba-1 with TBEV antigen. Microglia are highly efficient phagocytes and actively scavenge cell debris in the event of tissue damage. Therefore, it remains unclear whether detected signal resulted from infection of Iba1-positive cells or from phagocytosis. There is evidence that neurotropic flaviviruses can influence the behavior of microglia and astrocytes and either assist in viral clearance or augment neuropathogenesis by producing toxic neuroinflammatory mediators (17,20,28–30). To avoid any permanent damage, inflammation within the CNS is highly regulated and is primarily mediated by microglia and astrocytes. In response to any perturbation caused within the CNS, these glial cells proliferate and migrate to the affected areas and provide neuroinflammatory mediators and neurotrophic factors to limit the damage (31–33). This phenomenon coined reactive gliosis (34) is characteristic of neural tissue damage caused by injury or infection, which could be also have occurred within the present study. In mock immunized mice after TBEV-Hypr challenge, we also observed gliosis, particularly in those areas with increased viral load. Interestingly, mild areas of gliosis have also been observed in certain brain regions of the TP21 and TBEV-280 immunized mice after TBEV-Hypr challenge, indicating reactive changes within CNS even after immunization. Since we could not detect TBEV antigen in these brains it is difficult to conclude that the areas with gliosis directly correspond to those with assumed previous virus infection. Taken together these findings are highly indicative of TBEV gaining access to the CNS and triggering inflammation despite of pre-existing TP21 or TBEV-280 induced immunity. We have observed the presence of PVI in cerebral cortex, thalamus, hypothalamus and midbrain regions of mock immunized mice. In contrast, PVIs were limited to certain brain regions in immunized mice. This again may reflect differences in virus distribution within the CNS of mock immunized and immunized mice.
Extensive neuronal necrosis was found in infected brains of mock immunized mice. There was severe loss of neurons especially in the olfactory bulb, cerebral cortex and midbrain regions accompanied by detection of numerous neurons staining positive for TBEV antigen by IF as well as inflammatory and reactive changes. However, it is not clear if the observed neuronal damage represents a direct effect of virus infection or a secondary consequence of inflammation. It has been shown for other flaviviruses that infection and virus replication can trigger apoptosis and necrosis in neurons (19,35–40). Similarly, inflammation as a consequence of neuroinfection can also caus substantial neuronal loss (41,42).
Probably the most intriguing finding from this study is the detection of neuronal necrosis combined with inflammation and gliosis despite of vaccine-induced protective immunity to TBEV in some mice. This phenomenon was more prominent in TP21 than in TBEV-280 immunized mice. This is of special interest, considering breakthrough TBE in vaccinated individuals. The observation that TBEV may still enter the CNS despite vaccine induced protective immunity while causing inflammation and neuronal damage may at least in part explain vaccination breakthroughs in TBE vaccinated individuals.
In conclusion, this study demonstrates for the first time the presence of TBEV in the CNS of immunized mice that are protected from lethal or clinically manifest infection. However, absence of detectable TBEV antigen in these brains via IF hints at an important role of local inflammatory mediators in checking the virus spread within the CNS in the event virus escapes host immune response in the periphery. Nonetheless, the actual cause of the neuronal necrosis in immunized mice remains unclear. Whether this is caused by virus infection directly or a consequence of subsequent inflammation or even the action of infiltrating virus-specific T cells remains a matter of further investigation. Recently Garber et al. demonstrated that ZIKV specific T cells through microglia mediate neuronal loss and hence result in cognitive dysfunction (17). Furthermore, CD8+ T cells that clear infected cells also cause neuronal damage (18,19,43). Finally, it is tempting to speculate whether subtle neuronal necrosis as observed in immunized mice is the reflection of a broader challenge of raising sufficient immune mediated protection of the CNS from TBEV invasion, which may also be the basis of current human vaccination failures.
Material and Methods
Mice and ethics
Six-week old female C57BL/6JOlaHsd (BL6) mice were obtained from Envigo, Inc. (Indiana, USA). Mice were housed in isocage systems with individually ventilated cages. Experiments were done in biosafety level 3 laboratories of the Institute of Parasitology, Biology Center of Czech Academy of Sciences, České Budějovice, Czech Republic. The protocol was approved by the Departmental Expert Committee for the Approval of Projects of Experiments on Animals of the Czech Academy of Sciences and the Committee on the Ethics of Animal Experimentation at the Institute of Parasitology (permit No. 29/2016). All experiments were done in accordance with Czech national law guidelines (animal Welfare Act No. 246/1992 Col.) and European Union guidelines for work with animals.
