Effect of human secretory calcium-binding phosphoprotein proline-glutamine rich 1 protein on Porphyromonasgingivalis and identification of its active portions

The mouth environment comprises the second most significant microbiome in the body, and its equilibrium is critical in oral health. Secretory calcium-binding phosphoprotein proline-glutamine rich 1 (SCPPPQ1), a protein normally produced by the gingival epithelium to mediate its attachment to teeth, was suggested to be bactericidal. Our aim was to further explore the antibacterial potential of human SCPPPQ1 by characterizing its mode of action and identifying its active portions. In silico analysis showed that it has molecular parallels with antimicrobial peptides. Incubation of Porphyromonas gingivalis, a major periodontopathogen, with the full-length protein resulted in decrease in bacterial number, formation of aggregates and membrane disruptions. Analysis of SCPPPQ1-derived peptides indicated that these effects are sustained by specific regions of the molecule. Altogether, these data suggest that human SCPPPQ1 exhibits antibacterial capacity and provide new insight into its mechanism of action.

Effect of human SCPPPQ1 on P. gingivalis. To test the antibacterial propensity of the synthetic human SCPPPQ1 on P. gingivalis, we have attempted (i) to determine the minimum inhibitory concentration (MIC) using a broth microdilution assay, and (ii) to perform a killing assay of planktonic cells. However, given that the SCPPPQ1 has to be solubilized in a urea buffer, these assays did not allow any conclusions since the buffer itself was found to be bactericidal in these specific conditions (see Supplementary materials). As an alternative method, a qualitative bactericidal assay on a blood agar plate was performed. In this assay, P. gingivalis was mixed In silico analysis of the human SCPPPQ1 protein using the APD3 software provided information on the distribution and relative percentage of amino acids with hydrophobic properties, positive and negative charge, and other amino acids. (c) Tridimensional structure model from i-TASSER structure prediction software for human (red) and rat (pink) SCPPPQ1 proteins and their overlay. Secondary structure elements are included (H Helix, S Sheet) and labeled according to their position in the structure (H1 to H3, S1 to S3).
Scientific Reports | (2021) 11:23724 | https://doi.org/10.1038/s41598-021-02661-w www.nature.com/scientificreports/ with SCPPPQ1 and then applied on the surface of the agar plate. In the negative control (buffer without the protein), growth of the bacteria resulted in a uniform layer that appeared black on the blood agar plates, while areas without bacteria appeared white. As seen in the representative assay (Fig. 2), the extent of disruption of the uniform layer was more important when the SCPPPQ1 concentration was increased. As reported in Table 1, P. gingivalis was clearly affected at concentrations of 20 μM and above, although some sensitivity was also apparent with concentrations as low as 5 μM.
To characterize in greater detail the antibacterial propensity of the synthetic human SCPPPQ1 on P. gingivalis, we analysed the effect of the protein on the number of bacteria. Qualitative (Fig. 3a) and quantitative analysis of scanning electron microscope (SEM) images (Fig. 3b) revealed a significant decrease of 44% in the bacterial number after 20 min of incubation with respect to the buffer without the protein (negative control). The images also showed the presence of aggregates following incubation with SCPPPQ1 (Fig. 3a). Fluorescence-activated cell sorting (FACS) analysis was also used to evaluate the effect of SCPPPQ1 on P. gingivalis over time. There was a significant reduction in the number of FACS events as soon as the protein was added (40% reduction at 0 min incubation) (Fig. 3c). During the first hour, this decrease gradually reached 75% and then remained constant over time (Fig. 3c). In contrast, as expected, the evolution of the bacterial population in the buffer without the protein progressively doubled during the same time lapse analysis (Fig. 3c), despite the presence of urea. Together, these two independent methods demonstrate a significant variation in bacterial behavior in presence of SCPPPQ1.
