The first part focused on analysing the microbial communities on skin and milk with TCEO treatment in a defined timespan, forerun by routine bacteriological analysis. The second part of the study was instead focused on determining if topical treatment with TCEO had any impact on milk quality by evaluating milk composition, protein and fat content (lipidomic), hydrodynamic diameters of casein micelles, milk viscosity, and sensory analysis. Related inflammatory parameters (including somatic cells, IL-8, and APP content) were presented in the final part.
Evaluation of TCEO
The TCEO antibacterial activity was tested in vitro before the in vivo experiment, compared to popular antibiotics and other EO. Figure 1 shows the robust antibacterial efficacy of TCEO against E. coli and S. aureus, as evidenced by the substantial zone of inhibition measuring 25.3 mm and 36.8 mm, respectively. These values were found to be significantly higher compared to several other tested components (P = 0.05). Consequently, TCEO was chosen as the optimal candidate for further investigation in the in vitro experiments.
Part 1: Microbiota
Bacteriological Analysis of Milk Samples
Table 1 presents the microbiological content of composite milk during the study and the total colony counts (cfu/mL). On day 28, no bacteria were identified in the milk of 4 quarters, 2 in each group. There were no statistically significant changes related to the TCEO application.
Table 1:
Table 1
Evolution of the bacteria species in milk from Control (CT) and Treated (TR) groups. The number of colonies is determined by total colony counts (cfu/mL). CNS: coagulase-negative Staphylococci.
Quarter | Identification of the quarter | Day 0 | Day 7 | Day 21 | Day 28 |
Control | CT1A | Corynebacterium spp. (> 1 600) | Corynebacterium spp. (> 1 600) | Corynebacterium spp. (> 1 600) | Corynebacterium spp. (330) |
CT1B | Corynebacterium spp. (> 1 600) | Corynebacterium spp. (> 1 600) | Corynebacterium spp. (> 1 600) | Corynebacterium spp. (330) |
CT2A | CNS (> 1 600) | CNS (> 1 600) | CNS (> 1 600) | CNS (330) |
CT2B | Microccocus spp. (> 1 600) | Sterile | Sterile | Sterile |
CT3 | CNS (> 1 600) | CNS (> 1 600) | CNS (> 1 600) | CNS (> 1 600) |
CT4 | CNS (> 1 600) | CNS (> 1 600) | CNS (> 1 600) | CNS (> 1 600) |
CT5 | CNS (> 1 600) | CNS (> 1 600) | Sterile | Sterile |
CT6 | CNS (> 1 600) | CNS (> 1 600) | CNS (> 1 600) | CNS (330) |
Treated | TR1A | CNS (> 1 600) | CNS (> 1 600) | CNS (> 1 600) | CNS (> 1 600) |
TR1B | CNS (1 600) | Sterile | CNS (> 1 600) | Sterile |
TR2A | CNS (830) | CNS (332) | CNS (500) | CNS (332) |
TR2B | CNS (124) | CNS (> 1 600) | CNS (216) | CNS (34) |
TR3 | CNS (> 1 600) | CNS (> 1 600) | CNS (> 1 600) | CNS (> 1 600) |
TR4 | CNS (> 1 600) | CNS (> 1 600) | CNS (> 1 600) | CNS (> 1 600) |
TR5 | CNS (1 600) | CNS (> 1 600) | CNS (> 1 600) | CNS (1 330) |
TR6 | CNS (580) | CNS (332) | CNS (> 1 600) | CNS Sterile |
Microbiota analysis
This part of the study determined the changes in the milk microbiota by meta-transcriptomics. Further in the text, the changes in skin microbiota after external topical treatment with EO are also presented. The definiton. at the phyla level, originates from the Silva 132, which did not integrate some updates about bacterial nomenclature. Therefore, readers will find, e.g., Bacillus as Firmicutes, Pseudomonadota as Proteobacteria, Actinomycetota as Actinobacteria, and Bacteroidota as Bacteroidetes.
