Whole saliva from two males and one female between the ages of 20 and 40 was collected according to protocols described previously (9). After centrifugation (14,000 × g), the clear whole-saliva supernatants (0.4 μg/μl) collected from the three volunteers were pooled, digested immediately with trypsin (20 ng/μl) overnight (20), and then mixed 1:2 with alpha-cyano-4-hydroxycinnimic acid (7 mg/ml) for analysis by liquid chromatography-MALDI-TOF-TOF mass spectrometry (MS) (4800 TOF-TOF Analyzer; Applied Biosystems, Foster City, CA) (2, 21). Sixty-three saliva peptides corresponding to 22 proteins were identified from tryptic digests (data not shown). Nine saliva peptides ending with a PQ C terminus (Table ​(Table1)1) were derived from basic salivary proline-rich protein 2, salivary acidic proline-rich phosphoprotein (1/2), and basic proline-rich peptide P-E (IB-9).

The GPPPQGGRPQ peptide was synthesized and labeled with fluorescein (GenScript Corp., Piscataway, NJ). The fluorescein-labeled peptide was obtained by coupling 5(6)-carboxyfluorescein to the amino terminus of GPPPQGGRPQ (10). Two gram-positive oral bacteria, Porphyromonas gingivalis (ATCC 33277) and Porphyromonas acnes (ATCC 6919) (11, 16), were chosen for interaction with GPPPQGGRPQ. Both bacteria (10⁸ CFU) were incubated with fluorescein-labeled or unlabeled GPPPQGGRPQ (1 and 10 μM) for 1 h. The green fluorescence derived from the binding of peptides to the bacteria was detectable only when fluorescein-labeled peptide was incubated with P. acnes (Fig. ​(Fig.1),1), but not P. gingivalis (Fig. ​(Fig.1F),1F), indicating that GPPPQGGRPQ selectively adhered to P. acnes. The green fluorescence was visible when P. acnes was incubated with 1 μM (Fig. ​(Fig.1C)1C) or 10 μM (Fig. ​(Fig.1D)1D) of fluorescein-labeled GPPPQGGRPQ, but not unlabeled GPPPQGGRPQ (Fig. ​(Fig.1A)1A) and a fluorescein-labeled CGKRK (10 μM) (Fig. ​(Fig.1B)1B) (a gift from Zhang Lianglin, Moores Cancer Center at the University of California, San Diego). Notably, the green fluorescence generated by 1 μM of fluorescein-labeled peptide was entirely quenched (Fig. ​(Fig.1E)1E) when 1 mM of the unlabeled GPPPQGGRPQ was present in the reaction mixture of fluorescein-labeled peptide with P. acnes, indicating that the binding of GPPPQGGRPQ to P. acnes is peptide specific.

We next examined if the GPPPQGGRPQ peptide influences the growth of P. acnes. P. acnes (10⁵ CFU) was preincubated with phosphate-buffered saline (PBS) (Fig. ​(Fig.2A)2A) or GPPPQGGRPQ at concentrations of 1 μM (Fig. ​(Fig.2C),2C), 10 μM (Fig. ​(Fig.2D),2D), and 100 μM (Fig. ​(Fig.2E)2E) for 3 h. Preincubation of P. acnes with an antibiotic mixture (500 units/ml of penicillin G and 0.5 mg/ml of streptomycin sulfate) for 3 h significantly inhibited the 2-day growth of P. acnes (2.1 × 10³ ± 6.3 × 10³ CFU) in comparison with that after incubation of P. acnes with PBS (20.5 × 10⁵ ± 2.5 × 10⁵ CFU) (Fig. 2B and F). Treatment of P. acnes with 1, 10, and 100 μM of GPPPQGGRPQ considerably attenuated the colonization (Fig. 2C to E) and the growth (26.5 × 10⁴ ± 6.3 × 10⁴, 18 × 10³ ± 2.6 × 10³, and 9.5 × 10² ± 2.1 × 10² CFU, respectively) of P. acnes. These results demonstrated that the GPPPQGGRPQ peptide exerts a bactericidal activity against P. acnes.

Although we failed to identify low-abundance proteins, such as cytokines and defensins, that normally are considerably elevated during oral inflammation (15), an advanced electrospray ionization tandem MS with higher sensitivity may make it possible to identify lower-abundance saliva proteins (7). Since the intensity of each peptide signal in an MS spectrum does not necessarily correlate with its biological abundance, quantitative MS using isotopic-tag labeling (19) will provide a means to determine the differential abundances of peptides under different biological conditions. Although it is unclear if the GPPPQGGRPQ peptide is derived from fragmentation of bacterium-bound proline-rich proteins, it has been documented that proline-rich proteins can be cleaved by proteases from oral Streptococcus and Actinomyces species and converted into peptides with bacterium-binding PQ C termini (13). Intriguingly, several peptides with PQ C termini were identified in undigested human parotid saliva (8), suggesting that peptides with PQ C termini may exist naturally in human whole saliva. When oral bacteria predominate, these peptides with PQ C termini may serve as innate-immunity-like peptides (5, 13) to bind and/or kill oral bacteria instantly. The PQ-rich repeats also exist in other proteins, such as diacylglycerol kinase, an enzyme involved in the regulation of signal transduction (18). It has also been reported that peptides with polyprolyl or polyglutamine sequences could contribute to amyloidogenic diseases (14) and display the activity of cellular permeation, as well as binding to heat shock proteins (1). Thus, peptides with PQ C termini in saliva may have biological functions other than acting as antimicrobials.

Although the target molecules of the GPPPQGGRPQ peptide in P. acnes are undetermined, it has been documented that an RGRPQ peptide could be similar to the ERGMT peptide signal that affects intra- or extracellular receptors and gene expression in Bacillus subtilis (12). Future studies will include determining the eukaryotic toxicity and MICs of the GPPPQGGRPQ peptide against P. acnes. The complete genome of P. acnes has been sequenced (3, 4). P. acnes is involved in many infectious diseases, including acne vulgaris and biofilm formation on implanted biomaterials (17). Therefore, the future applications of the GPPPQGGRPQ peptide may include monitoring of the distribution of P. acnes and/or treatment of P. acnes-associated diseases by inhibiting bacterial growth.