Effect of phytoderivatives on the growth of homologous benecial vaginal lactobacilli (BVL) strains and their compatibility for the design of phytobiotics for the vaginal tract health

Background: Lactobacilli are the predominant bacteria in the vaginal tract, and while they are present, can prevent infectious situations, being used as probiotics for the health of the tract. Phytoderivatives are used historically to prevent or maintain the health of the tract, and is possible to combine both bioactives for novel vaginal formulas. The aim of this work was to study the effect of phytoderivatives (individual or combined extracts) on the growth of homologous vaginal lactic acid bacteria and their compatibility, to go further in the design of a combined pharmabiotic formula for the prevention or treatment of urogenital tract infections. Methods: the effect of phytocompounds approved by Pharcopoeias was evaluated on the growth of benecial lactic acid bacteria previously isolated from the vaginal tract. Statistical methods, as CART, were applied to determine the stimulatory or inhibitory effect of the vegetal sources on bacterial growth by applying different methodology. Compatibility between phyto and lactobacilli was determined, and the phenolic compounds contents quantied, to dene the optimal conditions for the formula design. Results: The BVL strains showing highest growth values with most of the phytoderivatives under study were L. gasseri CRL 1361, L. ga. CRL 1509 and L. ga. CRL 1263, and L. jensenii CRL 1333 and L. jen. CRL 1317, indicating the behavior is strain-dependent, and vegetable-compound-affected. The most adequate phytoderivatives were: Carica, Centella, Plantago, Uva ursi*, Zarzaparrila*, Cola de caballo, Echinaceae, Ortiga and Palo Azul. Conclusions: The BVL strains showing optimal resistance to the phytocompounds and compatibility with them were L. gasseri CRL 1320, 1307, 1509, L. salivarius CRL 1296 and L. rhamnosus CRL 1332 combined with Uva ursi*, Zarzaparrilla* and Echinaceae that will be used for the design of optimal phytobiotic formulas.


Introduction
The new concept of microbiome on the human and animal mucosa demonstrated during the last decade and the high variety of related functions described, supports the application of probiotic formulations designed with host and tract homologous microorganisms. Lactobacilli are a diverse group of more than 200 different species that occupy diverse nutrient-rich niches associated with humans, animals, plants and food [1]. They are used widely in biotechnology and food preservation, and are being explored as therapeutics [2]. Lactobacilli and other Generally Regarded as Safe (GRAS) and Quali ed Presumption of Safety (QPS) genera are widely used as probiotics, which can be administered to help the host in the restoration of the indigenous microbiota, stimulate the immune system, or either protect against the income of pathogenic microorganisms. On the other side, there is an ever-growing interest on natural ingredients, including vegetal sources, derivatives and formula, both by consumers and producers in the food and pharma industries. In fact, people are looking for those products in the market which are free from arti cial and synthetic additives and can promote their health, many of them with a very long application supported by concepts or custom-derived uses. In the ethnopharmacological area, the knowledge related to the traditional uses of medicinal plants is totally in the custody of elder community members and local herbalists [3]. A long list of them is included in the Pharmacopeia resources, indicating they can be used safely for different situations [4,5]. Historically, they were used with different proposals, effects reviewed by different authors [6]. Also, the limitation of the therapeutic options for emerging multidrug resistance microorganisms, and the urgent need of new (or old-uses) natural and safe combinations is emerging [7]. Then, the concept of phytobiotics has been conceived, to combine safe phytoderivatives and probiotic bacteria for the design of new formulations, resulting in the combination/synergy of the two bene cial effects. These bioactive ingredients should be formulated in such a way that they are compatible, protects them against harsh process and environmental conditions and could be delivered safely to the target organs and cells.
On the probiotic area, our research group has a long history of experiments performed "in vitro" or in animal models, complemented by studies on the design of different type of vaginal probiotic formula [8,9,10]. Then, the aim of this work was to study the effect of phytoextracts (macerated or powders) on the growth of homologous vaginal probiotic lactobacilli (by applying different methods) and their compatibility to select those than can be combined for the design of phytobiotic formulas. The growth kinetics and compatibility assays were designed and evaluated through the application of speci c statistical models, to select those with an optimal behavior. Also, the phenolic content was evaluated, and the best combination de ned to go further in the design of different type of formulas for the reproductive tract health.
The method applied to identify the BVL selected by their bene cial characteristics by the application of molecular-related techniques is described in previous works [13]. The strains, as identi ed by this method, are included in Table 1. phytocompounds under study, settled for diffusion, and later incubated in microaerophyllic conditions at 37°C. The inhibitory halos were measured after 48 h (in millimeters). Controls included the solvent used for each one of the phytocompounds. The assays were performed for triplicate [18].