Sequencing metrics of milk samples
Sequencing the V3-V4 regions of the bacterial 16S rRNA gene of the 64 milk samples produced 2,871,971 assembled reads (joined R1-R2 paired-end reads). After quality filtering, 749,772 sequences were removed, leaving 2,122,199 sequences for subsequent analyses (77,4% average retention rate, maximum 99,9%, minimum 20,6%). On average, there were 33,159 (± 24,007) sequences per sample in the Control group and 33,302 (± 23,780) in the Treated group. The initial number of identified OTUs was 6,874; after filtering out OTUs with less than 10 counts in at least 2 samples, 1,880 distinct OTUs were left.
Milk core microbiota at phylum and genus level characterization
Three main phyla were detected in the milk microbiota and were shared within all the samples: Actinomycetota (formerly named Actinobacteria, 11.7%), Bacillota (formerly named Firmicutes, 29.25%), and Pseudomonadota (formerly named Proteobacteria, 58.95%). At the genus level, 4 genera were detected in the milk microbiota, including Cutibacterium (10,1%), Halomonas (8.31%), Methylobacterium (49.33%), and Pseudomonas (27.78%). The remaining 4.48% was composed of the "uncultured or unknown" group, which was artificially composed for the aim of the statistical analysis, including all the genera that the database (SILVA v. 132) retrieved as "uncultured" or "uncultured bacterium" or "Other" or "uncultured organism" and similar.
Alpha- and beta-diversities in milk microbiota
The estimated alpha diversity indices for describing the richness, diversity, and evenness of the milk microbiota between the two experimental groups are reported in Supplementary Table S3.
Once corrected to a baseline equal to T0 (milk samples collected before treatment application), alpha indices were compared in 3 different linear models to assess their behavior: i) over time points (within the group, Eq. (1)), ii) between treatments (within time point, Eq. (2)) and iii) between treatments accounting for a time point (Eq. (3)) as described in the materials and methods section. Figure 2 reports the scatterplots of i) the significance (P) of treatment at T7, T21, and T28 (Fig. 2A), ii) the significance (P) of the time point within treatment (Fig. 2B), iii) the significance (P) of time point and treatment from the model (3) (Fig. 2C), for the different alpha diversity indices.
From model (1), the effect of time is significant: three indices (Shannon, Simpson, and InvSimpson) have a P < 0.05 on T21, and 6 indices in the significance range between 0.05–0.10 (Shannon, Obseved_otus, Simpson, Fisher, ACE, and Chao1) plus one index with a P < 0.05 in T28 (InvSimpson). From model (2), significance clusters are similar for Treated and Control groups, with 1 index (InvSimpson) belonging to T21, statistically significant in the 0.05–0.10. From model (3), treatment has a clear effect over time points, with six diversity indices found with P lower than 0.05 (Observed_OTUs, Chao1, ACE, Shannon, InvSimpson, and Fisher) for the Treated group compared with the Control, influenced by time effect.
The relationships between samples were assessed based on Bray-Curtis dissimilarities from the beta diversity analysis. Figure 2D shows the distribution of samples along the first two dimensions from the multidimensional scaling (MDS) of Bray-Curtis dissimilarities: a clustering was observed between time points (P = 0.01211), while no clear clustering by treatment nor by animal or quarter was detected (P = 0.09531, P = 0.06507 and P = 0.16199 from PERMANOVA between treatment, time, animals and quarters respectively, 999 permutations).
Effects of the treatments and time points on milk microbiota at phyla and genera level
Five main phyla were found in the milk microbiota with a relative abundance higher than 1% along different time points (Fig. 2E): Firmicutes, Proteobacteria, Actinobacteria, Bacteoridetes, and Patescibacteria. On the other hand, thirty-four main genera were found in the milk microbiota with a relative abundance higher than 1% along time points (Fig. 2F).
Only 3 phyla (Proteobacteria, Euryarchaota, and Cyanobacteria) were found to significantly differ between treatments along time points, interestingly decreasing all at T28 in the Treated group compared to the Control group (Fig. 2G and 2H).
Twenty-one genera were found to differ significantly between treatments within time points. In particular, the genus Sphingobium and Acinetobacter were increased in the Treated group compared to the Control group at T21 and T21 and T28, respectively (Fig. 2I and 2J). Interestingly, the genus Romboutsia has a bimodal trend: it decreases in the Treated group compared to the Control group at T21 but increases at T28 (Supplementary Table S4). The microbial milk community behaved significantly differently along two or more time points and described their behavior in the Treated group compared with the Control.
