Research ArticleAtrial Fibrillation

Cardiac glial cells release neurotrophic S100B upon catheter-based treatment of atrial fibrillation

See allHide authors and affiliations

Science Translational Medicine  22 May 2019:
Vol. 11, Issue 493, eaav7770
DOI: 10.1126/scitranslmed.aav7770

Treating the nervous ticker

Catheter ablation induces tissue damage in specific regions of the heart in patients with atrial fibrillation (rapid irregular heartbeat) to correct abnormal electrical signals. Although effective, the underlying mechanism remains incompletely understood. Scherschel et al. demonstrated that cryothermal ablation induced release of S100B, a marker of neuronal injury, from glial cells within mouse hearts. S100B stimulated nerve growth and reduced neuronal electrical activity. In patients receiving ablations, higher serum concentration of S100B after the procedure was associated with lower recurrence of atrial fibrillation. Further study is warranted to understand how ablation-induced injury to the intrinsic cardiac autonomic nervous system contributes to patient outcomes.


Atrial fibrillation (AF), the most common sustained heart rhythm disorder worldwide, is linked to dysfunction of the intrinsic cardiac autonomic nervous system (ICNS). The role of ICNS damage occurring during catheter-based treatment of AF, which is the therapy of choice for many patients, remains controversial. We show here that the neuronal injury marker S100B is expressed in cardiac glia throughout the ICNS and is released specifically upon catheter ablation of AF. Patients with higher S100B release were more likely to be AF free during follow-up. Subsequent in vitro studies revealed that murine intracardiac neurons react to S100B with diminished action potential firing and increased neurite growth. This suggests that release of S100B from cardiac glia upon catheter-based treatment of AF is a hallmark of acute neural damage that contributes to nerve sprouting and can be used to assess ICNS damage.


Atrial fibrillation (AF) is an epidemic heart rhythm disorder resulting in increased morbidity and mortality (1). Catheter ablation is currently the most effective therapy for patients with symptomatic AF. During this treatment, the pulmonary veins are electrically isolated from the atrial myocardium by thermal tissue damage [pulmonary vein isolation (PVI)] (2). Besides abolishment of abnormal electrical impulses originating from the pulmonary veins, the mechanisms by which catheter ablation is beneficial are not fully understood because both myocardial and neural structures can be affected (3). It has been suggested that PVI modulates the intrinsic cardiac nervous system (ICNS), which could explain the beneficial effect of ablation because some of the key elements of the ICNS are located within epicardial fat pads in close proximity to ablation sites near the pulmonary veins (4). The ICNS—a complex network of multiple types of neurons, glial cells, and interconnecting and end organ–targeting fibers—regulates cardiac physiology on a beat-to-beat basis (5, 6). Collateral damage of this delicate network during catheter ablation of AF is well known (2), but how this affects patient outcome remains controversial. The relevance of ICNS activity in the pathophysiology of AF has long been established (7), and some studies have even suggested that targeted damage of the ICNS additionally to PVI reduces AF recurrences (8, 9). However, others have shown that targeted disruption of the ICNS has no additional benefit or might even be proarrhythmic (10, 11). To elucidate the underlying mechanisms, sensitive methods are needed to evaluate acute intracardiac neural damage.

Neural injuries lead to the release of damage-associated molecular pattern proteins (DAMPs) (12). S100B is a DAMP that is abundantly expressed in glial cells and some neuronal populations in the central and peripheral nervous system (13). It exerts a variety of intra- and extracellular functions, mediating neurotrophic activity, survival, as well as neuronal electrical activity (14, 15). S100B is secreted constitutively but has also been characterized as a marker for several forms of neural damages because of its acute release immediately after injury (14, 16). S100B is required for regeneration in the peripheral nervous system (17), exerting effects via the receptor of advanced glycosylation end products (RAGE) (14, 18). Whether these observations applied to intracardiac neurons was previously unknown.

Nerve fibers disrupted by AF ablation can functionally recover (19). Thus, we aimed to characterize the neurotrophic factor S100B in intracardiac neural damage induced by AF ablation to elucidate the influence of intracardiac neural damage on patient outcome. Here, we show that S100B is expressed in glial cells of the ICNS and is specifically released into the blood of patients after ablation of AF. Patients with higher concentrations of S100B after ablation had less AF recurrences in a 6-month follow-up, supporting a beneficial effect of ICNS damage. S100B diminished action potential (AP) firing and induced neurite growth in intracardiac neurons in vitro, indicating a potentially proarrhythmic contribution to postprocedural nerve sprouting.


S100B expression in glial cells within the ICNS hierarchy

Key components of the ICNS are located in the proximity of the pulmonary veins (Fig. 1A), where their activity has been found to trigger AF (7) and where AF ablation is performed (Fig. 1B). Patients with AF receiving a catheter ablation (PVI group, n = 112) did not show differences in baseline plasma S100B concentration compared with controls without AF receiving a sole electrophysiological study (EPS, n = 19) or an ablation for typical atrial flutter in the right atrium [cavotricuspid isthmus (CTI) ablation, n = 20; Fig. 1C]. S100B measured in human atrial pectinate muscles revealed a total content of 422 ± 65.4 ng/g atrial tissue, confirming relevant amounts of S100B were present in cardiac tissue (Fig. 1D). To investigate S100B expression within the ICNS, we used immunohistochemical costaining with the neural marker β-III-tubulin (TUBB3; Fig. 1E). This revealed that S100B-expressing cells adjoin neural structures as follows: In intracardiac ganglia, S100B-expressing glial cells densely surround large neuronal cell bodies. In large cardiac nerves of the left atrial myocardium, S100B+ glial cells are found within the nerve ensheathing the single fibers. On a cellular level, delicate nerve fibers innervating single cardiomyocytes were found adjacent to glial cells (fig. S1, A and B). Staining for Sox10, which is expressed in the nuclei of glia at all developmental stages (20), confirmed the identity of S100B+ cells (fig. S1C). The S100B receptor RAGE was found to be expressed in the somata of intracardiac neurons as well as in nerve fibers, colocalizing with TUBB3, verifying the presence of the S100B receptor in intracardiac nerves (Fig. 1E).

