Development and Growth of Human Salivary Stones by Neutrophil Extracellular Traps

Salivary gland stones, or sialoliths, are the most common cause of the obstruction of salivary glands. Symptomatic sialolithiasis has a prevalence of 0.45% in the general population, is characterized by recurrent painful periprandial swelling of the affected gland and often results in sialadenitis with the need for surgical intervention. The mechanism behind the formation of sialoliths has been elusive. Here we show that neutrophil extracellular traps (NETs) initiate the formation and growth of sialoliths. The deposition of neutrophil granulocyte extracellular DNA around small crystals results in their dense aggregation, and the subsequent mineralization creates alternating layers of dense mineral, predominantly calcium salt deposits and DNA. Further agglomeration and appositional growth of these structures promotes the development of macroscopic sialoliths that nally occlude the efferent ducts of the salivary glands, causing clinical symptoms and salivary gland dysfunction. These ndings provide an entirely novel insight into the mechanism of sialolithogenesis in which an immune system mediated response essentially participates in the physico-chemical process of concrement formation and growth.


Introduction
Salivary gland stones, or sialoliths, are the most common cause for the obstruction of the salivary glands 1,2 . Symptomatic sialolithiasis has a prevalence of 0.45% in the general population and is characterized by recurrent painful periprandial swelling of the affected gland 3 . Sialolithiasis accounts for one third of all salivary gland disorders 4 and typically occurs in middle-aged adults. In 80% and 20% of patients with sialolithiasis, the stones affect the excretory duct system of the submandibular and parotid glands, respectively 5 . 25% of the patients develop more than one stone. Postmortem investigations in humans detected small concrements even in asymptomatic patients, particularly in the submandibular glands 6,7 .
These small concrements may provide a basis for the development of larger stones which nally become clinically symptomatic. Sublingual and other small salivary glands rarely harbor symptomatic sialoliths 8,9 . Stones found within the submandibular glands are usually larger than those of the parotid gland. The shapes of the stones depend on the sites of their formation, ranging from elongated to more round or oval when having developed in the duct system or in the parenchyma, respectively 10 . Ultrasound imaging and sialendoscopy are preferentially employed to detect the concrements 11,12,13,14 . The latter often need to be removed by sialendoscopy-assisted interventions such as basket extractions, mechanical fragmentation, or transoral or endoscopic-transcutaneous surgical approaches 15,16,17,18,19,20 .
The mechanism that leads to the formation of sialoliths has still been elusive. So far, various concepts have been discussed, ranging from altered ion concentrations in saliva 21 , the presence of bacteria or foreign bodies in the ducts 22 to micro-calci cations of cell debris ejected from salivary gland cells 23 . Saliva of parotid glands contains less calcium ions than those of submandibular glands, making the former less susceptible for the formation of concrements 24 . This observation is re ected by the reported differing prevalence of sialolithiasis for both glands 5,25 . Parotid sialoliths contain more organic and less inorganic compounds than submandibular stones 4,26 . In addition to calcium ions, phospholipids and secretory glycoproteins contribute to sialolith formation 23,27,28,29 . At a certain size, the concrements may obstruct the ducts. The secretory activity of the gland, predominantly controlled by the autonomous nervous system, seems to in uence this process 30,31 . The administration of the β-adrenoreceptor agonist isoprenaline and calcium gluconate into rats induced sialoliths and sialadenitis 32 , while parasympathetic stimulation facilitates saliva secretion. Along with reduced salivary ow rates, the presence of leucocytes in saliva was observed in patients with recurrent parotitis 33,34 . A leucocyte in ux into the ducts may be further aggravated by several factors, including a microbial ascend 21,22,23,35 . Salivary stasis and ongoing in ammation may nally result in glandular dysfunction, atrophy or sclerosis 23,36,37 .
While such concepts appear sound, the reason for the growth of sialoliths is unknown 23 . Current models seeking to explain the formation of sialoliths diverge substantially 6,7,21,22,23,33,34 . Our study shows that neutrophil extracellular trap (NET) formation is an essential step for sialolith development. Neutrophils can externalize their chromatin, decorated with granular proteins, a process usually referred to as the formation of NETs 38 . NET formation is a type of in ammatory response which can be triggered by various factors, among them the contact of neutrophils with crystals such as calcium salts 39 , cholesterol 40 or urate 41 , and also by pH variations 42 , foreign bodies 43 and bacteria 44 . NETs tend to aggregate and form aggNETs 45,46,47,48,49 that incorporate particular matter, cellular debris and viable immune cells 50 . In synopsis with former observations and our present results, we hypothesize that aggNETs serve as a "glue" that agglomerates salivary calcium crystals and proteins, thus driving the assembly of macroscopic sialoliths.

