Potassium carbonate improves milk quality by enhancing rumen metabolism in Holstein cows

Abstract

Mid-lactation is a key stage in dairy production characterized by high milk yields and metabolic stress in cows. Dietary potassium carbonate may enhance milk quality, but its response mechanisms, particularly the link between rumen microbiome changes and production performance, remain poorly understood. To address this knowledge gap, a total of 60 multiparous Holstein cows (parity = 2.47 ± 1.06, body weight = 601 ± 25 kg, and days in milk = 127.83 ± 31.91) were divided into four groups (n = 15 cows per group) using a randomized complete block design and fed the corresponding diets for 84 days. The feed treatments were as follows: a control group (CON, basal diet), a low dose group (LD, basal diet + 250 g/d K₂CO₃·1.5H₂O per head), a medium dose group (MD, basal diet + 500 g/d K₂CO₃·1.5H₂O per head), and a high dose group (HD, basal diet + 750 g/d K₂CO₃·1.5H₂O per head). The results showed that potassium carbonate supplementation significantly influenced rumen fermentation patterns, particularly by increasing acetate (P = 0.008) and isovalerate concentrations (P < 0.001). Milk fat (P = 0.004) and protein percentage (P = 0.006) exhibited the most pronounced effects in the MD group. The rumen microbiota and metabolome revealed significant alterations in microbial community structure and function. Notably, the results indicated that in the MD group, there was an increase in the abundance of Kyoto Encyclopedia of Genes and Genomes (KEGG) genes associated with crucial metabolic pathways: amino acid biosynthesis, long-chain fatty acid biosynthesis and fatty acid elongation pathways. These findings suggest that dietary supplementation with 500 g/d of potassium carbonate optimizes milk composition by modulating the rumen microbiota and associated metabolic pathways, supporting the potential for targeted nutritional strategies in dairy management.

1. Introduction

Mid-lactation is a critical stage in the dairy cow production cycle, characterized by sustained high milk yield but also significant metabolic stress and health challenges (Zhao et al., 2023). With the growing global demand for dairy products, optimizing nutritional strategies to enhance milk production performance and quality has become a key research focus in animal husbandry (Adesogan and Dahl, 2020). Improving milk quality not only enhances the market competitiveness of dairy products but also plays a crucial role in maintaining cow health and welfare (Martin et al., 2023). The use of feed additives has shown great potential in improving both milk yield and quality. Among these, potassium carbonate (K₂CO₃) has gained increasing attention as an effective nutritional regulator.

The application of K₂CO₃ in dairy cow diets is primarily based on its multiple biological functions (Nutrition, 2001). First, as an alkaline salt, K₂CO₃ effectively regulates rumen acid-base balance, helping to mitigate ruminal acidosis caused by high-concentrate diets (West et al., 1986). Second, K₂CO₃ has been shown to enhance heat stress tolerance in dairy cows, especially in hot summer conditions when reduced milk quality and health issues are common (Nzeyimana et al., 2023). Additionally, previous studies have demonstrated the positive effects of K₂CO₃ on improving dairy cow production performance. A study showed that dairy cows fed diets containing 1.5% to 2.1% potassium exhibited higher milk yield (Mallonée et al., 1985). Moreover, in a study by Catterton and Erdman (2016), K₂CO₃ supplementation has been shown to elevate ruminal acetate levels, which serve as crucial precursors for milk fat synthesis (Zhao et al., 2021). Subsequent research has revealed that cows fed K₂CO₃ had higher milk fat percentages than controls (West et al., 1987). The specific mechanisms underlying its impact on milk quality during mid-lactation showed that rumen lipid biohydrogenation typically converts linoleic and linolenic acids into cis-9,trans-11 conjugated linoleic acid and trans-11 18:1. However, this pathway can shift to produce trans-10,cis-12 conjugated linoleic acid—a known suppressor of milk fat synthesis (Baumgard et al., 2000). Research revealed that supplemental potassium increased beneficial cis-9, trans-11 conjugated linoleic acid and trans-11 18:1, while reducing production of milk fat-depressing trans-10, cis-12 conjugated linoleic acid and trans-10 18:1 (Harrison et al., 2012), explaining the 0.4% higher milk fat observed in cows supplemented with K₂CO₃ compared to controls (Jenkins et al., 2014).