Quanti cation of polyphenolic content in natural compounds
The photometric assay was applied to quantify the phenolic concentration in the phytocompounds under study [19] and the absorbance was determined at 750 nm using a UV-VIS spectrophotometer (Spectronic 20, Baush and Lomb, Roscherter, NY). The calibration curve was obtained with Gallic Acid (GA) (Sigma Aldrich, Argentina) as standard (concentration range 2.5-1000μg) with ethanol as control (Sigma-Aldrich; Merck KGaA, Darmstadt, Germany). The regression equation was: y = 0.0395x + 0.1633, and the coe cient of correlation R2 = 0.9907. The results were expressed in mg GAE·g−1 extract. The experiments were performed by triplicate.
Statistical evaluation.
The bacterial growth parameters were estimated applying the 4-parameter modi ed Gompertz-model: where OD is the optical density at time t (time of growth in hours), N0 the OD at t = 0, A the difference between the nal and the initial ODs, µ the maximum speci c growth rate (h -1 ), λ the lag phase time in hours and e the base of the neperian logarithm.
For the estimation of the parameters constrained nonlinear regression was performed. This method uses a sequential quadratic programming algorithm. For the parameter standard errors and con dence intervals the method of Bootstrapping was applied, using repeated samples from the original data set. For each growth curve 100 bootstrap samples were taken.
To identify subgroups of combinations of strains with phytocompounds with optimal growth a Classi cation and Regression Tree (CART) analysis was performed. The CART analysis is capable of detecting complex interactions of variables that might be missed by standard statistical approaches. The estimated growth parameter A was included as independent variable and the parameters strain and phytocompound were included as predictors in the analysis. Only experiments with an estimated parameter A (O.D.) of at least 0. 8 and an estimated lag phase shorter than 8 h were selected for the CART analysis.
For the analyses and graphical presentations the statistical programs SPSS 25, S-Plus 8.1 and STATISTICA 12 were used. The CART analysis was performed with Salford Predictive Modeler 8.2. [20].

Growth of BVL in presence of different extracts
The growth of many strains was affected by the phytocompound under assay, because they were either stimulated of inhibited. The complete list of phytoderivatives assayed is shown in Table 2. Some examples of the effect of phytocompounds on the BVL growth are indicated in Figure 1 and 2 that means the effect of each one of the vegetal compounds on the different BVL is strain-speci c, and no related with the metabolic groups in which they were originally classi ed. The claimed effects of all the phytoderivatives under assay are shown in Table 2  In most of the BVL evaluated, Betula (Bet*) and Carica (Car) extracts did not show an inhibitory effect on their growth.
Related to the optimal concentration of vegetal extracts to be used, Figure 2 shows that different concentration of Allium sativa (Fig 2B, 2C, 2D), Hamamelis (Fig 2E, 2F, 2G) and Uva ursi (Fig 2H, 2I, 2J) exert an inhibitory effect on the BVLAB growth related with the increased concentration of the phytoextracts added, being higher when adding 3 mg/ml than 0.75 mg/ml. Further studies are required to determine if the strains would be able to grow after an induction or adaptation phase, previously to their use for the technological application.
Liquid Phytocompounds and BVL: "Classi cation and regression tree analysis" With the experimental data obtained from the combination of all the phytocompounds (Table 2, 3) and lactic acid bacteria strains (Table 1), the growth parameters of each one of the strains associated with each phytoderivative was calculated. The cut point established were O.D.: 0. 8, lag phase shorter than 8 h, and incubation time 24 hours, with these conditions 184 trials were selected for the analysis. The statistical evaluation performed through the CART (Classi cation and regression tree analysis) software, showed the following results: The selected trials show a mean estimated O.D. of 1.007 growth (Fig. 3). The algorithm then divides the "root node" containing all selected trials into two nodes using the variable "strain", including 128 assays with an average growth of 0.938 O.D. (BVL in the box "Terminal Node 1"), while on the right 56 trials are included with an average growth of 1.163 O.D. ("Node 2"). Node 2 is then divided using the predictor "phytocompound". The resulting "Terminal node 2" includes 29 trials with an average growth of 1.026 O.D., while "Node 3" includes 27 trials showing an average growth of 1.309 O.D.
"Node 3" is then split into two terminal nodes using the predictor "strain": "Terminal Node 3" includes 18 assays with an average growth of 1.182 O.D. and "Terminal node 4" contains 9 trials with an average growth of 1.564 O.D.
Hence the CART analysis identi es four subgroups of trials (terminal nodes) with an average growth ranging from 0.938 O.D. to 1.564 O.D.