Results
Human sialoliths are macroscopically polymorphic.
Sialoliths show individual characteristics in size, shape and color. Their surface structure ranges from smooth to ssured, underlining the pleomorphic nature of the stones. Many appear to be composites of smaller spheric building blocks (Fig. 1). Detailed analyses of the mineral composition of sialoliths were described elsewhere 51,52 .
Sialoliths consist of an onion skin-like shell structure of extracellular DNA and calci ed layers.
Micro-computed tomography (µCT) showed concentric alternating radiodense (calci ed) and translucent (organic) layers in submandibular and parotid sialoliths (Fig. 2a-c). The 3D reconstruction of complete stones con rmed their layered structure, illustrating this alternating pattern which extended in all examined stones from the center to the surface (Fig. 2d). The center of all examined stones was found to be radiolucent, indicating a lower degree of mineralization and a higher degree of organic compounds. Von-Kossa staining identi ed the radiodense layers as crystalline calcium deposits, surrounded again by more radiolucent, organic layers (Fig. 2e).
Propidium iodide staining of hard-cut methacrylate embedded stones indicated that the radiolucent, organic regions contain extracellular DNA (Fig. 2f, displayed in red). Merged images con rmed that these extracellular DNA containing layers alternated with the dense calcium crystal layers (Fig. 2f, displayed in green). The extracellular DNA signals were predominantly detected as extended patches of aggregated chromatin.
The calcium containing layers of sialoliths are responsible for their tightness. The incubation of whole native sialoliths in different staining solutions objecti ed the compactness of the calci ed outer shells (acridine orange, MW = 265 g/mol; trypan blue, MW = 873 g/mol; propidium iodide, MW = 668 g/mol). A subsequent sanding of the stones revealed that parts of the interior were protected from penetration of any employed dye ( Fig. 3a-f). These dye spared regions were found in every single stone, particularly in the center. Occassionally, small caves in the stones´ interior communicated with the surface via small clefts (Fig. 3e). The surfaces of the inner caverns also stained, but their walls were as impermeable for all employed dyes as were the outer shells.
The detected propidium iodide signals in uorescent microphotographs illustrated a diffuse distribution of extracellular DNA on the surface of native stones (Fig. 3f, left panel). After decalci cation, propidium iodide penetrated into the stones´ inner layers, and the staining was then detected throughout the entire sialolith ( Fig. 3f, right panel). The center of the sialolith displayed a strong auto uorescence that distinguished the regular shaped, homogenous core from the jagged and ssured outer layers. However, DNA was to be detected all over the sialolith including the core ( Fig. 3f; Fig. 1S). This con rmed that the calci cation was responsible for the dye impermeability of the stones´ outer layers. To verify that the propidium iodide signals in the stones´ interior were caused by extracellular DNA, we incubated sialolithic gravel with propidium iodide. The obtained intense staining signals were DNase 1 sensitive and considerably reduced after the digestion, identifying the origin of the signal as extracellular DNA (Fig. 3g).
The majority of sialoliths harbour bacterial DNA.
As bacteria 53 and calcium salt crystals are established inducers of NETs 43, 47, 49, 54 , we tested whether bacteria are to be found in the sialoliths. The DNA was extracted from gravel obtained from 28 submandibular concrements. A subsequent bacteria speci c 16S rRNA PCR showed that 23/28 (82.1%) of the sialoliths contained bacterial DNA. To identify the involved species, we genotyped the four preparations with the highest yields of bacterial DNA and established the microbiome of these stones. We detected various bacterial taxa typically present in the oral [Streptococcus dentisani 55  reported to be present in the middle ear 61 , putatively being a commensal of the upper aerodigestive tract.
All these taxa were present in every stone we tested. Interestingly, none of these bacteria were detected in the microbiome of gallstone controls. These results suggest that oral bacteria, in addition to calcium salt crystals, are inducers of the NETs that drive sialolith formation.
NETs are involved in sialolithogenesis.
Next, we asked whether the patchy propidium iodide signals originated from NETs, and co-stained gravel specimens for extracellular DNA, neutrophil elastase and citrullinated histone H3. Neutrophil elastase and citrullinated histone H3 (citH3) could be robustly detected in sialoliths (Fig. 4a) and were shown to colocalize with the extracellular DNA. This nding strengthened the hypothesis that the DNA patches observed in the sialoliths originate from NETs. To con rm the presence of enzymatically active neutrophil elastase within the sialoliths, we selectively harvested gravel material from the stones´ centers, intermediate parts and surfaces, which was subsequently ground and analyzed for neutrophil elastase activity. We detected a considerable enzyme activity in all stones and in all areas. A tendency to higher activities at the surfaces was noted, the latter representing the newest parts of the sialoliths (Fig. 4b).

