Effect of Chitosan Microparticles on the Uterine Microbiome of Dairy Cows with Metritis

Third-generation cephalosporins, such as ceftiofur, are commonly used to treat metritis in dairy cows. Chitosan microparticles has been shown to have a broad spectrum of activity in vitro and to be effective against uterine pathogens in vivo; therefore, they have been hailed as a possible alternative to traditional antibiotics. Nonetheless, in the present study, we saw that chitosan microparticle treatment slowed the progression of the uterine microbiome of cows with metritis toward a healthy state, whereas ceftiofur treatment hastened the progression toward a healthy state. Given the lack of an effective alternative to traditional antibiotics and an increased concern about antimicrobial resistance, a greater effort should be devoted to the prevention of metritis in dairy cows.

ABSTRACT

The objective of this study was to evaluate the effect of chitosan microparticles on the uterine microbiome of cows with metritis. Dairy cows with metritis (n = 89) were assigned to 1 of 3 treatments: chitosan microparticles (n = 21), in which the cows received an intrauterine infusion of chitosan microparticles at metritis diagnosis (day 0), day 2, and day 4; ceftiofur (n = 25), in which the cows received a subcutaneous injection of ceftiofur on day 0 and day 3; and no intrauterine or subcutaneous treatment (n = 23). Nonmetritic cows (n = 20) were healthy cows matched with cows with metritis by the number of days postpartum at metritis diagnosis. Uterine swab samples collected on days 0, 3, 6, 9, and 12 were used for 16S rRNA gene sequencing and 16S RNA gene copy number quantification by quantitative PCR. Principal-coordinate analysis showed that the microbiome of the ceftiofur-treated and metritic untreated groups progressed toward that of the nonmetritic group by day 3, whereas that of the chitosan microparticle-treated group remained unchanged. The differences on day 3 were mainly due to a greater relative abundance of Fusobacteria, particularly Fusobacterium, in the chitosan microparticle-treated group than in the ceftiofur-treated and metritic untreated groups. Furthermore, the microbiome of the ceftiofur-treated group became similar to that of the nonmetritic group by day 9, whereas the microbiome of the chitosan microparticle-treated and metritic untreated groups became similar to that of the nonmetritic group only by day 12. The total bacterial 16S rRNA gene counts in the chitosan microparticle-treated group were greater than those in the metritic untreated controls on days 6 and 9, whereas the ceftiofur treatment group was the only group in which the total bacterial 16S rRNA gene count became similar to that in the nonmetritic group by day 12. In summary, chitosan microparticles slowed the progression of the uterine microbiome toward a healthy state, whereas ceftiofur hastened the progression toward a healthy state.

IMPORTANCE Third-generation cephalosporins, such as ceftiofur, are commonly used to treat metritis in dairy cows. Chitosan microparticles has been shown to have a broad spectrum of activity in vitro and to be effective against uterine pathogens in vivo; therefore, they have been hailed as a possible alternative to traditional antibiotics. Nonetheless, in the present study, we saw that chitosan microparticle treatment slowed the progression of the uterine microbiome of cows with metritis toward a healthy state, whereas ceftiofur treatment hastened the progression toward a healthy state. Given the lack of an effective alternative to traditional antibiotics and an increased concern about antimicrobial resistance, a greater effort should be devoted to the prevention of metritis in dairy cows.

INTRODUCTION

Improvements in genetics, housing, nutrition, and management have produced dramatic increases in milk production in the last 70 years, going from ∼2,000 kg/year in 1944 to over 10,000 kg/year in 2016 (1–3). Nonetheless, the improvements in milk production have not been accompanied by improvements in health. Approximately 45% of dairy cows experience a disease or disorder in the first 60 days postpartum (2), and metritis, the inflammation of all layers of the uterus, is one of the most prevalent and costly diseases to the dairy industry (4, 5).

Metritis is associated with polymicrobial infection with both Gram-negative bacteria (e.g., Fusobacterium, Bacteroides, and Porphyromonas) and Gram-positive bacteria (e.g., Trueperella, Peptostreptococcus, and Helcococcus) (6–8); therefore, metritis is commonly treated with broad-spectrum antibiotics. Ceftiofur, a third-generation cephalosporin approved for the treatment of metritis, is the broad-spectrum antibiotic of choice for treating metritis because it does not result in milk withdrawal (9–11). Nonetheless, because third-generation cephalosporins are an important class of antimicrobials in human medicine, the U.S. Food and Drug Administration has banned their use for disease prevention in food animals. This highlights the need for alternatives to traditional antibiotics for the prevention and treatment of diseases in food animals.

Chitosan microparticles are derived from chitosan, which is a linear polysaccharide produced by the deacetylation of chitin. Chitosan has antimicrobial properties at acidic pH, whereas chitosan microparticles maintain broad-spectrum antimicrobial activity even at neutral pH (12). Chitosan microparticles exerts their antimicrobial activity by binding to outer membrane protein A (OmpA) and by electrostatic interactions (12). Preliminary studies with a limited number of cows showed that the intrauterine infusion of chitosan microparticles was as effective as ceftiofur in reducing Escherichia coli counts (12) and the relative abundance of Fusobacterium necrophorum and appeared to be even more effective than ceftiofur in reducing the relative abundance of Bacteroidetes (13). E. coli, F. necrophorum, and Bacteroides spp. are believed to be important pathogens associated with the development of metritis (8); therefore, chitosan microparticles seemed to be a promising alternative to traditional antibiotics. The ultimate goal of antibiotic treatment is to return the organ and organism to a healthy state; therefore, we hypothesized that chitosan microparticles would be as effective as ceftiofur in returning the uterine microbiome of metritic cows to a healthy state. Health is the state of being free from disease or injury, and in our previous work, we have shown that uterine disease (i.e., metritis) is associated with a dysbiosis of the uterine microbiome characterized by a decreased bacterial richness, decreased heterogeneity, a greater total bacterial count, a greater relative abundance of Fusobacteria and Bacteroidetes, and a lower relative abundance of Proteobacteria and Tenericutes (6, 8). Hence, a reversal of those shifts characterizes the microbiome of cows that are cured of metritis (14). Therefore, the objective of this study was to evaluate if chitosan microparticles would be as effective as ceftiofur in returning the uterine microbiome of metritic cows to a healthy state.