Sequencing metrics of skin swab samples
Sequencing the V3-V4 regions of the bacterial 16S rRNA gene of the 32 udder skin swabs samples produced 675938 assembled reads (joined R1-R2 paired-end reads). After quality filtering, 105 300 sequences were removed, leaving 570 638 sequences for subsequent analyses (85.4% average retention rate, maximum 98.1%, minimum 57.6%). On average, the Control group had 17 421 (± 9 085) sequences per sample and 17 832 (± 9 029) in the Treated group. The initial number of OTUs identified was 11,255; after filtering out OTUs with less than 10 counts in at least 2 samples, 3326 distinct OTUs were left.
Alpha- and beta-diversity in skin microbiota
Figure 3 reports the scatterplots of i) the significance (P) of treatment at T7, (Fig. 3A), ii) the significance (P) of the time within treatments (Fig. 3B), iii) the significance (P) of time and treatments from the model (3) (Fig. 3C), for the different alpha diversity indices. Details about indices are reported in Supplementary Table S5.
From model (1), the effect of time appears not significant when comparing T7 against T0, which is used as the baseline. On the contrary, from the model (2), it is possible to see that, considering treatments only, Control groups show significance while the Treated group did not (6 indices in P < 0.05: Observed_OTUs, Chao1, ACE, Shannon, Simpson, and Fisher). From model (3), it was possible to observe that combining time and treatment effect, no significance is evidenced, suggesting that the significance in Fig. 3A is strictly related to treatment. Bray-Curtis dissimilarities method was used to assess the relationship between samples for the beta-diversity analysis (Fig. 3D): results showed that samples tend to cluster significantly by time point, while not per treatment nor when considering time and treatment effect together (p = 0.001 for time, 0.311 for treatment and 0.18 for time and treatment combined effect, from PERMANOVA, 999 permutations).
Effects of the treatments and time points on skin microbiota at phyla and genera level
Phyla. Seven main phyla were found in the skin microbiota with a relative abundance higher than 1% along time points (Fig. 3E): Firmicutes, Proteobacteria, Actinobacteria, Bacteoridetes, Euryarchaota, Verrucomicrobia, and Patescibacteria. However, no taxa had a significant difference between treatments along time points.
Genera. Thirty-one main genera were found in the skin microbiota with a relative abundance higher than 1% along time points (Fig. 3F). Twenty-six genera significantly differed between treatments along time points, of which 16 belonged to T0 and 10 to T7 (Supplementary Table S6) (Fig. 3G and 3H).
Part 2: Milk Quality
Composition, physical, and organoleptic properties
The overall composition of milk was not affected by the treatment with TCEO. No-significant differences (p = 0.05) were detected in the fat or the protein content in milk, as shown in Fig. 4A. Also, the effects on casein micelles and milk viscosity, two important physical parameters for milk processing, were evaluated. The average hydrodynamic diameter of the casein micelles varied between 140 and 170 nm. No significant difference was found between the two milk groups (p = 0.05), as shown in Fig. 4B. The viscosity between samples from the same group (Treated or Control) varies from 2.2 to 3.4 mPa∙s at 20°C. No significant differences in viscosity were observed between the 2 groups of samples (p = 0.05). The same conclusions were drawn concerning the organoleptic properties. The panel of experts didn't notice any significant difference in the milk samples' visual, olfactory, and taste aspects (p = 0.05). Figure 4C details the average score the panel of experts gave on the different kinds of milk (fresh and pasteurized).
Part 3: Lipidomic analysis
Lipidome profile and analysis
This study aimed to ascertain the changes in the untargeted milk lipidome. The lipidome was determined following a liquid chromatography-quadrupole time-of-flight mass spectrometry approach. In more detail, it was possible to characterize the untargeted lipidome of milk, identifying 2450 lipid species. Only the species over a cut-off of 5000 (684 species) were considered for further analysis. The lipids were classified into 18 classes, namely acyl steryl glycosides (ASG [10 species]), N-Acylethanolamine (NAE [25 species]), triacylglycerols (TAG [23 species]), diacylglycerols (DAG [97 species]), monoacylglycerols (MAG [12 species]), phosphatidylserine (PS [48 species]), sphingomyelin (SM [52 species]), phosphatidylinositol (PI [35 species]), phosphatidylethanolamine (PE [128 species]), phosphatidylcholine (PC [82 species]), lysophosphatidylethanolamine (LPE [14 species]), lysophosphatidylcholine (LPC [10 species]), lysophosphatidylinositol (LPI [3 species]), ceramides (Cer [27 species]), hexosylceramides (HexCer [35 species]), galactosylceramide sulfate (SHexCer [18 species]), acylcarnitines (AcCarn [5 species]) and fatty acids (FA [60 species]). A list of the species is also reported in Table 2.