Discussion
Although the development of a clinical sialadenitis is not per se associated with the presence of sialoliths and vice versa, these two constellations occur in parallel, mutually reinforcing each other 23 . Altered ion concentrations in saliva 21 leading to crystal formation as well as the presence of cellular debris, foreign bodies or bacteria have been considered to contribute to sialolith formation in the past 22,23,24,62,63,64 . However, it was unknown to date whether these mechanisms are interconnected and how microscopic crystals in the salivary ducts can form and grow to macroscopic concrements. Crystals and other danger signals (i.e. bacteria) are potent attractors for neutrophils, inducing their activation and NET formation 43,47,49,54 . NETs were already shown to promote the development of depositions in other organs in vivo 40,46 . As salivary stones consist of calcium based crystals (hydroxylapatite, brushite and whitlockite) 65 , we hypothesize that NET formation fuels the process of sialolithogenesis.
We show that the surface of salivary stones displays robust neutrophil elastase activity and is covered by an abundance of extracellular DNA, suggesting neutrophil recruitment and NET formation. The high concentrations of bicarbonate ions present in saliva are known to be an important NET formation trigger 49 , additionally explaining the plenty of extracellular DNA that covers sialoliths. The clinical experience that patients with chronic recurrent sialadenitis respond favorably to the anti-in ammatory action of glucocorticoids instilled into the affected ducts 33 further strengthens the role of neutrophil-based in ammation in sialolith formation.
The onion skin-like distribution of calcium depositions, associated extracellular DNA accumulations, neutrophil elastase activity and the presence of citrullinated histone H3 argue for a step-wise, episodic growth of sialoliths. The low abundance of citrullinated histone H3 in the center of sialoliths can be explained by the fact that the less basic citrullinated molecules are lost from chromatin in the sake of time 66 . More intense signals can be found in the outer layers as these represent younger parts of the sialoliths. The development of the stones seems to be based on a repeated process of neutrophil attraction, presumably intensi ed by bacterial colonisation, NET-formation and the assembly of calcium containing depositions. The subsequent aggregation to larger concrements, the packaging, stabilization and further calci cation of the layers result in the appositional growth of the salivary stone.
In summary, we conclude that NETs promote calcium crystal aggregation in the salivary ducts and trigger concrement formation. Different previous models trying to explain the process of sialolithogenesis can be brought into line by the induction of an in ammatory reaction, causing an enhanced in ux of neutrophils into the salivary ducts. The activation of these neutrophils and the formation of NETs, enhanced by the presence of crystals and other danger signals as bacteria, results in the gradual development and appositional growth of salivary stones which nally congest the excretory ducts of the gland. This immune system based mechanism in the development of concrements may also apply for other organs within the body. In addition to the existing endoscopic or surgical treatment options, targeting neutrophils and NET-formation may thus become a valuable instrument to prevent the development of salivary stones.

EXPERIMENTAL METHODS
Ethics. An informed consent was obtained from each patient concerning diagnostic procedures, treatment to remove salivary stones and the subsequent use of these specimen for research purposes, including data analysis. The study was approved by the University´s ethical review committee (186_19 Bc).
Patients and clinical interventions. All patients of our study cohort with symptomatic sialolithiasis (n = 37) received a clinical examination, ultrasound imaging and a subsequent sialendoscopy to ensure the diagnosis. Diagnostic and therapeutic interventions were carried out in a tertiary referral center specializing in salivary gland diseases (FAU Medical School, Department of Otolaryngology, Head and Neck Surgery, University of Erlangen-Nuremberg, Erlangen, Germany), obtaining stones from submandibular (n = 30) and parotid glands (n = 7). Submandibular stones were removed by transoral surgery which allowed to obtain a complete unfractionated sialolith. Parotid stones were removed either in total during an open surgical approach in which the stone was removed as a whole, or by sialendoscopically assisted basket extraction (when the concrement was small enough to evacuate it through the duct). In cases of larger parotid concrements which could not be evacuated in total due to their size or which were not obtained during an open surgical approach, stone fragments were collected after sialendoscopically controlled intraductal pneumatic lithotripsy and subsequent basket extraction.