Effect of SCPPPQ1 on formation of bacterial aggregates. To evaluate the propensity of SCPPPQ1 to induce formation of bacterial aggregates, the turbidity of the culture medium under the experimental and control conditions was compared. Addition of the protein to the bacterial suspension decreased its turbidity and increased precipitation and accumulation of material at the bottom of the tube as compared to adding buffer without protein (negative control) (Fig. 4a). Super-resolution fluorescence imaging revealed that precipitates, only formed in the presence of SCPPPQ1, consisted largely of aggregated bacteria (Fig. 4b). Furthermore, FACS analysis indicated that the relative volume of aggregates increased from almost two-fold at 0 min to fourfold at 120 min during incubation with SCPPPQ1 compared to the negative control (Fig. 4c). To eliminate a possible bias that could be introduced by the self-aggregation of the protein, SCPPPQ1 alone was similarly analysed by FACS at 0, 60 and 120 min. None or few fluorescent aggregates were detected (Fig. 4c). Details of the volume of these different aggregates are presented in Fig. S1a. In addition to there being more aggregates when bacteria are incubated with SCPPPQ1, aggregate volumes above 2.5 K on the FSC-H axis are significantly represented at 60 min (Fig. S1b). Similar results were obtained at the other time intervals (data not shown). Altogether, these results indicate that SCPPPQ1 promotes the formation of bacterial aggregates. Determination of the working concentration of SCPPPQ1 that affect P. gingivalis growing conditions. Comparison of the difference in growth of the bacteria on a blood agar petri dish in (a) the presence of buffer alone (negative control) or (b-f) various concentrations of + SCPPPQ1. (b-f) Enlargements of the boxed areas in the blood agar plate (left panel). (a) Under control conditions the bacteria formed a relatively uniform layer that appeared black in photographs, (b-f) while in presence of proteins, this layer was interrupted producing a mosaic of black areas containing bacteria (white arrowhead) and white areas of various sizes that indicate absence/paucity of bacteria (white arrow). Qualitative evaluation suggests that the effect of SCPPPQ1 on the bacteria (white areas) started at 5 µM (weak) and plateaued at 20 µM (see Table 1). These results are representative of three experiments. super-resolution fluorescence imaging was applied (Fig. 5). Protein was observed on the majority of bacterial aggregates and it accumulated focally causing a beaded fluorescence appearance around the bacteria ( Fig. 5d-f).
Labeling was absent and only bacteria were visible in negative control incubations ( Fig. 5a-c). Colloidal gold immunolabelling for SCPPPQ1 was carried out and the presence of gold particles on the surface of bacteria was visualised by SEM (Fig. 6a). These preparations also allowed to observe the disruptive effect that the protein has on bacterial membrane integrity (Fig. 6b). Such membrane disruption was infrequently observed under control conditions (carrier buffer alone). Transmission electron microscope (TEM) analysis of pre-embedding colloidal gold labeled preparations confirmed the above SEM data (Fig. 7). In addition, TEM imaging revealed (a) the association of the protein with the bacteria surface/membrane (Fig. 7a), (b) the alteration of the membranes by the protein, which appeared disrupted and having an enlarged spacing (Fig. 7a,b), and (c) the presence of bacterial debris as well as fine granular extracellular material (Fig. 7b). Together, these results indicate a causal link between the interaction and surrounding of bacteria with SCPPPQ1 and their alteration.

Discussion
It was recently reported that rat SCPPPQ1 has a marked effect on the membrane integrity of P. gingivalis that ultimately affects its growth 11 . A previous prediction for the human sequence of SCPPPQ1 20 turned out to be somewhat different from the one decrypted using DNA extracted from human dental tissues and reported in GeneBank (#MK322956.1; https:// www. ncbi. nlm. nih. gov/ nucco re/ MK322 956.1). This sequence also has only 56% identity and 64% similarity with its rat homologue. Because of these differences, it was therefore important to validate and further explore the antibacterial potential and mode of action of human SCPPPQ1. The various assays carried out, showed that the P. gingivalis population rapidly and significantly decreases when incubated with human SCPPPQ1. Two mechanisms, aggregation of bacteria and membrane disruption, appear to be involved in this decrease. As we have previously shown, SCPPPQ1 tends to self-aggregate and intrinsically form molecular networks 9 , which we believe could promote bacterial 'clumping' and favor the interaction of SCPPPQ1 with the membrane of intact bacteria, ultimately affecting their integrity. In silico analyses, using two different prediction softwares, further indicated that like the rat protein, human SCPPPQ1 possesses potential antimicrobial peptide sequences. Some AMPs are part of the innate immune system of both humans and animals and act as a first line of defence against a hostile environment 21 . They have broad-spectrum antimicrobial activities and possess modes of action that cannot be easily hijacked by the pathogen. The combination of small peptides with other AMPs or antibiotics for antibacterial applications [22][23][24][25] is an emerging category of therapeutic agents. Such peptides have the advantages of being (a) easily synthesized and delivered, (b) more resistant, (c) less toxic and (d) more specific and selective 26 . Peptide 2-16 and peptide [17][18][19][20][21][22][23][24][25][26][27][28][29][30] were responsible for aggregation and, together, may determine the overall higher aggregating capacity demonstrated by the full-length protein. Regarding membrane disruption, our data points to peptide [34][35][36][37][38][39][40][41] (FPLPPQPP) as the active portion of the protein. The resulting cell lysis can also help to reinforce bacterial clumping by releasing DNA extracellularly. Because of the "sticky" nature of free DNA, it can act as an agglutinin that promotes the formation of larger bacterial clumps 27 . Altogether, our data suggest a concerted antibacterial action caused by bacterial aggregation and membrane disruption sustained by different active portions of the molecule. In addition to P. gingivalis, we also have generated new evidence that other major bacteria implicated in PD are susceptible to human SCPPPQ1. There are examples of AMPs with both antibacterial and antifungal properties such as Histatins produced by the parotid and submandibular glands 28 . In this context, we explored whether SCPPPQ1 has any effect on Candida albicans (C. albicans), a fungus present in the oral environment. Initial results show that samples exposed to SCPPPQ1 showed fewer fungi as compared with incubation with buffer only (Fig. S3). Hyphae were also more abundant in the control (Fig. S3). This is interesting since they have an important role in causing disease 29 . Thus, SCPPPQ1 could potentially also have both antibacterial and antifungal properties.