Fig. 1 S100B in the human ICNS.

(A) Photograph of a human heart showing a posterior view of the left atrium (LA) and the pulmonary veins where cardiac ganglia and nerves are located. (B) Anatomical three-dimensional map of the LA (posterior view). The red dots show the areas in which AF ablation via PVI is performed. (C) Preprocedural S100B plasma concentrations in patients receiving a sole diagnostic EPS versus patients receiving an ablation of the muscular CTI versus AF ablation (PVI) [EPS: n = 19, CTI: n = 20, PVI: n = 112; data presented as box plots (minimum to maximum); P < 0.05 was considered statistically significant; compared using Kruskal-Wallis test]. ns, not significant. (D) S100B content of human atrial pectinate muscles per gram of tissue (n = 5 tissue samples, data presented as means ± SEM). (E) Immunohistological staining of S100B, TUBB3, and RAGE in intracardiac ganglia, cardiac nerves, and delicate nerve fibers within the atrial myocardium. LSPV, left superior pulmonary vein; RSPV, right superior pulmonary vein; LIPV, left inferior pulmonary vein; RIPV, right inferior pulmonary vein; LAA, left atrial appendage; RA, right atrium; LV, left ventricle; RV, right ventricle.

Increase in S100B plasma concentrations in patients after AF ablation

To investigate whether intracardiac neural damage leads to changes in S100B plasma concentrations, we analyzed blood samples before and immediately after catheter-based treatment of AF. Baseline characteristics of all patients are given in table S1. In the PVI group, S100B plasma concentrations increased from 33.8 ± 1.5 to 81.9 ± 4.1 pg/ml (P < 0.0001; Fig. 2A). In contrast, only small changes in S100B plasma concentrations were detectable in patients receiving an EPS (no ablation) (31.3 ± 4.0 to 41.0 ± 6.4 pg/ml, P = 0.0276). This was also the case in the CTI group receiving ablation of a sparsely innervated muscular sleeve in the right atrium, between the inferior caval vein and the tricuspid valve (21) (29.5 ± 2.5 to 36.3 ± 3.6 pg/ml, P = 0.0032). To compare myocardial damage between the groups, we used a high-sensitivity detection assay to quantify troponin I (hsTnI) (22). In the PVI group, hsTnI increased from 4.1 ± 0.3 to 151.5 ± 11.39 pg/ml (P < 0.0001, Wilcoxon test; Fig. 2B) after ablation, which was comparable to hsTnI change in the CTI group (from 14.6 ± 3.1 to 188.5 ± 26.27 pg/ml, P < 0.0001, Wilcoxon test). Only small increases were detected in the EPS group, from 7.84 ± 2.1.23 to 14.2 ± 3.6 pg/ml (P = 0.0017, Wilcoxon test). Normalization of ΔS100B to ΔhsTnI shows a 4.5 times higher S100B concentration in the PVI group compared with the CTI group, suggesting more neural damage in AF ablation with the same amount of myocardial damage (0.05 ± 0.02 to 0.33 ± 0.03, P < 0.0001, Fig. 2C). Corresponding to the well-known cholinergic denervation resulting from AF ablation, the mean heart rate (HR) increased from 60 ± 1 to 69 ± 1 bpm (P = 0.0001, n = 72; Fig. 2D) within 24 to 48 hours after the procedure but did not increase in the control groups (EPS: 61 ± 3 to 62 ± 3 bpm, P = 0.8956, n = 15; CTI: 60 ± 3 to 66 ± 3 bpm, P = 0.0646, n = 9). To further characterize autonomic function, heart rate variability (HRV) was analyzed before and after PVI in short-term electrocardiogram (ECG) recordings (table S2). Parameters indicating parasympathetic function were reduced (SD of normal RR intervals: 32.2 ± 10.1 to 13.3 ± 9.8 ms, P = 0.0020; the square root of the mean squared differences of successive NN intervals: 19.4 ± 8.8 to 8.8 ± 8.3 ms, P = 0.0049). Low-frequency (LF) component, a parameter used in several studies as an indirect parameter of sympathetic function (23, 24), was lower after PVI (257.8 ± 166.7 to 74.8 ± 187.5 ms2, P = 0.0322), indicating ablation damaged sympathetic structures in addition to parasympathetic structures. Reduction in the LF/HF ratio, a parameter for sympathovagal balance, was reduced (from 2.8 ± 2.4 to 1.1 ± 0.7, P = 0.0426), indicating a stronger damage of parasympathetic cell bodies and fibers. Supporting this, immunohistochemical costaining of S100B with the sympathetic marker tyrosine hydroxylase (fig. S2) in the human atrial myocardium showed that sympathetic fibers are also ensheathed by S100B-expressing glial cells. This is in line with other studies showing intertwined sympathetic and parasympathetic nerve fibers in human atrial myocardium (25, 26).

Fig. 2 Plasma S100B in patients undergoing AF ablation.

