Supplementing branched-chain volatile fatty acids in dual-flow cultures varying in dietary forage and corn oil concentrations. II: Biohydrogenation and incorporation into bacterial lipids

To maintain membrane homeostasis, ruminal bacteria synthesize branched-chain fatty acids (BCFA) or their derivatives (vinyl ethers) that are recovered during methylation procedures as branched-chain aldehydes (BCALD). Many strains of cellulolytic bacteria require 1 or more branched-chain volatile fatty acid (BCVFA). Therefore, the objective of this study was to investigate BCVFA incorporation into bacterial lipids under different dietary conditions. The study was an incomplete block design with 8 continuous culture fermenters used in 4 periods with treatments (n = 4) arranged as a 2 × 2 × 2 factorial. The factors were high (HF) or low forage (LF, 67 or 33% forage, 33:67 alfalfa: orchardgrass), without or with supplemental corn oil (CO; 3% dry matter, 1.5% linoleic fatty acid), and without or with 2.15 mmol/d (5 mg/d 13 C each of isovalerate, isobutyrate, and 2-methylbutyrate). After methylation of bacterial pellets collected from each fer-menter’s effluent, fatty acids and fatty aldehydes were separated before analysis by gas chromatography and isotope ratio mass spectrometry. Supplementation of BCVFA did not influence biohydrogenation extent. Label was only recovered in branched-chain lipids. Lower forage inclusion decreased BCFA in bacterial fatty acid profile from 9.45% with HF to 7.06% with LF and decreased BCALD in bacterial aldehyde profile from 55.4% with HF to 51.4% with LF. Supplemental CO tended to decrease iso even-chain BCFA and decreased iso even-chain BCALD in their bacterial lipid profiles. The main 18:1 isomer was cis -9 18:1, which increased (P < 0.01) by 25% from CO (data not shown). Dose recovery in bacterial lipids was 43.3% lower with LF than HF. Supplemental CO decreased recovery in the HF diet but increased recovery with LF (diet × CO interaction). Recovery from anteiso odd-chain BCFA and BCALD was the greatest; therefore, 2-methylbutyrate was the BCVFA primer most used for branched-chain lipid synthesis. Recovery in iso odd-chain fatty acids (isovalerate as primer) was greater than label recovery in iso even-chain fatty acids (isobutyrate as primer). Fatty aldehydes were less than 6% of total bacterial lipids, but 26.0% of 13 C recovered in lipids were recovered in BCALD because greater than 50% of aldehydes were branched-chain. Because BCFA and BCALD are important in the function and growth of bacteria, especially cellulolytics, BCVFA supplementation can support the rumen microbial consortium, increasing fiber degradation and efficiency of microbial protein synthesis.


INTRODUCTION
Microbial membrane homeostasis is mediated through genetic and biochemical mechanisms and is necessary for microbial ability to adjust to different conditions (Kaneda, 1991;Zhang and Rock, 2008).Because desaturation is typically an O 2 -using process, anaerobic bacteria, including those in the rumen, rely on a methyl branch of fatty acids (FA) to increase fluidity, with those in an anteiso configuration more fluidizing than FA in an iso configuration.Because saturation and desaturation cannot occur after the membrane is formed as in aerobes, the membrane in ruminal bacteria needs to alter the lipid profile de novo to manage fluidity, regulate passive permeability, and to anchor membrane-bound proteins.Eubacterium cellulosolvens 5494 (a cellulolytic member of the Clostridiales order) increased mediumchain FA and unsaturated FA, especially 18:1 derived from the medium, to increase fluidity when switched to cellulose compared with glucose, cellobiose, or fructose as energy sources (Moon and Anderson, 2001).Similarly, both Fibrobacter succinogenes and Ruminococcus albus decreased their membrane hydrophobicity (i.e.,

Supplementing branched-chain volatile fatty acids in dual-flow cultures varying in dietary forage and corn oil concentrations. II: Biohydrogenation and incorporation into bacterial lipids
increased fluidity) when exposed to cellulose, and thickening of the cell wall facing the adjacent cellulose was noted for F. succinogenes (Burnet et al., 2015).Adaptation to cellulolysis requires enough membrane flexibility for translocation of large polymers coinciding with enough rigidity to maintain trans-membrane ion gradients.A model describes how F. succinogenes translocates and assembles the extracellular (or inner membrane for gram-negative bacteria) proteins needed for assembly of complexes, adherence, depolymerization of cellulose and hemicellulose, and transport of oligosaccharides (Raut et al., 2019).
Cellulolytic bacteria primarily use branched-chain FA (BCFA), which require branched-chain VFA (BCVFA) as a primer, to maintain membrane fluidity (Miyagawa et al., 1979;Or-Rashid et al., 2007;Roman-Garcia et al., 2021b).In addition, butyrivibrios occupy an important hemicellulolytic and biohydrogenation niche, and members of that group have high inclusion of iso and anteiso BCFA in their membrane structure (Kopečný et al., 2003;Maia et al., 2010).The butyrivibrios have a very thin membrane structure, so greater inclusion of the more fluid BCFA should make them more sensitive to inhibitory fluidization from PUFA imbedding in the membrane (Hackmann and Firkins, 2015).
The BCVFA isobutyrate, 2-methylbutyrate, and isovalerate are formed by the degradation of branchedchain AA are primers for synthesis of iso even-chain, anteiso odd-chain, and iso odd-chain FA, respectively (Parsons and Rock, 2013).In addition to BCFA synthesis, Allison et al. (1962) documented that Ruminococcus flavefaciens incorporated 14 C from isovalerate into a C15 aldehyde (ALD).Additionally, R. albus incorporated 14 C from isobutyrate into C14 and C16 ALD.The authors presumed that the ALD with 14 C label were branched-chain aldehydes (BCALD) from plasmalogens.Plasmalogens are formed by the reductive conversion of an ester bond in a FA in the sn-1 position of a phospholipid to a vinyl ether bond in anaerobic bacteria (Jackson et al., 2021).
Unlike alkaline hydrolysis (Allison et al., 1962), acid hydrolysis can release plasmalogen components.During the standard methylation process to prepare for GC analysis, FA are converted to FAME, but the vinyl ethers in plasmalogens are converted to dimethylacetals (DMA).With GC analysis, individual DMA and FAME can coelute.The ALD are a much smaller proportion of bacterial lipids and have few commercial standards, which would underrepresent quantification of ALD.These DMA can be important products in mixed ruminal bacteria (Alves et al., 2013) and sinks for BCVFA precursors (Allison et al., 1962).Plasmalogen lipids are more tightly packed than phospholip-ids, which decreases the permeability of membranes (Goldfine, 2017).In support, Megasphaera elsdenii (not one of the ruminal strains) decreased plasmalogens that corresponded with increased production of more hydrophobic saturated FA (Kaufman et al., 1988).An emerging role for disruption of reactive oxygen damage in anaerobic bacteria has not been verified in ruminal bacteria but is likely based on a high proportion of genes in plasmalogen synthesis recovered after searching genomes from Firmicutes and Actinobacteria as compared with Bacteroidetes and Proteobacteria (Jackson et al., 2021).Thus, BCVFA as precursors likely play an emergingly important role in rumen bacterial plasmalogens containing branched lipids.
Our objective was to investigate the effect of dietary conditions and BCVFA supplementation on composition of branched-chain lipids, both BCFA and BCALD, from bacteria harvested from dual-flow continuous cultures.Supplemental BCVFA was expected to increase the extent of biohydrogenation of supplemented PUFA because butyrivibrios could benefit from increased supply of BCVFA primers for BCFA synthesis.Supplemental CO provides PUFA, primarily linoleic acid, but also provides oleic acid (Gunstone, 1996).The PUFA might disrupt cell membrane integrity and require more de novo synthesis of FA (Hackmann and Firkins, 2015), including BCFA derived from BCVFA; however, incorporation of more 18:1 isomers into membranes might lessen the need for BCFA to maintain membrane fluidity, including lessening the need for plasmalogens (Kaufman et al., 1988).

