2.1 Antimicrobial activity screening

A screening of the antimicrobial activities of 17 of the 50 plants initially selected was undertaken to pinpoint plants presenting potential as cosmetic preservatives. For this, solvent crude extracts were prepared at respectively 2% (massic concentration in the raw extract; results displayed in Table 1) and 0.2% (data non shown): EtOH/H₂O 75:25 (v/v) extracts were preferentially tested because such a solvent system is non-toxic and enables the extraction of a broad range of metabolites. Their antimicrobial activity was evaluated in compliance with the regulation against the yeast Candida albicans, the filamentous fungus Aspergillus niger and the bacteria Staphylococcus aureus and Pseudomonas aeruginosa, using 96-well microplate assays based on microbial growth inhibition assessed after respectively 24 and 48 h for yeast and fungus, and 48 and 72 h for bacteria. Tests were performed on several extracts from plants collected at different seasons over a 2–3 years period, and the results shown in Table 1 represent the mean activity of each extract.

As shown in Table 1, almost all of the tested extracts display good to excellent S. aureus growth inhibition activities. On the contrary, only the Pistacia lentiscus extract (EtOH/H₂O 70:30 v/v) seems to present a good activity against the growth of P. aeruginosa after 48 h.

Interesting activities against the three remaining micro-organisms have been observed for the Laurus nobilis L. and Hypericum perforatum L. extracts (EtOH/H₂O 75:25 v/v). The prominent antibacterial activity of the ethanolic extract of L. nobilis L. fresh leaves and twigs was already assessed [15]. Similarly, the effective antibacterial activity of the methanolic extract of L. nobilis L. seed oil was observed by Ozcan et al. [16] and one of the supercritical fluid extracts of L. nobilis L. leaves was observed by Santoyo et al. [17]. In a similar way, the antibacterial activity of various extracts of H. perforatum L. has already been assessed [18–20].

The EtOH/H₂O 75:25 (v/v) crude extract of the flowered aerial parts of Santolina chamaecyparissus display the best unprecedented results against these micro-organisms (inhibition superior to 90% of the growth of C. albicans, A. niger and S. aureus), except for P. aeruginosa.

S. chamaecyparissus L., commonly known as cotton lavender, is a species of flowering plant belonging to the Asteraceae family. It is native to Southeastern (Croatia and Italy) and Southwestern Europe (France and Spain) and is mainly used as an ornamental plant [21].

This extract was analysed by using a High Performance Liquid Chromatography-Diode array Detector-Evaporative Light Scattering Detector (HPLC-DAD-ELSD; Fig. 1): besides polar (e.g., sugars and amino acids) and some hydrophobic compounds, this extract is mainly constituted by flavonoids and spiroketal derivatives, among which a major unknown compound eluting after 26 min is detected. This extract was then selected for further investigation and optimization for practical cosmetic formulation.

2.2 Bioguided fractionation of the S. chamaecyparissus extract

In order to identify the active compounds, the EtOH/H₂O 75:25 (v/v) extract of S. chamaecyparissus was then fractionated over silica gel (normal phase). The fractionation of 2.2 g of the bulk extract led to the recovery of five distinct fractions: F1 (cyclohexane; 8 mg), F2 (cyclohexane/AcOEt 1/1; 249 mg), F3 (AcOEt; 88 mg), F4 (EtOAc/MeOH 1/1; 781 mg) and F5 (MeOH; 412 mg).

The resulting fractions were further evaluated for their antimicrobial activity. Fraction solutions were concentrated at 100 mg/mL in DMSO and their activities were assessed the same way as one of the crude extracts.

Table 2 shows the inhibition percentages of the five resulting fractions against four bacteria, yeast and fungi at a concentration of 0.4% (w/v) (the results obtained for the extract at a concentration of 0.04% are not shown in the present article). Methyl-paraben at 0.4% (w/v) was used as the positive control.

No activity was observed for the cyclohexane (F1) and MeOH fractions (F5). The EtOAc/MeOH fraction (F4) was only active at 0.4% (w/v) against A. niger (inhibition percentage higher than 60%). The ethyl acetate fraction (F3) was significantly active at 0.4% (w/v) against A. niger and C. albicans, but no activity was observed against P. aeruginosa and S. aureus.