(A) Pre- to postprocedural S100B concentrations (ΔS100B) in patients receiving AF ablation (PVI) compared with sole diagnostic EPS or ablation of the muscular CTI [EPS: n = 19, CTI: n = 20, PVI: n = 112; data presented as box plots (minimum to maximum); ****P < 0.0001, compared using Kruskal-Wallis test]. (B) Changes of pre- to postprocedural troponin I (ΔhsTnI) between the groups (EPS: n = 14, CTI: n = 19, PVI: n = 101; ****P < 0.0001, compared using Kruskal-Wallis test). (C) Normalization of ΔS100B/ΔhsTnI was performed to account for tissue damage (****P < 0.0001, compared using Mann-Whitney test). (D) Change in patients’ HR after AF ablation, EPS, or CTI (only patients in sinus rhythm; EPS: n = 15, CTI: n = 9, PVI: n = 72; ***P = 0.0008, compared using Kruskal-Wallis test with Dunn’s posttest for multiple comparison against EPS group). (E) S100B concentrations in patients with and without recurrences of AF [number of patients: none, n = 67; 0 to 3 months, n = 29; 3 to 6 months, n = 19; data are presented as box plots (minimum to maximum); *P = 0.0076, compared using Kruskal-Wallis test]. (F) Kaplan-Meier curve of freedom from AF stratified by plasma concentrations of ΔS100B (n = 103; P = 0.028).

Six months of follow-up was available for 103 patients in the PVI group (92.0%). Of these, 29 patients (28.2%) had a documented AF episode within the first 3 months after ablation. A total of 19 patients (18.4%) had a documented AF episode between 3 and 6 months after ablation. Independent of the time of recurrences, patients without AF episodes had a higher S100B release after ablation (no recurrences: 57.3 ± 5.1 pg/ml; recurrence 0 to 3 months after ablation: 35.0 ± 5.8 pg/ml; recurrence 3 to 6 months after ablation: 30.7 ± 5.9 pg/ml; P = 0.0135; Fig. 2E). Outcome of ablation for the 6 months of follow-up is depicted in the Kaplan-Meier curve (Fig. 2F). Patients with higher ΔS100B showed a better outcome compared with the patients with lower ΔS100B concentrations (P = 0.028).

S100B release from the ICNS

Because S100B released from the central nervous system might be a potential confounder for the increase in plasma concentrations, we used ex vivo mouse hearts without central innervation to verify S100B release specifically from cardiac nerves. We confirmed the expression of S100B in nerves (Fig. 3A) and ganglia (Fig. 3B) within murine hearts. Next, we induced cryothermal damage to mouse hearts at the posterior wall of the left atrium, where neural structures are located (boxed area in Fig. 3C). S100B concentrations in the perfusion fluid of these hearts increased acutely after thermal damage (4.7 ± 0.7 pg/ml versus 13.2 ± 2.7 pg/ml, P = 0.0045; Fig. 3D). Hematoxylin and eosin (H&E) staining of cardiac sections after acute cryodamage showed some ganglia with acute signs of thermal damage and necrosis, whereas others appeared inconspicuous (Fig. 3E). Immunohistochemical detection of ChAT and S100B in ganglia with cryodamage was diffuse (Fig. 3F), indicating loss of structural integrity (27).

Fig. 3 S100B release from the ICNS in murine hearts ex vivo.

S100B expression in the murine ICNS in nerve fibers (A) and intracardiac ganglia (B), as shown by the costaining with choline acetyltransferase (ChAT). White boxed region is shown at higher magnification in the rightmost panel. S100B+ glial cells are marked with white arrows in the magnification. (C) Photograph of a murine heart showing the area surrounding the pulmonary veins where cryogenic damage was generated (black box). (D) S100B measured in the perfusion fluid before and after cryogenic damage in ex vivo mouse hearts (n = 8; **P = 0.0045, compared using paired t test). (E) Histological characterization of a murine heart after acute cryoablation. H&E staining of a murine cardiac ganglion without (left) and with acute signs of thermal damage (right) from a heart after ex vivo cryoablation. Cellular shrinkage marked by black arrows. (F) Staining of ChAT and S100B in cryodamaged ganglia. Glial cells are marked by white arrows. White boxed region in the merged panel is shown at higher magnification in the rightmost panel.

The release of S100B by peripheral nerves was further verified by data from one patient in whom a temporary phrenic nerve damage occurred during ablation. Nerve damage was detected during routine continuous monitoring throughout the procedure (fig. S3). S100B concentration in this patient was elevated compared with the mean increase in the PVI group (326.9 pg/ml versus 47.9 pg/ml mean increase in the PVI group; fig. S3). To rule out adipocyte damage as a potential source for S100B (28), we investigated activin A, a marker for cardiac fat tissue (29). Activin A was increased in the PVI as well as in the CTI control group after the procedure (PVI: n = 34, 243.2 ± 19.4 to 2035.1 ± 70.8 pg/ml, P < 0.0001; CTI: n = 5, 274.3 ± 49.3 to 1897.4 ± 238.0 pg/ml, P = 0.0014; fig. S4). Because S100B was increased in the PVI but not in the CTI group, this indicates that the release is specific to nerve and not to adipocyte damage.