Viruses
LGTV strain TP21 (referred to as TP21) was isolated 1956 from a pool of hard ticks (Ixodes granulatus) of forest rats near Kuala Lumpur, Malaysia (44). TP21 was grown in Vero B4 and Vero E6 cells and titers were determined by TCID50 assay on Vero E6 cells. Virus was provided the Department of Molecular Immunology, Helmholz Centre for Infection Research, Braunschweig, Germany. Highly virulent TBEV strain TBEV-Hypr (further on described as TBEV-Hypr) was isolated in 1953 from a diseased child in Brno (former Czechoslovakia). The virus was passaged eight times in suckling mice brains before its use in this study. A naturally avirulent strain TBEV-280 was isolated in 1987 from a pool of I. ricinus ticks near Kaplice (former Czechoslovakia). The virus was plaque-purified three times, and passaged three times in suckling mice brains, one time in UFK-NB4 cells and one times in SK-N-SH cells before its use in this study. TBEV titers were determined by plaque assay as described in (45). The TBEV strains were provided by the Collection of Arboviruses, Institute of Parasitology, Biology Centre of the Czech Academy of Sciences, České Budějovice, Czech Republic.
Immunization study
Subcutaneous (s.c) administration of 500 pfu of TBEV-Hypr results in 100% lethality in mice. Here, we utilized this mouse model to test the efficacy of TP21 and TBEV-280 in providing protection against lethal TBEV-Hypr challenge. At 28 days prior to TBEV-Hypr challenge, mice were administered s.c with medium (mock), TP21 (104 pfu) or TBEV-280 (104 pfu). Following a s.c. challenge with 500 pfu of TBEV-Hypr, development of clinical signs in mice was monitored for a period of 14 days and when clinical score corresponding to humane endpoint was reached, they were sacrificed.
RNA isolation and quantitative real-time PCR (qPCR)
To determine viral loads in organs, half the brain, spinal cord and spleen were dissected, weighted and homogenized in AVL Buffer (Qiagen). RNA was isolated using QIAamp viral RNA Mini QC Kit (Qiagen) and the QIAcube machine. Viral load was quantified by Taq-Man qPCR using OneStep RT-PCR Kit (Qiagen) and AriaMx Real-time PCR Systems (Agilent). TBEV forward primer (5’-3’ GGGCGGTTCTTGTTCTCC), TBEV reverse primer (5’-3’ ACACATCACCTCCTTGTCAGACT) and TBEV probe (5’-3’ TGAGCCACCATCACCCAGACACA) were designed by (46). AriaMX software (version 1.5; Agilent) in combination with intraassay TBEV standard curve was used to analyse the data and viral copies were extrapolated by individual organ weight in RNA copies/gram.
Histology
Brains were cut midsagittal and fixed for 48 hours in 4% Histofix (Carl Roth). After fixation organs were stored in PBS till paraffin embedment and cut of serial sections on a microtome (2-3 µm; Leica RM 2035; Leica Instruments GmbH, Nuβloch, Germany). Sections were stained by hematoxylin and eosin (HE) and immunofluorescence (IF).
Immunofluorescence
IF was performed as described before (47). Briefly, sections of formalin-fixed, paraffin-embedded brains were dewaxed and rehydrated using graded alcohols. For antigen retrieval, sections were heated in a microwave oven (800 W) for 20 min in 0.01 M citrate buffer, followed by application of inactivated goat serum or horse serum, respectively. Afterwards, sections were incubated with primary antibodies (see Table 1) targeting TBEV E-protein (clone 1493) (48), glial fibrillary acidic protein (GFAP, astrocytes) and ionized calcium-binding adapter molecule (Iba-1, microglia/macrophages) overnight at 4°C. For detecting TBEV-infected astrocytes and microglia, double labeling of TBEV E-protein with GFAP and Iba-1 was performed. Negative control sections were incubated with equally diluted rabbit or mouse serum. For visualization of antigen-antibody reactions, sections were treated with Alexa Fluor 488- or Cy3-labeled secondary antibodies (1:200) for 1 hour at room temperature. Nuclei were stained with bisbenzimide (Hoechst 33258, 0.01 % in methanol, 1:100; Sigma-Aldrich, Taufkirchen, Germany) and mounted with fluorescence mounting medium (Dako Diagnostika, Hamburg, Germany). Sections were evaluated with a fluorescence microscope (Keyence BZ-9000E, Keyence Deutschland GmbH, Neu-Isenburg, Germany).