Finally, the cyclohexane/AcOEt fraction (F2) was the most interesting one, exhibiting a strong antimicrobial activity against the four evaluated strains over 48 h (C. albicans and S. aureus) and over 72 h (P. aeruginosa and A. niger).

2.3 Isolation and identification of the active metabolite

HPLC-DAD-ELSD analysis of both the crude extracts of S. chamaecyparissus and the five resulting fractions showed the presence of a major metabolite in F2 (Fig. 2) that is absent in other fractions: this major metabolite was suspected to be responsible for the antimicrobial activity of S. chamaecyparissus crude extracts or at least a contributor to it.

Semi-preparative HPLC was then performed on this active fraction in order to confirm this assumption: multiple injections of this fraction concentrated at 50 mg/mL were carried out and 9 mg of the molecule were obtained.

Its molecular formula C₁₅H₁₂O₄ was deduced from UPLC-ESI-HRMS experiments ([M+H]⁺ m/z 257.0816, C₁₅H₁₃O₄). The UV spectrum showed the band of maximum absorption centred at 310 nm.

¹H and ¹³C NMR, associated with COSY, HSQC and HMBC experiments revealed the presence of an acetoxy group, as suggested by the signal at δ_H 2.1 (3H,s, H-15), and at δ_C 20.5 (C-15) and 170.5 (C-14). Then δ_H and δ_C signals suggested the presence of a polyacetylene chain (C₂C₃C₄C₅) while the four quaternary carbons present a δ_C from 64.8 to 80.5; the interpretation is further confirmed using HMBC. Complete correlations are presented in Table 3. The structural analyses enabled the identification of (5S,7Z)-7-(hexa-2,4-diyn-1-ylidene)-1,6-dioxaspiro[4.4]nona-2,8-dien-4-yl acetate, from the polyacetylene family, already found in Santolina species (Fig. 3) [22–25]: commonly named spiroketalenol, it is referred to as SKE in the present article. No mention of any kind of the assessed antimicrobial activity was reported for this substance in the literature.

2.4 SKE antimicrobial activity

Pure SKE appeared to be unstable once isolated, particularly in aqueous and acidic conditions. Since antimicrobial assays are performed in water, its antimicrobial activity could not be directly evaluated as for crude plant extracts or fractions. Thus there was a need to develop a methodology to establish a relationship between the biological activity of crude extracts and their SKE content.

Such a quantitation of SKE was performed by HPLC-UV at 310 nm (SKE absorption maximum): a calibration curve was constructed using (E)-cinnamic acid as a standard and the active substance concentration was expressed as mg equivalent of (E)-cinnamic acid per gram of the crude extract.

Once the SKE concentrations in various extracts were determined, the corresponding antimicrobial activities were assessed. As shown in Table 4, a relationship between the SKE concentration and the antimicrobial activity, especially against A. niger and C. albicans can be evidenced: no activity against these two strains was detected when no SKE was quantified in the extract. On the contrary, as the SKE concentration increases, antimicrobial activity is observed and appears to be maximal when the SKE concentration is higher than 9.1 mg equiv/g of the extract.

However, the activity against S. aureus seems to be independent of the SKE presence: in fact, almost all the extracts containing SKE or not were active against this strain.

Finally, it was highly difficult to conclude about the activity against P. aeruginosa as only one extract containing SKE at 17.6 mg equiv/g was active against this last strain.

2.5 Stability assessment

The instability of isolated SKE has already been stated; it appeared crucial to test its stability in the S. chamaecyparissus extract incorporated into a cosmetic formulation before any further ingredient development.

Creams formulated with 0.5 (data not shown) and 2% of the S. chamaecyparissus extract together with various cosmetic supports were then subjected to stability tests: these cosmetic products were kept either at room temperature or at 42 °C for 13 weeks. The evolution of the SKE concentration in these finished products was followed by HPTLC, using an appropriate mobile phase to properly separate SKE from the other ingredients (retention factor Rf = 0.42).

