The intrathecal expression and pathogenetic role of Th17 cytokines and CXCR2-binding chemokines in tick-borne encephalitis

Tick-borne encephalitis (TBE) is caused by a Flavivirus species (TBE virus, tick-borne encephalitis virus (TBEV)) transmitted by Ixodes ticks. The European TBEV subtype frequently causes a mild flu-like disease, but in some cases, it is able to penetrate the blood/brain barrier (BBB) causing the second, neurologic phase of the disease, which may take a severe course and leave permanent neurologic sequelae [1‐5]. Both the animal models and the results of human studies point to the involvement of the host immune/inflammatory response in the pathogenesis. TBEV entry into the central nervous system (CNS) probably occurs at least partially via transcellular pathway [6], but animal models of Flavivirus encephalitis suggest it may be facilitated initially by the disruption of BBB by the excessive systemic inflammation [7, 8] and subsequently by the virus-infected leukocytes migrating to the intrathecal inflammatory focus [9]. The further CNS tissue damage results from a combination of TBEV cytopathic effect on neurons [10] and of a vivid intrathecal inflammatory response, involving primarily Th1-subtype CD4+ and cytotoxic CD8+ lymphocytes [11‐15]. Interestingly, mouse strains more sensitive to TBEV present with a higher expression of the pro-inflammatory cytokines and chemokines during encephalitis [16].

The cerebrospinal fluid (CSF) of TBE patients is characterized by the increased albumin concentration and albumin CSF/serum index persisting during the convalescent period, indicating protracted BBB disruption, elevated IgG index pointing to the intensive intrathecal humoral response [17], and a tendency for a relatively low initial pleocytosis with a high neutrophil fraction [4, 18]. The neutrophils may constitute a majority of CSF leukocyte population at the initial diagnostic examinations, analogously to the infections with related neurotropic Flavivirus species, West Nile virus (WNV) and Japanese encephalitis virus (JEV) [19, 20]. This is typically accompanied by a moderate peripheral neutrophilia, a likely source of cell population penetrating into CSF. However, the factors driving this neutrophilic response and its role in the pathogenesis are not well characterized.

Neutrophils are involved in the early unspecific response and elimination of pathogens, which is typically accompanied by a significant local tissue damage [21, 22]. Through the secretion of cytokines including interleukin-12 (IL-12) and chemokines acting preferentially on Th1 (CXCL9, CXCL10, CXCL11) and Th17-lymphocytes (CCL2, CCL20), they participate in the initiation and shaping of the later stages of the response, favoring the development of Th1 and/or Th17-type milieu [21‐24]. Data from animal studies show that neutrophils are involved mainly during the onset of viral meningitis/encephalitis, after which they are replaced by mononuclear cells [25‐28]. In WNV infection in mice, they are upregulated both in periphery and in CNS at the onset of encephalitis analogously to human TBE [28]. In this model, infected neutrophils play an important pathogenic role as an early reservoir and a vehicle of virus entry into CNS and were found to co-localize with WNV in meninges, thus contributing prominently to the onset and severity of encephalitis [9, 27]. Later, during the established encephalitis, they seem to contribute to its control and, accordingly, their depletion is beneficial when occurring directly before, but harmful after an exposition to the virus [27]. The balance of the protective and harmful effects of the neutrophil infiltration in TBE is unknown and the existing data on other human Flavivirus infections are few and conflicting. For example, the activation of neutrophils in contact with JEV and an effective degradation of viral proteins and RNA has been observed in vitro suggesting a protective role [20], but intrathecal expression of a chemokine for neutrophils, IL-8, is associated with a poor prognosis in Japanese encephalitis (JE) in vivo [29]. Intrathecal neutrophils should influence the development of the next stages of the local inflammatory/immune response, which is consistent with the prevailing Th1-type response observed in TBE [13, 14, 30], and hints at the possibility of the coexistence of Th17-mediated response as well.

