B. fasciatus crude venom and venomous glands were collected from Guang Xi Province, China. The SMART™ PCR cDNA synthesis kit was purchased from Clontech, USA. Chromatography media Sephadex G-50 and CM-Sephadex C-25 were obtained from Amersham Bioscience, Sweden. Trifluoroacetic acid (TFA, HPLC grade) was from Perkin-Elmer. Acetonitrile (ACN, HPLC grade) was bought from Fisher Chemicals. 2,2,2-trifluroethanol-d3 98% (TFE-d3), sodium dodecyl-d₂₅ sulfate (SDS-d25) 98.7%, trimethylsilyl-2,2,3,3-tetradeuteropropionic acid (TSP)-d₄ 98% and D₂O 99% were purchased from Cambridge Isotope Laboratories. Reverse Phase High Performance Liquid Chromatography (RP-HPLC) C4 column (30 cm×0.46 cm) was from Agilent. The pMD18-T vector was from Takara, Dalian, China. All other reagents were of analytical or sequencing grade. The animals used for the experiments were treated according to the protocols evaluated and approved by the ethical committee of Kunming Institute of Zoology.

The purification procedure was according to our previously report [33], [34]. 0.4 g B. fasciatus crude venom was first fractionated using gel filtration chromatography Sephadex G-50 column (26 cm×100 cm), equilibrated with 50 mM Tris–HCl, 50 mM NaCl (pH 7.8). The elution was performed with the same buffer and monitored at UV absorption of 280 nm. The peak having antimicrobial activity was collected and further dialyzed against PBS (pH 6.0). The dialyzed product was next subjected to the cation-exchange CM-Sephadex C-25 column (1.6 cm×30 cm). The elution was achieved with a linear NaCl gradient, at a flow rate of 1 ml/min. The peak with antimicrobial activity was collected and finally purified by reverse phase high performance liquid chromatography (C4), equilibrated with 0.1% (v/v) TFA/water. The elution was performed with a liner gradient of acetonitrile at a flow rate of 0.7 ml/min.

The amino acid sequence of the N-terminus was determined by the automated Edman degradation using an Applied Biosystems pulsed liquid-phase sequencer, model 491. Electrospray ionization mass spectrometry (ESI-MS) was used to determine the molecular weight by a Finnigan LCQ ion trap mass spectrometer (ThermoFinnigan, San Jose, CA, USA) in positive-ion mode. The sample solutions (50%H2O/50%ACN) were infused into the mass spectrometer via a Harvard syringe pump (Holliston, MA, USA). The spray voltage was set to +4.5 kV. Spectra were acquired by summing 30 scans.

Circular dichroism (CD) spectra were recorded at 298 K on a JASCO J-810 spectrometer (Jasco, Japan). Samples were prepared by dissolving the peptide powder to a concentration of 90 µM in TFE/H₂O mixtures or in SDS micelles of different concentrations. The spectra were measured between 190 and 250 nm using 0.1 cm path-length cell with 1 nm bandwidth, 1 sec response time, and a scan speed of 100 nm/min. Three consecutive scans per sample were performed, added and averaged followed by subtraction of the signal of the solvent. The secondary structure elements of the peptides were estimated according to the Yang formula [35].

Samples for nuclear magnetic resonance (NMR) measurements contained 4 mM cantheicidin-BF in TFE-d3/H₂O (9∶1, v/v) at pH 6.5, or in 300 mM SDS-d25 at pH 6.5. All NMR spectra were recorded at 298 K on a Varian Unity INOVA 600 MHz spectrometer equipped with three RF channels and a triple resonance z-axis pulsed-field gradient probe. The 2D ¹H-¹H TOCSY spectra were acquired with a mixing time of 75 ms, while ¹H-¹H NOESY spectra were acquired with mixing times of 200 and 300 ms. The watergate approach was employed for water suppression. Data were collected with 256 and 1024 complex data points in t1 and t2 dimensions, respectively. Signals were averaged over 64 transients. All NMR spectra were processed and analyzed using the NMRPipe/NMRDraw software and the Sparky program [36], [37]. Linear prediction in the t1 dimension was used before the Fourier transformation. Assignments of the proton resonances were achieved using both TOCSY and NOESY spectra. The ¹H chemical shifts were referenced to TSP. The secondary structure was predicted using the H_α Chemical Shift Index approach [38].

Total RNA was extracted using TRIzol (Life Technologies, Ltd.) from the venomous glands of B. fasciatus. cDNA was synthesized by SMART™ techniques by using a SMART™ PCR cDNA synthesis kit (Clontech, Palo Alto, CA). The first strand was synthesized by using cDNA 3′ SMART CDS Primer II A, 5′-AAGCAGTGGTATCAACGCAGAGTACT (30) N-1N-3′ (N = A, C, G or T; N-1 = A, G or C), and SMART II An oligonucleotide, 5′-AAGCAGTGGTATCAACGCAGAGTACGCGGG-3′. The second strand was amplified using Advantage polymerase by 5′ PCR primer II A, 5′-AAGCAGTGGTATCAACGCAGAGT- 3′.