Table 2
Table 2
Number of species detected in milk from subclinical mastitis.
Class | Acronym | Number of species |
acyl steryl glycosides | ASG | 10 |
N-Acylethanolamine | NAE | 25 |
triacylglycerols | TG | 23 |
diacylglycerols | DG | 97 |
monoacylglycerols | MG | 12 |
phosphatidylserine | PS | 48 |
sphingomyelin | SM | 52 |
phosphatidylinositol | PI | 35 |
phosphatidylethanolamine | PE | 128 |
phosphatidylcholine | PC | 82 |
lysophosphatidylethanolamine | LPE | 14 |
lysophosphatidylcholine | LPC | 10 |
lysophosphatidylinositol | LPI | 3 |
ceramides | Cer | 27 |
hexosylceramides | HexCer | 35 |
galactosylceramide sulfate | SHexCer | 18 |
acylcarnitines | AcCarn | 5 |
fatty acids | FA | 60 |
The list of lipids identified in milk samples is presented in Supplementary Table S7.
https://unimibox.unimi.it/index.php/s/zPws6CxNWRYFXTd
Figure 5 presents the Partial Least Squares - Discriminant Analysis (PLS-DA) that measured the difference in the milk lipidome from Treated and non-treated mammary gland quarters along the time points T0, T7, T21, and T28 (Panel A). The results showed that the scatters of the two groups were limitedly separated across PC1 and PC2 at T0, indicating a quite homogeneity of the two groups. The effect of the treatment became more evident with time, as presented in panel A. Panel B shows the Heat maps that include the important features identified as statistically significant (p < 0.05) after a t-test. The most important lipids, shown to be differentially abundant between Treated and Control quarters, were presented in panel C as variable importance in the projection (VIP) score.
At T7, most of the lipids that changed between EO-treated and Control groups (8 out of ten) had a lower concentration in the milk of Treated cows. Of the ten lipids whose abundance changed, the most noticeable feature is an increase of PC 36:3. At T21, 15 lipids changed their abundance: two decreased, and 13 increased with treatment. Most changes were reported in the family of diglycerides, namely DG 33:3, DG 36:8, DG 35:3, DG 33:2, DG 35.2, DG 40:8, and PC 34:2 (VIP score > 1). At 28, the effects of TCEO on milk lipidome were limited to only 9 lipids (four increased and five decreased their abundance), the most interesting feature being MG 15:0, which exhibited an increased abundance after TCEO.
Part 4: Milk inflammatory parameters
SCC of milk after TCEO treatment
The changes in somatic cells in composite milk after TCEO are presented in Supplementary Figure S8. No statistically significant differences were detected after treating mammary gland quarters with TCEO (P = 0.05).
Milk concentration of IL-8 and the acute phase proteins Lf and Hp
The concentration of IL-8 was not homogeneous at T0. Only 2 milk samples from the Control and 3 from the Treated groups contained a concentration of IL-8 higher than 50 pg/mL. At the end of the experiment, the concentration of IL-8 decreased in both groups, as shown in Fig. 6A, but this difference was not significant (p = 0.05).
As reported in Fig. 6B, the concentrations of Lf found in quarter milk ranged between 143 and 217 µg/mL: when tested with Two-way ANOVA, interaction between time points and treatment groups resulted significant, with P = 0.0002, suggesting an influence of the EO on Lf even if not in a specific time point alone. On the other hand, as shown in Fig. 6C, the concentrations of Hp ranged between 1 and 0.7 µg/mL for the Control group and between 1.4 and 0.7 µg/mL for the Treated group, but this difference was not significant (P = 0.05, tested with one-way ANOVA)