A challenge for antibiotics is to diffuse through biofilm and/or bacterial membranes, therefore biofilms bacteria show greater resistance to antibiotics 30 . In the case of the oral environment, P. gingivalis was found in up to 85% of dental plaque biofilms sampled from patients with PD [31][32][33] . Due to the natural presence of SCPPPQ1 in the mouth, its affinity to mineral tooth surface 20,34 and its action against P. gingivalis and other dental plaque bacteria, this protein appears to be a prospective candidate to fight dental plaque biofilms. Its distribution in slow delivery systems and tooth paste, for instance, could be exploited to prevent bacterial attachment and accumulation on exposed tooth surfaces to limit biofilm formation. Since phosphoserine residues interact strongly with hydroxyapatite 35,36 , modifying our most effective peptides with a tooth-binding domain of diphosphoserine (Ser(p)-Ser(p)-) should probably increase their potential against dental biofilm 36 .
The findings we report here are novel and point to an interesting antimicrobial potential, however, our study has some limitations. While we have demonstrated that human SCPPPQ1 has an impact on both the population number and cell membrane integrity of P. gingivalis, we were unable to carry out a dose-dependent effect using a conventional MIC assay 37 because we could not discriminate the effect of the protein from that of the Na 2 HPO 4 buffer containing urea at final concentrations ≤ 1 M. However, a concentration-dependent response, albeit qualitative, was observed using a blood agar growth plate assay, which validated the 20 μM SCPPPQ1 concentration used in our various experiments.
The reason we have used a urea-containing buffer is that SCPPPQ1 does not dissolve readily. Urea is well known to have antibacterial properties 38 and is often used for extracting proteins from the bacterial cell surface 39 , www.nature.com/scientificreports/ albeit at substantially higher molarity (e.g. 8 M vs ≤ 1 M) and with longer incubation period (e.g. 1 h vs 20 min) than we have used. Clearly this confounds interpretation of our results, but in all cases, they were contrasted to the urea-containing buffer control diluted in the same proportions. Also, the distinctive results obtained with the various peptides highlights an effect above and beyond any contribution by the buffer. Since it cannot be excluded that the observed antibacterial effect relates in part to urea, optimisation of the antibacterial activity of SCPPPQ1 and particularly therapeutic use will require a better adapted delivery vehicle. In fact, urea is used to isolate the crude bacterial envelope and bacterial outer membrane 39 and it could even actually 'prime' the surface for SCPPPQ1 binding for a synergistic action. Irrespective of these limitations, the evidence derived from our multiple experimental approaches used shows that SCPPPQ1 has an effect on bacterial number and integrity.
In conclusion, our results brought evidence that human SCPPPQ1, a structural extracellular matrix protein, can affect various periodontal pathogens, more specifically P. gingivalis. Its mode of action against P. gingivalis has been demonstrated for the first time and implicates bacterial aggregation and membrane disruption. These lead to major membrane breaches that are bound to cause cell death. Using engineered peptides, we have further identified the active portions of SCPPPQ1 respectively implicated in these two processes. The integration of SCPPPQ1 or these peptides in oral hygiene products may offer novel therapeutic strategies for controlling the formation of biofilms associated with oral surfaces and dental implants 40 , limiting development of dental carries and PD, and ultimately for alleviating linked systemic complications. In a broader context, it could also offer an additional prospective to deal with the increasing challenge of bacterial resistance.