The Th17-type specific response, essential for the development of a sustained neutrophil-dominated inflammation, is driven by a subset of immune cells (including γδ lymphocytes, a differentiated Th17 subpopulation of classical CD3+CD4+ lymphocytes, and a specialized fraction of NK cells) able to secrete a set of cytokines including interleukin-17A (IL-17A), IL-17F, and IL-22 [23, 31‐36]. Th17-type cytokine response may be boosted in an autocrine/paracrine manner by IL-17 acting on CD4+ lymphocytes [34]. Having no receptors directly on neutrophils, IL-17 and IL-22 promote a neutrophil-dominated inflammation indirectly via mediators released from tissue cells like hepatocytes, epithelial, and endothelial cells [9, 23, 37]. Especially, they upregulate the expression of the subset of chemotactic cytokines (chemokines) specific for neutrophils, including IL-8 (CXCL8), CXCL1, CXCL2, and CXCL5, signaling through receptors CXCR1 (binding mainly IL-8) and CXCR2 (binding IL-8, CXCL1, CXCL2, and CXCL5) [23, 36‐40]. The neutrophils support the activation and migration of Th17 cells in a positive feedback loop, with the resultant response having a strong pro-inflammatory and self-sustaining character [23]. These characteristics make Th17-dependent inflammation deeply involved in the pathogenesis of autoimmunity, which was especially well documented in autoimmune arthritis [34, 41, 42].

Although originally linked to the response against extracellular pathogens in skin and mucous membranes, Th17 cells may contribute to the inflammation and immunopathology within CNS. IL-17 and IL-22 act on BBB endothelial cells, increasing secretion of IL-6, CCL2, CXCL1, CXCL2, and IL-8; increasing BBB permeability; and facilitating transmigration of CD4+ cells [9, 43]. In that way, they may participate in the onset of encephalitis promoting BBB disruption and CNS inflammation. Moreover, human Th17 lymphocytes preferentially transmigrate across BBB and may have a cytolytic effect on neurons through the expression of granzyme B, directly contributing to the CNS damage [43]. The role of Th17 response in human CNS pathology was confirmed in multiple sclerosis [43, 44], and increased concentrations of CXCL1, IL-8, and granulocyte colony-stimulating factor (G-CSF) were detected in CSF of patients with acute disseminated encephalomyelitis [45]. In TBE and other Flavivirus neuroinfections, the involvement of Th17 lymphocytes and related cytokines might explain both the presence of the substantial neutrophil infiltrate and the significant immunopathology. The study by Wang et al. on WNV-infected mice confirmed the intrathecal expression and pathogenetic role of IL-22 and CXCL1 [9]. Pietikäinen et al. detected highly increased concentrations of CXCL1, IL-8, IL-17, and G-CSF in CSF of a group of TBE patients [46]. IL-8 was also detected in TBEV-infected astrocytes and in the serum of TBE patients, but its chemotactic gradient towards CSF in vivo has not been directly studied [14, 47].

We have attempted to gather more data on the role of neutrophils in the pathogenesis of TBE in humans and on the expression of the cytokines which might contribute to their recruitment into CNS. We have analyzed retrospectively the correlations between the CSF neutrophil count and other laboratory and clinical parameters, including the disease manifestation and severity, in TBE patients hospitalized in six consecutive years in a reference center. We have also analyzed concentrations of chemokines for neutrophils IL-8, CXCL1, and CXCL2, as well as Th17 cytokines IL-17A, IL-17F, and IL-22 in CSF and serum in prospectively recruited TBE patients, to assess their peripheral and intrathecal expression, concentration gradients, and relation to the CSF inflammatory parameters.

The results of the retrospective study are in accordance with previous data and also give additional insights into the mechanism and consequences of the neutrophil CSF infiltration. The CSF neutrophil fraction was similar to previously described by Mickiené et al. in TBE, and somewhat lower than that reported in WNV-infected patient groups, which was 41% in meningitis and 45% in encephalitis according to Tyler et al., and 50% decreasing to 37% over 3 days since onset according to Rawal et al. [4, 19, 48]. In agreement with the trend observed in the later study, the presence of neutrophils in CSF of TBE patients was transient, giving a way to a convalescent phase pleocytosis of almost entirely mononuclear cells. The CSF neutrophil count did not correlate with a C-reactive protein concentration, suggesting it was not directly dependent on the systemic inflammatory response, and although it could be facilitated by high numbers of circulating neutrophils, it was not evidently dependent on neutrophilia and probably driven by other, likely intrathecal, factors. The presence of neutrophils in CSF was not associated with the BBB disruption, and it did not correlate with CSF lymphocyte count, pointing to independent mechanisms behind the migration of these two cell populations into CSF. In the WNV encephalitis model described by Bai et al., neutrophils co-accumulated with T CD8+ lymphocytes and the migration of the later was partly dependent on CXCL1 and CXCL2, either because of their direct response to these chemokines or because of the cooperation and co-migration with neutrophils [27]. Although such a relation with a particular population of lymphocytes would be of pathogenetic significance, it could not be evaluated in our retrospective data and requires further studies. Taken together, our results suggest an early massive recruitment of neutrophils into CSF, not dependent directly on the peripheral inflammation, lymphocyte influx, and BBB function, and possibly facilitated by, but not critically dependent on the presence of an increased population of circulating neutrophils.