Experimental Design and Treatments
The experiment was a 2 × 2 × 2 factorial arrangement of 8 treatments of high (HF,67:33) or low (LF, 33:67) forage: concentrate, without or with supplemental 3% corn oil (CO), and without or with twice-daily dosed BCVFA (2.15 mmol/d each of isobutyrate, isovalerate, and 2-methylbutyrate).Dietary treatments, continuous culture conditions, and sampling conditions are described in Mitchell et al. (2023c) except that FA concentration and profile of the diet are presented in Table 1.Two Jersey cows were housed according to Institutional Animal Care and Use Committee standards at the Waterman Dairy in Columbus, Ohio.There were 8 anaerobic dual-flow continuous culture systems administered these 8 treatments in 4 periods, 12 d long, in a randomized incomplete block (n = 4).The buffer was infused to maintain fermenter pH to be comparable between HF and LF (Weller and Pilgrim, 1974).After background sampling (d 5) was completed, 5 mg/d of 13 C from each BCVFA were added (Mitchell et al., 2023c).The source of labeled 2-methylbutyrate is racemic.
Sample Collection and Processing.Effluents from the continuous cultures were collected on ice on d 9 to 12, and one 250-mL subsample was dosed with 8 mL of 5 N HCl to bring the pH down to 2 and stored at 4°C for 24 h before differential centrifugation to recover a bacterial sample that was enriched with particulateassociated bacteria, lyophilized, and pooled by fermenter, as described previously (Mitchell et al., 2023c).
Approximately 0.50 g of lyophilized effluent or feed samples and 0.10 to 0.20 g of bacteria pellet were hydrolyzed and methylated as described by Sukhija and Palmquist (1988) with the adaptations described by Jenkins (2010).In addition, 17:1 cis-10 heptadecenoic acid (Sigma-Aldrich) was the internal standard, replacing 17:0 or 19:0 as standards because bacteria also synthesize small amounts of these FA.With every period of bacteria pellet, 2 samples of octadecanal standards (18:0 ALD, Apollo Scientific) were also methylated because no such commercial DMA standards were available.
Separation of FAME and DMA.After methylation of bacterial pellets, FAME and DMA were separated by thin-layer chromatography (TLC) as outlined previously (Alves et al., 2013) with minor modifications.The addition of 75 mL of methylene chloride and a 1 h incubation period allowed for full saturation of the TLC chamber.Nitrogen was used to evaporate 0.5 mL of each sample.Samples were redissolved in 200 µL of methylene chloride and added to individually labeled lanes on glass TLC plates one drop at a time under a constant flow of N 2 .On each plate, lane 1 was cis-10-heptadecenoic acid methyl ester (17:1, Sigma-Aldrich), lanes 2 to 5 were samples, and the final lane was the DMA standard (Figure 1).The TLC plate was placed in the saturated chamber for approximately 30 min, allowing the solvent to travel 10 to 15 cm vertically on the plate.Once removed from the chamber, the TLC plate was dried with N 2 and sprayed with 0.1% (wt/vol) of 2′,7′-dichlorofluoroscein in methanol.An UV lamp at 254 nm was used to illuminate and identify the FAME and DMA spots for collection from lanes 2 to 6 on each plate.
The FAME and DMA for each sample were outlined in pencil and scraped from the TLC plate into individual collection tubes (Figure 1).Samples had 1.5 mL of methanol added and were sonicated for 10 min.Following the addition of 2 mL of hexane and 1.5 mL of 5% NaCl, samples were vortexed and centrifuged at 400 × g for 5 min at 4°C.The organic phase was removed from each tube and transferred to a new tube.With the addition of another 2 mL of hexane, centrifugation and organic phase collection were repeated.Solvent was evaporated under N 2 .The FAME samples were redissolved with 1 mL of hexane and added to GC vials.The DMA samples were redissolved with 0.2 mL of hexane with cis-10-heptadecenoic acid methyl ester (Sigma-Aldrich), which provided approximately 90 µg of methylated standard.The cis-10-heptadecenoic acid standard was pipetted into the DMA samples, including the DMA standard samples from lane 6, to serve as the internal standard to correct for injection variation.
The DMA samples were redissolved with only 0.2 mL of hexane to ensure the DMA was concentrated enough to allow for adequate peaks during GC analysis.
Gas Chromatography and Isotope-Ratio Mass Spectrometry Procedure.Effluent and feed FAME samples were injected into a Hewlett-Packard 5890 GC (Agilent Technologies) equipped with a SP-2560 capillary column (100 m × 0.25 mm × 2.0 µm).The carrier gas (He) flow was 20 cm/s with a 20:1 split, the inlet temperature was 280°C, and the detector temperature was 280°C.The initial oven temperature of 60°C was held for 4 min, and the temperature was then increased 25°C/min to 150°C and held for 20 min.The temperature was then raised to 160°C, 1°C/min, and held for 50 min, or until 18:1 isomers were eluted.The temperature was increased at 5°C/min to 240°C and was held for 20 min.
Bacterial FAME, DMA, and DMA standards were injected into a GC (Trace 1300; Thermo Fisher Scientific) with a SP-2560 capillary column (100 m × 0.25 mm × 2.0 µm) and equipped with isotope-ratio MS (Delta V Advantage; Thermo Fisher Scientific).Methyl esters were separated using He as a carrier at 20 cm/s.The injector temperature was 280°C with splitless mode; initial oven temperature was 60°C and held for 4 min, the temperature was then increased 50°C/min to 150°C and held for 20 min.The temperature was then raised to 170°C, 1°C/min, and held for 30 min.The final temperature increase was 5°C/min to 240°C and was held for 5 min.The sample was then combusted through a combustion reactor as described by Roman-Garcia et al. (2021a).Isotope-ratio MS was calibrated for 13 C enrichment using standards of FAME with known 13 C enrichment (n18M, USGS71, and USGS72; Reston Stable Isotope Laboratory).
Individual FAME were identified using the standard mix GLC-68D (Nu-Chek Prep Inc.), GLC-110 (Matreya LLC), and single FAME standards 18:1 t11 and 18:2 t10 c12 (Nu-Chek Prep Inc.).Standard mixes of Bacterial Acid Methyl Esters CP Mixture (Matreya LLC) and 37 Component FAME Mix (Sigma-Aldrich) were used for peak identification and FAME response factor calculations (area/concentration).There were 2 samples that were identified as having 16:0 and 18:0 FA in the DMA fraction; the TLC procedure was repeated for these samples, and FA were not detected.
Chromatograms from Molkentin and Precht (1995) and Alves et al. (2013) were used for the order of elution of 18:1 isomers and DMA, respectively.For DMA, there were no commercially available appropriate internal standards at the time we performed this research.Even though 18:0 ALD standard was available, this was an inappropriate internal standard because 18:0 ALD is present in bacteria and was influenced by dietary conditions (Alves et al., 2013;Ventto et al., 2017;Mannelli et al., 2018).Therefore, the 18:0 ALD standards were methylated separate from the bacteria samples and used for response factor calculations to account for DMA losses during methylation and as a recovery factor for the TLC separation procedure as shown for Lane 6 in Figure 2. First, the DMA standards were used to determine a response factor for 18:0 DMA (area/ concentration).The 18:0 ALD standard was not pure, but the 18:0 and 18:1 FA impurities were quantified, as described previously, and were mathematically removed in our calculations.Then the ratio of the response factor acquired for 18:0 DMA: 18:0 FAME was used as a conversion factor on the other FAME response factors to convert them to the corresponding DMA response factors.Samples were analyzed via GC-MS to confirm separation and correct spot identification for the TLC procedure by Lin Xi, North Carolina State University.
Additionally, in a separate study, branched-chain lipids were further confirmed as iso even-chain, iso oddchain, or anteiso lipids from bacterial pellets from dualflow continuous cultures that were individually dosed with 13 C labeled isobutyrate, isovalerate, 2-methylbutyrate, Val, Leu, and Ile (data not shown).The single dosage confirmed that BCALD identified as above were indeed derived (without crossover) from the respective parent compounds.
The methylation procedure adds 1 methyl group per FA and 2 methyl groups per ALD to form the respective FAME and DMA (Jackson et al., 2021).Therefore, FAME and ALD flows were converted to an unmethylated basis using their molecular weights (MW): FA flow = FAME mg/d × FAME MW/FA MW, and ALD flow = DMA mg/d × DMA MW/ALD MW Biohydrogenation index was calculated as described by Tice et al. (1994): 100 -{100 × [(18:1 Flow + (18:2 Flow × 2) + (18:3 Flow × 3)]/18 Flow )/[(18:1 Intake + (18:2 Intake × 2) + (18:3 Intake × 3)]/18 Intake }, where 18:1 is all C18 FA with 1 unsaturated bond, 18:2 is all C18 FA with 2 un- .Diagrammed in step 1, bacteria pellets and 18:0 aldehyde (ALD) standard were methylated with 300 µg of cis-10 heptadecenoic acid (17:1) internal standard (IS).In step 2, the FAME, DMA, and 18:0 DMA standard were separated using thin-layer chromatography (TLC).The FAME standard was only used for verification of the procedure; the spot in lane 1 was not collected.In step 3a, spots in lanes 2 to 6 were identified, scraped, and extracted from the silica.In step 3b, the FAME samples with 17:1 IS from lanes 2 to 5 were redissolved with 1 mL of hexane.In step 3c, the scraped DMA from lanes 2 to 5 and scraped DMA standard from lane 6 were redissolved with 0.2 mL of hexane that included 90 µg of cis-10 heptadecenoic methyl ester.The original 17:1 from methylation was recovered exclusively in the FAME band; therefore, this 17:1 FAME IS is necessary to account for injection error in step 3c.In step 4, the FAME and DMA samples were injected into the GC/isotope-ratio mass spectrometry.The 18:0 DMA standards and 17:1 IS were used to calculate a response factor (RF, area/concentration).The 18:0 ALD standard is not pure, but the 18:0 FA and 18:1 FA impurities were quantified, and the 18:0 ALD concentration was corrected accordingly.The RF for 18:0 FAME was calculated using standard mixes of bacterial acid methyl esters CP mixture and 37 component FAME mix.The RF for 18:0 ALD was scaled based on the same assumptions for all DMA (e.g., 18:0 FAME RF/18:0 DMA RF × 16:0 FAME RF = 16:0 DMA RF).The scraped 18:0 DMA standard (corrected for impurities) containing the 17:1 IS in step 3c was used to calculate a recovery factor for DMA from each TLC plate run [sample DMA concentration/(actual concentration/ expected concentration)].saturated bonds, 18:3 is all C18 FA with 3 unsaturated bonds, and 18 is the flows or intakes of total C18 FA isomers.Respective bacterial N flows were converted to FA and ALD flows based on the FA:N or ALD:N ratios of bacterial samples.Recovery of 13 C in bacterial lipid was calculated as the product of atom fraction excess [( 13 C atom percent-background 13 C atom percent)/100] and C flows of FAME or DMA, which were then divided by total 13 C dosed and multiplied by 100.The distribution of recovered dose was the 13 C recovered in FA and ALD/total 13 C recovered in the bacteria C outflows reported in Mitchell et al. (2023c).