RESULTS

Effect of treatment on uterine microbiome richness, diversity, and evenness.

There was a main effect of treatment group on the Chao1 richness index (P < 0.01), the Shannon diversity index (P = 0.01), and evenness (P < 0.01). The Chao1 index was greater (P < 0.01) for the nonmetritic group than for the metritic groups (Fig. 1A). On the other hand, the Shannon index (Fig. 1B) and evenness (Fig. 1C) were lower (P ≤ 0.05) for the nonmetritic group than for the metritic groups. There was no difference (P > 0.20) among the metritic groups. There was also an effect of time on the Chao1 index (P = 0.01), the Shannon index (P = 0.04), and evenness (P < 0.01) (Fig. 1A to C), but there was no effect of the interaction (P > 0.45) between treatment and time on the microbial community richness, diversity, or evenness.

FIG 1.

Chao1 richness index (A), Shannon diversity index (B), and evenness (C) of the uterine microbiota at enrollment (day 0), day 3, day 6, day 9, and day 12. CM (n = 21), intrauterine infusion of 24 g of chitosan microparticles at the time of diagnosis (day 0), day 2, and day 4; CEF (n = 25), injection of ceftiofur at day 0 and day 3; UNT (n = 23), no intrauterine or subcutaneous treatment; NME, (n = 20), nonmetritic healthy cows matched with metritic cows based on the number of days postpartum at the time of metritis diagnosis. *, significant (P ≤ 0.05) differences between each group of metritic cows and the group of nonmetritic healthy cows; +, significant (P ≤ 0.05) differences between ceftiofur-treated and nonmetritic healthy cows in panel A and between chitosan microparticle-treated and nonmetritic healthy cows in panel C. d, day.

Effect of treatment on uterine microbiome structure at the phylum level.

Principal-coordinate analysis (PCoA) paired with permutational multivariate analysis of variance (PERMANOVA) based on the Bray-Curtis dissimilarity of relative abundance at the phylum level showed that there was an effect of treatment group (P < 0.01), time (P < 0.01), and the interaction between treatment group and time (P < 0.01) on the structure of the uterine microbiome. The main effect of treatment showed that the uterine microbiomes of all metritic groups were different (P < 0.01) from the uterine microbiome of the nonmetritic group and that the uterine microbiome of the chitosan microparticle-treated group was different from the uterine microbiomes of the ceftiofur-treated (P = 0.03) and the metritic untreated (P = 0.04) groups, but there was no difference in the uterine microbiomes between the ceftiofur-treated and metritic untreated groups (P = 0.34) (Fig. 2A). Nonetheless, the interaction between treatment group and time showed that the uterine microbiomes of the metritic groups had different progressions toward a healthy uterine microbiome over time. On day 0, the uterine microbiomes of the metritic groups were not different (P > 0.30) among themselves, but they were different (P < 0.01) from the uterine microbiome of the nonmetritic healthy group (Fig. 2B). The microbiomes of metritic cows were very homogeneous, whereas the uterine microbiomes of nonmetritic cows were heterogeneous. On day 3, the uterine microbiomes of all the metritic groups were different (P < 0.01) from the uterine microbiomes of the nonmetritic group and the uterine microbiome of chitosan microparticle-treated group was different from the uterine microbiomes of the ceftiofur-treated (P < 0.01) and the metritic untreated (P = 0.03) groups, but there was no difference in the uterine microbiomes between the ceftiofur-treated and the metritic untreated groups (P = 0.68) (Fig. 2C). The uterine microbiomes of both the ceftiofur-treated and the metritic untreated groups became more heterogeneous, but the uterine microbiome of the chitosan microparticle-treated group remained very homogeneous. On day 6, the uterine microbiomes of the metritic groups were not different (P > 0.13) among themselves, but they were different (P < 0.01) from the uterine microbiome of the nonmetritic group, although the uterine microbiome of the chitosan microparticle-treated group was still less heterogeneous than the uterine microbiomes of the ceftiofur-treated and metritic untreated groups (Fig. 2D). On day 9, the uterine microbiomes of the metritic groups were not different (P > 0.11) among themselves and the uterine microbiomes of the chitosan microparticle-treated and the metritic untreated groups were still different (P < 0.01) from the uterine microbiome of the nonmetritic group, but the uterine microbiome of the ceftiofur-treated group was not different (P = 0.09) from that of the nonmetritic group (Fig. 2E). On day 9, the uterine microbiomes of all groups became more heterogeneous, but the progression of the uterine microbiomes of the chitosan microparticle-treated and metritic untreated groups toward a healthy state lagged that of the uterine microbiome of the ceftiofur-treated group. On day 12, the uterine microbiomes of none of the groups were different from each other (P > 0.10), although the uterine microbiome of the chitosan microparticle-treated group was still less heterogeneous than the uterine microbiomes of the other groups. These results show that the chitosan microparticle treatment delayed the progression of the uterine microbiome toward a healthy state, whereas ceftiofur treatment hastened the progression of the uterine microbiome toward a healthy state.