BVL and phytocompounds compatibility
The behavior of 24 BVL classi ed into different metabolic groups were assayed with 25 different natural alcoholic extracts. The degree of interaction between BVL and the vegetal extracts is summarized in Table 5, showing the compatible or non-compatible combinations: stimulatory (green) or inhibitory effect (red) of the phytocompounds on the growth of each strain. The BVL growing with the phytoderivatives is indicated with different colors: green boxes: "High Stimulation", (higher of 70% growth than in control MRS broth); light-green color "Medium Stimulation" (between 70% and 35% growth higher than control) and "Low Stimulation" (35% higher growth than the control). At the same time, the red boxes indicated "High Inhibition" (70% decrease of the maximal growth compared with control), while orange boxes represent "Medium Inhibition" (those strains inhibited between 70% and 35%); pink boxes "Low Inhibition" (BVL inhibited lower than 35%); The yellow boxes, "No Effect", did not modify their growth with phytoderivatives when compared with control MRS.
The BVL strains showing highest growth, indicating their compatibility with most of the phytoderivatives under study were L. gasseri CRL 1361, L. gasseri CRL 1509 and L. gasseri CRL 1263; L. jensenii CRL 1333, L. jensenii CRL 1317. The strains strongly inhibited by most of the phytocompounds were L. fermentum CRL 1287 and L. salivarius CRL 1296.
L. gasseri CRL 1261 showed a better results, because the growth was favored by most of the phytocompounds evaluated (supplementary material).
The pure extract of Yerba meona (Yme), Palo azul (Paz) and Echinaceae (Ech), as well as the mixtures corresponding to Carica (Car), Calendula (Cal) and Betula (Bet*) did not signi cantly affect the growth of lactic acid bacteria.

Phenolic content in the phytocompounds
The phenolic compounds were quanti ed trying to de ne it they are related with their bene cial effect or compatibility with BVL. All the phytoextracts showed to contain phenolic compounds in different quantities, showed in Fig 4, where eleven extracts with content higher than 500 µg eq Galic acid/ml are indicated. Hamamelis showed the higher content of phenolics. Manzanilla and Zarzaparrilla did not show signi cant differences among them, as well as Echinaceae, Chelidonia and Zarzaparrilla*. Echinaceae and Yerba meona did not show signi cant differences between them. Betula and Perilla did not have a signi cant difference, however, Plantago presented a signi cant difference with Betula, but not with Perilla.