Statistical Analysis
Daily FA flow, bacterial FA and ALD flow, and profile measurements were analyzed using PROC MIXED in SAS 9.4 (SAS Institute Inc.) according to the model described in Mitchell et al. (2023c).Briefly a mixed model had random effects of period and fermenter and fixed effects of diet, CO, BCVFA, and their interactions.Recovery of 13 C dose was analyzed with the same model, but an additional covariate was included because LF had a higher RDP than the HF diet and different branched-chain AA supply (Mitchell et al., 2023c), which would influence BCVFA pool sizes.Therefore, BCVFA production was included as the covariate for total dose recovery responses.
Extents of biohydrogenation of linolenic and linoleic acids increased (P ≤ 0.02) with CO supplementation and tended to decrease (P = 0.07 each) with BCVFA supplementation (Table 2).However, CO interacted (P = 0.03) and BCVFA tended (P = 0.10) to interact with diet.Biohydrogenation of linoleic acid increased more when CO was supplemented with LF diets.Biohydrogenation of linoleic acid decreased when BCVFA were supplemented to LF diets but did not influence biohydrogenation with HF treatments.Biohydrogenation index, which was calculated as a percentage of unsaturated bonds that were saturated in the vessels, increased (P < 0.01) with LF versus HF and with CO versus without supplemental CO.