A stability test was performed over 6 weeks at room temperature; then one week later (day 49), the formulated creams were either stored at room temperature or at 42 °C. As shown in Fig. 4, the SKE concentration remained stable at room temperature with the three cosmetic supports tested. However, SKE degradation was observed in a cream placed at 42 °C after 7 weeks at room temperature (only 40% of SKE was measured in the cream after 20 weeks): one can suggest that SKE, stable at room temperature, is heat sensitive, but further investigations should confirm this observation and appropriate storage conditions could be defined.

Furthermore, it would be necessary to push these stability tests forward once the definite formulation has been determined: tests at room temperature and at 42 °C should be supplemented with tests in an ageing chamber where the effect of temperature variations mimicking the ones a finished product is subjected to, from formulation to consumer's use, could be monitored.

2.6 Natural ingredient design

Once the stability of the active molecule has been established in cosmetic formulae, the corresponding natural cosmetic ingredient can be developed.

The necessary requirements for such a development are quite numerous, as evidenced in Fig. 5: natural substances that show antimicrobial activity are usually not directly incorporable into a formulation as they may display undesirable qualities.

First, there is a necessity to respect the integrity of the active molecule during the extraction process, and hence to optimize this extraction process via the development of an appropriate non-toxic extraction solvent system or via the use of modern extraction techniques such as microwave assisted extraction, ultrasonic assisted extraction or supercritical fluid extraction. Secondly, some substances may display offensive odours and/or colours that would be unacceptable in skin care products [7] and therefore need to be subjected to further processing, including notably a discolouration procedure. Finally, the addition of a cosmetic support, either liquid or solid, may be necessary to facilitate its incorporation into a cosmetic formulation.

As the antimicrobial activities appeared to be correlated with extract's SKE concentration, optimization of the extraction process was in the first place performed via the modification of the solvent system used. Several extraction replicas were performed using each solvent system tested to testify individual repeatability: the extraction repeatability was good as the mean standard deviation per solvent system never exceeded 10%.

A comparison of EtOH/H₂O gradients was initially performed, testing gradients ranging from 100% EtOH to 100% H₂O. As shown in Table 5, the highest SKE concentration (19.9 mg equiv/g) was obtained using EtOH/H₂O 80:20 (v/v). The more the water ratio in the extracting solvent system increased, the weaker the SKE concentration was. Similarly, using absolute EtOH rather than EtOH 96% leads to a decrease in the SKE concentration. An optimal ratio was thus established containing between 60 and 96% EtOH (v/v).

Other solvents were also compared to the extract of the plant. Methanol appears to be the extraction solvent giving the highest concentration of metabolite (44.3 mg equiv/g). Dichloromethane, acetone and ethyl acetate were also good candidates for the extraction of S. chamaecyparissus (40.8, 31.0 and 20.3 mg equiv/g, respectively). All these four solvents gave higher active concentration than ethanol/water 80:20. Methanol was subsequently selected for the following optimization of the extraction process to produce a cosmetic ingredient based on the S. chamaecyparissus extract [26]. However, caution should be taken as methanol as well as dichloromethane is CMR (carinogenic/mutagenic/reprotoxic) solvent, and its use in the extraction of the cosmetic ingredient is highly contraindicated and shall be progressively banned: any residual trace of extraction solvent in the finished product is scrutinized [27–29].

The antimicrobial activity of the methanolic extract was tested (see Table 6): a significant activity was observed against S. aureus, A. niger and C. albicans, whereas only a moderate activity was noted against P. aeruginosa and no activity at all was observed against Escherichia coli at 0.4%.

As already stated, the presence of pigments (e.g., chlorophylls, anthocyanins, etc.) in solvent crude extracts dedicated to the development of cosmetic ingredients is unacceptable. Discolouration of such extracts can be performed using either activated carbon or molecular distillation. Both techniques were tested in the second phase of the optimization of a natural cosmetic ingredient based on the S. chamaecyparissus extract obtained using EtOH/H₂O 80/20 (v/v).