The cDNA synthesized by SMART™ techniques was used as template for PCR to screen the cDNAs encoding serine protease inhibitor. Two oligonucleotide primers, BFS₁ 5′-AA(A/G)TT(T/C)TT(T/C)AG(A/G)AA(A/G)(C/T)T(A/T/C/G)AA(A/G)AA (A/G)-3′, in the reverse direction, a specific primer designed according to the amino acid sequence determined by Edman degradation and primer II A as mentioned in “SMART cDNA synthesis” in the sense direction were used in PCR reactions. The DNA polymerase was Advantage polymerase from Clontech (Palo Alto, CA) The PCR conditions were: 2 min at 94°C, followed by 30 cycles of 10 sec at 92°C, 30 sec at 50°C, 40 sec at 72°C. Finally, the PCR products were cloned into pGEM®-T Easy vector (Promega, Madison, WI). DNA sequencing was performed on an Applied Biosystems DNA sequencer, model ABI PRISM 377.

Reverse transcription-polymerase chain reaction (RT-PCR) was carried out to analyze gene expression of cathelicidin-BF in B. fasciatus. Total RNA extraction from different tissues and first-strand cDNA synthesis were the same as described above. The primers were, forward primer, 5′-cathelicidin: 5′-ATGGAAGGGTTCTTCTGGA AGACC-3′, and reverse primer, 3′-cathelicidin: 5′-CAAATTAGAAGGGGATGGAG ACC-3′. PCR conditions were: 95°C (3 min), and 30 cycles of 95°C (30 s), 56°C (30 s), 72°C (3 min) followed by a 15 min extension period at 72°C. The control PCR was performed using the specific primers (forward primer, actin-s 5′-GGGTGTGATGGT TGGCATGG-3′, and reverse primer, actin-as 5′-TGGCTGGAAGAGGGCTTCTG-3′) for snake actin, using the same conditions as above.

Cathelicidin sequences were obtained from the protein database at the National Center for Biotechnology Information. The phylogenetic tree is constructed by neighbor-joining analysis, using the ClustalW program (version 1.8).

Antimicrobial activities of cathelicidin-BF and cathelicidin-BF15 (VKRFKKFFRKLKKSV) were tested according to our previous methods [39]–[42]. Ampicillin, benzylpenicillin (Amresco) and Imipenem and Cilastatin Sodium for Injection (ICS, Merck) were used as positive controls. The details were provided in the Materials and Methods S1.

In vitro bacteria killing kinetics of cathelicidin-BF, ICS (its minimal inhibitory concentration (MIC) for Escherichia coli 08A866 is 0.15 µg/ml), and HDW (an antimicrobial peptide from the frog of Rana nigrovittata, with a amino acid sequence of FIGPVLKIATSILPTAICKIFKKC, its MIC for E. coli 08A866 is 18.7 µg/ml), respectively, were determined according to the methods described by Mygind et al [43]. The details were provided in the Materials and Methods S1.

Hemolytic activity was checked by incubating the tested samples with human red blood cells to determine hemoglobin releasing ability by measuring the absorbance at 540 nm, using 1% Triton X-100 as a positive control. Cytotoxicity and serum stability were measured according the methods described by Mygind et al [43]. The details were provided in the Materials and Methods S1.

All of the peptides used for the bioactivity assays and NMR analysis in this paper were synthesized by the peptide synthesizer (433A, Applied Biosystems) in AC SCIENTIFIC (Xi An) INC. (Xi An, China) and analyzed by HPLC and MALDI-TOF mass spectrometry to confirm that the purity was higher than 95%. All peptides were dissolved in water.

The crude snake venom was separated into four fractions by Sephadex G-50 gel filtration as our previous report (Figure S1a) (Fig. S1a) [33], [34]. The fraction III, containing antimicrobial activity was further subject to CM-Sephadex C-25 cation-exchange column, and nine sub-fractions were collected (Figure S1b). The fraction VI with both trypsin-inhibitory and antimicrobial activities was further purified using RP-HPLC. The peak with antimicrobial activity is marked with an arrow in Figure S1c. The purified antimicrobial peptide was named cathelicidin-BF. The molecular mass and purity of purified cathelicidin-BF was further analyzed by a ESI mass spectrometry, giving a [M+7H]⁷⁺, [M+7H]⁶⁺, [M+7H]⁵⁺and [M+7H]⁴⁺of 521.1, 607.6, 729.1 and 991.5 (Figure S2), indicating that purified cathelicidin-BF has a molecular weight of 3637.5–3638.5.