Bacterial Fatty Acid Concentration and Profile
The FA concentrations of bacteria were reported on an OM basis because of ash contamination during bacterial harvesting (Table 3).The FA concentration tended to be lower (P = 0.10) with HF at 9.25% compared with LF at 10.7%, but bacterial FA flow was not influenced by any main effect or interactions.The weight percentage of even straight-chain FA (ECFA) decreased when CO was supplemented but to a greater degree in HF treatments (P = 0.01, diet × CO).Supplemental BCVFA also decreased (P < 0.01) ECFA profile.Both 12:0 and 14:0 profiles were lower (P ≤ 0.01) with LF versus HF.The weight percentage of 14:0 and 16:0 decreased (P ≤ 0.06) with BCVFA supplementation.Profile of 18:0 decreased (P = 0.05) with CO.Arachidic acid (20:0) decreased with CO supplementation but more with HF (P = 0.04, diet × CO).The profile of odd straightchain FA (OCFA) was lower (P < 0.01) with LF versus HF and decreased (P = 0.01) with CO versus without CO.Although we detected no interactions with OCFA, 15:0, which was the major OCFA, tended to decrease when CO was supplemented but more so with LF than with HF diets (P = 0.07, diet × CO).Additionally, 17:0 concentration was lower (P < 0.01 each) with LF versus HF and with CO versus without CO, and 17:0 tended to decrease (P = 0.09) with BCVFA versus without BCVFA.Also, 19:0 was lower (P < 0.01) with LF versus HF and tended to decrease (P = 0.06) with BCVFA versus without BCVFA.The total 18:1 isomers in the bacterial FA profile increased (P < 0.01 each) from 14.7% of total FA with HF to 18.5% with LF diets, from 14.7% without CO to 18.5% with CO, and from 15.4% without BCVFA to 17.8% with BCVFA.The major 18:1 isomers (18:1 trans-11 and 18:1 cis-9) also were greater (P ≤ 0.05) with LF versus HF, with CO versus without CO, and with BCVFA versus without BCVFA.The 18:1 trans-10 weight percentage did not experience any main effects; however, this isomer tended to increase when BCVFA were supplemented in HF diets but tended to decrease when BCVFA were supplemented with LF (P = 0.09, diet × BCVFA).The concentration of linoleic acid in bacterial FA tended to be greater (P = 0.09) with LF versus HF, increased (P = 0.03) with BCVFA versus without BCVFA, but was unaffected by CO.The total BCFA was greater (P < 0.01) with HF compared with LF (9.46 vs. 7.06%, Table 3).The total iso even-chain BCFA also was greater (P < 0.01) with HF at 1.72% of total FA versus LF at 1.09% and tended to decrease (P = 0.10) from 1.53% without CO to 1.28% with CO.The percentages of FA as 12:0 iso, 14:0 iso, and 16:0 iso were lower (P < 0.01) with LF compared with HF.Both 14:0 iso and 18:0 iso decreased (P ≤ 0.04) with CO supplementation.When BCVFA were supplemented, 12:0 iso increased (P = 0.01) and 18:0 iso tended to increase (P = 0.06) in the FA profile.The total anteiso-BCFA concentration tended to be lower (P = 0.06) with HF compared with LF (4.64 vs. 3.95%).The profiles of 13:0 anteiso and 17:0 anteiso were lower (P ≤ 0.03) with LF versus HF.The total concentration of iso odd-chain BCFA was lower (P < 0.01) with HF, 3.10%, compared with LF, 2.01%.The percentages of total FA as 15:0 iso and 17:0 iso were lower (P ≤ 0.01) with LF versus HF.Supplemental BCVFA tended to decrease (P < 0.10) 11:0 iso and decreased (P = 0.03) 17:0 iso concentration in bacterial FA.