Methods
In silico analysis. CLUSTAL multiple sequence alignment tool Multiple Sequence Comparison by Log-Expectation, (MUSCLE version 3.8, https:// www. ebi. ac. uk/ Tools/ msa/ muscle/; EMBL-EBI, Hinxton, England) 41 was used to compare the sequence of rat and human SCPPPQ1. The amino acid composition and the antimicrobial potential of SCPPPQ1 was determined using both AMP database APD3 (https:// aps. unmc. edu; University of Nebraska, Omaha, NE, USA) 14 and different prediction algorithms (iAmpPred (http:// cabgr id. res. in: 8080/ amppr ed/ index. html), and AmPEP and Deep-AmPEP30 (https:// cbbio. online/ AxPEP/)) 12,13 . The Grand Average Hydropathicity Value (GRAVY) was measured using the software Sequence Manipulation Suite 42 to obtain the hydropathic character of the SCPPPQ1 protein 43 . To predict the tridimensional conformation of proteins  Protein and peptide production. All tests were performed using human SCPPPQ1 protein and derived peptides synthesised by LifeTein Inc (Somerset, NJ, USA) with amidation in the C-terminal for peptides. To be consistent with the rat protein produced in bacteria, we have included a methionine at the beginning of the protein sequence 11 . In some cases, two arginines were added at the C-terminal end of peptides to increase their total net charge. The bulk synthetic products were solubilized in 50 mM Na 2 HPO 4 (Sigma-Aldrich) containing 6 M urea (Sigma-Aldrich) buffer at pH 7. Several concentrations of protein ranging from 5 to 150 µM and peptides from 10 to 200 µM were screened to determine a minimal working concentration that yielded apparent effects. It was determined that a final concentration of 20 µM for the protein and of 150 µM for the peptides would be used for the analyses described below.
Several lots of SCPPPQ1 were commercially synthesized and used throughout the study period. For every lot, following solubilisation, the protein concentration was evaluated using a Biodrop (Montreal Biotech Inc., Kirkland, QC, Canada) and ranged between 114-259 µM. To achieve a final 20 µM concentration, the 'stock' proteins dilutions ranged between 5.7-12.9X. Consequently, the final urea concentration to which bacteria were exposed in the analyses ranged between 0.47-1.05 M, for both the protein incubations and matched negative control (buffer only). with damaged membrane in percentage from the negative control and + peptides treated samples. Two peptides were synthesized in which two arginines were added on the C-terminal end; these are referred to as peptide  and peptide  . Data are represented as mean ± standard error of mean (n = 5). Significance was determined by two-tailed Student's t test analysis (ns: p > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001).
Scientific Reports | (2021) 11:23724 | https://doi.org/10.1038/s41598-021-02661-w www.nature.com/scientificreports/ Growth plate assay. The test was carried out on 10% Sheep Blood agar petri plate (Thermo Fisher Scientific). As for the liquid cultures, the petri dishes were reduced beforehand for 48 h under anaerobic conditions. Bacteria in Todd-Hewitt broth were mixed with different concentrations of SCPPPQ1 (see above) or protein solubilisation buffer only, as negative control. For each condition, the exact same liquid volume was quickly spread on the surface of a blood agar petri plate. The plates were then incubated at 37 °C under anaerobic conditions for 4 days to allow bacterial growth. were incubated with a 1% aqueous osmium tetroxide (Electron Microscopy Sciences) for 1 h at room temperature (RT). They were then dehydrated through a graded ethanol series (30%, 50%, 70%, 80%, 90%, 95% and twice in 100%) and finally dried using a Critical Point Drier CPD300 (Leica Biosystems, Concord, ON, Canada). A Regulus 8220 field-emission SEM (Hitachi, High-Technologies, Tokyo, Japan) operated at 1 kV was used to visualize the effect of the protein or peptides on the bacteria. ImageJ (NIH, Bethesda, MD USA) was used to count bacteria on at least ten representative images per sample derived from five different experiments for a total of 50 images analysed. Each condition on a disc incubated with SCPPPQ1 had a corresponding negative control disc for comparison. Fig. 3b, data are represented in percentage of bacteria present on the surface of titanium discs for a total area of 7560 µm 2 . The total number of bacteria on the negative control disc was counted and considered as the 100% reference. Then, the bacteria of the disc incubated with the protein were also counted and compared to its negative control to determine the ratio between the control and the treated sample. For Fig. S2b, to determine bacterial aggregation following incubation of bacteria with peptides, we used the same method of counting as described above to represent the percentage of bacteria in aggregate. Briefly, all bacteria (single and in aggregates) present on the disc were counted and then those present in aggregate were separately recounted. A ratio was then determined between the bacteria present in aggregate and the number of total bacteria on the disc for each condition.