FIG 2.

Assessment of the overall effect of each treatment on the uterine microbiome (A) and its progression from the time of enrollment (day 0) (B) to day 3 (C), day 6 (D), day 9 (E), and day 12 (F), using principal-coordinate analysis of the Bray-Curtis dissimilarity of relative abundance at the phylum level. CM (n = 21), intrauterine infusion of 24 g of chitosan microparticles at the time of diagnosis (day 0), day 2, and day 4; CEF (n = 25), subcutaneous injection of ceftiofur at day 0 and day 3; UNT (n = 23), no intrauterine or subcutaneous treatment; NME (n = 20), nonmetritic healthy cows matched with metritic cows based on the number of days postpartum at the time of metritis diagnosis. The letters a, b, c, and d indicate significant (P ≤ 0.05) differences between treatment groups. D, day.

Linear discriminant analysis effect size (LEfSe) showed that the differences in the relative abundance of the different phyla that most likely explained the differences between the metritic groups and the nonmetritic group on day 0 were a greater relative abundance of Bacteroidetes and Fusobacteria in metritic cows and a greater relative abundance of Proteobacteria and Tenericutes in nonmetritic cows (Fig. 3A). On day 3, the differences in the relative abundance of the phyla most likely to explain the differences between the chitosan microparticle-treated and the ceftiofur-treated groups were a greater relative abundance of Fusobacteria in the chitosan microparticle-treated group and a greater relative abundance of Firmicutes in the ceftiofur-treated group (Fig. 3B). On day 3, the differences in the relative abundance of phyla most likely to explain the differences between the chitosan microparticle-treated group and the metritic untreated group were a greater relative abundance of Fusobacteria in the chitosan microparticle-treated group and a greater relative abundance of Verrucomicrobia in the metritic untreated group (Fig. 3C). In order to develop a more intuitive understanding of the microbiome composition and dynamics, we also present stacked bars for each day showing the average relative abundances of the phyla for each treatment (see Fig. S1 in the supplemental material) and the phylum-level compositions of all day 0 and day 12 samples by treatment (Fig. S2).

FIG 3.

Linear discriminant analysis effect size (LEfSe) was used to determine the phyla most likely to explain differences between metritic cows and healthy cows (A), cows treated with chitosan microparticles and ceftiofur (B), and cows receiving chitosan microparticle and no treatment (C) 3 days after enrollment. A logarithm base 10 linear discriminant analysis (LDA) score cutoff of 3 was used. CM (n = 21), intrauterine infusion of 24 g of chitosan microparticles at the time of diagnosis (day 0), day 2, and day 4; CEF (n = 25), subcutaneous injection of ceftiofur at day 0 and day 3; UNT (n = 23), no intrauterine or subcutaneous treatment; NME (n = 20), nonmetritic healthy cows matched with metritic cows based on the number of days postpartum at the time of metritis diagnosis.

Effect of treatment on uterine microbiome structure at the genus level.

The results of PCoA paired with PERMANOVA based on the Bray-Curtis dissimilarity of the relative abundance at the genus level mostly agreed with the results at the phylum level, showing that chitosan microparticle treatment delayed the progression toward a healthy state, whereas ceftiofur treatment hastened the progression of the uterine microbiome toward a healthy state (Fig. S3). The results of LEfSe at the genus level mostly agreed with results at the phylum level, with metritic cows having a greater relative abundance of Fusobacterium, Bacteroides, and Porphyromonas than nonmetritic cows on day 0 and with cows receiving the chitosan microparticle treatment having a greater relative abundance of Fusobacterium than cows receiving the ceftiofur treatment and untreated control cows on day 3 (Fig. S4).

Effect of treatment on total bacterial 16S rRNA gene counts.

There was an effect of treatment group (P < 0.01), time (P < 0.01), and the interaction between treatment group and time (P = 0.01) on total bacterial 16S rRNA gene counts in the uterus. The main effect of treatment showed that total bacterial 16S rRNA gene counts for all metritic groups were different (P < 0.01) from the count for the nonmetritic group, but there was no difference (P ≥ 0.06) among the metritic groups (Fig. 4). Nonetheless, the interaction between treatment group and time showed that the total bacterial 16S rRNA gene counts of the metritic groups had a different progression toward a healthy uterine microbiome over time. On day 0, the total bacterial 16S rRNA gene counts were not different (P > 0.20) among the metritic groups, but the counts for all groups were greater (P < 0.01) than the count for the nonmetritic group (Fig. 4). On day 3, the total bacterial 16S rRNA gene counts were not different (P > 0.13) among the treatment groups (Fig. 4). On day 6, the group receiving the chitosan microparticle treatment had greater (P ≤ 0.05) total bacterial 16S rRNA gene counts than the other groups, among which the total bacterial 16S rRNA gene counts were not different (P > 0.60) (Fig. 4). On day 9, the group receiving the chitosan microparticle treatment had greater (P ≤ 0.05) total bacterial 16S rRNA gene counts than the untreated control and nonmetritic groups, and the count for that group was not different (P = 0.08) from that for the group receiving the ceftiofur treatment, whereas the counts for the ceftiofur-treated, untreated control, and nonmetritic groups were not different (P > 0.06) among themselves (Fig. 4). On day 12, there was no difference (P > 0.35) in the total bacterial 16S rRNA gene counts among the chitosan microparticle-treated, ceftiofur-treated, and untreated control groups, but the chitosan microparticle-treated and untreated control groups had greater (P ≤ 0.03) total bacterial 16S rRNA gene counts than the nonmetritic group, whereas there was no difference (P = 0.16) in total bacterial 16S rRNA gene counts between the ceftiofur-treated group and the nonmetritic group. These results show that chitosan microparticle treatment delayed the progression of the uterine microbiome toward a healthy state.