Bacterial Fatty Aldehyde Concentration and Profile
The concentration of ALD in bacteria pellets and total ALD flow tended to have interactions between diet and CO (P ≤ 0.09, Table 4).When CO was supplemented with HF, ALD concentration and flow decreased by 20.1% and 23.6%, respectively, compared with HF -CO treatments, whereas when CO was supplemented with LF, bacterial ALD concentration and flow increased by 15.8% and 13.6, respectively, compared with LF -CO treatments.Overall, the sum of ALD that was straight even-chain ALD (ECALD) was greater (P < 0.01, main effect) with LF compared with HF treatments and tended to increase with supplemental BCVFA in HF -CO, HF + CO, and LF + CO diets but not with the LF -CO (P = 0.08, diet × CO × BCVFA).The profile of 14:0 ALD increased with BCVFA supplementation in HF -CO and LF + CO diets but decreased when supplemented with HF + CO and LF -CO (P = 0.02, diet × CO × BCVFA).Supplementation of BCVFA decreased 12:0 ALD but increased both 16:0 and 18:0 ALD in the profile with HF diets, whereas BCVFA with LF did not influence these percentages (P ≤ 0.10, diet × BCVFA).The percentage of 16:0 ALD increased (P = 0.02) with CO supplementation.The percentage that was straight odd-chain ALD tended to increase (P = 0.07) with LF compared with HF.Supplementation of BCVFA with HF increased 13:0 ALD percentage but decreased 15:0, whereas with LF treatments, 13:0 did not change and 15:0 ALD decreased with BCVFA supplementation (P ≤ 0.02, diet × BCVFA).The percentage of 15:0 ALD decreased (P = 0.03) with CO supplementation.Finally, 17:0 ALD increased (P ≤ 0.05) with LF versus HF and with CO versus without CO.
The total BCALD in the total bacterial ALD profile decreased (P = 0.01) from 55.4% with HF to 51.4% with LF diets (Table 4).The total iso even-chain BCALD in the profile decreased (P ≤ 0.03) by 12.2% with LF versus HF and by 8.03% with CO versus without CO.Supplementation of BCVFA increased iso 12:0 ALD in the profile with HF -CO and LF -CO but did not influence 12:0 iso ALD with HF + CO or LF + CO diets (P = 0.05, diet × CO × BCVFA).The percentage of 16:0 iso ALD tended to increase when BCVFA were supplemented to LF diets more than when supplemented to HF diets (P = 0.07, diet × BCVFA).The main effect of CO decreased (P = 0.04) 16:0 iso with CO versus without CO.The sum of unknown iso evenchain BCALD, which were ALD with enrichment that were confirmed with individually dosed 13 C labeled Val and isobutyrate (data not shown), was lower (P < 0.01) with LF versus HF.The sum of anteiso BCALD in bacterial ALD profile was not influenced by any main effects or interactions.The percentages of 11:0 anteiso and 13:0 anteiso decreased when BCVFA were supplemented with HF treatments, whereas 15:0 anteiso and 17:0 anteiso increased with BCVFA (P < 0.10, diet × BCVFA).However, BCVFA supplementation with LF 11:0 anteiso and 17:0 anteiso increased, but 13:0 anteiso and 15:0 anteiso-ALD percentages decreased.The sum of unknown anteiso BCALD, which were ALD with enrichment that were confirmed with individually dosed 13 C labeled Ile and 2-methylbutyarte (data not shown), were not influenced by any main effects or interactions.The total iso odd-chain BCALD in the profile was 11.1% with HF but was 8.95% with LF (P < 0.01).The weight percentages of 11:0 iso and 13:0 iso tended to decrease with BCVFA supplementation and HF diets but increased with LF diets (P ≤ 0.08, diet × BCVFA).With HF diets, the 11:0 iso tended to decrease when CO was supplemented, but this isomer was not affected when CO was added to LF diets (P = 0.09, diet × CO).The percentage of 15:0 iso was lower (P = 0.01) with LF versus HF.The profile of  8 Interaction of diet × BCVFA supplementation has (P ≤ 0.09).The interaction was P > 0.10 for all other data.9 Total branched-chain fatty acids.Even straight-chain aldehydes.

8
BCALD that were unknown but enriched in 13 C and confirmed as iso even-chain, iso odd-chain, or anteiso BCALD with individually dosed 13 C labeled sources (isobutyrate, isovalerate, 2-methylbutyrate, Val, Leu, and Ile; unpublished data) in dual-flow continuous cultures.17:0 iso ALD increased with BCVFA supplementation in HF -CO, HF + CO, and LF -CO treatments but decreased when supplemented with LF + CO and (P = 0.04, diet × CO × BCVFA).The sum of unknown iso odd-chain BCALD, which were ALD with enrichment that were confirmed with individually dosed 13 C labeled Leu and isovalerate (data not shown), tended to decrease when BCVFA were supplemented with HF diets, but increased with LF diets (P = 0.07, diet × BCVFA).

Isotope Recovery in Bacterial Lipids
The recovery of 13 C in bacterial lipid C flow was greater (P < 0.01, Table 5) by 42.2% with HF versus LF diets, but CO supplementation decreased recovery with HF but increased recovery with LF (P = 0.02, diet × CO).We observed no appreciable 13 C enrichment outside of branched-chain lipids.Recovery in total BCFA and anteiso odd-chain FA was higher (P = 0.01 each) with HF compared with LF by 48.3% and 42.0%, respectively.Dose recovery in iso even-chain and iso odd-chain BCFA tended to be higher (P ≤ 0.08) with HF compared with LF.Supplemental CO decreased recovery of total BCFA and anteiso odd-chain FA with HF treatments, but CO increased recovery with LF treatments (P = 0.03, diet × BCVFA).Similarly, we detected a tendency for this interaction for dose recovery in iso odd-chain BCFA (P = 0.10, diet × BCVFA).Dose recovery in total BCALD was higher (P = 0.01) and tended to be higher (P = 0.07) for iso even-chain BCALD with HF versus LF treatments by 29.0% and 44.3%, respectively.Supplementation of CO with HF treatments decreased recovery in total BCALD, whereas CO supplementation with LF increased dose recovery (P = 0.03, diet × BCVFA).The distribution of recovered dose in total BCFA and anteiso BCFA, which is the proportion of total 13 C recovered in bacteria pellets and reported in Mitchell et al. (2023c), decreased (P < 0.10, diet × CO) when CO was supplemented with HF but increased when CO was supplemented with LF.

