Syncollin is an antibacterial polypeptide

Syncollin is a 16‐kDa protein found predominantly in the zymogen granules of pancreatic acinar cells, with expression at lower levels in intestinal epithelial cells and neutrophils. Here, we used Strep‐tagged syncollin isolated from the supernatant of transiently transfected mammalian cells to test the hypothesis that syncollin has antibacterial properties, which might enable it to play a role in host defence in the gut and possibly elsewhere. We show that syncollin is an exceptionally thermostable protein with a circular dichroism spectrum consistent with a predominantly beta‐sheet structure. Syncollin binds to bacterial peptidoglycan and restricts the growth of representative Gram‐positive (Lactococcus lactis) and Gram‐negative (Escherichia coli) bacteria. Syncollin induces propidium iodide uptake into E. coli (but not L. lactis), indicating permeabilisation of the bacterial membrane. It also causes surface structural damage in both L. lactis and E. coli, as visualised by scanning electron microscopy. We propose that syncollin is a previously unidentified member of a large group of antimicrobial polypeptides that control the gut microbiome.

Syncollin is expressed not only in the exocrine pancreas but also in the gut and the spleen (Edwardson et al., 1997;Tan & Hooi, 2000).
In the rat gut, it is present principally in epithelial cells lining the proximal duodenum and the colon. Expression of syncollin in the gut is almost undetectable until rats begin suckling, when expression rises rapidly; further, syncollin expression falls upon starvation and then rebounds upon refeeding (Tan & Hooi, 2000). Syncollin is also found in azurophilic granules of human neutrophils and is secreted from these cells upon stimulation (Bach et al., 2006). These observations suggest that syncollin might play a role in host defence, as first proposed by Bach et al. (2006).
There is evidence for the involvement of syncollin in a number of disease states, particularly related to the gut. For instance, syncollin expression was found to be strongly down-regulated in the colon when a bacterial suspension was administered to germ-free mice (Fukushima, Funayama, Ogawa, Takahashi, & Sasaki, 2003) and in mice with chemically induced colitis-associated cancer (Li et al., 2014).
In addition, syncollin mRNA is significantly reduced in intestinal epithelia taken from patients with ulcerative colitis (but not Crohn's disease; Fukushima et al., 2003). Finally, syncollin levels are significantly elevated in both pancreatic juice (Makawita et al., 2011) and serum taken from pancreatic cancer patients (Makawita et al., 2013), suggesting that it might be a useful pancreatic cancer biomarker.
Since syncollin is a component of the pancreatic juice and is implicated in a number of gut disease states, it is tempting to speculate that it may be active in the gut. We set out to test this hypothesis.
Here, we show that syncollin is a very stable protein, consistent with a function in the pancreatic juice. It binds to bacterial peptidoglycan and restricts the growth of representative Gram-positive and Gramnegative bacteria, reducing the viability of the latter. We also show that syncollin causes profound structural changes to the surface of bacteria. We propose that syncollin might be a previously unidentified member of a large group of antimicrobial polypeptides that control the gut microbiome. Further, the fact that syncollin is known to be secreted in response to activation of human neutrophils suggests that it might be involved in host defence more widely.

| Purification of syncollin-Strep
tsA-201 cells were transfected with a syncollin-Strep construct in pcDNA3.1, and syncollin-Strep was purified from cell supernatants using the Strep-Tactin XT™ system. Following biotin elution of syncollin from Strep-Tactin XT™ beads, the protein ran at 20 kDa and appeared as a triple band on silver-stained gels and immunoblots ( Figure 1a). Analysis of these bands by mass spectroscopy indicated that all three are syncollin. We could not find any evidence for posttranslational modification of the protein. We therefore propose that the three bands represent differentially folded (and possibly differentially disulfide bonded) forms.