The data on the pathogenetic role of the neutrophilic CNS infiltration in Flavivirus encephalitis come mainly from animal models and remain conflicting. In the mouse model of viral encephalitis studied by Zhou et al., neutrophils constitute the first population of cells infiltrating CNS and their depletion leads to an increased virus replication and impaired migration of mononuclear cells. The authors hypothesized that neutrophils initiate the protective accumulation of mononuclear cells within CNS by first contributing to BBB disruption and then by secreting chemokines including CCL3, CCL4, CXCL9, CXCL10, and CXCL11 [25], some of which have actually been detected in TBE CSF [13, 49]. On the other hand, a number of studies suggest a role of neutrophils in the development of the severe Flavivirus encephalitis, which could be mediated either by an increased viral load entering CNS with migrating neutrophils or by their direct contribution to the intrathecal immunopathology. In mice infected with Murray Valley encephalitis virus, both the depletion of the neutrophils dominating in the brain infiltrate and the inhibition of the inducible nitric oxide synthase (iNOS) prolong the survival, suggesting an immunopathology dependent on the neutrophil-generated NO [26]. In the JE model, the severity of encephalitis is related to skewing of the specific cellular response from regulatory (Treg) to Th17-type, with the increased IL-17, CXCL1, and CXCL2 expression promoting the CNS infiltration by granulocytes, monocytes, and cytotoxic CD8+ lymphocytes [50]. In mice infected with Langat virus, a prolonged neutrophil influx into CNS is a hallmark of a severe encephalitic phenotype [51]. The Bai et al. study on WNV encephalitis mouse model suggests a more complicated scenario, with infected neutrophils contributing to the virus entry into CNS, but later playing a protective role during an established CNS infection [27]. However, as the survival of WNV infection in mice lacking IL-22 and CXCR2 signaling is improved, the overall effect of neutrophil infiltration seems unfavorable [9, 29, 52].

The significance of these observations for the pathogenesis of human Flavivirus encephalitis remains unclear. In WNV patients described by Tyler et al., the CSF parameters did not differentiate meningitis from encephalitis and although a higher pleocytosis, neutrophil fraction, and CSF protein concentration correlated with a worse outcome, the prognostic value of individual factors was weak [19]. In the study of Mickiené et al., there was no significant association of CSF neutrophils with the severity of TBE, although the data presented by the authors may suggest a non-significant trend towards higher neutrophil fraction in more severely ill patients [4]. The sensitivity of our own analysis was limited by its retrospective character and by a small number of patients with severe meningoencephalitis. The peripheral neutrophilia tended to be the highest in patients with the most severe neurologic presentation and with permanent neurologic sequelae, consistent with a possible role of circulating neutrophils in a virus spread and migration into CNS in accordance with animal models [10, 27], but a small number of patients and possible confounding factors in severely ill subjects make this observation inconclusive. The CSF neutrophils on admission did not correlate with clinical presentation, severity, or outcome of the CNS involvement, with an exception of a higher neutrophil count in patients with myelitis, hinting at the specific pathogenesis of the spinal involvement in TBE. The presence of the residual neutrophils in the convalescent CSF was associated with encephalitis and with worse neurologic outcome, which may reflect a prolonged intrathecal inflammation in more severe TBE cases, but it is not clear if neutrophils played any specific pathogenetic role at this late stage of the disease. Interestingly, it resembles the results of a Langat virus study by Michlmayr et al. in which neutrophil counts within CNS were initially similar between resistant and sensitive mice, but differed at the later stage of the encephalitis [51].

To identify mediators of the neutrophilic response in TBE, we have first verified the synthesis of IL-8 described in previous studies. IL-8 was previously reported in CSF of patients with aseptic meningitis of various etiology [53, 54], and its upregulation has been well documented in JEV infection: it has been detected in serum of persons receiving live attenuated anti-JEV vaccine [55] and in serum and CSF of JE patients, in whom it correlates both with the neutrophil counts and with the disease severity [29, 56, 57]. Recently, Palus et al. have confirmed increased IL-8 concentration in serum of TBE patients [47] and Pietikäinen described its moderate increase in their CSF [46], although no direct comparison of serum and CSF levels was made. Our study confirmed that IL-8 is upregulated mainly intrathecally, creating a steep concentration gradient between CSF and serum, consistent with its role as a chemotactic factor.