FIG 4.

Total uterine bacterial 16S rRNA gene copy number (log₁₀) at enrollment (day 0), day 3, day 6, day 9, and day 12. CM (n = 21), intrauterine infusion of 24 g of chitosan microparticles at the time of diagnosis (day 0), day 2, and day 4; CEF (n = 25), subcutaneous injection of ceftiofur at day 0 and day 3; UNT (n = 23), no intrauterine or subcutaneous treatment; NME (n = 20), nonmetritic healthy cows matched with metritic cows based on the number of days postpartum at the time of metritis diagnosis. *, significant (P ≤ 0.05) differences between each group of metritic cows and the group of nonmetritic healthy cows; +, significant (P ≤ 0.05) differences between chitosan microparticle-treated cows and all the other groups; ‡, significant (P ≤ 0.05) differences between chitosan microparticle-treated and untreated cows and between chitosan microparticle-treated and nonmetritic healthy cows; †, significant (P ≤ 0.05) differences between chitosan microparticle-treated and nonmetritic healthy cows and between untreated and nonmetritic healthy cows.

Effect of chitosan microparticles on F. necrophorum and B. pyogenes in vitro.

Chitosan microparticles were able to inhibit the growth of or to eliminate both F. necrophorum and B. pyogenes, depending on the dose used (Fig. 5). For F. necrophorum (Fig. 5A), compared with the bacterial concentrations for the control at 24 h, treatment with chitosan microparticles resulted in lower (P < 0.01) bacterial concentrations when the microparticles were used at 0.1, 0.2, and 0.4%, although the bacterial concentration achieved with 0.1% chitosan microparticles increased (P = 0.004) from 0 to 24 h and the bacterial concentration achieved with 0.2% chitosan microparticles was unchanged (P = 0.41) from 0 to 24 h. The concentration of F. necrophorum bacteria in the group treated with 0.4% chitosan microparticles was greatly reduced (P < 0.001) from 0 to 24 h. For B. pyogenes (Fig. 5B), compared with the bacterial concentrations for the control at 24 h, chitosan microparticles resulted in lower (P < 0.001) bacterial concentrations when the microparticles were used at 0.2 and 0.4% but not when they were used at 0.1% (P = 0.64). The bacterial concentration achieved with 0.1% chitosan microparticles increased (P < 0.001) from 0 to 24 h and was reduced (P < 0.001) when the chitosan microparticles were used at 0.2 and 0.4%. In summary, chitosan microparticles inhibited F. necrophorum growth at all concentrations tested but were able to eliminate F. necrophorum only at a dose of 0.4%. In the case of B. pyogenes, the chitosan microparticles were able to eliminate B. pyogenes when they were used at a dose of 0.2 and 0.4%, although a dose-response was not observed.

FIG 5.

In vitro bacterial counts (log₁₀) for F. necrophorum (A) and B. pyogenes (B) in chopped meat broth (Anaerobe Systems, Morgan Hill, CA) containing 0.0%, 0.1%, 0.2%, and 0.4% chitosan microparticles at 0 and 24 h in culture. The letters a, b, c, and d indicate significant (P ≤ 0.05) differences in the chitosan microparticle concentrations at 24 h. *, significant (P ≤ 0.05) differences between 0 and 24 h for each chitosan microparticle concentration.

Utilization of N-acetylglucosamine as a carbon source by F. necrophorum and B. pyogenes.

The carbon source utilization assay showed that F. necrophorum and B. pyogenes did not preferentially utilize N-acetylglucosamine as a carbon source (Table S1). N-Acetylglucosamine was utilized only minimally by F. necrophorum and not at all by B. pyogenes. The use of N-acetylglucosamine by F. necrophorum and B. pyogenes ranked 55th and 79th out of the 95 carbon sources tested, respectively. The preferred carbon sources for F. necrophorum were l-glutamine, pyruvic acid methyl ester, l-alanyl–l-histidine, pyruvic acid, and β-hydroxybutyric acid (BHBA). The preferred carbon sources for B. pyogenes were urocanic acid, l-malic acid, d-galacturonic acid, l-glutamine, and fumaric acid. In summary, F. necrophorum did not preferentially utilize N-acetylglucosamine as a carbon source, and B. pyogenes was not able to utilize N-acetylglucosamine as a carbon source.

DISCUSSION

Previous studies showed that the intrauterine infusion of chitosan microparticles was as effective as ceftiofur in reducing E. coli counts (12) and the relative abundance of F. necrophorum and appeared to be even more effective than ceftiofur in reducing the relative abundance of Bacteroidetes (13); therefore, chitosan microparticles seemed to be a promising alternative to traditional antibiotics. Therefore, in this study we assessed the effectiveness of chitosan microparticle treatment in returning the uterine microbiome of metritic cows to a healthy state compared with that of ceftiofur treatment and no treatment as a control.