| Syncollin quality control
As a quality control measure, the purified syncollin-Strep was checked for two properties known to be hallmarks of native syncollinthe F I G U R E 1 Isolation and characterization of syncollin-Strep. (a) Analysis of isolated syncollin-Strep by sodium dodecyl sulfatepolyacrylamide gel electrophoresis (SDS-PAGE) followed by Coomassie blue staining (left) or immunoblotting (right). Molecular mass markers (in kDa) are indicated. (b) Immunoblot of syncollin-Strep after SDS-PAGE in the absence and presence of β-mercaptoethanol (10% [v/v]). (c) Interaction of syncollin-Strep with syntaxin 2. Syncollin-Strep was incubated with either glutathione S-transferase (GST)-syntaxin 2, or GST alone, bound to glutathione-Sepharose™, and loaded, bound and unbound samples were analysed by immunoblotting. Syncollin-Strep eluted from GST-syntaxin 2 with free reduced glutathione (15 mM) was also analysed. Since the polyclonal anti-syncollin antibody used also recognises GST, only the region of the blot where syncollin migrates is shown, for clarity. (d) Circular dichroism (CD) spectra of syncollin-Strep at various temperatures. The results presented in (a) are representative of those from 60 syncollin-Strep purifications; results in (b-d) are representative of those from two experiments presence of disulfide bonds  and the ability to bind to syntaxin 2 (Edwardson et al., 1997). Syncollin was analysed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) in either the presence or absence of β-mercaptoethanol followed by immunoblotting. The immunoblot (Figure 1b) shows that the protein migrated more slowly in the presence than in the absence of β-mercaptoethanol, consistent with unfolding upon reduction of its disulfide bonds. As shown in Figure 1c, syncollin-Strep bound to glutathione S-transferase (GST)-syntaxin 2 (but not to GST alone) attached to GSH-Sepharose™ beads and could be eluted upon addition of reduced glutathione. Hence, at least two key characteristics of the native protein are retained in purified recombinant syncollin-Strep.
Following secretion into the pancreatic juice, syncollin will enter an environment rich in proteolytic activity. One might expect, therefore, that its structure would be highly stable. To test this hypothesis, the thermal stability of the protein was assessed using circular dichroism (CD) spectroscopy. The CD spectrum of syncollin-Strep (10 μM in 10 mM sodium phosphate, pH 8.0 containing 0.1% [w/v] CHAPS), had two peaks: a negative peak at approximately 218 nm and a positive peak at approximately 234 nm (Figure 1d), consistent with a predominantly beta-sheet structure. When the protein was subjected to a temperature ramp up to 90 C, both peaks were reduced in amplitude, although complete unfolding did not occur, indicating that the protein does indeed have a very high thermal stability.

| Syncollin binds to bacterial peptidoglycan
To explore a potential role for syncollin in host defence, we first examined whether syncollin-Strep binds to bacterial peptidoglycan.
Syncollin-Strep or BSA was incubated with peptidoglycan (320 μg), and the peptidoglycan was pelleted by centrifugation. The majority of the syncollin-Strep was found in the pellet, whereas all of the BSA remained in the supernatant ( Figure 2). In the absence of peptidoglycan, syncollin-Strep remained in the supernatant. These results suggest that syncollin is likely to bind to bacterial cell walls.

| Interactions between syncollin and antibiotics
The effects of syncollin-Strep (110 μg/ml) on the growth of L. lactis in combination with the antibiotics ampicillin (bactericidal) and tetracycline (bacteriostatic) were assessed (Figure 6a,b). The concentrations of the two antibiotics (0.05 and 2 μg/ml, respectively) were chosen so as to produce a clear effect on bacterial growth but not to inhibit growth completely. As shown in Figure 6c, syncollin-Strep, ampicillin and tetracycline, when added alone, all failed to cause a significant reduction in the span of the growth curve for L. lactis. However, syncollin-Strep in combination with either ampicillin or tetracycline did significantly reduce the span. None of the treatments used, either alone or in combination, caused a significant effect on the T 50 of the F I G U R E 2 Interaction of syncollin-Strep with peptidoglycan. Syncollin-Strep and bovine serum albumin (BSA) (33 μg in both cases) were incubated with insoluble peptidoglycan (PG; 320 μg). In a control experiment, syncollin-Strep was also incubated without PG. The PG was then pelleted by centrifugation, and the presence of protein in the pellet and the supernatant was  2.6 | Syncollin reduces the viability of E. coli