Molecular distillation of cosmetic raw materials enables removing undesirable odours, colour and some of the impurities that are present in the starting materials. This distillation is performed in an extreme vacuum environment at high temperature for a very short contact time. This technique was applied on S. chamaecyparissus crude extracts using liquid medium-chained triglyceridic support: the resulting liquid product presents a translucent yellow to orange colour, quite interesting for cosmetic formulation. Unfortunately, once tested for anti-microbial activity, it appeared that the resulting extract do not display any interesting microbicidal activity. Furthermore, to perform such a distillation process, it is necessary to use an almost non-volatile solvent, traces of which, unacceptable in cosmetics, are quite difficult to remove thereafter.

Discolouration by adsorption of the undesirable molecule on activated carbon is quite easy as the powdered activated carbon can easily be added to the liquid extract, and then removed by settling and filtration. Discolouration of S. chamaecyparissus crude extracts was performed with 10% of active carbon. The active compound content was then dosed in discoloured extracts and in almost all cases, enrichment in SKE was observed as presented in Table 7.

The next step of this optimization procedure consisted in the addition of an adequate cosmetic support. Maltodextrine, a solid cosmetic support, was tested but it appears that after solvent evaporation, the S. chamaecyparissus extracts on maltodextrin were thick and doughy, inducing the necessity for a tedious solubilization step for subsequent incorporation into cosmetic formulations. Furthermore, this support only leads to the “dilution” of extract's antimicrobial activity, as maltodextrin itself do not display any interesting microbicidal potential.

Polyols and monoglyceride esters, liquid cosmetic supports, were then judiciously selected for the development of such a cosmetic preservative based on the S. chamaecyparissus extract, as they present potentially interesting antimicrobial activity, which could improve the antimicrobial activity of the final ingredient [7,30–32]. The solubility of the S. chamaecyparissus extract was hence evaluated in the following cosmetic supports: ethylhexylglycerin (S1), a mixture of ethylhexylglycerin and caprylyl glycol (S2), caprylyl glycol (S3), glyceryl caprate (S4), capryloyl glycine, glyceryl caprylate, glycerin, propylene glycol, butylene glycol and propanediol. The S. chamaecyparissus extract presented the highest solubilities in S1, S2, S3 and S4, and were only moderately soluble in the other raw materials tested. The first three were selected for further investigations and the resulting ingredients corresponding to the mixture of the crude extract of S. chamaecyparissus with either S1, S2 or S3 were subsequently named ingredient 1 to ingredient 3.

The antimicrobial activities of these three new ingredients were tested and compared to one of the crude extracts alone: results are presented in Table 8. Optimal activity (>90% of microbial inhibition) was observed for all ingredients against A. niger. The activity of ingredient 2 appears to be very good (>80% of inhibition) against E. coli and C. albicans and good (>60% of inhibition) against S. aureus, whereas it remained optimal with ingredients 1 and 3. Finally, ingredient 1 only displays a medium to poor activity against P. aeruginosa, whereas ingredient 2 presents a good one and ingredient 3, an optimal one against the latter strain.

So, all ingredients evaluated at this step of the study were highly active against the micro-organisms tested and were effectively more active than the crude extract alone: they were then selected for the next steps of the preservative development.

2.7 Cosmetic evaluation

The microbial safety of a cosmetic product constitutes a pre-requisite for its production and commercialization: the results of challenge tests are required to prove its microbiological stability [33].

To comply with this requirement, the three optimized ingredients were incorporated into various cosmetic formulations (O/W emulsions), one of which is described in Table 9. The features of the optimized cream were in accordance with those previously described by Kerdudo et al. 2015 [34].