First, we did not observe any difference among the metritic groups in microbial richness, diversity, or evenness. However, PCoA and the total bacterial count showed that, in contrast to our preliminary findings (12, 13), treatment with chitosan microparticles was less effective than treatment with ceftiofur in returning the uterine microbiome of metritic cows to a healthy state. In fact, untreated cows showed a faster recovery than chitosan microparticle-treated cows, which showed that chitosan microparticles hindered the progression of the uterine microbiome toward a healthy state. This could be clearly observed in the PCoA at the phylum and genus levels on day 3 and day 6 and the total bacterial counts on day 6 and day 9. LEfSe showed that differences between the chitosan microparticle-treated and ceftiofur-treated groups and between the chitosan microparticle-treated group and untreated controls on day 3 were mainly due to a greater relative abundance of Fusobacteria, particularly Fusobacterium, in the chitosan microparticle-treated group than in the ceftiofur-treated group and the untreated controls. Fusobacterium from the phylum Fusobacteria is one of the major pathogens associated with metritis in dairy cows (15); therefore, our data indicate that chitosan microparticle treatment altered the uterine microbiome in favor of Fusobacterium.

The negative effects of chitosan microparticle treatment on the uterine microbiome of metritic cows were unexpected because chitosan is categorized as generally recognized as safe (GRAS) for general use in foods by the U.S. Food and Drug Administration. Because chitosan is safe in humans, we expected that it would be safe in animals. Nonetheless, in our larger study, we observed that chitosan microparticle treatment resulted in increased culling from the herd (i.e., loss through sale or death) in the first 60 days postpartum (16). When we looked at the reasons for culling, chitosan microparticle treatment resulted in a greater proportion of cows being culled because of pelvic inflammation and peritonitis. Others have also reported increased inflammation after the infusion of chitosan hydrogel in the mammary gland of cows (17). Therefore, the negative effects on the microbiome observed here may have been a result of exacerbated inflammation after chitosan microparticle administration. The reasons for the discrepancies between the findings of this study and those of our previous study evaluating the effect of chitosan microparticles on the uterine microbiome are not clear (13), but they may be a result of having a very small sample size in our earlier study (two to three cows per group), which could have led to spurious observations. Another difference was the dose administered, which was increased 3-fold in this study compared with the dose used in our previous study. Hence, our finding of a delayed progression of the uterine microbiome toward a healthy state, marked by a greater relative abundance of Fusobacteria, particularly Fusobacterium, corroborates the finding of poorer recovery after chitosan microparticle treatment. Some possible confounders that were not investigated in this study are the effect of chitosan microparticles on viruses, protozoa, and fungi and the interaction between chitosan microparticles and mucin, which could have affected its overall effect on the uterine microbiome. Therefore, these factors should be investigated in further studies.

Because the efficacy of chitosan microparticles against Fusobacterium had not been previously tested in vitro, we tested the efficacy of chitosan microparticles against F. necrophorum, the main Fusobacterium species (15). We observed that although chitosan microparticles inhibited F. necrophorum growth at all concentrations tested, chitosan microparticles were able to significantly reduce F. necrophorum counts only at a dose of 0.4%. Here, we aimed for a concentration of 0.6% chitosan microparticles in the uterus because, although chitosan microparticles were effective at eliminating E. coli when they were used at a concentration of 0.2% in vitro (12), a dose of 0.6% chitosan microparticles was required ex vivo when uterine fluid from cows with metritis was used (18). Therefore, it is possible that a dose higher than 0.6% chitosan microparticles would be needed to significantly reduce F. necrophorum counts in vivo. In the case of B. pyogenes, chitosan microparticles were able to reduce B. pyogenes concentrations when they were used at a dose of 0.2%; therefore, a dose of 0.6% may have been able to inhibit its growth in vivo.

Because of the potential for the conversion of chitosan microparticles into glucosamine and N-acetylglucosamine in the uterus, we evaluated if F. necrophorum could utilize N-acetylglucosamine as a carbon source; however, we observed that N-acetylglucosamine was not a preferred carbon source for F. necrophorum, although a minor utilization was observed. Similar to our findings, the utilization of N-acetylglucosamine by an F. necrophorum strain isolated from sheep was not observed unless iron was restricted (19). Iron restriction is not expected to happen, given the extensive tissue damage and hemolysis in the case of metritis. N-Acetylglucosamine was not utilized by B. pyogenes. The Biolog AN MicroPlate does not include glucosamine as a carbon source. However, because of the similarities and congruent metabolism of these amino sugars, it is unlikely that glucosamine could be utilized as a carbon source. Nonetheless, the lower growth rate of E. coli on glucosamine than on N-acetylglucosamine has been reported (20); therefore, further investigation is needed to completely rule out the utilization of glucosamine by F. necrophorum. Finally, although E. coli can utilize glucosamine and N-acetylglucosamine as a carbon source, we did not observed any shift in the uterine microbiome toward a greater abundance of Escherichia in the chitosan microparticle-treated group; therefore, the conversion of chitosan microparticles into glucosamine or N-acetylglucosamine or the utilization of glucosamine or N-acetylglucosamine may be a minor contributor to bacterial growth in vivo.

An interesting finding was the predilection of F. necrophorum to use BHBA as an energy source. Cows undergo a period of negative energy balance at about the time of calving, which leads to lipid mobilization and the increased production of BHBA in the liver, and these compounds end up in the blood circulation (21) and, hence, can reach the uterus because of vascular degeneration postpartum (22). More specifically, cows that develop metritis have higher BHBA concentrations than healthy cows (23), which could be the reason for the greater abundance of Fusobacterium in cows that develop metritis. The preference of F. necrophorum for l-glutamine and pyruvic acid seems to be of lesser relevance for its rapid growth in the uterus during early lactation because the amounts of glutamine and pyruvic acid are actually decreased in early lactation because of mammary gland uptake for milk production (24, 25). Therefore, energy metabolites and their availability in the uterus may regulate uterine pathogen growth and deserve further research.

