Dosage‐dependent effects of monensin on the rumen microbiota of lactating dairy cattle

Abstract We examined the dose‐dependent effects of feeding lactating dairy cows a standard diet supplemented with monensin at 175, 368, or 518 mg cow‐1 day‐1on the rumen microbiota. For each dosage, 3 animals were randomly assigned into groups and fed the same basal total mixed ration diet supplemented with monensin, at the respective dose. After 20 days, rumen samples were taken and the effect on the microbiota was examined by 16S rRNA gene sequence analysis and qPCR. At the lowest dose no significant change in 16S rRNA gene sequences associated with any bacterial phyla was observed; however, at the medium and high dosages, we observed significant reductions in sequences associated with gram‐positive bacteria and significant increases in those associated with gram‐negative bacteria that were dosage dependent. All dosages reduced the levels of sequences associated with methanogenic archaea in the rumen, with the medium dosage showing the largest decline. No significant difference was observed for the 18S rRNA gene sequences associated with protozoa in any of the libraries. Our results indicate that with this diet the medium dosage of monensin was most efficacious for the reduction in methanogenic archaea in the rumen of lactating dairy.

the same basal total mixed ration diet supplemented with monensin, at the respective dose. After 20 days, rumen samples were taken and the effect on the microbiota was examined by 16S rRNA gene sequence analysis and qPCR. At the lowest dose no significant change in 16S rRNA gene sequences associated with any bacterial phyla was observed; however, at the medium and high dosages, we observed significant reductions in sequences associated with gram-positive bacteria and significant increases in those associated with gram-negative bacteria that were dosage dependent. All dosages reduced the levels of sequences associated with methanogenic archaea in the rumen, with the medium dosage showing the largest decline. No significant difference was observed for the 18S rRNA gene sequences associated with protozoa in any of the libraries. Our results indicate that with this diet the medium dosage of monensin was most efficacious for the reduction in methanogenic archaea in the rumen of lactating dairy.

K E Y W O R D S
16S rRNA, dairy cattle, microbiota, monensin, Rumen Monensin's ability to effect these changes is due to its ability to alter the microbial populations within the rumen. Monensin achieves this by inserting itself into susceptible cell membranes where it acts as a metal/proton antiporter (Pressman, 1976), eliminating the H + , Na + , and K + ion gradients across the cell membrane, resulting in the collapse of the proton motive force and eventually cell death (Russell, 1987;Russell & Strobel, 1989). Monensin is most effective against gram-positive bacteria, because they lack the protective outer membrane of gram-negative cells (Russell & Strobel, 1988). Thus, animals fed monensin have higher levels of gram-negative bacteria that are more likely to produce propionate; and reduced levels of gram-positive bacteria, that are more likely to degrade dietary protein (Russell & Strobel, 1988) and supply methanogens with hydrogen and formate for methanogenesis (Russell & Strobel, 1989).
However, studies have shown that monensin supplementation does not always produce these effects (Hamilton, DePeters, McGarvey, Lathrop, & Mitloehner, 2010;Hook, Northwood, Wright, & McBride, 2009;Odongo et al., 2007). Several studies have suggested that monensin's ability to alter the microbial populations within the rumen is dependent on both animal diet (Grainger, Williams, Eckard, & Hannah, 2010;Guan, Wittenberg, Ominski, & Krause, 2006) and monensin dosage (Duffield, Merrill, & Bagg, 2012;Ellis et al., 2012). However, there have not been any studies to date that have examined the dose-dependent effects of monensin on the rumen microbiota of lactating dairy cows. We hypothesized that we could identify a dosage of monensin that is effective at altering the bacterial population structure of the rumen and thus reduce the levels of methanogenic archaea by feeding lactating cows a standardized diet supplemented with increasing levels of monensin and measuring the changes in the rumen microbiota. In this study, we fed lactating dairy cattle a standardized diet supplemented with three different dosages of monensin and monitored the changes in their rumen microbiota via 16S gene sequence analysis and qPCR.