| DISCUSSION
We have shown here that syncollin hinders the growth of Grampositive (L. lactis) and Gram-negative (E. coli) bacteria, reducing the viability of the latter in the plateau phase of growth. Syncollin also causes obvious structural damage to the bacterial capsule in both L. lactis and E. coli. Syncollin has two properties that are likely be involved in its antibacterial activitythe ability to bind to peptidoglycan, as demonstrated in this study, and the ability to permeabilise biological membranes (Wäsle et al., 2004) and lipid bilayers (Geisse et al., 2002). We suggest that the slowing of the growth of L. lactis (but not E. coli) in log phase might be caused primarily through binding to the exposed peptidogycan-based cell wall in the former, whereas the permeabilisation of the E. coli (but not L. lactis) in the plateau phase might be caused by an action of syncollin on the membranes of the Gram-negative bacterium. Further studies will be required to confirm this suggestion.
Syncollin-knockout mice are viable and fertile, although gut physiology and sensitivity to bacterial challenge have not been addressed.
Given that a plethora of antibacterial polypeptides are known to be secreted into the gut (see below), it is perhaps unlikely that the absence of just one (syncollin) would cause an obvious gut phenotype.
Signs of abnormality in the pancreas of the knockout mice were F I G U R E 7 Effect of syncollin-Strep on bacterial viability in the plateau phase of growth. Bacteria were incubated with buffer alone, syncollin-Strep (300 μg/ml), ampicillin (0.05 μg/ml) or tetracycline (2 μg/ml). When bacteria reached the early plateau phase of growth, they were incubated in phosphate-buffered saline (PBS) containing propidium iodide (1 μg/ml; red) and 4 0 ,6 0 -diamidino- Intriguingly, syncollin shares features with other small antimicrobial polypeptides that are known to play roles in host defence in the gut, including ZG16p, RegIII and members of the defensin family.
These features include a predominantly beta-sheet structure, the presence of intramolecular disulfide bonds, and in some cases, an ability to form homo-oligomers. ZG16p, like syncollin, is located on the luminal surface of the pancreatic ZG membrane (Kalus et al., 2002). It has a predominantly beta-sheet structure and contains a single intramolecular disulfide bond (Kanagawa et al., 2011). Members of the regenerating islet-derived (Reg) protein family also adopt a predominantly beta-sheet structure. The family member RegIIIγ is secreted into the intestinal lumen, where it has antibacterial actions (Shin & Seeley, 2019). RegIIIα is capable of forming hexamers on bacterial membranes, resulting in the formation of a transmembrane pore and membrane leakiness (Mukherjee et al., 2014). Defensins are small proteins (3-4 kDa), which have a predominantly beta-sheet structure (Min et al., 2017), and six disulfide-bonded cysteines. Defensins, like RegIIIα, are able to form pores and permeabilise bacterial membranes.
As mentioned above, syncollin is also expressed in (and secreted from) neutrophils (Bach et al., 2006), again consistent with a role in host defence. Indeed, the authors of this previous paper speculated that syncollin might have an antibacterial effect but were unable to demonstrate this property. We suggest that this failure might have been a result of the elaborate method used to prepare native syncollin (removal from the ZG membrane at high pH followed by selective precipitation in an aggregated state after dialysis to neutral pH; An et al. , 2000).
F I G U R E 8 Scanning electron microscopy (SEM) analysis of the effects of syncollin-Strep and control proteins on bacterial structure. Bacteria were incubated in either buffer alone or in buffer containing protein (0.3 mg/ml) before SEM imaging. Scale bar, 1 μm. The results are representative of those from three experiments In conclusion, given the structural similarities between the properties of syncollin and those of other known antimicrobial proteins, together with the functional properties reported here, we propose that syncollin is a previously unidentified member of the antimicrobial protein superfamily. It is known that pancreatic juice has antibacterial activity (Rubinstein et al., 1985), which is likely to be important not Beads were pelleted by centrifugation for 5 min at 21,000g, and the supernatant was discarded. The beads were resuspended in wash buffer (10 mM sodium phosphate, pH 8.0) supplemented with cOm-plete™ protease inhibitors (Sigma) and 0.1% (w/v) 3-[(3cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS) and loaded into an Econo-Pac™ disposable chromatography column (BioRad). The column was washed with 20 column volumes, and bound protein was eluted using 7 x 2 column volumes of wash buffer containing 50 mM biotin. All fractions were analysed using SDS-PAGE followed by either Coomassie Blue staining or immunoblotting using a rabbit polyclonal anti-syncollin antibody .