We also observed that ceftiofur treatment hastened the progression of the uterine microbiome of metritic cows toward health. Using PCoA, we observed that the uterine microbiome of the ceftiofur treatment group was not different from that of the nonmetritic group by day 9 and was the closest to that of the nonmetritic group by day 12. We also observed that the ceftiofur treatment group was the only group in which the total bacterial count was not different from that in the nonmetritic group by day 12. This finding corroborates the finding of an increased cure rate for the ceftiofur treatment compared with the rates for the chitosan microparticle treatment and no treatment as a control by day 12 (16). Interestingly, although an advantage for the ceftiofur treatment in returning the uterine microbiome to health could be observed, the differences between ceftiofur treatment and no treatment were not as marked as one would expect. This finding is intriguing, given the clear advantage in cure rates observed for ceftiofur treatment compared with the cure rate observed for no treatment as a control in our companion study (16) and in previous studies (9, 10). Nonetheless, our findings presented here agree with our previous findings, where we showed only minor differences in the microbiomes of ceftiofur-treated and untreated cows (14). Therefore, it seems that even minor differences in the uterine microbiome are associated with significant changes in cure rates.

Lastly, as it has been shown before (6, 15, 26–28), metritic cows had a greater relative abundance of Bacteroidetes, particularly Bacteroides, Porphyromonas, and Fusobacteria, particularly Fusobacterium, whereas healthy cows had a greater relative abundance of Proteobacteria and Tenericutes on day 0. Similar to previous findings (6, 15, 29), we also showed that metritic cows had decreased richness (i.e., a lower Chao1 richness index) on day 0. On the other hand, contrary to what we observed previously (15, 29), metritic cows were shown to have increased diversity (i.e., a higher Shannon diversity index), which was a result of having an increased evenness of the uterine microbiota compared with that for healthy cows. Others have also shown no difference in richness (numbers of operational taxonomic unit), diversity (Shannon diversity index), or evenness (28) or even an increased alpha diversity (i.e., the measure was not specified) in metritic cows compared with that in healthy cows (27). It was clear from the PCoA that the uterine microbiome of metritic cows on day 0 was very homogeneous, whereas the uterine microbiome of healthy cows was heterogeneous; therefore, our interpretation of decreased richness but increased evenness in metritic cows is that the uterine microbiome of metritic cows is consolidated into fewer genera with a more similar relative abundance than that of healthy cows.

To conclude, despite promising preliminary results, chitosan microparticle treatment slowed the progression of the uterine microbiome toward a healthy state, thus failing as an alternative to traditional antibiotic treatment. Therefore, our study indicates with a high degree of confidence that chitosan microparticles, tested as described here, are detrimental to uterine health. As expected, but not previously reported, ceftiofur treatment hastened the progression of the uterine microbiome toward a healthy state. Therefore, we provide data on the expected shift in the uterine microbiome when an ineffective treatment and an effective treatment are administered. Finally, given the continuous rise in antibiotic resistance, further research is needed to develop effective alternative therapies to traditional antibiotics.

MATERIALS AND METHODS

Cows, housing, and feeding.

All animal procedures were approved by the University of Florida Institutional Animal Care and Use Committee (IACUC protocol number 201509189).

The study was conducted in a dairy herd located in north central Florida from May 2016 to June 2017. The herd consisted of ∼4,400 Holstein cows milked three times daily, and the rolling herd average milk yield was ∼12,000 kg. Postpartum cows were housed in tunnel-ventilated six-row barns with sprinklers over the feeding areas that were activated when the ambient temperature rose above 21°C. The stalls were deep bedded with sand. The postpartum diet was formulated to meet or exceed the dietary nutrient requirements for a lactating cow weighing 680 kg and producing 45 kg of 3.5% fat-corrected milk and 3.0% protein (30), and it was delivered as a total mixed ration twice daily.

Diagnosis of metritis.

Cows were examined for the diagnosis of metritis at 5, 7, and 9 days postpartum using a Metricheck device (Simcro, Hamilton, New Zealand), as previously reported (6, 15). Discharge retrieved from the vagina was scored as follows: 1 for a not fetid normal lochia with viscous, clear, red, or brown fluid; 2 for a cloudy mucoid discharge with flecks of pus; 3 for a not fetid, mucopurulent discharge with <50% pus; 4 for a not fetid mucopurulent white, yellow, or reddish brownish discharge with ≥50% pus; and 5 for a fetid, thin, serous, or watery discharge that may have been reddish-brownish with or without pieces of necrotic tissue present (10). Cows with a vaginal discharge score of 5 were classified as having metritis.

Treatment groups.