S100B affects electrophysiology of intracardiac neurons but not cardiomyocytes

S100B has been shown to alter central neuronal cellular electrophysiology in mollusks (30). To evaluate whether this also holds true for intracardiac neurons or the myocardium itself, we analyzed the effects of S100B on isolated murine intracardiac neurons, murine whole hearts, and human atrial myocardial tissue. The exposure of intracardiac neurons to recombinant S100B strongly diminished the AP firing rate during long-lasting depolarizing pulses (n = 5; Fig. 4A). This effect developed within minutes and was not reversible upon washout during the duration of the experiment (Fig. 4B). Normalized firing frequency showed a significant reduction after 7 min of S100B treatment with a saturating effect after 15 min, when firing frequency was reduced to 52 ± 18% of time-matched controls (P < 0.05; Fig. 4C). The AP firing frequency in controls remained stable over a 20-min period (n = 6), excluding rundown effects (Fig. 4C). We did not detect changes in AP shape in isolated human atrial pectinate muscles after 20 min of incubation with S100B (Fig. 4, D and E, and table S3). Similarly, electrophysiological parameters in murine whole hearts did not change upon exposure to S100B (Fig. 4F).

Fig. 4 S100B diminishes neuronal, but not myocardial, AP firing.

(A) Exemplary trains of APs evoked by long-lasting (500 ms) depolarizing current pulses (150 pA, amplitude) in control conditions (upper trace), and 10 min after application of recombinant S100B (100 ng/ml) (lower trace) in single murine intracardiac neurons. (B) Plot showing the time course of the effect of S100B on firing frequency in single murine intracardiac neurons. (C) Averaged normalized AP frequency from intracardiac neurons as a function of time in the absence (TimeCTRL) and the presence of S100B [TimeCTRL n = 6 cells, S100B n = 5 cells; *P < 0.05, compared using two-way repeated-measures analysis of variance (ANOVA)]. (D) Exemplary AP shape in human atrial pectinate muscles at basal state and upon exposure to S100B (100 ng/ml) (n = 3 independent experiments). (E) Continuous measurement of AP duration at 90% of repolarization (APD90) in human atrial myocardium (paced at 1 Hz) from 5 min before to 20 min after the superfusion with S100B (100 ng/ml). (F) Cardiac electrophysiological parameters in control hearts and hearts infused with recombinant S100B [n = 6 per group; data are presented as box plots (minimum to maximum); compared using Mann-Whitney or unpaired t test as appropriate; P < 0.05 was considered statistically significant]. ARP, atrial refractory period; WBP, Wenckebach period; AVNRP, AV nodal recovery period; VRP, ventricular refractory period.

S100B induces neurite growth in intracardiac neurons

Because S100B is known to positively affect neurite growth (18), we aimed to evaluate its role in the ICNS. After infusion of S100B into murine hearts, an enrichment of S100B was found in cardiac nerves and ganglia, but not in the surrounding adipose tissue (Fig. 5A). We then cultivated intracardiac neurons from adult mice with recombinant S100B. After 16 hours in the presence of S100B (100 ng/ml), neurons had longer neurites compared with the negative control (812.1 ± 41.9 to 616.1 ± 37.2 μm, P = 0.0009; Fig. 5, B and C). In patients, we were still able to detect elevated S100B plasma concentrations on the day (19.8 ± 4.1 hours) after the ablation procedure (39.8 ± 11.5 pg/ml versus 33.6 ± 11.5 pg/ml at baseline, n = 18, P = 0.002; Fig. 5D).

Fig. 5 S100B is taken up by intracardiac neurons and increases neurite outgrowth.

(A) Ex vivo murine hearts stained for S100B showing enrichment in intracardiac nerves and ganglia after infusion of recombinant S100B (100 ng/ml, 90 min). Boxed regions are shown at higher magnification. (B) Images and (C) quantification of neurite lengths in murine intracardiac neurons cultivated for 18 hours with recombinant S100B (100 ng/ml) [positive control: 5% fetal calf serum n = 29, negative control n = 108, S100B (100 ng/ml) n = 113; data are presented as means ± SEM and represent five independent experiments with two to four replicates per condition; **P = 0.0011, ***P = 0.0009; groups were compared with Kruskal-Wallis and Dunn’s posttest for multiple comparisons against negative control]. (D) Quantification of S100B in plasma from patients before and 1 day after ablation (n = 17; **P = 0.002, compared using paired t test).


More than 100,000 catheter-based AF ablations are performed annually in Europe alone (31), and the number is increasing (32). Although the relevance of the ICNS in the pathophysiology of AF has long been established (7), the impact of intracardiac neural damage on cardiac electrophysiology during AF ablation remains unclear. We found that the neural damage marker S100B is released from cardiac glial cells upon ICNS damage and that S100B increases neurite growth, thus potentially contributing to intracardiac nerve sprouting after ablation.

The S100 protein family consists of more than 20 calcium-binding proteins, each with unique characteristics such as cell-specific expression and specific intra- and extracellular functions (14, 15). Among the first to be discovered in 1965 was S100B (33), a protein expressed and secreted by glial cells, which was shown to functionally modulate neuronal survival and outgrowth 20 years later (34). Because of its immediate release after neural damage, S100B is the member of the family that is most extensively characterized as a biomarker for central neural damage (14) and was recently added to the Scandinavian guidelines for the management of traumatic head injuries (35). Even though most clinical studies focus on the release of S100B in central neural damage, it is also widely considered a marker for glial cells of the peripheral nervous system (36). Here, we established that S100B is prominently expressed in glial cells adjoining neural structures throughout the ICNS hierarchy in humans and mice. Together with the release of S100B into the blood of patients after ablation of the left atrium, this indicates that S100B is released because of ICNS damage. Still, S100B from central sources, for example due to asymptomatic cerebral emboli as a side effect of ablation, might be a potential confounder (37). Although the rate of such events varies broadly among studies—from 1.7% (38) to 38% (39) depending on catheter type, energy source, and anticoagulation—more than 98% of patients in our cohort presented an increase in S100B concentration. Several other arguments point toward a relevant cardiac origin of S100B: (i) We showed that S100B is released upon damage from ex vivo mouse hearts devoid from central innervation; (ii) one patient showed a temporary attenuation in phrenic nerve activity during the ablation procedure, and this patient had the highest S100B increase among all patients, fortifying relevant S100B releases from peripheral nerves; (iii) releases of S100B up to 48 hours after coronary bypass procedures were shown to indicate central damage (40). Our study focused on the immediate increase in S100B after AF ablation; thus, the timing supports that S100B originates from the ICNS and not the central nervous system.