Biohydrogenation
Decreasing forage and increasing PUFA increased biohydrogenation based on the calculation described by Tice et al. (1994), which is expressed as a percentage of unsaturated bonds in C18 FA.With additional ground corn and CO, we expected that biohydrogenation extent and intermediates of the alternate trans-10 pathway would increase because of the increased 18:2 cis-9,cis-12 supply (Duckett et al., 2002;Zhang et al., 2019).In our study, pH was similar between diets, whereas a lower pH with LF and more starch would have additively increased 18:1 trans-10 accumulation (Fuentes et al., 2009).
Our hypothesis that dosing BCVFA would increase the index of biohydrogenation of PUFA was not supported, and BCVFA actually decreased biohydrogenation of 18:2 and 18:3.Butyrivibrios, which biohydrogenate via the trans-11 pathway, are characterized as including BCFA in their membrane structure but also are proteolytic and could use exogenous branched-chain AA rather than synthesize them; the stearate producers represented by B. proteoclasticus have particularly high BCFA concentrations (Hackmann and Firkins, 2015).Previously, BCVFA and valerate supplementation did not increase biohydrogenation to 18:0 in vitro (Wu and Palmquist, 1991).A diet designed to depress milk fat increased 18:1 trans-10 and 18:2 trans-10,cis-12 percentages of rumen FA with isoacid supplementation (Lee et al., 2021), which is consistent with our results.However, the isoacid treatment still alleviated milk fat depression and increased milk concentration of de novo FA (Copelin et al., 2021).Milk fat concentration increased with 2-methylbutyrate and isobutyrate supplementation in multiparous cows (Mitchell et al., 2023b).The supplementation of BCVFA increased de novo fatty acid synthesis in the mammary gland, as supported by increased expression of mammary FA synthesis genes (Liu et al., 2018).Although those authors suggested that isoacid supplementation increased NDF digestibility and thereby acetate supply as a mechanism for increased de novo synthesis of FA in the mammary gland, NDF degradability did not increase in the study of Copelin et al. (2021), and acetate: propionate only numerically increased (Lee et al., 2021).In our companion study (Mitchell et al., 2023c), BCVFA stimulated NDF degradability but decreased acetate production, apparently because more acetyl-CoA was routed to support more bacterial growth.Moreover, CO supplementation also stimulated NDF degradability.Thus, further work is needed to understand potential interactions when PUFA are supplemented with BCVFA, potentially involving the FA and plasmalogen components of fibrolytic bacteria compared with carbon used for protein synthesis.
Although microbes have some ability to incorporate some FA into membranes, fat also can accumulate in lipid droplets primarily as free FA in particulate-phase bacteria (Bauchart et al., 1990).Though bacterial 18:1 FA concentrations did increase with CO, the bacteria FA percentage of OM was not affected, potentially because the linoleic acid dose was measured to be only an additional 0.353 g/d with HF + CO versus HF -CO but was 0.799 g/d greater with LF + CO versus LF -CO.Making small batches of pellets is C. The model also included a covariate (BCVFA, mmol/d production) for the differences between diet for RDP and feed AA profile. 3 The 13 C dose was provided as BCVFA (5 mg/d of 13 C from each isobutyrate, 2-methylbutyrate, and isovalerate). 4 Branched-chain lipids are branched-chain fatty acids (BCFA), which is the sum of iso even-chain, anteiso, and iso odd-chain fatty acids, and branched-chain aldehydes (BCALD); BCALD is the sum of iso even-chain, anteiso, and iso odd-chain fatty aldehydes.challenging, and we discarded crumbled pellets before feeding.We expect that our discarded crumbled pellets from the HF + CO diet had higher FA concentration and biased our pellets that were actually fed.The FA concentration in the diets from Bauchart et al. (1990) ranged from 1.90 to 18.2% FA, compared with 2.69 to 4.01% in this study.The FA supply provided in our diet was designed to be physiologically relevant yet at low enough inclusion to avoid unphysiological depressions in NDF degradability.Thus, our FA supply did not likely facilitate accumulation of lipid droplets in particulate-phase bacteria.The increased FA concentration with decreased forage: concentrate in this study are more likely to be the result of prokaryotic profile changes, especially major shifts from Bacteroidetes to Proteobacteria populations previously described in Mitchell et al. (2023c).