Counting analysis methods. For
For Fig. 9c,d, to determine the percentage of bacteria with damaged membrane after incubation of the bacteria with peptides, all the bacteria (intact and damaged) were counted on the disc, and only those with damaged membrane were separately recounted. For each condition, a ratio was calculated to determine the percentage of bacteria with damaged membrane comparing to the total number of bacteria on the disc.

Flow cytometry analysis, Fluorescence-activated cell sorting (FACS). Suspensions of bacteria
were incubated with either a 20 µM protein solution or buffer only (negative control) for up to 2 h at 37 °C. At 0, 20, 40, 60, 90 and 120 min, 50 µL of the mixture were sampled and fixed with 1% glutaraldehyde. To evaluate the propensity of the protein alone to form aggregates the protein solution at a final concentration of 20 µM was also incubated for up to 2 h at 37 °C in the supplemented Todd-Hewitt broth and then fixed as for the bacteria at 0, 60 and 120 min. The suspensions were then stained using 5 µg/mL FM 4-64 dye final (Thermo Fisher Scientific) before analysis with the analyser FacsAria III SORP (Becton Dickinson Life Sciences, Franklin Lakes, NJ, USA). The data were analysed using the FlowJo program version 10.6.1 (Becton Dickinson Life Sciences). The data obtained in Figs. 3c and 4c are represented as fold change over the negative control (buffer only) at 0 min due to the sensitivity of the FACS and to the variations between experiments.
For the aggregate volume experiments (Fig. 4c and Fig. S1), the total of aggregates was subdivided into different volume categories in the same way for all the samples in order to arbitrarily and consistently characterize various size of aggregates (Fig. S1a). To determine variation in total aggregates volume, the percentage of number of events for each subdivided part were multiplied by the number designated by the FSC-H axis in order to obtain a value for each section (smaller aggregates 1 to bigger aggregates 8). Detailed comparative analysis of aggregate formation under all the tested conditions (protein alone, bacteria alone, and protein + bacteria) was only carried out, and data at 60 min was presented. The independent values were added to obtain the total aggregates volume.   Immunogold studies. For immunogold studies analysed by SEM, bacteria were incubated with 20 µM (final concentration) of protein or the same volume of buffer only (negative control) on polished titanium discs at 37 °C for 20 min and then fixed 30 min with 4% PFA and 0.1% glutaraldehyde. After rinsing samples three times with 0.1 M PB, samples were blocked 1 h with 5% milk in 0.1 M PB. They were then incubated 1 h in 0.5% milk in 0.1 M PB with 1:500 rabbit anti-human SCPPPQ1 antibody at RT. They were rinsed three times in 0.1 M PB, followed by a second 1 h blocking step and incubated 30 min with 1:50 20 nm protein A-gold beads (UMC, Utrecht, Netherlands) at RT. After rinsing three times with 0.1 M PB, samples were again fixed 30 min with 4% PFA and 0.1% glutaraldehyde, rinsed, incubated in 1% aqueous osmium tetroxide for 1 h at RT, dehydrated at RT through an ethanol series (30%, 50%, 70%, 80%, 90%, 95% and twice in 100%) for 15 min at each step and finally dried using a CPD300. A Regulus 8220 operated at 1 kV to detect the protein on the bacteria. For preembedding immunogold studies, bacteria were incubated with 20 µM (final concentration) of SCPPPQ1 or buffer only (negative control) at 37 °C for 20 min. Then, the same immunolabeling protocol as described above for titanium discs was applied, except that between each step the sample was centrifuged for 4 min at 7,500 rcf. After the 100% ethanol step, the bacteria were processed for embedding in LR-white (Electron Microscopy Sciences) as follows: 3 parts of pure ethanol and 1 of resin (6 h, 4 °C), 2 parts pure ethanol and 2 parts resin (overnight, 4 °C), 1 part pure ethanol and 3 parts resin (12 h, 4 °C), pure resin (overnight, 4 °C), 100% resin (6 h, 4 °C), and 100% resin (72 h, 65 °C for curing). Ultrathin sections of 80-110 nm in thickness were cut with a diamond knife and transferred onto Formvar-coated (polyvinyl formate) 200-mesh nickel grids (Electron Microscopy Sciences) for imaging. Grids were stained with uranyl acetate and then examined with a Tecnai 12 (FEI, Eindhoven, Netherlands) transmission electron microscope (TEM) operated at 80 kV.
Statistical tests. For SEM and FACS, mean values and standard error of mean were calculated from at least three independent experiments. p values were obtained by a two-tailed Student's t test analysis of each condition from data in excel spreadsheet. Statistical significance was defined as p < 0.05 (*), p < 0.01 (**) and p < 0.001 (***).