| Animals used in the study
Twelve multiparous, lactating Holstein dairy cows from the University of California, Davis dairy herd were used in this study.
The animals were housed in a facility approved by the Association for Assessment and Accreditation of Laboratory Animal Care and all protocols were approved by the UC Davis Institutional Animal Care and Use Committee. The 12 animals were randomly assigned into four groups (three cows per group) and were fed the same basal total mixed ration diet, twice daily, that was top-dressed with monensin at each respective dose: Control (0.0 mg cow −1 day −1 ), Low (175 mg cow −1 day −1 ), Medium (368 mg cow −1 day −1 ), or High (518 mg cow −1 day −1 ) for 20 days. The nutritional composition of the diet (Table 1)

| Rumen sampling
Rumen samples were collected from each cow before and after 20 days of monensin treatment to ensure the monensin treatment had enough time to take effect. Rumen fluid was collected from each animal via an oral stomach tube as described by Lodge-Ivey, Browne-Silva, and Horvath (2009). Briefly, approximately 2.5 hr after the feeding, cows were placed in a chute and a steel bovine mouth speculum was placed in the animal's mouth. A plastic stomach tube (0.6 cm I.D. and 3 m length) was inserted through the speculum into the rumo-reticulum and approximately 250 ml of fluid was collected from each animal and transferred into 50 ml tubes (Becton Dickenson, Franklin Lakes, NJ, USA) that were immediately sealed, placed on ice and transported to the laboratory for analysis. Precautions were taken to prevent saliva contamination; however, it is possible that some contamination occurred.

| DNA sequence and statistical analysis
DNA sequences were edited manually to correct falsely called bases and trimmed at both the 5′ and 3′ ends using the Seqman software (DNASTAR Inc., Madison, WI, USA) and analyzed for chimeras using UCHIME (Edgar, Hass, Clemente, Quince, & Knight, 2011).
Each OTU was assigned to a phylum using the Classifier software (Wang, Garrity, Tiedje, & Cole, 2007), available at the Ribosomal Database Project II (http://rdp.cme.msu.edu/index.jsp). Once classified, pair-wise comparisons of the OTU were performed using the Ribosomal Database Project II Library Compare software (Wang et al., 2007). Sequences of all OTU were deposited in GenBank under submission SUB2986750. Simpson entropy, Chao1 estimates, and evenness were calculated using Estimate S (Colwell, 2006).

| Quantitative PCR
Quantitative polymerase chain reaction (qPCR) for the methanogenic archaea DNA in rumen fluid was performed as described by Ohene-Adjei et al. (2008) and for protozoa, it was as described by Sylvester, Karnati, Yu, Morrison, and Firkins (2004). For both experiments, standard curves of template DNA (methanogen or protozoan) were made to contain 1.0 × 10 8 -1 × 10 5 copies per µl in 10-fold serial dilutions. All qPCR was performed in triplicate using an

| 16S rRNA analysis
To determine the effects of monensin on the bacterial populations within the rumen we constructed 16S rRNA gene libraries from DNA extracted from rumen fluid of cows fed an identical diet (Table 1) supplemented with either no, low (175 mg cow -1 day -1 ), medium (368 mg cow -1 day -1 ), or high (518 mg cow -1 day -1 ) doses of monensin for 20 days ( Table 2). The library derived from the rumen fluid of the cows fed the control diet (i.e., no monensin) contained sequences representative of 10 phyla ( Table 2). The majority of these sequences were associated with the phyla Bacteroidetes (58.9%), Firmicutes These dose-dependent changes in the bacterial populations are consistent with in vitro studies that showed the antibacterial activity of monensin is most effective against gram-positive bacteria (Chow & Russell, 1990;Russell, 1987). While monensin is known to attach equally well to both gram-positive and gram-negative bacteria (Chow, Kessel, & Russell, 1994), its effectiveness against the gram-positive bacteria is believed to be due to their lack of an outer membrane (Russell & Strobel, 1988). However this may be an oversimplification, as some gram-positive bacteria have been shown to become adapted to monensin in vitro (Simjee, Heffron, Pridmore, & Shryock, 2012;Weimer, Stevenson, Mertens, & Hall, 2011) aswell-as in vivo (Weimer, Stevenson, Mertens, & Thomas, 2008). The adaptation of gram-positive bacteria to monensin exposure has been associated with alterations in protein production, cell wall structure (Simjee et al., 2012) and the production of extracellular polysaccharides (Rychlik & Russell, 2002;Weimer et al., 2008). However, there are no resistance genes associated with this phenotype, and it is rapidly lost when selective pressure is removed (Simjee et al., 2012), suggesting it is epigenetic.
Monensin treatment also resulted in changes in the bacterial diversity within the rumen (Table 3). Feeding all doses of monensin reduced the number of operational taxonomic units (OTU), as estimated by Chao1 analysis, by approximately 25%. The Simpson indices, which incorporate species richness (or in this case OTU richness) and evenness, were also affected by the addition of monensin to the diet. At the low dose an increase in these indices was observed, and as the doses increased these indices decreased in a dose-dependent manner. This effect on the microbial diversity in the gut of animals fed antibiotics has been observed previously (Looft et al., 2012;Suchodolski et al., 2009).