Milk fat and protein content are crucial indicators for assessing milk quality, and their synthesis is closely associated with rumen microbial metabolism. Research has shown that the diversity and functional status of rumen microbial communities directly influence the generation of milk components (Xue et al., 2020). Therefore, investigating the effects of K₂CO₃ on rumen microbial communities and their metabolic functions is of significant importance for elucidating the mechanisms by which it improves milk quality (Jenkins et al., 2014). In recent years, the rapid advancement of metagenomics and metabolomics technologies has enabled researchers to delve deeper into the regulatory effects of K₂CO₃ on rumen microbial communities and their metabolic pathways. Although previous studies have confirmed that K₂CO₃ can improve dairy cow productivity and speculated that it may exert its effects by modulating the rumen microbiota (Harrison et al., 2012), the specific regulatory mechanisms remain to be further investigated. Therefore, investigating the effects of K₂CO₃ on rumen microbial communities and their metabolic functions is of great significance for elucidating the mechanisms by which it improves milk quality (Jenkins et al., 2014). Against this backdrop, the present study aims to investigate the impact of K₂CO₃ supplementation in the diet of mid-lactation dairy cows on milk quality and its underlying mechanisms, and further to determine the optimal supplementation dosage. By analyzing the effects of K₂CO₃ on rumen microbial community structure, metabolic functions, and milk component synthesis, this research will provide a theoretical foundation for optimizing nutritional regulation strategies for mid-lactation dairy cows, as well as practical guidance for improving dairy product quality and cow health.

2. Materials and methods

2.1. Animal ethics statement

All experimental procedures involving animals used in this study were approved by the experimental animal welfare and animal ethics committee of China Agricultural University (protocol number: AW60105202-1-03).

2.2. Cows and experimental design

The experiment was conducted at the Shounong Farm (Beijing, China). The experiment was conducted with 60 multiparous lactating Holstein cows (601 ± 25 kg). The parity and days in milk were 2.47 ± 1.06 and 127.83 ± 31.91 days, respectively. The experiment was conducted as a randomized complete block design, with cows allocated to four treatment groups (n = 15 cows per group). The feed ingredients and nutrient levels are listed in Table 1. Treatments were as follows: (1) control diet (CON), (2) control diet supplemented with K₂CO₃ at 250 g/d per head (LD), (3) control diet supplemented with K₂CO₃ at 500 g/d per head (MD), and (4) control diet supplemented with K₂CO₃ at 750 g/d per head (HD). A study has shown that dietary supplementation of potassium inhibited the absorption of magnesium (Jittakhot et al., 2004). Therefore, each cow was supplemented with 50 g of magnesium oxide per day throughout the trial period. The cows were housed in a free stall barn for a 15-d adaptation period followed by an 84-d experimental period.

Table 1.

Ingredients and nutrient levels of the mid-lactation TMR (dry matter basis,%).

Items	Groups1
	CON	LD	MD	HD
Ingredients
Steam-flaked corn	8.56	8.47	8.38	8.29
High yield cow concentrate2	37.76	37.24	36.86	36.48
Spraying corn husk	1.14	1.13	1.12	1.11
DDGS	1.14	1.13	1.12	1.11
Soybean husk	1.14	1.13	1.12	1.11
Cottonseed	6.67	6.60	6.54	6.45
Beet granules	1.52	1.51	1.49	1.48
Corn silage	30.50	30.08	29.76	29.46
Alfalfa silage	1.25	1.24	1.22	1.21
Alfalfa hay	6.48	6.41	6.35	6.28
Oat hay	1.91	1.89	1.87	1.85
Fatty powder	0.69	0.68	0.67	0.67
Fatty acid calcium	1.24	1.22	1.21	1.20
Potassium carbonate	0.00	1.05	2.07	3.08
Magnesium oxide	0.00	0.22	0.22	0.22
Total	100.00	100.00	100.00	100.00
Nutrient levels3
CP	16.55	16.45	16.30	16.21
EE	5.13	5.10	5.09	5.06
NDF	32.18	31.80	31.20	31.25
ADF	16.31	16.15	16.02	15.90
OM	92.50	91.71	90.16	89.47
Ca	0.73	0.73	0.68	0.67
K	1.33	1.84	2.47	3.01
P	0.42	0.41	0.38	0.38
NEL, Mcal/kg	1.74	1.73	1.73	1.72

TMR = total mixed ration; DDGS = distillers dried grains with solubles; CP = crude protein; EE = ether extract; NDF = neutral detergent fiber; ADF = acid detergent fiber; Ca = calcium; K = kalium; P = phosphorus; NE_{L =} net energy for lactation; OM = organic matter.