Discussion
The use of phytoderivatives is generally transmitted as a consequence of their ancient applications, most of them lately rediscovered, or under stronger evaluation. Scientists have published some experimental data related with isolated vegetal derivatives, for example on speci c infections of urogenital pathogens, as Candida vaginal infections [21], some viruses [22], and parasites as Trichomonas [23] or bacteria [24]. In addition, they are widely recommended and included in the design of vegetal-derived formulas for postmenopausal women or for reproductive disorders [25].
It is widely known that Uva ursi and Ortiga have shown antibacterial properties against different microorganisms [26] showing results that were expected. But Hamamelis and Aesculus that are natural derivatives used by their vascular properties, mainly to improve the venous circulation, have shown some results that were not expected. Many of them were used for the inhibition of pathogens related with the tract, as shown in Table 2 and 3. On the other side, the use of Lactic acid bacteria as probiotics at different mucosal sites is widely recommended [27,28]. Then, is possible to combine those phytoderivatives currently or historically applied to the urogenital tract with bene cial lactic acid bacteria with probiotic properties in a way to design different formula directed to exert or produce a synergistic effect on the host. In this way, is of main importance to determine if the phytoderivatives affect the BVL growth and if they are compatible to be used in a unique bene cial formula. The results of this work shows the interactions of a long list of phytoderivatives approved in the Pharmacopeias to be used either for oral or local administration with probiotics BVL characterized in our laboratory. There is no many data on the compatibility of phytoderivatives and vaginal lactobacilli strains published. The paper of Murina [29] reports the administration of a vaginal gel containing Thymus vulgaris and Eugenia caryophillus in conjunction with two Lactobacillus strains speci cally formulated in slow-release capsules in treating bacterial vaginosis, or recurrent vulvovaginal candidiasis disease, recommended their use in the acute treatment. But there is no previous studies reporting the compatibility between the two bioactives constituents of the commercial formula. Then, is of main importance, and as the rst step for the formula design, to evaluate the effect of the phytoderivatives, either as liquids or extracts on the BVL growth. The results obtained in this work indicate that the effect is phytoderivative and strain dependent, because there is no general rules. Each strain and each extract must be asssayed in a way to de ne their optimal or adequate combinations. The classi cation of lactobacilli into metabolic groups supported by their sugar-catabolic ways, either as homofermentative or heterofermentative, does not predict or indicate any type of compatibility, resistance or behavior when combined with the wide variety of vegetal derivatives under assay. The evaluation of their combinations was performed in this work through the growth kinetics, and by MIC assays, allowing to de ne which of them could be combined adequately.
The very high number of experimental protocols applied to determine the compatibility of the two bioactives under study did not allowed to de ne and decide promptly the optimal or most adequate combination. Then, the use of statistical models, and the CART analysis is a widely applied tool to de ne the most adequate combinations, when a huge number of analytical or experimental data are available [30]. The application of statistical methods, and CART analysis, have contribute to go further in the decision and selection of the most adequate BVL strains and phytoderivatives in a way to combine them in a phytobiotic formula. The results indicate that the most adequate phytoderivatives to be combined with BVL are Carica, Centella, Plantago, Uva ursi*, Zarzaparrilla*, Cola de Caballo, Echinaceae, Ortiga and Palo azul.
The concentration of the phenolic compounds on the extract is not directly correlated with the effect on the BVL growth or in the compatibility between them. Further studies must be performed to determine if some speci c type of components is responsible of the stimulatory or inhibitory effect.

Conclusion
The BVL strains showing compatibility with the phytoextracts were L. gasseri CRL 1320, 1307, 1509, L. salivarius CRL 1296 and L. rhamnosus CRL 1332 with Uva ursi*, Zarzaparrilla* and Echinaceae, which were selected supported by their ethnopharmacological effects (antibacterial, anti-in ammatory and healing) for the further design of phytobiotic formula Declarations Ethics statements: not applicable.
Data statements: the data included in the manuscripts are original, and had not been published previously. They were not submitted to other scienti c editorial. All the authors agree on the data and results included in the manuscript.
Availability of data and materials: please contact author for data request.
Funding statements: this work was supported by grants received from CONICET PIP 545 and ANPCYT (PICT 1187 and 4324) referred to design of the study, collection of data and writing the manuscript. Also, was carried out under the Cooperation between Mincyt (Argentina) and BMBF (Germany), project AB02 that supports the analysis and interpretation of data.
Competing interest: The authors declare that they have no competing interests.    (4,5,14,15,16,17). Uses: way of administration. Pharmacopoeia: year of publication and annexes in which they were accepted (15,16,17). All the extracts were obtained to FITOT laboratory, Tucuman-Argentina. Figure 2 Kinetics of BVL growth added with different phytocompounds. Fig A-  Cart tree of BVL and liquid phytocompounds obtained from the Statistical analysis carried out by SSPP, as indicated in materials and methods section. *represents the extracts assayed as mixture.