Histological evaluation
HE stained slides were evaluated by semi-quantitative scoring of neuronal necrosis, gliosis and inflammation for individual brain regions. The applied scoring system included three categories. Neuronal necrosis scores ranged from 0 to 3 (0 normal; 1 single necrotic neurons, 2 less than 30 % affected neurons; 3 more than 30 % affected neurons). Gliosis, interpreted as hyperplasia and hypertrophy of glial cells, ranged from 0 to 3 (0 normal; 1 multifocal, mild; 2 multifocal, moderate; 3 multifocal, severe) and inflammation scores ranged from 0 to 3 (0 normal; 1 single perivascular infiltrates; 2 2-3 layers of perivascular infiltrates; 3 more than 3 layers of perivascular infiltrates). TBEV scoring ranged from 0 to 3 (1 <30 % positive cells, 2 30-60 % positive cells; 3 >60 % positive cells). All scorings were performed using a defined area of high power fields (HPFs; 400x).
Statistics
Data was analyzed by GraphPad Prism Software 9. Survival curves are displayed by Kaplan-Meier curves using log rank test. Body weights are expressed by mean ± standard error (SD) in graphs and were analyzed using unpaired, two-tailed student’s - t test. Differences between viral load or HE scorings were analyzed using Mann-Whitney U test unless described differently. A p-value of < 0.05 was considered significant and indicated by (*).
Supplementary information
Supplementary Figure S1. Taq-Man qPCR analysis of the spinal cord of mock, TP21 and TBEV-280 immunized mice that were infected with TBEV-Hypr.
Supplementary Figure S2. Evaluation of neuronal necrosis by HE staining in other brain regions. Regions were scored from 0 to 3 (see methods) and statistical analyzed by Mann-Whitney test (*p < 0.05; n=6/group).
Supplementary Figure S3. Evaluation of inflammation using HE staining in different brain regions. Regions were scored from 0 to 3 (see methods) and statistical analyzed by Mann-Whitney test (*p < 0.05; n=6/group).
Supplementary Figure S4. Evaluation of gliosis using HE staining in different brain regions. Regions were scored from 0 to 3 (see methods) and statistical analyzed by Mann-Whitney test (*p < 0.05; n=6/group).
Abbreviations
- CNS
- Central nervous system
- dpi
- Days post infection
- EMEM
- Eagle’s Minimum Essential medium
- GFAP
- Glial fibrillary acidic protein marker
- HE
- Haematoxylin and eosin staining
- HPF
- High power fields
- Iba-1
- Ionized calcium binding adaptor molecule 1 marker
- IHC
- Immunohistochemistry
- LAV
- Live-attenuated vaccine
- LGTV
- Langat virus
- pfu
- Plaque-forming unit
- PVI
- Perivascular infiltrates
- qPCR
- Quantitative real-time PCR
- s.c.
- Subcutaneous
- TBE
- Tick-borne encephalitis
- TBEV
- Tick-borne encephalitis virus
- ZIKV
- Zika virus
Acknowledgements
We thank Prof. Matthias Niedrig former employee of the Department of Virology of the Robert-Koch-Institute Berlin, Germany (niedrigm{at}gmx.de) for providing the antibody against TBEV for these studies. We thank the Federal Ministry of Education and Research for funding this study within the TBENAGER grant for AO. This work was funded in part by the Alexander von Humboldt Foundation in the framework of the Alexander von Humboldt Professorship endowed by the German Federal Ministry of Education and Research. We thank the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – 398066876/GRK 2485/1, VIPER for funding parts of the study. We thank for a support from the Czech Science Foundation (grant No. 20-14325S to DR, and No. 20-30500S to MP).