| Syncollin quality control
To check for the presence of disulfide bonds, purified syncollin-Strep was analysed by SDS-PAGE in the presence or absence of β-mercaptoethanol, followed by assessment of band migration on an immunoblot.
To check for syntaxin binding, GST-syntaxin 2 (C-terminal domain, residues 181-264; Edwardson et al., 1997) was bound to glutathione-Sepharose™ (Cytiva), and the beads were incubated with purified syncollin-Strep for 3 hr at 4 C. The supernatant was removed for analysis, and the beads were washed with 10 bed volumes of 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid-(HEPES)buffered saline, pH 7.6 (HBS). A sample of the beads was then taken, and the bound GST-syntaxin 2 was eluted with HBS containing 15 mM reduced glutathione. Syncollin in the bound, unbound and eluted fractions was detected using SDS-PAGE followed by immunoblotting, using a polyclonal anti-syncollin antibody. As a negative control, GST was attached to the glutathione-Sepharose™ in place of syntaxin 2.

| Circular dichroism spectroscopy
Syncollin-Strep was purified in 10 mM sodium phosphate, pH 8.0, containing 0.1% (w/v) CHAPS, and diluted to 10 μM in the same buffer, before 300 μl was added to a quartz cuvette for analysis using a circular dichroism (CD) spectrophotometer (Applied Photophysics).
Far-UV CD data were collected between 200 and 280 nm using a 1-mm path length and a 2-nm bandwidth. A temperature ramp (25 to 95 C) was applied, with 0.5 s per point. All readings were corrected for the signal with buffer alone.

| Peptidoglycan binding
Syncollin-Strep (100 μl of a 0.33 mg/ml solution in 10 mM sodium phosphate, pH 8.0 containing 0.1% [w/v] CHAPS) was incubated overnight at 4 C, with agitation, with or without 320 μg insoluble peptidoglycan (Sigma). Peptidoglycan was precipitated by centrifugation at 21,000g for 3 min at 4 C. The supernatants were retained, and the peptidoglycan pellets were washed twice with ice-cold HBS. Supernatants and pellets were resuspended in equal volumes of sample buffer and analysed by SDS-PAGE followed by immunoblotting, using a polyclonal anti-syncollin antibody. As a control, bovine serum albumin (BSA) was used at the same concentration as syncollin in the binding assay and visualised on gels by Coomassie Blue staining.

| Bacterial growth curves
Lactococcus lactis (Gram-positive) and Escherichia coli (Gram-negative) were used to test for a potential antibacterial effect of syncollin. washed once with phosphate-buffered saline, pH 7.4 (PBS) before resuspension in PBS containing 1 μg/ml propidium iodide and 1 μg/ml DAPI. Resuspended bacteria were incubated in the dark for 20 min at room temperature and then added to poly-L-lysine coated 18-well flat imaging plates (Ibidi) before confocal imaging.
Confocal imaging settings were identical for all conditions for each bacterial strain. Raw images were incorporated into image composites using ImageJ™. Brightness was adjusted identically for all images within an experiment so that red and blue staining could be clearly identified. Red staining and red plus blue staining was quantified using set colour thresholding and area measurement using ImageJ. The percentage of red (dead) cells relative to red plus blue (total) cells were then calculated. For each experiment, at least three images were analysed, and a mean percentage dead cells was determined. Data from independent experiments were then combined. Statistical analysis used one-way ANOVA with Dunnett's multiple comparisons with buffer control.

| Scanning electron microscopy imaging of syncollin-treated bacteria
Lactococcus lactis and Escherichia coli were incubated with syncollin-Strep, or either soybean trypsin inhibitor or ribonuclease A as control proteins, as described above. For L. lactis, 13-mm coverslips were washed with acetone for 5 min before coating with a 50:1 (v/w) acetone:Vectabond™ solution for 5 min. For E. coli, coverslips were washed with water before coating with 0.1% (w/v) poly-L-lysine (Sigma) for 30 min at room temperature. Once coated, coverslips were washed three times with water and dried in a stream of nitrogen. Bacteria in the plateau phase of their growth curves were allowed to adhere to coated coverslips for 30 min at room temperature. Coverslips were very briefly dipped twice in cold, de-ionised water to remove any buffer salts and then quickly plunge-frozen by dipping into liquid nitrogen-cooled ethane. Samples were transferred to liquid nitrogen-cooled brass inserts and freeze-dried overnight in a liquid nitrogen-cooled turbo freeze-drier (Quorum K775X). Samples were mounted on aluminium SEM stubs using conductive silver paint (TAAB) and sputter-coated with 20-nm gold using a Quorum K575X sputter coater. Samples were viewed using an FEI Verios 460 scanning electron microscope run at 1 keV and 25 pA probe current. Images were acquired in SE-mode using an Everhard-Thornley detector.