| Methanogenic archaea qPCR
Monensin has been shown to have little or no direct effect against methanogenic archaea (Russell & Houlihan, 2003); however, the decreases in gram-positive bacteria within the rumen has been shown to reduce the concentration of methanogenic substrates (i.e., hydrogen and formate) needed by these microorganisms (Haney & Hoehn, 1968;Russell & Strobel, 1989). To determine the effect of monensin on the number of methanogens in the rumen, we performed qPCR on DNA extracted from the rumen fluid (Table 4). In the control group, no significant change was observed in methanogen sequences present after 20 days. However, significant decreases were observed for all levels of monensin treatment (p < 0.01). At the low and medium dose, we observed a 3.9-and 7.5-fold decrease in these sequences, respectively. Interestingly, the high dose only decreased these sequences by ~twofold. These data are consistent with those of Hook et al. (2009), who observed a decrease in the number of methanogen sequences in rumen fluid after 20 days of monensin treatment. However, after 90 days of treatment, the methanogens recovered and no significant effect was observed over their 180-day experiment. Likewise, Guan et al. (2006) reported a significant decrease in methane production after short-term monensin treatment; however, normal methane production resumed after 4 weeks, leading the authors to speculate that the rumen bacteria had adapted to monensin.

| Protozoa qPCR
It is estimated that up to 25% of rumen methanogens are associated with protozoa (Newbold, Lassalas, & Jouany, 1995) that supply them with H 2 and CO 2 via their hydrogenosomes (Embley, Giezen, Horner, Dyal, & Foster, 2003) that they convert to methane, water, and energy (Wolin, 1974). To determine if the decreases in methanogens were related to the anti-protozoan activity of monensin, we performed qPCR to quantify the number of protozoan 18S rRNA genes extracted from rumen fluid before and after monensin treatment. Our results showed that monensin had no significant effect on the number of protozoan 18S rRNA genes in the rumen fluid at any dosage tested (Table 5).
Overall, the high dosage of monensin produced the greatest increase in the gram-negative phyla Bacteroidetes and Proteobacteria and the greatest decrease in the gram-negative phylum Firmicutes.
However, the middle dosage also produced significant alterations in the phyla Bacteroidetes and Firmicutes and was more effective at reducing the methanogenic archaea than the high dosage. From these data, we conclude that for lactating dairy cows fed this diet the middle dosage was the most efficacious. Future studies are needed to examine the effects of these dosages on milk production and animal health to determine the economic return of using these dosages with this diet and to examine the possibility of bacterial adaptation to monensin over time.

ACK N OWLED G EM ENTS
This work was funded by the United States Department of Agriculture, Agricultural Research Service, National Program 108.

CO N FLI C T O F I NTE R E S T
No conflict of interest declared.

AUTH O R S CO NTR I B UTI O N
JM, SP, JP, and RN carried out bench work, data analysis and were involved in the manuscript preparation. SP and FM designed the study and performed all work related to sampling and care of the animals.
All authors have read and approved the final manuscript.

E TH I C S S TATEM ENT
The animals were housed in a facility approved by the Association for Assessment and Accreditation of Laboratory Animal Care and all protocols were approved by the UC Davis Institutional Animal Care and Use Committee.

DATA ACCE SS I B I LIT Y
All sequences were deposited in GenBank and are available under submission SUB4783994 accession numbers MK161521-MK163307.
TA B L E 5 Protozoan 18S rRNA gene copies ng −1 DNA before and after monensin treatment