Chemicals and Biochemicals. If not stated otherwise, we obtained all chemicals and biochemicals from Sigma-Aldrich/Merck (Darmstadt, Germany).
Salivary stones storage and processing. Sialoliths were stored at room temperature (for µCT, calcium staining) or dry at -20°C (immuno uorescence) until analysis. Some stones were embedded in methacrylate and cut into 2 µm sections with a rotary microtome (RM2265, Leica, Wetzlar, Germany); others were xed for 4-12 hours with paraformaldehyde and decalci ed in Teitel buffer (140 g EDTA free acid; 90 ml NH 3 [30%] in 500 ml H 2 O; pH = 7.2; adjusted with NH 4 OH) for up to 4 weeks. The stones used for dye penetration assays were incubated with trypan blue (100 µg/ml), propropidium iodide (10 µg/ml), or acridin orange (10 µg/ml) at ambient temperature for 10 minutes and subsequently ground with a diamond le. By mechanical grinding with ceramic or diamond tools we obtained gravel of sialoliths, and material from the surface, the intermediate and the center of the stones was received by selectively skimming off the layers.
Digital reconstruction of volumetric µCT data. Based on the volume data generated in CT, a physically based volume rendering algorithm using a Monte Carlo path tracing method simulated the complex interactions of photons (emission, absorption, scattering) within the scanned specimens 67,68 . Cinematic Rendering generated photorealistic images by calculating realistic lighting by light transport simulation along hundreds or thousands of photons paths per pixel, using a stochastic process. Thus, even complex effects such as ambient occlusion or tissue density could be modeled.
Macrophotography. Macro images were taken with a Nikon 700 camera (Nikon, Tokyo, Japan) with a CMOS sensor in FX format (36.0 x 23.9 mm, 12.87 million pixels) and two different objectives (Nikon AF-S Nikkor 60mm/2.8G ED; Nikon AF-S Nikkor 105mm 1:2,8G VR). For illumination, we employed white light.
Microscopy. We used a BZ-X700 automated video microscope for uorescence microscopy, equipped with Z-stack and stitching technology to increase the depth of eld and the size (Keyence Corporation, Osaka, Japan). For bright eld microscopy, oblique illumination was available. We used Photoshop CS5 or CC2018 (Adobe, Munich, Germany) for post-processing of pictures and morphometry. Areas for morphometry were selected by a blinded investigator.
Optical clearing and Light sheet uorescence microscopy (LSFM) of submandibular sialoliths. We xed and decalci ed submandibular sialoliths with paraformaldehyde and Teitel buffer, respectively, as described above. DNA of both decalci ed and non-decalci ed sialoliths was stained with propidium iodide (10 μg/ml) in PBS for 5 days at ambient temperature. We performed optical clearing according to established protocols 69 . Brie y, samples were dehydrated in ethanol series of 50%, 70%, and twice 100% (v/v). Each dehydration step was carried out at room temperature for 2 days in gently shaking 5 ml tubes. After dehydration, samples were transferred to ethyl cinnamate, optically cleared for one day at ambient temperature and imaged with a LaVison BioTec Ultramicroscope II including a LaVision BioTec Laser Module (LaVision BioTec GmbH, Bielefeld, Germany), an Olympus MVX10 zoom body (Olympus Germany, Hamburg, Germany), and an Andor Neo sCMOS Camera (Andor, Belfast, UK) with a pixel size of 6.5 μm. We used detection optics with fourfold optical magni cation and NA 0.5. Auto uorescence was excited by a 488 nm optically pumped semiconductor laser (OPSL) and detected at 525/50 nm. For propidium iodide excitation, a 561 nm OPSL and a 620/60 nm emission lter were employed. For 3D reconstruction and optical clipping of LSFM data we used Imaris software (Version 9.1, Bitplane AG, Zurich, Switzerland).
Von Kossa staining for mineralized areas. We rinsed methacrylate sections several times with distilled water, incubated them with a 5% aqueous silver nitrate solution in the dark for 20 minutes and nally exposed them to UV-light until a black precipitate had formed. The remaining ionic silver was removed by incubation with 5% sodium thiosulfate for 5 minutes and extensive rinsing with distilled water. DNA staining. We stained the extracellular DNA of stones and histological samples with 10 µg/mL and 1 µg/mL propidium iodide in isotonic PBS (both from Thermo Fisher Scienti c, Waltham, USA) for at least 10 minutes at ambient temperature. Digestions with DNase 1 (1U/ml; 10mM Mg 2+ ; 5mM Ca 2+ ) were performed for 120 minutes at 37°C to con rm extracellular DNA as the origin of the observed propidium iodide signals.