Although the destruction of neuronal cell bodies is permanent, neurite regrowth is an established concept in the peripheral nervous system and leads to reinnervation of the target organ and regain of functional control after damage (41). Glial cells are involved in peripheral nerve sprouting via secretion of neurotrophic factors (42). Reinnervation in the ICNS after AF ablation is evidenced via indirect measurements of autonomic function as HRV and HR turbulence (19); nevertheless, specific mechanisms were not known yet. Our data indicate that S100B contributes to this process: We provide evidence that S100B is taken up by cardiac nerves and can induce neurite outgrowth in intracardiac neurons. The neurotrophic effect of S100B has been attributed to downstream signaling of its receptor RAGE (18), which we show is expressed in human intracardiac neurons. It is well described that S100B influences neuronal electrical activity (30, 43, 44). Enes et al. (45) demonstrated that AP firing is an intrinsic negative signal suppressing neurite growth, which is in line with our findings. However, intracellular and extracellular S100B were shown to have different downstream effects: direct binding and phosphorylation of cytoskeletal and growth-associated proteins (15), as well as direct inhibition of EAG1 potassium channels regulating neuronal excitability (43). Therefore, reduced firing and neurite growth are not necessarily connected.

S100B released from glial cells might serve as a first-line signal to induce nerve sprouting in the ICNS. This has been described, for example, for nerve growth factor, which is released 1 day after AF ablation (46) and leads to outgrowth of sympathetic nerves. S100B+ glial cells were found in the presence of parasympathetic and sympathetic nerves, indicating that S100B release marks damage of both types of nerves. Disruption of sympathetic neurons during ablation of AF has been previously described by us (47) and other groups (48) and is also in line with anatomical studies showing that adrenergic fibers proceed, together with parasympathetic fibers, via intracardiac ganglia to supply the ventricles (25, 26). Because local neurons in the heart are mostly parasympathetic, these were the focus of our study.

The benefits of accidental or targeted disruption of the ICNS on patient outcome are a matter of discussion that has not been resolved to date (8, 10, 11). This could be due to challenges associated with evaluating acute intracardiac neural damage, because clinically feasible methods are limited: Autonomic responses evoked by ablation itself or high-frequency stimulation can only detect the approximate localization of ganglia (49) and are therefore limited during subsequent ICNS modification. Recently, intraprocedural parameters such as RR interval, Wenckebach cycle length, and atrial-His interval shortening have been suggested as end points for the targeted ablation of ganglionated plexi (50) but can only reflect selected parts of the ICNS immediately during the procedure. Even though direct recordings of intrinsic cardiac nerve activity in humans have recently been reported (51), this method required electrode placement to be performed during open heart surgery. In contrast, the use of blood-based markers is reliable and has been shown to be useful, for example, blood-based marker analysis predicted successful sympathetic denervation of renal arteries during 6-month follow-up (52). A marker for neuronal damage induced by ablation is of utmost interest. Higher S100B release after ablation was associated with less AF recurrences during follow-up, supporting a beneficial effect of ICNS damage. Still, to decide whether targeted ICNS disruption, in addition to PVI, is desirable was beyond the scope of the current study. Nerve sprouting and heterogeneity after cardiac injury in the ventricles are related to arrhythmogenesis (53). Whether the neurotrophic effects of S100B are sufficient to create potential proarrhythmic substrates in patients after AF ablation and how that affects outcome remain to be elucidated.

Our study has some limitations: First, we chose patients with an ablation of the CTI as a control group because functional data indicate sparse innervation in that region (21), but detailed neuroanatomical studies are lacking. S100B concentrations increased marginally in this group, supporting that innervation of this area is scarce. Second, although S100B is not expressed in healthy cardiomyocytes, its expression is initiated in ventricular cardiomyocytes after myocardial infarction (54). We cannot exclude the possibility that atrial cardiomyocytes react similarly; however, no differences were detectable in S100B concentrations from patients with and without a history of myocardial infarction in our cohort. Last, although reduced neuronal AP firing has been mechanistically linked to neurite outgrowth after peripheral damage (45), we do not know whether this is relevant during the regeneration processes in our patients. The correlation between S100B and patient outcome shown in our study might also be based on the amount of nerve damage rather than regeneration processes. Therefore, whether and to what degree subcellular effects of S100B influence patient outcome cannot be fully assessed yet.

In summary, our findings suggest that S100B release from cardiac glial cells is a hallmark of acute intracardiac neural damage in AF ablation. S100B contributes to sprouting of local neurons, reducing neuronal electrical activity, and increasing neurite growth. Further studies should assess the value of targeting the ICNS during ablation and impact on patient outcomes.


Study design

The current study aimed to characterize the neurotrophic factor S100B in intracardiac neural damage. To this end, we used three different approaches: (i) In neuroanatomical studies, we characterized S100B expression in the human and murine ICNS. (ii) In an observational study, we analyzed S100B plasma concentration and autonomic function in patients undergoing catheter-based treatment of AF. (iii) Using in vitro experiments, we studied release and uptake of S100B as well as its cellular effects on human and murine tissues and cells.