Following the IL-8 detection, we have studied expression of two additional chemokines for neutrophils (CXCL1 and CXCL2), as well as of Th17 cytokines previously detected in animal models of Flavivirus encephalitis [9, 27, 50, 58, 59]. In general, our findings confirm the intrathecal synthesis of all these mediators and suggest their contribution to CNS neutrophilic infiltration. The concentrations of CXCL1 in CSF, of IL-17F in both serum and CSF, and of CXCL2 in serum correlated specifically with CSF neutrophil count, but not with the lymphocyte count or markers of BBB integrity. As particular cytokines presented with different dynamics of expression and concentration gradients between serum and CSF, their role in the pathogenesis might vary. Increased concentration of IL-17A in serum suggests its involvement in the peripheral response, while IL-22 expression was mainly intrathecal. Of the CXCR2-agonistic chemokines, CXCL1, but not CXCL2, created a chemotactic gradient towards CSF and correlated significantly with the CSF neutrophil count.

Besides IL-8, there are few data on the expression of Th17 cytokines and chemokines for neutrophils in other human CNS infections to provide context for our TBE observations. According to Pietikäinen et al., the concentration of IL-17 is higher in CSF of TBE patients in comparison with neuroborreliosis, a condition characterized by almost exclusively lymphocytic pleocytosis [46]. Zhang et al. described increased IL-22 and decreased IL-17A levels in serum of children with enterovirus encephalitis qualitatively different from the pattern of response we detect in TBE [60]. In our study, of cytokines compared between TBE and non-TBE meningitis groups, the CSF concentration of IL-8 and IL-17A was lower in TBE, of IL-17F tended to be higher, and of CXCL2 did not differ. Thus, although non-TBE patients presented with a clearly different clinical picture of mild, quickly resolving, and uncomplicated meningitis, with no peripheral neutrophilia and mononuclear CSF infiltrate, the observed profile of the expression of Th17-related cytokines do not explain these differences. The differentiating factors, however, could be pro-neutrophilic cytokines not investigated in our study, like CXCL5, or not included in this part of the analysis: IL-22 and CXCL1.

Both our data and previous animal results point to an important role of IL-22 in the pathogenesis of Flavivirus neuroinfection, particularly in mediating the neutrophil migration. In the mouse model of WNV encephalitis, the lack of IL-22 expression decreased CNS viral load, infiltration by neutrophils, and mortality rate, attenuating all the aspects of the intrathecal infection and inflammation. However, when WNV was injected intrathecally, IL-22 was moderately protective [9]. These results show that IL-22 signaling promotes the onset of Flavivirus encephalitis but later contributes to the intrathecal protective response, which resembles the sequential role of neutrophils in WNV infection as described by Bai et al. [9, 27]. These effects of IL-22 are mediated by an upregulation of CXCR2 on neutrophils and the intrathecal expression of CXCL1 and CXCL5 [9]. According to Michlmayr et al., an increase of mRNA levels for CXCL1 and CXCL2 in the brains of WNV-infected mice may be as high as 1000-fold [58]. In vitro, human brain microvasculature endothelial cells secrete CXCL1 and CXCL5 and favor transmigration of neutrophils under IL-22 stimulation [9], confirming that findings from animal experiments may apply to human encephalitis. In TBE, we have observed an intrathecal expression of IL-22 and especially of CXCL1, the later associating significantly both with the encephalitic presentation and with high CSF neutrophil counts and thus reproducing important features of these animal and in vitro models.