Cosmetics containing antimicrobial extracts at concentrations ranging from 0.5 to 2.0% were exposed, as recommended by the FDA, to challenge tests to ensure the safety (e.g., the ability of the preservative system to protect the product from microbial growth) of both the ingredients alone and the finished cosmetic products prior to marketing and consumer use. As expected, the control cream, e.g., without preservative of any sort, was not in accordance with the European Pharmacopoeia standards [35]. The three assays containing 2% of the respective ingredient produced interesting results. The cream containing ingredient 3 appears to be resistant enough to conform to criteria A of the European Pharmacopoeia against all microorganisms [26,35]. Furthermore, the cream containing ingredient 2 appears to be in compliance with criteria A of the European Pharmacopoeia against P. aeruginosa, S. aureus, E. coli and C. albicans and with the criteria B of the European Pharmacopoeia against Aspergillus brasiliensis. Finally, the finished product containing ingredient 1 was also of interest as it was active against all microorganisms (criteria A of the European Pharmacopoeia) except against A. brasiliensis.

These assays were stopped after 14 days of challenge tests.

3.1 Materials

Water (Chromasolv^® for HPLC, purity > 99.8%), acetonitrile (Chromasolv^® for HPLC, purity > 99.9%), 2-propanol (Chromasolv^® for HPLC, purity > 99.9%), formic acid (Puriss, purity > 98%), (E)-cinnamic acid, ethanol, methanol, acetone and dichloromethane were purchased from Sigma–Aldrich (Saint-Quentin Fallavier, France). Deuterated chloroform (CDCl₃) for NMR experiments was purchased from Euriso-top (Saint-Aubin, France). Glycerin (99.5%), butylene glycol and propylene glycol were purchased from Interchimie (Bobigny, France). Sensiva SC50 (ethylhexylglycerin stabilized with vitamin E) and Sensiva SC10 (caprylyl glycol and ethylhexylglycerin stabilized with vitamin E) were obtained from Schülke (Gentilly France). Microcare Emollient CLG (caprylyl glycol) was distributed by Coptis (Croissy-sur-Seine, France). Dermosoft GMC (glyceryl caprate) and Dermosoft GMCY (glyceryl caprylate) were purchased from Dr. Straetmans (Hamburg, Germany). Zemea (propanediol) was obtained from Dupont Solvents.

3.2 Plant material

All the plants initially selected were collected in 2010 in the Provence-Alpes-Côte d’Azur region, in the Alpes-Maritimes, Var and Vaucluse departments.

Aerial parts of S. chamaecyparissus L. were collected from Cabris, Isola, Mouans-Sartoux in the Alpes-Maritimes department, and from Mont Caume and Mont Coudon in the Var department.

The plant material was botanically authenticated by J.-L. Polidori and F. Boillot, both botanists, and a specimen of S. chamaecyparissus L. collected from Mont Caume, Var, in April 2013 has been deposited at the Regional Botanical Conservatory of Nice (voucher n° C 3746).

3.3 Plant extraction

Aerial parts of S. chamaecyparissus L. were harvested, dried, ground into powder and extracted under mechanical stirring using organic solvents. Two consecutive 2 h-extractions were performed at 25 °C and the resulting extracts were filtered, gathered together and vacuum-concentrated.

When cosmetic support was added, solvent extraction was performed as previously: the raw cosmetic material was added before solvent extraction in proportion 1:1 (w/w).

3.4 Sample preparation

Crude extracts and fractions were diluted at 20 mg/mL in methanol (MeOH) and filtrated over 0.45 μm PTFE syringe filter before HPLC analysis.

For HPTLC analyses, crude extracts and creams were dissolved in MeOH at 2 and 20 mg/mL respectively, ultrasonicated for 15 min and centrifuged: 2 μL of each supernatant were systematically applied on the HPTLC plate for analysis.

3.5 HPLC

Crude extracts and fractions were analysed using an HPLC Agilent 1200 system (Courtaboeuf, France) equipped with a DAD and an ELSD operating under the following conditions: injection volume: 20 μL, flow rate: 1.0 mL/min, UV-detection performed at 310 nm. The mobile phase consisted in a gradient of water (A), acetonitrile (B) and 2-propanol (C), all acidified with 0.1% acid formic: 0–5 min, 5% B; 5–35 min, 5–100% B; 35–45 min, 100% B; 45–50 min, 100–40% B and 0–60% C, 50–60 min, 40% B and 60% C, 60–65 min, 40–5% B and 0–95% A, 65–70 min, 5% B and 95% A.