The cows used for this study represented a subset of 89 cows with metritis that were randomly sampled from a larger study with 826 cows with metritis that had been randomly assigned to 1 of 3 treatments: chitosan microparticles, which consisted of the intrauterine infusion of 24 g of chitosan microparticles dissolved in 40 ml of sterile distilled water at the time of diagnosis (day 0), day 2, and day 4; ceftiofur, which consisted of the subcutaneous injection of 6.6 mg/kg of body weight of ceftiofur crystalline-free acid (Excede; Zoetis, Parsippany, NJ, USA) in the base of the ear at day 0 and day 3; or no intrauterine or subcutaneous treatment (16). A group of healthy nonmetritic cows matched by the number of days in milk to that for cows with a metritis diagnosis was added for comparison. The chitosan microparticles were prepared as previously reported (12). Briefly, a 2% (wt/vol) chitosan (molecular weight, 50 to 190 kDa; deacetylation degree, 75 to 85%; catalog number 448869-250G; Sigma-Aldrich, Saint Louis, MO, USA) solution was prepared with 2% (vol/vol) acetic acid and 1% (vol/vol) Tween 80. For cross-linking, the chitosan solution was stirred and 10% (wt/vol) sodium sulfate was added dropwise during 25 min of sonication. The chitosan microparticles were collected by centrifugation (8,200 × g) and washed with sterile water. The weight of the chitosan microparticles was measured after freeze-drying. For quality control, each batch of chitosan microparticles was confirmed to have antimicrobial activity against E. coli in vitro at a concentration of 0.2% chitosan microparticles. The amount of chitosan microparticles infused in the uterus (24 g) was calculated to give a final concentration of at least 0.6% (assuming a uterine content in cows with metritis of 4 liters), which was the lowest concentration shown to effectively eliminate E. coli from uterine fluid ex vivo (18). The sample size was estimated based on our previous studies, in which a sample size of approximately 15 cows per group provided adequate power to evaluate the uterine microbiome shift after antibiotic the treatment of cows with metritis (14). Hence, uterine swab samples were collected from chitosan microparticle-treated (n = 21), ceftiofur-treated (n = 25), metritic untreated (n = 23), and nonmetritic (n = 20) cows on days 0, 3, 6, 9, and 12.

Sample collection and DNA extraction.

Uterine swab samples were collected right before the first treatment on day 0 from ceftiofur-treated and chitosan microparticle-treated cows. The swab samples were collected using a 30-in. double-guarded sterile culture swab (Continental Plastics Corporation, Delavan, WI). Briefly, the instrument was gently passed through the cervix and positioned in the uterine body, where the internal sheath and the swab were exposed, and the swab was gently rolled against the uterine wall. The swab was retracted within the double sheath before removal from the cow. The swab was stored in a 15-ml conical sterile tube and placed on ice until its return to the laboratory, where the tip of the swab was cut off, immersed in 1,000 μl of phosphate-buffered saline, and vortexed, and then the swab tip was discarded and the fluid was stored at −80°C until DNA extraction.

The isolation of genomic DNA was performed using a QIAamp DNA minikit (Qiagen, Valencia, CA, USA) according to the manufacturer’s instructions for buccal swabs. To maximize bacterial DNA extraction, 5 μl of lysozyme (50 mg/ml; Thermo Fisher, Waltham, MA, USA) and 6 μl of mutanolysin (14.3 kU/ml; Sigma-Aldrich, Saint Louis, MO, USA) were mixed with 200 μl of the thawed fluid from the swab sample, and the mixture was incubated at 37°C for 1 h before starting the QIAamp DNA minikit protocol. The DNA concentration and purity were evaluated by determination of the optical density (OD) using a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies, Rockland, DE, USA) at wavelengths of 230, 260, and 280 nm, and the ratio of the OD at 260 nm (OD₂₆₀)/OD₂₈₀ for DNA ranged from 1.7 to 2.1.

16S rRNA gene amplification and sequencing.

For each metagenomic DNA sample, amplification of the V4 hypervariable region of the bacterial/archaeal 16S rRNA gene was performed using the 515F and 806R primer set according to previously described methods and by optimization for the Illumina MiSeq platform (31). The PCR products were tagged with 280 different 12-bp error-correcting Golay barcodes (32). PCRs were carried out in triplicate in 25-μl reaction mixtures using 12 to 300 ng of template DNA, 1× EconoTaq Plus Green master mix (Lucigen, Middleton, WI), and 10 μM each primer. Thermal cycling consisted of an initial denaturing step of 94°C for 3 min, followed by 35 cycles of 94°C for 45 s, 50°C for 1 min, and 72°C for 90 s and a final elongation step of 72°C for 10 min. Replicate amplicons were visualized on a 1.2% agarose gel stained with 0.5-mg/ml ethidium bromide and then pooled and purified using a gel/PCR DNA fragment extraction kit (IBI Scientific, Peosta, IA).

The amplicon DNA concentration in each sample was measured using a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies, Rockland, DE, USA) at wavelengths of 260 and 280 nm. Amplicons were standardized to the same concentration and pooled into three different sequencing runs according to the individual barcode primers for the 16S rRNA gene. Final equimolar libraries were sequenced using a MiSeq reagent kit (version 2; 300 cycles) on a MiSeq platform (Illumina, Inc., San Diego, CA). The generated 16S rRNA gene sequences were processed using a MiSeq Reporter proprietary metagenomics work flow based on the Greengenes database (33) as previously reported (6). Illumina sequencing of the V4 region of the 16S rRNA gene provides an excellent sequencing depth and coverage, but the shorter reads (reads of ∼300 bp) may provide an inaccurate species classification, especially for species with high 16S rRNA gene similarity; therefore, only classification to the genus level was attempted and is presented (34).

Total bacterial 16S rRNA gene copy number quantification using quantitative PCR.

Total bacterial 16S rRNA gene copy number quantification was performed using a Femto bacterial DNA quantification kit targeting the 16S rRNA gene (Zymo Research, El Cajon, CA, USA) following the manufacturer’s protocol. Briefly, 18 μl of the master mix supplied with the kit was added to each well with 2 μl of each sample or standards ranging from 2 fg to 20 ng of DNA. The cycling parameters were as indicated by the manufacturer’s protocol. Samples were run in duplicate. The amount of DNA in each sample was calculated based on the standard curve. Then, the 16S rRNA gene copy number was calculated using the following formula, provided by the company: 16S rRNA gene copy number = amount of DNA (in grams)/constant for the number of grams to the number of base pairs × genome size, where the constant for the number of grams to the number of base pairs was 1.096 × 10⁻²¹ g and the genome size was 5.5 × 10⁶ bp. Then, logarithms to the base 10 conversions of the raw values were determined.