Bacterial Lipids
In this study, total ECFA in the bacterial profile did not change but shifted toward shorter chain ECFA with decreased forage: concentrate ratio.Longer chain FA (14:0, 15:0, and 18:0) decreased concomitantly with increased profile of shorter even-chain FA (12:0) with lower forage: concentrate (Alves et al., 2013) or decreased pH (Roman-Garcia et al., 2021b), likely as a response to decrease membrane fluidity (Or-Rashid et al., 2007).Other reports of forage: concentrate on OCFA have been mixed.Vlaeminck et al. (2006) theorized the varied results could be due to differing propionate supply, the primer for OCFA synthesis.We expected that decreasing forage would increase OCFA in the FA profile as the diet shifted the microbiome toward a population that prefers to use OCFA for membrane homeostasis rather than BCFA (Kaneda, 1991).However, in our study, 18:1 FA isomers replaced BCFA and OCFA in the bacterial FA profile in addition to decreasing ECFA chain length to maintain membrane fluidity with decreased forage: concentrate.
In addition to decreasing chain length of FA to fluidize membranes, BCFA in iso and particularly anteiso configurations spread the FA of phospholipids to increase fluidity of bacterial membranes (Parsons and Rock, 2013).Decreasing pH did not change total BCFA concentration because even though iso evenchain and iso odd-chain FA decreased with low pH, anteiso odd-chain FA increased (Roman-Garcia et al., 2021b).Alves et al. (2013) also reported increased iso FAME in rumen contents of lambs with increased forage (alfalfa):concentrate but decreased anteiso FAME with decreased forage: concentrate.They did not report ruminal pH in their results, but lambs fed the high alfalfa diet likely had higher pH than the higher concentrate diet.Additionally, the proportion of 15:0 anteiso in total odd branched-chain FA in bacteria was negatively correlated with forage inclusion (Vlaeminck et al., 2006).In our study, decreasing forage decreased all BCFA in the bacterial FA profile.Therefore, an increase in anteiso BCFA in bacteria probably is a specific adaptation to decreasing pH or to a shift in bacterial FA profile that was not replicated with only changing forage: concentrate without a change in pH.
Decreased forage: concentrate ratio decreased ECALD, especially 16:0 ALD, in our study.Similarly, 16:0 DMA (the methylated form of the ALD) decreased in the rumen DMA profiles reported by Alves et al. (2013) and Ventto et al. (2017).Both studies reported lower 13:0 DMA, 15:0 DMA, and total odd-chain DMA with decreased forage: concentrate.Their decrease contrasted with our study in which the shorter chain ECALD and longer chain odd-chain ALD replaced BCALD in the bacterial ALD profile to maintain membrane homeostasis.Using different combinations of ALD is likely a result of species preference for the BCVFA precursor for iso-and anteiso BCFA (Miyagawa et al., 1979).
With the confirmation that sn-1 FA are formed into phospholipids before conversion to plasmalogens (Jackson et al., 2021), we sought to understand if bacteria prefer specific FA for this conversion depending on their dietary or environmental conditions.Our results are generally consistent with increased BCALD in the ALD profile (or branched-chain DMA in the DMA profile as reported by others) with increased forage: concentrate (Alves et al., 2013;Ventto et al., 2017).Because Alves et al. (2013) also did not observe anteiso DMA being affected by forage: concentrate ratio, our work suggests a more critical role or preference for anteiso ALD.Ventto et al. (2017) recorded increased anteiso and iso DMA with increased forage: concentrate.Our BCALD were 50.0 to 56.3% of total ALD across treatments.Although DMA profiles ranged from 17.5 to 34% branched-chain DMA in previous reports (Alves et al., 2013;Ventto et al., 2017;Mannelli et al., 2018), those authors did not evaluate lipid profiles of isolated bacteria.Their analysis of rumen contents would include dietary, protozoal, and bacterial lipids, whereas our study only analyzed bacterial lipids because protozoa were not retained in the system.Dietary lipids would not influence the DMA profiles, but rumen protozoa do have plasmalogens in their membrane structure (Prins and Van Golde, 1976), perhaps from consumed bacteria, which could explain some differences between our results and the previously reported data.Though increasing forage: concentrate did increase BCALD, the increase in our study (~8% increase) is much smaller than the difference reported (~49% increase) in Alves et al. (2013), who varied forage: concentrate more extensively than in our study and did not maintain ruminal pH to be similar between diets.Therefore, inclusion of BCALD in the structure with increasing forage: concentrate may also be a response to increased pH, which will require a more fluid membrane (Roman-Garcia et al., 2021b).
Supplementation of CO only decreased total iso evenchain BCFA and BCALD in bacterial lipid profiles without any changes in total iso-and anteiso lipids.With 10% soybean oil supplementation, total DMA flow did not change, but 15:0 anteiso DMA in the rumen DMA profile increased (Alves et al., 2013), whereas total DMA omasal flow decreased with supplementation of 5% sunflower oil (Ventto et al., 2017).Both oil doses were much more than what was dosed in our study.We still expected that PUFA supplementation would inhibit cellulolytics and butyrivibrios, which would otherwise have relatively high branched-chain lipids in their membranes.However, as we discussed previously (Mitchell et al., 2023c), CO actually improved NDF digestibility, likely because of lower inclusion of CO under our conditions.Thus, without disruption, the greater bacterial uptake of cis-18:1 isomers from CO or its biohydrogenation likely increased fluidity, was sensed cellularly, and decreased need for BCFA.We can only speculate why only isobutyrate elongation was decreased because this response has not been reported in other papers.When dosed singly, both Leu and Ile inhibited growth of mixed rumen bacteria in vitro, whereas Val (produced from isobutyrate) did not (Kajikawa et al., 2002), perhaps because Val is readily excreted by many bacteria, but Leu and Ile are not (Stevenson, 1978).The critical need for 2-methylbutyrate for anteiso lipids will be explained subsequently, whereas Leu (and its precursor, isovalerate) appear to be prioritized for bacterial AA (Mitchell et al., 2023a).
Despite greater availability of BCVFA primers with supplementation, dosing BCVFA elicited minimal differences in the total inclusion of branched-chain lipids in bacterial membranes.Supplementation of BCVFA increased numerous 18:1 isomers in the bacterial FA profile and outflow but dosed BCVFA decreased the percentages of ECFA, particularly 16:0.Previously, Roman-Garcia et al. (2021b) reported decreased 13:0 anteiso and increased 14:0 iso with BCVFA supplementation, so the total BCFA did not change.Generally, the ALD profile shifted toward a more fluid profile, which could either be due to a preferential positioning of the higher fluidity FA in the sn-1 position of phospholipids or the preferential conversion of phospholipids with these FA to plasmalogens.Similarly, 14:0 iso and 15:0 anteiso BCFA were preferentially converted to plasmalogens (Alves et al., 2013;Ventto et al., 2017).Plasmalogens are more rigid than their phospholipid counterparts (Goldfine, 2010); therefore, inclusion of more BCALD and shorter ECALD in the structure compared with longer chain ALD may help moderate this stiffening response to maintain membrane fluidity.The role of plasmalogens in rumen microbes has not been defined, but plasmalogens do decrease permeability of biological membranes (Chen and Gross, 1994;Zeng et al., 1998).In humans, plasmalogens are common in numerous tissues and therefore have varying roles in addition to membrane structure, including protection against reactive oxygen species (Braverman and Moser, 2012).The reactive vinyl ether is considered responsible for this function (Dean and Lodhi, 2018).Sacrificial oxidation of plasmalogens might help prevent cellular damage, but the functional roles of plasmalogens still require further study (Jackson et al., 2021).

Isotope Recovery
The recovery of 13 C in bacterial FA total C flow did not differ by pH (Roman-Garcia et al., 2021b), but this recovery increased with greater forage inclusion in our study.Presumably, fibrolytics are more likely to incorporate BCVFA primers into bacterial components (Vlaeminck et al., 2006).However, increased pH also increased the relative abundance of fibrolytic bacteria in the prokaryotic profile but not the incorporation of BCVFA into bacterial components (Roman-Garcia et al., 2021b).Eubacterium cellulosolvens 5494 increased membrane fluidity by increasing medium-chain and 18:1 FA compared with 18:0 when grown on a medium with cellulose versus glucose, cellobiose or fructose (Moon and Anderson, 2001).With CO supplementation, recovery decreased in the high forage: concentrate, perhaps because the bacterial profile shifted toward a population that was more sensitive to PUFA.
Roman-Garcia et al. (2021b) used a shorter (60-m) column than ours (100-m) and reported notable isotope recovery in ECFA, primarily 16:0; however, our current results suggest that these conclusions were more likely a result of coelution of FAME that were much more abundant and thus overshadowed highly enriched DMA.We noted that the highly enriched anteiso-15:0 DMA can coelute with 14:0 FAME even on the longer column.Anteiso-15:0 ALD generally is abundant in numerous strains of ruminal bacteria (Miyagawa et al., 1979;Vinogradov et al., 2001).In the current study with the longer column, we detected no significant enrichment except for that in identified branched-chain lipids.Therefore, recovery in anteiso lipids must come from 13 C dosed as 2-methylbutyrate, iso even-chain from dosed isobutyrate, and iso odd-chain from dosed isovalerate.These minor DMA peaks were not initially discovered until the DMA sample was injected with a higher volume on the column, particularly for the very small peaks eluting between 12:0 and 14:0 iso DMA.These peaks were not identified in any other papers that analyzed DMA probably because those samples were from methylated rumen contents from which these minor compounds were likely further diluted by dietary FA below detection limits.
The 13 C recovery in iso odd-chain BCFA was greater than the recovery in iso even-chain BCFA, but recovery in iso even-chain BCALD was greater than the recovery in iso odd-chain BCALD.Thus, there appears to be preferential conversion of iso even-chain BCFA in phospholipids to the vinyl ethers in plasmalogens compared with conversion of iso odd-chain BCFA.Just as in our current study, Roman-Garcia et al. (2021b) also had greater 13 C recovery in anteiso BCFA than iso odd-chain or iso even-chain BCFA.In both studies, the dosed 2-methylbutyrate was a racemic mix; therefore, the requirement for 2-methylbutyrate for bacterial lipid synthesis apparently is even more significant than our isotope recoveries indicate because of stereo preference for the (S)enantiomer of anteiso odd-chain BCFA (Eibler et al., 2017).