For the observational study, plasma samples were obtained from femoral venous sheaths before and after ablation from patients receiving catheter-based treatment of AF [PVI group, n = 113; 1 patient was excluded from all analyses because of reduction in compound motor action potential (CMAP) amplitude; fig. S3]. Two groups served as controls: patients who underwent an EPS without ablation (n = 19) and patients with typical atrial flutter undergoing sole CTI (n = 21; 1 patient was excluded from all analyses because of a diagnosis of AF), due to the limited autonomic innervation in that region (21). Sample sizes were chosen on the basis of pilot data in the beginning of the study. For in vitro experiments, Langendorff perfusion of murine hearts was performed to study the effect of S100B release and uptake without central neural influences. Effects of recombinant S100B on electrophysiology were studied in murine hearts, human atrial pectinate muscles, and murine isolated intracardiac neurons. Studies on neurite outgrowth were performed in cell culture using murine intracardiac neurons. Analyses of neurite lengths were performed blinded with regard to the treatment. We excluded outliers that were identified with the robust regression and outlier removal method. All specimens were assigned randomly to their respective treatment.

Patient recruitment

Patients eligible for this study ("> NCT03388333) had symptomatic AF and an indication for AF ablation without previous catheter-based or surgical ablations (PVI group), a diagnosis of right atrial flutter (CTI group), or an indication for an EPS (EPS group) for suspected atrial or ventricular tachyarrhythmia. Patients who received an ablation upon successful stimulation were excluded. Informed consent was obtained from all patients. Human tissues used were anonymized. This study was approved by the review committee of the Medical Association of Hamburg, Germany, and conforms to all principles outlined by the Declaration of Helsinki.

EPSs and catheter ablation

All procedures in patients were performed under sedation with propofol, fentanyl, and, if necessary, midazolam as described previously (55). For the EPS group, diagnostic procedures were performed in patients with suspected supraventricular or ventricular arrhythmias according to institutional protocol. For the CTI group, diagnostic of typical atrial flutter, as well as CTI ablation, was performed using irrigated tip ablation catheters at 38 W with complete bidirectional block of the CTI as end point.

AF ablation group

AF ablation was performed via isolation of the PVs with either radiofrequency (RF; n = 62) or cryoballoon (n = 50). RF ablation was performed three dimensionally navigated point by point around ipsilateral veins with an adequate distance from the ostia using a 3.5-mm irrigated tip ablation catheter with a maximal power up to 25 W at the posterior wall and up to 30 W at the anterior wall. Esophageal temperature was measured throughout the procedure using a multipolar temperature-sensing catheter. For cryoballoon-based AF ablation, target application time in this study was 240 s per application. To assess integrity of the phrenic nerve during ablation, electrical stimulation was performed from the superior vena cava evoking a CMAP. The freezing cycle was immediately terminated if CMAP amplitude decreased by one-third, indicating loss of phrenic nerve capture or if temperature in the esophagus fell below 15°C. Electrical isolation for RF- and cryoballoon-based ablation was confirmed using a decapolar diagnostic catheter for circumferential mapping of the PVs.

HR and HRV

HR was analyzed from resting ECGs performed before and within 24 hours after the procedure in patients in sinus rhythm. To assess autonomic modulation, HRV analysis was performed according to the guidelines of the European Society of Cardiology and established protocols (23, 56) in randomly selected patients with AF (n = 10) in sinus rhythm before and 24 hours (23.8 ± 7.3 hours) after the procedure. Patients had a 5-min resting period before a 5-min ECG was recorded. HRV analysis was performed using the public domain software Kubios HRV [Biosignal Analysis and Medical Imaging Group, Department of Physics and Mathematics, University of Eastern Finland, Finland (57)].


Patients were scheduled for visits in our outpatient clinic within 6 months after the procedure. Twelve-lead ECG and/or 24-hour Holter monitoring was performed when patients complained about symptoms. If patients missed a visit, they were contacted via phone to inquire for symptoms and asked to present at an outpatient clinic. Recurrence of AF was defined as an ECG-based episode of AF lasting at least 30 s.

Neuroanatomical methods

Human tissues for neuroanatomical studies were obtained from patients receiving open heart surgery (left atrial appendages and left appendage–atrial junction (58)] as well as from an explanted heart of a patient receiving transplantation (atrioventricular groove). Specimens were cut into 1- to 2-cm blocks before fixation in neutral-buffered formalin for at least 48 hours at room temperature (RT). Murine hearts were Langendorff perfused to remove erythrocytes before fixation for 24 to 48 hours. After routine processing for paraffin embedding, specimens were sectioned at 4 μm.

Immunohistochemistry was performed as described previously (47, 59). Sections were deparaffinized in Histo-Clear and a descending series of ethanol. Heat-induced antigen retrieval was performed with citrate buffer (pH 6) except for Sox10 (tris-EDTA, pH 9). Autofluorescence was quenched with 0.25% Sudan black/70% ethanol for 30 min. After permeabilization in 0.2% Triton X-100 in tris-buffered saline (TBS) for 10 min, sections were blocked with 3% bovine serum album (BSA) in TBS for 1 hour at RT. Primary antibodies (table S4) were incubated overnight at 4°C. After washing three times for 10 min each in TBS, secondary antibodies (table S5) were incubated for 2 hours at RT in TBS. Slides were mounted in DAPI Fluoromount-G (SouthernBiotech).

Images were taken using a Leica TCS SP5 confocal microscope (Leica Microsystems) with ×10 numerical aperture (NA) = 0.3 HC PL Fluotar, ×20 NA = 0.7 HC PL APO CS Imm/Corr oil, and ×40 NA = 1.3 HCX PL APO CS objectives. Three-dimensional images were collected over the full range of the signal, and a maximum projection image was created using the Leica LAS AF software.