¹CON, the control group (basal diet); LD, the low dose group (basal diet + 250 g/d K₂CO₃·1.5H₂O per head); MD, the medium dose group (basal diet + 500 g/d K₂CO₃·1.5H₂O per head); HD, the high dose group (basal diet + 750 g/d K₂CO₃·1.5H₂O per head).

²Main ingredients of the high-yield cow concentrate: ground corn (41.0%), soybean meal (23.0%), canola/rapeseed meal (8.0%), corn gluten feed (5.0%), wheat bran (10.0%), beet pulp pellets (7.0%), molasses (2.0%), limestone (CaCO₃; 1.5%), dicalcium phosphate (DCP; 1.0%), sodium chloride (NaCl; 0.5%), and a vitamin–trace mineral premix (1.0%). The vitamin–trace mineral premix was formulated to contain the following (perkilogram of premix): vitamin A, 600 KIU; vitamin D₃, 150 KIU; vitamin E, 5 KIU; niacin, 2 g; Ca, 176.5 g; P, 85 g; Zn, 2480 mg; Mn, 2260 mg; Cu, 955.5 mg; I, 45 mg; Se, 30 mg.

³NE_L was calculated according to the NRC (2001).

2.3. Sampling and measurements

The respiratory rate was estimated by observing thoracoabdominal movements for 1 min and expressed in breaths/min. The rectal temperature was measured by inserting a digital thermometer (VT-1831; Microlife AG, Widnau, St. Gallen, Switzerland) to the depth of approximately 3 cm into the rectum of the cow and the result was read once the beep sounded. Feed intake and milk production were monitored using the Roughage Intake Control system (RIC; Insentec B.V., Marknesse, Flevoland, the Netherlands) on a daily basis and the ALPROTM system (DeLaval International AB, Tumba, Stockholm County, Sweden) at each milking, respectively. Total mixed ration (TMR) samples were collected weekly and dried at 65 °C for 72 h to determine dry matter intake (DMI) per cow. The dried samples were ground in a feed mill with a 1-mm screen for nutrient analysis. Feed components, including dry matter, crude protein, ether extract, and organic matter, were analyzed according to the Association of Official Analytical Chemists (AOAC, 2012): methods 930.15, 984.12, 920.39, and 924.05, respectively. Neutral detergent fibre (NDF) and acid detergent fibre (ADF) contents were measured using Van Soest et al.'s methods (Van Soest et al., 1991). Additionally, potassium, calcium, and phosphorus levels were quantified using inductively coupled plasma optical emission spectrometry (ICP-OES; iCAP 7400, Thermo Fisher, Waltham, MA, USA) following AOAC (2012): methods 985.01, 991.25 and 984.27, respectively. Individual milk samples were collected daily during three consecutive milkings and mixed at a ratio of 4:3:3 on d 84 and analyzed for milk fat, protein, and lactose contents. Blood samples were collected from the tail vein before morning feeding and then centrifuged for 20 min at 3000 × g and 4 °C, and serum was stored at −20 °C until further analysis of biochemical indicators. The levels of glucose (GLU), total cholesterol (TC), triglycerides (TG), aspartate aminotransferase (AST), and alanine aminotransferase (ALT), total protein (TP) and urea nitrogen (UN), β-hydroxybutyric acid (BHBA), non-esterified fatty acids (NEFA) were determined using an automatic biochemistry analyzer (CLS880, Jiangsu Zecheng Biotechnology Co., Ltd., Taizhou, Jiangsu, China), following the protocols of commercial kits (Nanjing Jiancheng Bioengineering Institute Co., Ltd., Nanjing, Jiangsu, China). Immunoglobulin A (IgA), immunoglobulin G (IgG), immunoglobulin M (IgM) and tumor necrosis factor-alpha (TNF-α) were measured using commercial enzyme-linked immunosorbent assay (Shanghai Enzyme-linked Biotechnology Co., Ltd., Shanghai, China). For the assessment of cows' complete blood count index, blood samples were also taken into ethylenediaminetetraacetic acid vacutainer tubes for processing. The blood routine analysis was performed using an automatic hematology analyzer (BC-2800, Shenzhen Mindray Bio-Medical Electronics Co., Ltd., Shenzhen, Guangdong, China). Rumen fluid samples were collected using an oral stomach tube 2 h after morning feeding at 07:00 on d 84 (Shen et al., 2012). The ruminal fluid pH was promptly assessed with a digital pH meter (Sartorius PB-10, Beijing Sartorius Instrument Systems Ltd., Beijing, China) after filtering through four layers of sterile cheesecloth. The remaining rumen fluid was immediately frozen in liquid nitrogen for the subsequent determination of volatile fatty acid (VFA), metagenome sequencing and metabolomics analyses. The VFA were measured using gas chromatograph (6890N, Agilent Technologies, Inc., Santa Clara, CA, USA) following the methods described by Cao et al. (2008).