Effect of chitosan microparticles on Fusobacterium necrophorum and Bacteroides pyogenes in vitro.

Because the analysis of the uterine microbiome using 16S rRNA gene sequencing revealed that chitosan microparticles were ineffective against Fusobacteria, particularly Fusobacterium, we tested the effect of chitosan microparticles against Fusobacterium necrophorum in vitro. We also used Bacteroides pyogenes, another highly abundant uterine pathogen (7), for comparison. F. necrophorum and B. pyogenes were inoculated into chopped meat broth (Anaerobe Systems, Morgan Hill, CA) and incubated under anaerobic conditions with a GasPak anaerobic system (BD, Franklin Lakes, NJ) at 37°C overnight. Then, the culture was diluted to approximately 5 × 10⁴ CFU/ml and incubated in chopped meat broth with 0, 0.1, 0.2, and 0.4% chitosan microparticles. After 24 h of anaerobic incubation at 37°C, the bacterial cultures were serially diluted and plated on tryptic soy agar supplemented with 5% sheep blood to enumerate the bacteria. The assay was conducted in triplicate. Because ceftiofur is known to reach the MICs in the endometrium and lochia of cows with metritis that are required to inhibit the growth of 90% of F. necrophorum bacteria and other uterine pathogens, such as Prevotella melaninogenica and Trueperella pyogenes (0.125 μg/ml) (35, 36), ceftiofur was not tested.

Utilization of N-acetylglucosamine as a carbon source by F. necrophorum and B. pyogenes.

Similar to what happens in the gastrointestinal tract, it is possible that the complex microflora of the uterus may hydrolyze chitosan microparticles into glucosamine and N-acetylglucosamine, which would then be available for bacterial growth (37). The lysozyme released after cell death or leukocyte degranulation may also hydrolyze chitosan microparticles into glucosamine and N-acetylglucosamine (38). Hence, we tested whether F. necrophorum could use N-acetylglucosamine as a carbon source for growth using an AN MicroPlate (Biolog, Inc.) containing 95 carbon sources. We also used B. pyogenes for comparison. The AN MicroPlate does not include glucosamine as a carbon source. The assay was carried out in triplicate according to the manufacturer’s instructions. Briefly, F. necrophorum KG34 (GenBank accession no. SRX5402669) and B. pyogenes KG32 (BioSample accession no. SRS5092514) were grown on Wilkins-Chalgren agar medium (Sigma-Aldrich) and suspended in 14 ml of AN inoculating fluid (Biolog) to make bacterial suspensions with transmittance levels of about 65%. A 100-μl bacterial suspension was quickly plated into each well of the AN MicroPlate, and the plate was incubated at 37°C in a GasPak EZ anaerobic pouch system (BD). Utilization of the carbon sources by F. necrophorum and B. pyogenes was indicated by tetrazolium violet forming a blue color, which was read at 3, 6, 12, and 24 h postinoculation at an OD₅₉₀ using a SmartSpec 3000 spectrophotometer (Bio-Rad). The well containing water in the Biolog AN MicroPlate was used as a blank control, and the actual OD for the carbon source was obtained by subtracting the measured OD value from the OD at time zero.

Data analysis.

The swab samples resulted in 13,638,041 high-quality reads after filtering for size, quality, and the presence of chimeras. Four swab samples from three ceftiofur-treated cows did not pass the filtering criteria and were excluded from the study.

Principal-coordinate analysis (PCoA) of the Bray-Curtis dissimilarity at the phylum and genus levels was used to compare treatment groups and time points using the PAST (version 3.18) package. The effects of treatment, time, and the interaction between treatment and time on the uterine microbiome structure were evaluated with a two-way PERMANOVA with 9,999 permutations, based on the Bray-Curtis dissimilarity at the phylum level, using the PAST (version 3.18) package. Linear discriminant analysis effect size (LEfSe) was used to determine the phyla and genera most likely to explain the differences between treatment groups (39). A logarithm base 10 linear discriminant analysis (LDA) score cutoff of 3 was used. Alpha diversities at the genus level were compared among the groups using the Chao1 richness index and the Shannon diversity index. Evenness was also evaluated. Evenness was calculated by dividing the Shannon diversity index by the natural logarithm of the genus richness (i.e., the number of genera in each sample). Continuous outcomes, such as the relative or absolute abundances of bacteria, Chao1 richness index, Shannon diversity index, and evenness, were compared by performing analysis of variance (ANOVA) for repeated measures using the MIXED procedure of SAS (version 9.4) software (SAS Institute, Cary, NC). The ANOVA models included the effects of treatment, time, parity (multiparous versus primiparous), and the interaction between treatment and time. Parity was dropped from the model if P was >0.05. Normality and the homoscedasticity of residuals were assessed using residual plots. For repeated measures, unstructured, autoregressive 1, and compound symmetry covariance structures were tested, and the one with the smallest Akaike information criterion was chosen. For all models, Tukey’s honestly significant difference test for multiple comparisons was used. Significance was considered when P was ≤0.05.

Data availability.

All the sequences were uploaded to the National Center for Biotechnology Information under the Sequence Read Archive (https://www.ncbi.nlm.nih.gov/sra) using two BioProjects: BioProject accession number PRJNA643565 with 165 samples and BioProject accession number PRJNA643556 with 280 samples.