Purified cathelicidin-BF was subjected to amino acid sequence analysis using automated Edman degradation. Its amino acid sequence is KFFRKLKKSVKKRAKEFFKKPRVIGVSIPF. Cathelicidin-BF is composed of 30 amino acid residues including 12 basic residues (9 Lys and 2 Arg), 5 phenylalanines, and only one acidic amino acid residue (Glu). It is a lysine-rich and phenylalanine-rich peptide. Analysis using the ExPASy MW/pI tool (http://www.expasy.ch/tools/pi_tool.html) showed that cathelicidin-BF had the predicted pI (isoelectric point) of 11.79 and a predicted molecular weight of 3637.5 that matched well with the observed mass by ESI mass spectrometry (Figure S2). By BLAST search, no similar sequence was found in GenBank.

Several positive clones, which contained an insert of 750 bp were identified and isolated from B. fasciatus venomous gland cDNA library. The complete nucleotide sequence of cDNA (GenBank accession EU753183) and deduced amino acid sequence of cathelicidin-BF precursor are shown in Figure 1. Unexpectedly, the cathelicidin-BF precursor displays the maximal similarity (47%) with predicted myeloid cathelicidin 3 from Ornithorhynchus anatinus (GenBank accession XP_001512130) by BLAST search. The protein precursor is composed of 191 amino acid (aa) residues, including a predicted signal peptide, a conserved cathelin domain and a mature cathelicidin-BF. Noticeably, four cysteines that are conserved in the cathelin domain of all mammalian cathelicidins are also invariantly spaced in cathelicidin-BF precursor, suggesting that the snake cathelicidin-BF precursor is a real mammalian cathelicidin.

The amino acid sequence of cathelicidin-BF determined by Edman degradation is identical with the amino acid sequence deduced from the cDNA sequence. There is a possible cleavage site (Valine¹⁵⁷) for elastase at the N-terminus of the mature cathelicidin-BF (Figure 1). Based on the possible cleavage site, a 34-aa peptide should be released from the precursor, but the purified cathelicidin-BF is only composed of 30 aa. Different from other cathelicidins, there is an acidic doman between the cathelin doman and the antimicrobial peptide in the cathelicidin-BF precursor (Figure 2).

Evolution analysis revealed that all vertebrate cathelicidins formed three distinct clusters with fish cathelicidins located in a separated clade from others. Supported by a bootstrap value of 77%, cathelicidin-BF was clustered with platypus CATH-3 (Figure 3). Although platypus is a mammal, it also has reptilian features, For instance, it lays eggs and it is venomous [44]. The close evolution relationship of cathelicidin-BF found in the venoms of B. fasciatus with platypus cathelicidins may provide further proof for platypus's reptilian features.

The secondary structure elements in different solvent environments were detected by CD spectroscopy (Figure S3, Table S1). In H₂O, the CD spectrum of cathelicidin-BF showed a strong negative band at 200 nm, indicative of a random-coil conformation. Interestingly, in TFE/H₂O mixtures, the CD spectra showed double minima at 208 and 222 nm, indicating a highly α-helical conformation. The signals at 208 and 222 nm were intensified gradually by increasing concentrations of TFE, which indicated that the helicity of the peptide was increased in more hydrophobic or membrane-mimetic environments. The CD spectra of the peptide in SDS micelles also showed a typical α-helix pattern and the content of the α-helix structure increased with the increasing SDS concentration.

NMR spectra recorded on the peptide in SDS micelles were of low quality and can not be used for structural analysis, might due to aggregation of the peptide in negative charged SDS micelles. Therefore, the helical structure of the peptide in TFE/H₂O mixture was investigated using NMR spectroscopy. Table S2 lists the nearly complete assignments of the proton chemical shifts of cantheicidin-BF in TFE/H₂O mixture (9∶1, v/v). Comparison of the HN-HN region of NOESY spectrum recorded on cantheicidin-BF in H₂O with that in TFE/H₂O (9∶1, v/v) illustrates that, the peptide adopts a stable secondary structure in TFE/H₂O mixture (Figure S4). The H_α CSI prediction is indicative of a helical structure in the N-terminal region comprising residues F2–F18 (Figure S5), although a well-defined three-dimensional structure of the peptide in TFE/H₂O mixture has not been obtained yet, mainly due to the deficiency of enough unambiguous conformational restraints for the exact structural analysis.

The amphipathic helical conformation is well known to be a crucial factor for many antimicrobial peptides to interact with membranes [45], [46]. On basis of the rough structural analysis described above, it is indicated that the N-terminal region of cantheicidin-BF adopts a typical amphipathic α-helical conformation (Figure S6) as many other cantheicidins.