CONCLUSIONS
Supplementation of BCVFA did not increase complete biohydrogenation of dietary FA or increase the total amount of BCFA or BCALD in bacterial membranes.Label was only recovered in BCFA and BCALD.Decreasing forage decreased the proportion of bacterial lipids that were BCFA and BCALD that decreased the incorporation of 13 C from supplemented BCVFA.Corn oil decreased iso even-chain branched-chain lipids in both the FA and ALD profiles, which indicated that isobutyrate was inhibited from incorporation into bacterial membrane structure with increasing PUFA.The greatest recovery of label was from anteiso BCFA and anteiso BCALD, indicating that 2-methylbutyrate was the BCVFA used the most as a primer for branchedchain lipids synthesis.Whereas isovalerate (iso oddchain) was incorporated more than isobutyrate (iso even-chain) for BCFA synthesis, isobutyrate was incorporated more in BCALD than isovalerate.Even though BCALD accounted for just ~6% of bacterial lipids, the recovery of isotope in BCALD was ~26% of the total dose in lipids due to the high percentage of ALD that were BCALD.Because BCFA and BCALD play an important role in the function and growth of bacteria, especially cellulolytics, BCVFA supplementation can support the rumen microbial consortium and increase fiber degradation across a variety of dietary conditions that require adaptation of bacterial membranes.

Figure 1 .
Figure 1.Example thin-layer chromatography plate used in the development and verification of the procedure for separation of FAME (red rectangle) and dimethylacetals (DMA, black rectangle) after methylation.Lanes 1 and 2 are different concentrations of cis-10 heptadecenoic acid.Lanes 3 and 4 contained only 18:0 aldehyde standard, which has minor contamination shown in the FAME band.Lane 5 was a mix of the 17:1 standard and 18:0 aldehyde.Lanes 6 and 7 were from methylated extracts from approximately 0.15 g of freeze-dried bacteria pellet.The faint DMA bands in lanes 6 and 7 are readily visualized under UV.All samples were methylated following the procedure described by Jenkins (2010) and then separated via thin-layer chromatography as described by Alves et al. (2013).

Figure 2 .
Figure2.Flowchart describing the procedure for methylation of fatty acids (FA) and vinyl ether in plasmalogens and the separation of FAME (red boxes) and dimethylacetals (DMA, black solid boxes).Diagrammed in step 1, bacteria pellets and 18:0 aldehyde (ALD) standard were methylated with 300 µg of cis-10 heptadecenoic acid (17:1) internal standard (IS).In step 2, the FAME, DMA, and 18:0 DMA standard were separated using thin-layer chromatography (TLC).The FAME standard was only used for verification of the procedure; the spot in lane 1 was not collected.In step 3a, spots in lanes 2 to 6 were identified, scraped, and extracted from the silica.In step 3b, the FAME samples with 17:1 IS from lanes 2 to 5 were redissolved with 1 mL of hexane.In step 3c, the scraped DMA from lanes 2 to 5 and scraped DMA standard from lane 6 were redissolved with 0.2 mL of hexane that included 90 µg of cis-10 heptadecenoic methyl ester.The original 17:1 from methylation was recovered exclusively in the FAME band; therefore, this 17:1 FAME IS is necessary to account for injection error in step 3c.In step 4, the FAME and DMA samples were injected into the GC/isotope-ratio mass spectrometry.The 18:0 DMA standards and 17:1 IS were used to calculate a response factor (RF, area/concentration).The 18:0 ALD standard is not pure, but the 18:0 FA and 18:1 FA impurities were quantified, and the 18:0 ALD concentration was corrected accordingly.The RF for 18:0 FAME was calculated using standard mixes of bacterial acid methyl esters CP mixture and 37 component FAME mix.The RF for 18:0 ALD was scaled based on the same assumptions for all DMA (e.g., 18:0 FAME RF/18:0 DMA RF × 16:0 FAME RF = 16:0 DMA RF).The scraped 18:0 DMA standard (corrected for impurities) containing the 17:1 IS in step 3c was used to calculate a recovery factor for DMA from each TLC plate run [sample DMA concentration/(actual concentration/ expected concentration)].

5
Percentage of recovered dose was calculated by 100 × (mg 13 C recovered in lipid C flow/mg 13 C recovered in bacterial total C flow).
Mitchell et al.: BRANCHED-CHAIN VOLATILE FATTY ACIDS AND BACTERIAL LIPID METABOLISM

Table 1 .
Mitchell et al.: BRANCHED-CHAIN VOLATILE FATTY ACIDS AND BACTERIAL LIPID METABOLISM Fatty acid composition of the diets varying in forage and corn oil supplementation 1Treatments are high forage (HF, 67% forage) or low forage (LF, 33% forage) and -CO (no additional supplemented fat) or + CO (3% DM supplemented fat as corn oil).

Table 3 (
Continued).Bacterial fatty acid (FA) content, flow rate, and FA profile in continuous cultures that were administered with either high-or low-forage diets that varied in corn oil and branched-chain VFA supplementation Mitchell et al.: BRANCHED-CHAIN VOLATILE FATTY ACIDS AND BACTERIAL LIPID METABOLISM

Table 4 .
Aldehyde (ALD) composition from bacteria harvested from continuous cultures that were administered with either high-or low-forage diets that varied in corn oil and branched-chain VFA supplementation 2P-values for main effect means for diet (HF vs. LF), CO, BCVFA, and interaction; interactions not shown have P > 0.10.3Interaction of diet × CO supplementation (0.07 ≤ P ≤ 0.10).4

Table 5 .
Mitchell et al.:BRANCHED-CHAIN VOLATILE FATTY ACIDS AND BACTERIAL LIPID METABOLISM Dose recovery in bacterial lipids in continuous cultures that were administered with either high-or low-forage diets that varied in corn oil supplementation Item for main effect means for diet (HF vs. LF), CO, and diet × CO interaction.Only the branched-chain VFA (BCVFA) treatments were dosed with 13 2P-values