Chromogenic staining for S100B was performed with the Ventana BenchMark XT (Roche) slide preparation system with antigen retrieval performed in EDTA buffer (pH 8.0). The antibody was diluted 1:500 and visualized using the ultraView Universal DAB Detection Kit (Roche).

H&E staining of paraffin sections from cryoablated hearts was performed according to routine procedures.

Recombinant S100B

Because neurotrophic effects of S100B depends on its dimeric form, recombinant protein (MyBioSource) was dimerized using 5 mM sodium tetrathionate and 1 mM CaCl2 (60). Dimerized protein was dialyzed against phosphate-buffered saline (PBS) for 3 days.

In vitro EPSs in human atrial preparations

Right atrial pectinate muscles obtained from patients receiving open heart surgery were used for AP measurements (61). All specimen were placed in cardioplegic solution (100 mM NaCl, 50 mM taurine, 20 mM glucose, 10 mM KCl, 5 mM MgS04, 5 mM MOPS, and 1.2 mM KH2PO4) containing 30 mM myosin adenosine triphosphatase inhibitor BDM (2,3-butanedione monoxime) at RT immediately after excision and transferred to the laboratory, where pectinate muscles were isolated. APs were recorded with standard intracellular microelectrodes. Bath solution contained the following: 127 mM NaCl, 5.4 mM KCl, 1.05 mM MgCl2, 1.8 mM CaCl2, 11 mM glucose, 22 mM NaHCO3, and 0.42 mM NaH2PO4, equilibrated with 95% O2/5% CO2 at 36.5 ± 0.5°C (pH 7.4). Preparations were field stimulated for at least 30 min (1 Hz) before data acquisition. S100B (100 ng/ml) was added to the superfusing solution for at least 20 min. APs were analyzed offline using the Lab-Chart software (ADInstruments).


C57BL/6 mice (8 to 16 weeks of age; stock number 000664, male and female, Jackson Laboratories) were used in accordance with the Guide for the Care and Use of Laboratory Animals published by the U.S. National Research Council Committee (8th edition, updated 2011) and approved by the regional regulatory authorities.

In vitro EPSs in murine whole hearts

Langendorff preparation was performed with male mice as previously described (47). A 2F octapolar electrophysiology catheter (CIB’ER Mouse, NuMED Inc.) was inserted into the right ventricle for electrical stimulation with a cycle length of 100 ms for the initial 20-min equilibration period. To determine a direct effect of S100B on cardiac electrophysiology, human recombinant S100B (100 ng/ml) (30) was added to the perfusion buffer (modified Krebs-Henseleit solution: 119 mM NaCl, 25 mM NaHCO3, 4.6 mM KCl, 1.2 mM KH2PO4, 1.1 mM MgSO4, 2.5 mM CaCl2, 8.3 mM glucose, and 2 mM Na pyruvate; pH 7.4, 95% O2/5% CO2) after equilibration. Hearts perfused without S100B were used as a negative control. Electrophysiological parameters were determined 10 min later using an established protocol (47). Programmed stimulation was applied using a designated digital stimulus generator (STG4002, Multi Channel Systems) at twice the atrial or ventricular pacing threshold.

In vitro cryoablation in murine whole hearts

An in vitro ablation was performed to evaluate S100B release in an ex vivo Langendorff heart without central innervation. After equilibration, cryogenic damage was induced using a nitrogen-cooled metal probe on the posterior wall of the atria (62). Perfusion buffer for S100B measurements was collected on ice and stored at −80°C. Hearts were formalin fixed for histology.

In vitro EPSs in murine intracardiac neurons

To evaluate the effect of S100B on the electrophysiology of intracardiac neurons, cells were isolated as described previously (63). For AP measurements, extracellular solution consisted of normal Tyrode’s solution, and the pipette solution contained the following: 110 mM K-gluconate, 30 mM KCl, 5 mM NaCl, 1.0 mM MgCl2, 0.22 mM amphotericin B, 10 mM Hepes, pH 7.2 (KOH). Trains of APs were elicited by applying long-lasting (500 ms) depolarizing current pulses of increasing amplitude. AP firing frequency was expressed as the count of the total number of APs occurring during the 500-ms lasting depolarization. AP frequency was normalized to t = −2-min values. APs were recorded at 36.5°C with the amphotericin B–perforated or ruptured patch-clamp technique, using an Axopatch 200B Clamp amplifier (Molecular Devices Corporation). Signals were low-pass filtered (cutoff frequency, 5 kHz) and digitized at 20 kHz, and potentials were corrected for liquid junction potential. To rule out the possibility of a time-dependent decrease in firing frequency contributing to the observed S100B effect, a set of time-matched controls was performed.

Culture of murine intracardiac neurons

Intracardiac neurons were isolated from six to eight mice per independent experiment (n = 5) according to a protocol adapted from Hasan and Smith (64). Animals were euthanized with CO2 before transcardiac perfusion with heparinized PBS (10 U/ml) to remove blood cells. Heart-lung packages were dissected in ice-cold 0.25% BSA (Serva) in Hanks’ buffered saline (Thermo Fisher Scientific). After removal of lungs, trachea, ventricles, and atrial appendages, the remaining tissue, ganglia-containing epicardial fat pats and atrial wall, was cut mechanically followed by enzymatic digestion with collagenase type 1 (1 mg/ml; Worthington Biochemical) and dispase 2 (5 U; Sigma-Aldrich) for 30 min at 37°C. Every 8 min, tissue was triturated 10 times using 10-ml pipettes. Final trituration was performed with Pasteur pipettes before the suspension was filtered through a 40-μm cell strainer, and cells were pelleted for 5 min at 350g. After removal of the supernatant, cells were washed with Neurobasal-A medium (NB-A; Life Technologies) supplemented with 0.5 mM l-glutamine, 1% penicillin/streptomycin, and B27 (20 μl/ml) (Thermo Fisher Scientific), pelleted again, and plated on Matrigel-coated μ angiogenesis slides (Ibidi).