Using actin as control, expression pattern of cathelicidin-BF was investigated by RT-PCR. Tissue distribution of cathelicidin-BF expression in snake tissues were illustrated in Fig. 4. All the selected tissues including stomach, trachea, skin, muscle, heart, kidney, lung, brain, intestine, spleen, liver, ovary and venomous gland can express this protein.

As listed in Table 1, cathelicidin-BF and its analogue, cathelicidin-BF15 showed strong antimicrobial activities against tested microorganisms. Of the 40 tested microorganism strains, cathelicidin-BF exerted potent antimicrobial ability against most of Gram-negative bacteria (either standard strains or clinically isolated drug-resistance strains). For most of E. coli, the MICs are lower than 2.3 µg/ml, while ampicillin, benzylpenicillin and ICS are effective only to standard strain with a MIC of 18.7, 37.5 and 0.15 µg/ml respectively. The lowest MIC for K. pneumoniae is 0.3 µg/ml, while ampicillin, benzylpenicillin and ICS are effective only to standard strain with a MIC of 150, 18.7 and 9.4 µg/ml respectively. In contrast, most of S. aureus are not so sensitive for cathelicidin-BF, only one strain could be killed by cathelicidin-BF with a low MIC (4.7 µg/ml). Another Gram-positive bacteria genus, Bacillus also seems to be sensitive for cathelicidin-BF and cathelicidin-BF15. A dangerous clinically isolated strain, Salmonella typhi could also be killed by cathelicidin-BF and cathelicidin-BF15 with a low MIC (1.2 µg/ml). Cathelicidin-BF and cathelicidin-BF15 are the same effective to some fungi as bacteria, for example, C. albicans ATCC2002 (with a MIC of 4.7 µg/ml).and P. pastoris (with a MIC of 0.3 µg/ml). Cathelicidin-BF exerted obvious antimicrobial activity against some saprophytic fungus such as A. terreus GIM3.34 (with a MIC of 18.7 µg/ml), A. niculans (with a MIC of 4.7 µg/ml), and C. globosum (with a MIC of 37.5 µg/ml). All the tested classic antibiotics including Ampicillin, Benzylpenicillin and ICS had no effect on these funguses. Several other cathelicidin-BF analogues, KF1–11 (KFFRKLKKSVK), KF12–19 (KRAKEFFK) and KF20–30 (KPRVIGVSIPF) had no any antimicrobial activity.

The antimicrobial activity of cathelicidin-BF in different solutions was also investigated as listed in Table 2. In 150 mM phosphate buffer solution (PBS) and 150 mM NaCl solution, cathelicidin-BF had stronger antimicrobial activities that in water. It suggested that salts could increase cathelicidin-BF's antimicrobial ability.

Using the antibiotics ICS as a positive control, antibacterial properties of the snake cathelicidin-BF were tested by the colony counting assay. As listed in Table 3 and Table 4, cathelicidin-BF could rapidly exert its antibacterial activities. It just took less than 1 minute to kill all the E. coli at the concentration of 1, 5 or 10 times of MIC. The antibacterial activity was proved to be lethal for E. coli. E. coli was not capable of resuming growth on agar plates after a 6-h treatment with concentrations above the corresponding MICs. In contrast, the antibiotics, ICS could not clean the bacteria within 6 h at the concentration of 1 or 5 times of MIC. Only 10 times MIC of ICS could clean all the E. coli within 6 h. Furthermore, E. coli treated by 1 time MIC of ICS was capable of resuming growth during 6 h.

In order to compare properties with other antimicrobial peptide, the frog antimicrobial peptide HDW was used as a control. Their bacteria killing kinetics during 30 min was listed in Table 4. Although HDW had a rapid bacteria killing ability, cathelicidin-BF is faster to clean E. coli than HDW. Cathelicidin-BF just took less than 1 minute to clean E. coli, while HDW took several minutes.

Cathelicidin-BF had little hemolytic activity on human red blood cells even with peptide concentrations up to 400 µg/ml. At the same concentration, cathelicidin-BF was neither cytotoxic for mouse macrophage (RAW264.7) nor for human liver tumor cell (HepG₂) (data not shown). Thus, it showed considerable selectivity for microorganisms over mammalian cells in vitro.

Serum stability was checked by incubating 100 µg/ml cathelicidin-BF and cathelicidin-BF15 with 90% fresh normal human serum at 37°C for 0, 1, 2, 3, 6, 10 and 24 hours. For cathelicidin-BF, antimicrobial activities against E. coli 08A866 could not be detected after 3 h-incubation, while cathelicidin-BF15 could keep its antimicrobial activity up to 10 h in 90% fresh normal human serum. Cathelicidin-BF15 seems to be more stable than cathelicidin-BF in serum.