Quantification of the active molecule was performed by HPLC-UV at 310 nm (wavelength corresponding to the absorption maximum of the active molecule) under the same analytical conditions and with the same solvent gradient (but stopped after 43 min) as the one described previously. A calibration curve was constructed using (E)-cinnamic acid dissolved in methanol at concentrations ranging from 0.0078 to 1 mg/mL as a standard. SKE concentrations, expressed as mg equivalent of (E)-cinnamic acid per gram of the crude extract, were determined thanks to the linear regression equation generated by the calibration curve after integration and area measurement of the corresponding peak in various extracts.

Semi-preparative HPLC of the cyclohexane/ethyl acetate fraction resulting from the column fractionation of the S. chamaecyparissus extract (see Compound isolation) was performed on a Luna C18 column (Phenomenex, 250 × 10 mm; 5 μm). Elution was performed using a gradient of non-acidified water (A) and acetonitrile (B): 0–5 min, 60% B; 5–35 min, 75% B and 35–45 min, 100% B.

3.6 HPTLC

High performance thin-layer chromatography (HPTLC) was performed on 10 cm × 20 cm HPTLC silica (0.20 mm) gel 60F₂₅₄ pre-coated glass plates (Merck). Standards and samples were applied as 8 mm large bands (15 tracks per plates, 50 nL/s delivery speed), 20 mm from the both plate edges by means of an automated ATS4 sampler (Camag). Two microlitres of each standard and samples were applied on silica gel plate. The mobile phase consisted of a solution of toluene/ethyl acetate/formic acid 98:2:1 (10 ml). The separation on the silica gel plate was performed in an automatic ADC2 developing chamber (Camag) with the tank previously saturated for 20 min with 25 ml of the developing solvent (humidity previously controlled for 10 min: 33–38%). Developing distance was 70 mm from the lower edge of the plate. Documentation of chromatographic plates (visualization and image capture) was performed at 254 nm, 366 nm and under white light by the TLC Visualizer documentation (Camag) operated with winCATS Version 1.4.6 software. Plates were scanned under the following conditions using the TLC scanner 4 controlled by the WinCATS software: scanning mode, reflectance mode at 315 nm, D2 (deuterium) and W (tungsten) lamps, slit dimension 8.00 mm × 0.40 mm, scanning speed 20 mm/s, and data resolution 100 μm/step.

SKE presents a good UV absorbance, reaching a maximum at λ at 315 nm, without any co-elution.

The quantification of SKE content in formulated creams was performed via calibration curves constructed with five relative concentrations of ingredients 1, 2 and 3 diluted in MeOH and applied in triplicate on the HPTLC plate (stock solution prepared at 1 mg/ml; application of respectively, 1, 2, 6, 8 and 12 μL of the corresponding solutions on the HPTLC plate).

The R² coefficients were all acceptable above 0.999.

3.7 Compound isolation

Open column chromatographic fractionation of the crude extract was performed using silica gel (normal phase) and yielded five fractions: cyclohexane fraction (F1), cyclohexane/ethyl acetate 1:1 v/v fraction (F2), ethyl acetate fraction (F3), ethyl acetate/methanol 1:1 v/v fraction (F4) and methanol fraction (F5).

3.8 ¹H and ¹³C NMR analysis

NMR analyses were carried out on Bruker Avance DRX500 (500 MHz) (Brucker, Wissembourg, France). Mono- (¹H and ¹³C) and bi-dimensional (COSY, HSQC, HMBC, NOESY) NMR spectra were recorded in CDCl₃ at 25 °C (Euriso-top, Saint-Aubin, France); chemical shifts (δ) are given in parts per million (ppm), and coupling constants (J) are given in hertz (Hz).

3.9 UPLC-ESI-HRMS

UPLC-ESI-HRMS analyses were performed on a QStar Elite QTOF mass spectrometer (Applied Biosystems SCIEX) operating under atmospheric pressure ionization, in positive-ion mode. Samples diluted in methanol at a concentration of 1 mg/mL, were eluted on an Acquity UPLC BEH C18 column (Waters, 100 × 2.1 mm, 1.7 μm, 35 °C); the flow rate was set at 0.6 mL/min and the injection volume was 1 μL.