Neuronal growth assay in murine intracardiac neurons

For morphometrical studies on neurite growth, factors were added 2 hours after plating in NB-A supplemented with B27 (0.4 μl/ml) in the following concentrations: S100B, 100 ng/ml; fetal calf serum, 5% (positive control), no additional factors (negative control). Cells were grown for 16 hours before fixation with 4% (v/v) formaldehyde/PBS (Thermo Fisher Scientific) for 10 min at RT. Permeabilization was performed in 0.2% Triton X-100/PBS. After blocking for 1 hour in 3% BSA/PBS, cells were incubated for 1 hour in ms α β-III-tubulin in 1% BSA/PBS. Secondary antibody incubation was performed for 1 hour in 1% BSA/PBS. Cells were mounted in DAPI Fluoromount-G. Cells were photographed using a Leica TCS SP5 (Leica Microsystems) with a ×20 NA = 0.7 HC PL APO CS Imm/Corr oil over the full range of the signal. Five to 20 cells were analyzed per well. Only cells that were clearly distinguishable from neighboring cells bearing at least one neurite with the length of two cell diameters were analyzed. Neurite lengths were measured using Fiji’s simple neurite tracer (65).

Protein analyses

Plasma samples were processed immediately and stored at −80°C for analysis. Human plasma, Langendorff perfusion fluid, and human atrial pectinate muscles were assessed using commercially available assays according to the manufacturer’s instructions. Tissues were lysed using a glass tissue homogenizer in PBS with 0.2% Triton X- 100 and protease inhibitors. S100B was measured using an enzyme-linked immunosorbent assay (ELISA; Merck Millipore) with a sensitivity of 2.7 pg/ml and an inter- and intra-assay variation between 1.9 and 4.8%. HsTnI concentrations were determined using a highly sensitive troponin I immunoassay (Architect i2000SR, Abbott Diagnostics). Activin A was measured using Quantikine sandwich ELISAs (R&D Systems).

Statistical analysis

All values are described in mean ± SEM if not stated otherwise. For human studies, continuous variables were described as quartiles, whereas categorical variables were described as absolute numbers and percentages. The Wilcoxon rank sum (for continuous variables) or the χ2 (for categorical variables) test was used for between-group comparisons. Experimental data were tested for normality using D’Agostino-Pearson omnibus test. Group comparisons for firing frequency in intracardiac neurons were performed using two-way repeated-measures ANOVA with the Holm-Sidak post hoc test. Survival curves stratified by terciles of ∆S100B were produced using the Kaplan-Meier method. Survival curve differences were tested with the log-rank test. P values <0.05 were considered statistically significant. Statistical analysis was performed using GraphPad Prism 6.07 (GraphPad Software).


Fig. S1. S100B-expressing glial cells ensheathing nerve fibers are interspersed within the human atrial myocardium.

Fig. S2. S100B-expressing glial cells ensheathing sympathetic nerve fibers.

Fig. S3. Patient with reduced phrenic nerve activity during ablation presents a higher increase in S100B plasma concentration.

Fig. S4. Activin A increase independent of ablation type.

Table S1. Baseline characteristics of patients.

Table S2. Changes in HR variability parameters in patients after AF ablation.

Table S3. Electrophysiological measurements in human atrial pectinate muscles after exposure to S100B.

Table S4. Primary antibodies used in this study.

Table S5. Secondary antibodies used in this study.

Reference (66)


Acknowledgments: We thank C. Meyer-Schwesinger for providing the PGP 9.5 antibody; J. P. Sperhake for providing the photograph of a human heart; the UKE Microscopy Imaging Facility (Umif), University Hospital Centre Hamburg-Eppendorf for providing microscopes and support; H. Wieboldt for excellent technical support; S. Krasemann and K. Hartmann (UKE-HEXT Mouse Pathology Facility, Hamburg, Germany) for chromogenic immunohistochemistry; S. Dünger and S. Gerth for troponin measurements; and the team of the electrophysiology laboratory for their help in sample collection. Funding: This work is supported by the German Centre for Cardiovascular Research (DZHK) (FKZ 81Z4710141 to S.W. and C.M. and 81X2710149 to C.M. and K. Scherschel) and the National Institutes of Health (NIH) (OT2OD023848 to K. Shivkumar). Author contributions: K. Scherschel designed and performed experiments (histology, ELISA, and cell culture) and wrote the manuscript. K.H. acquired data and wrote the manuscript. C.J. performed Langendorff experiments, M.D.L. and T.C. performed AP measurements in human atrial tissue, and N.R. performed statistical analyses. M.W.V. and S.C. performed AP measurements in intracardiac neurons. N. Klatt, C.E., and S.W. performed clinical procedures and critically discussed the manuscript. D.L., D.W., P.K., N. Klöcker, K. Shivkumar, T.C., and S.W. provided conceptional advice and critically discussed the results and the manuscript. T.Z. performed hsTnI measurements and provided conceptional advice. C.M. designed and supervised the study, performed clinical procedures, and wrote the manuscript. All authors critically discussed the results and reviewed and approved the manuscript before submission. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data associated with this study are present in the paper or the Supplementary Materials.

Stay Connected to Science Translational Medicine

Navigate This Article