These results may have implications for a better understanding of a still debatable route of TBEV entry into CNS, a crucial event differentiating between the mild peripheral and potentially severe neuroinvasive infection. In vitro TBEV is able to cross BBB by infecting and replicating persistently within brain endothelial cells, without directly affecting BBB structure and permeability [6]. This forms a likely route of the initial CNS entry during the peripheral phase, but additional migration mechanisms that could increase CNS viral load are possible and supported by mouse WNV encephalitis models [7‐9]. Previously, we have studied the postulated mode of CNS entry following BBB disruption by a systemic inflammatory response mediated by macrophage migration inflammatory factor (MIF) and tumor necrosis factor alpha (TNFα)––a mechanism that seems feasible but remains unproven and does not exclude the existence of other pathogenically and clinically relevant routes of infection [61]. The current results are consistent with the alternative possibility of the migration of the infected neutrophils dependent on intrathecal IL-22 and CXCL1 as a clinically important route of TBEV entry into CNS. This transition to the neurologic phase of TBE must be determined during a peripheral phase and a study of the cytokine expression in patients at that stage of the disease, as well as in patients with a flu-like TBEV infection not progressing to neurologic phase, could be highly informative as for factors determining the neuroinvasion. However, such study is very difficult to arrange in a clinical setting, as a TBEV etiology is only rarely suspected at that stage of the disease. Our data from 4 patients seen in the peripheral or at the onset of the neurologic phase, although limited, offer a unique insight into this key period of the TBEV infection. The peripheral phase cytokine and chemokine concentrations were at least comparable with neurologic phase values and for Th17 cytokines tended to be even higher. The early systemic Th17-type response could facilitate TBEV neuroinvasion in both postulated mechanisms, contributing to BBB disruption caused by the systemic inflammation while in the same time providing a pool of circulating neutrophils able to infiltrate CNS. In animal experiments, CXCR2 signaling contributes to the neutrophil migration from the bone marrow into blood under stress stimuli [62] and IL-22 causes neutrophilia mediated by CXCL1 release from hepatocytes [37, 43], and similar mechanisms could play a role in neutrophil mobilization and migration in TBE patients, contributing to a viral spread within infected cells. More surprisingly, the cytokines tended to be already upregulated intrathecally in 2 TBE patients with peripheral infection, before the onset of any changes in the general CSF examination, suggesting that the intrathecal inflammatory process undetectable with standard diagnostic techniques had already been initiated. Possibly, Th17 cytokines in CSF could be even applied as prognostic markers of a developing neuroinfection in patients with a suspected TBE and an unremarkable result of the general CSF examination.

We found the concentrations of pro-neutrophilic cytokines to decrease in the early convalescent period of TBE, confirming the association of a Th17 response with its early stages. However, this early response had a lasting influence possible to trace in the convalescent CSF parameters. Especially, the CSF concentration of IL-17A correlated negatively with the residual CSF neutrophil infiltrate, which in the retrospective study associated with a more severe presentation. According to Zhou et al. the protective mononuclear CNS infiltrate is facilitated by neutrophils, but once it is established, it may suppress further neutrophil migration, for example, through effects of interferon gamma (IFN-γ) on the local chemokine synthesis [25]. According to this model, the initial vivid Th17-type inflammation should be short-lived and associated with a good recovery, which is consistent with a coexistence of an initial high IL-17 expression, prompt elimination of neutrophils from CSF, and a good clinical outcome we observe. This may document a secondary protective effect of the neutrophilic response in encephalitis analogous to that described by Bai et al. in the animal model [27]. Thus, the complex roles played by neutrophils in animal models seem to be observable in a human TBE as well. This may explain the lack of a straightforward correlation of the neutrophilic response with the clinical severity, as different protective and pathogenic effects may be difficult to trace and disentangle in individual patients.

Neutrophils migrate into CSF of patients with TBE independently of the lymphocyte influx and BBB disruption. The CSF neutrophil count does not differ between meningitis and meningoencephalitis, but is higher in myelitis, and normalizes with delay in patients with encephalitis and with persistent neurologic complications, suggesting a complex involvement in the intrathecal pathology. The pro-neutrophilic Th17 cytokines and chemokines signaling through CXCR1 and CXCR2 are upregulated intrathecally, and at least some of them (IL-17A) also systemically, starting from the peripheral and continuing into the neurologic phase of TBE, which allows them to exert their influence during the critical events deciding about the onset and severity of the CNS involvement. Of Th17 chemokines, IL-22 is most vividly expressed intrathecally. Of chemokines attracting neutrophils at least IL-8 and CXCL1 create chemotactic gradients towards CSF and may be responsible for neutrophil recruitment, and CXCL1 concentration correlates with CSF neutrophil count and with encephalitic presentation. By analogy with animal models, a signaling axis involving IL-22, its mediator chemokines and CXCR1/CXCR2 receptors may increase the TBEV influx into CNS with infected neutrophils, thus providing an important route of neuroinvasion leading to a more severe CNS involvement. After the encephalitis is established, however, an intrathecal neutrophilic response may be protective, contributing to its control and eradication.