Immunostaining for neutrophil markers. Histological sections and gravels from sialoliths were analyzed for extracellular DNA, neutrophil elastase and citrullinated histone H3. Samples were xed with 4% paraformaldehyde for 10 minutes and blocked for 18 hours at 4°C in PBS containing 10% FBS (Merck Millipore, Billerica, Waltham, USA). Primary antibodies detecting neutrophil elastase or citrullinated histone H3 (Abcam, Cambridge, UK; ab21595 and ab1503, respectively) were used following manufacturer´s recommendations. A niPure Cy5-conjugated Goat-Anti-Rabbit IgG (H+L) secondary antibodies (Jackson Immuno Research Labs, West Grove, USA) were co-incubated with 1 µg/ml Hoechst stain or propidium iodide. Slides were embedded in DAKO uorescent mounting medium (Agilent Technologies, Santa Clara, USA) according to manufacturer´s recommendations.
Neutrophil elastase activity in human sialoliths. 1 mg of sanding dust from sialoliths (diamond le, grain size 300) was resuspended in 500 µl of PBS. 25 Figure 1 Sialoliths are unique in shape and surface composition. Submandibular stones under oblique white light, illustrating the individual surface differences between stones and even within the same sialolith. Note the different shapes and appearances that suggest a discontinuous and variable sialolithogenesis in differently shaped molds. Size bar: 10 mm. Sialoliths show onion skin-like layers, indicating a discontinuous formation process. a Micro-CT images reveal concentric layers of calci cations (bright) around the center of submandibular and parotid sialoliths, alternating with more radiotranslucent (dark) areas. b Magni cation of a micro-CT image of a submandibular sialolith shown in a. c Follow-up sections of arti cially stained complete parotid sialoliths (blue and pink re ect X-ray translucent and radiodense areas, respectively). d 3D CT scan reconstructions of a submandibular sialolith. Hounds eld units re ecting the radiodensity are represented by color codes (green re ects X-ray translucent and red re ects radiodense areas, respectively). e Von-Kossa staining of methacrylate sections of submandibular sialoliths revealed that mineralized, predominantly calcium containing areas (black) surround regions of organic depositions (grey). The observed onion skin-like layers are partially traversed by small septs, additionally enclosing organic compounds. f Calci cations (von-Kossa stain, negatively displayed in green) alternate with organic layers which harbour extracellular DNA (stained with propidium iodide, displayed in red), shown in a submandibular sialolith. Size bar a-f: 5 mm.

Figure 3
Sialoliths contain dye-impermeable, densely calci ed areas and extracellular DNA (ecDNA). Stones were immersed in different dyes and subsequently ground to reveal their interior. The calci ed surface is very compact and prevents the penetration of all employed dyes into the stones´ interior. a-c Trypan blue (MW = 873 g/mol) incompletely penetrates the native calculi, shown here in 11 submandibular sialoliths. d Morphometry of the submandibular sialoliths depicted in a-c: every single stone shows dye inaccessible parts, particularly in the inner layers, quanti ed by image analysis. e Acridin orange (MW = 265 g/mol) reveals that the surface of sialoliths occasionally communicates with cavities in the interior via small clefts (displayed in red, right panel). f LSFM (Light sheet uorescence microscopy) of a submandibular sialolith. In native stones, propidium iodide (red, MW = 668 g/mol) only stains the extracellular DNA (ecDNA) at the surface. After decalci cation, the dye penetrates the sialolith, demonstrating the role of the calci cation for the compactness of the surface and indicates the containment of extracellular DNA within the entire stone, including the core. g Propidium iodide staining of stone gravel reveals the widespread distribution of extracellular DNA within the stones. The signals attenuated after treatment with DNase 1, con rming extracellular DNA as the signal´s origin. Size bar a-c, e, f: 10 mm; size bar g: 200 µm.

Figure 4
Sialoliths contain extracellular DNA, neutrophil elastase and citrullinated histone H3. a Staining of extracellular DNA (ecDNA), neutrophil elastase (NE) and citrullinated histone H3 (citH3) in gravel of submandibular sialoliths (SMG: submandibular gland sialolith; IF: immuno uorescence; wo1st: without rst antibody, negative control). The co-localization of these molecules (overlay) indicate NETs as building blocks of the sialoliths. b Neutrophil elastase activity was present in all parts of the sialoliths (quanti ed with a speci c uorogenic substrate). The highest activities were detected in the super cial layers.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.