The mobile phase consisted of water (A) and acetonitrile (B), both acidified with 0.1% formic acid: 0–5 min, 0% B; 5–10 min, 0–100% B; 10–15 min, 100% B.

Compound 1: Powder, UV λ_max (MeOH) 310 nm, C₁₆H₁₂O₄, MW = 256.2534, HRMS m/z 257.0816 (calcd for C₁₆H₁₃O₄, 257.0816).

¹H NMR (500 MHz, CDCl₃) δ_H 6.8 (d, J = 5.7 Hz, 1H), 6.6 (d, J = 3.0 Hz, 1H), 6.3 (dd, J = 5.7, 1.6 Hz, 1H), 5.7 (d, J = 2.5 Hz, 1H), 5.2, (dd, J = 2.5, 3.0 Hz, 1H), 5.1 (br s, 1H), 2.1 (s, 3H) and 2.0 (s, 3H) ppm.

¹³C NMR (125 MHz, CDCl₃) δ_C 170.5, 168.9, 148.6, 133.8, 126.1, 117.3, 100.8, 83.1, 80.5, 77.7, 75.2, 70.3, 64.8, 20.5 and 4.8 ppm.

3.10 Antimicrobial activity

The crude extracts (tested at 2% and 0.2%), the subsequent fractions (tested at 0.4% and 0.04%) and the ingredients (tested at 0.4%) were evaluated for their antimicrobial activity against the fungus A. niger ATCC 16404, the yeast C. albicans ATCC 10231, the gram(−) bacteria P. aeruginosa ATCC 9027 and S. aureus ATCC and the gram(+) bacteria E. coli ATCC 8739 (only the ingredients were tested over E. coli). Bacterial and yeast trains were respectively grown in liquid nutrient broth (Difco, Surrey, England) at 37 °C, for 24 and 48 h before antimicrobial testing.

The assays were performed in 96-well plates as follows: extract samples were diluted in sterile water. Each concentration was assessed in triplicate with one supplementary control well containing no microbial strain. Negative control was constituted of the solvent mixture used to prepare the sample solutions, and positive control consisted of methyl-paraben, a synthetic preservative, at 0.4%. The 96-well plates were incubated at 37 °C, and the optical density (OD) reading was obtained at 620 nm, 24, 48 and 72 h after the beginning of the incubation to evaluate inhibition of microorganism growth. The results were expressed as a percentage of inhibition according to the formula:

3.11 Cosmetic formulation

All the formulations were prepared on a 100 g final mass-basis. Stirring was performed using a propeller mixer (Turbotest laboratory mixer equipped with an axial flow propeller, Rayneri, VMI). The cream formulations were prepared as described by Kerdudo et al. 2015 [34], using the ingredients listed in Table 9.

3.12 Challenge test

Antimicrobial preservative efficiency evaluation was performed by an external laboratory (Ideatest, Plouzané, France) following the procedure described in the European Pharmacopoeia [35].

Five potential pathogenic germs were chosen for this study (a cosmetic contamination by these microorganisms could notably cause skin infection or respiratory diseases): Pseudomonas aeruginosa ATCC 9027, S. aureus ATCC 6538, E. coli ATCC 8739, C. albicans ATCC 10231 and A. brasiliensis ATCC 16404 [36]. Tryptic soy agar and Sabouraud dextrose were respectively used as culture media for bacteria and for yeast/fungus. Cosmetic formulations were sampled in single-use sterile flasks and every flask was inoculated with strain suspension (at 22.5 ± 2.5 °C in the dark). The inoculation concentration was fixed at 10⁶ CFU/mL, 10⁵ CFU/mL and 10⁴ CFU/mL respectively, for bacteria, yeast, and fungus. The validation of the neutralization of the preservative was realized on the five strains, in the 1/10th and 1/100th dilution, with LT100 broth. The microbial densities were counted by inclusion (results in CFU/g or mL), and compared with the logarithmic reduction criteria of the European Pharmacopoeia presented in Table 10.