2.4. DNA extraction, metagenome sequencing, and data processing

Six cows from each group were randomly chosen for additional metagenomic analysis. DNA quality and concentration were evaluated using a Qubit 4 Fluorometer (Qubit 4, Thermo Fisher Scientific, Inc., Waltham, MA, USA) after extracting from each rumen fluid sample. Metagenomic shotgun sequencing libraries with an average insert size of approximately 400 bp were constructed using the Illumina TruSeq Nano DNA LT Library Preparation Kit (Illumina, Inc., San Diego, CA, USA). Metagenomic sequencing was conducted using Illumina TruSeq Nano DNA LT Library Preparation Kit (Illumina, Inc., San Diego, CA, USA) in Personalbio Company (Shanghai, China). For metagenomic data processing, the quality control of each metagenomic sequence reads was performed. The sequencing adapters, low-quality reads and host genome of cows were removed from sequencing reads using Cutadapt (v1.2.1) and Minimap2 (v2.24-r1122), respectively (Martin, 2011; Li, 2018). In addition, low-quality reads were trimmed with a sliding-window algorithm using fastp (v0.23.2) (Chen et al., 2018). After obtaining quality-filtered reads, Kaiju (v1.9.0) was used to classify metagenomic reads against a GTDB-derived database (v207) for each sample, performing taxonomical classifications. Reads annotated as Metazoa or Viridiplantae were removed. Next, the metagenomic assembly was performed using Megahit (v1.1.2). Subsequently, all generated contigs were clustered using Mmseqs2 (v15) in ‘easy-linclust’ mode with a sequence identity threshold of 0.95 and 90% coverage of the shorter contig. For the overlapping sequences, Prodigal (V2.6.3) was used to identify the open reading frames and predict the coding regions, thus obtaining the corresponding gene sequence files, protein sequence files, and protein sequence files (Hyatt et al., 2010). Coding DNA sequences were clustered using Mmseqs2 (v15) in ‘easy-cluster’ mode with a protein sequence identity threshold of 0.95 and 90% coverage of shorter sequences. Subsequently, the reads were aligned to the predicted gene sequences using Minimap2 (v2.27), followed by read quantification with featureCounts (v2.0.1), calculated as transcripts per kilobase per million mapped reads (TPM) (Liao et al., 2014). The resulting non-redundant gene set was functionally annotated using Mmseqs2's ‘search’ mode against the Kyoto Encyclopedia of Genes and Genomes (KEGG) and Carbohydrate-Active enzymes (CAZy) protein databases (Wang et al., 2025).

2.5. Rumen fluid untargeted metabolomics analyses

The cows used for rumen microbiome metagenomic analysis were also subjected to rumen fluid untargeted metabolomics analyses. Firstly, the cow rumen fluid, which had been slowly thawed at 4 °C, was combined with a precooled methanol-acetonitrile-water solution (2:1:1, v/v). The mixture was vortex - oscillated, followed by ultrasonication at a low temperature for 30 min. The mixture was vortexed, subjected to ultrasonication at low temperature for 30 min, incubated at −20 °C for 10 min, and then centrifuged at 14,000 × g and 4 °C for 20 min. Analyses were performed using Vanquish ultrahigh pressure liquid chromatography (UHPLC) (Thermo Fisher Scientific, Inc., Waltham, MA, USA) coupled with an Orbitrap Exploris 120 mass spectrometry (Thermo Fisher Scientific, Inc., Waltham, MA, USA). Next, the Thermo Orbitrap Exploris 120 mass spectrometer (Thermo Fisher Scientific, Inc., Waltham, MA, USA) was utilized to acquire data dependent acquisition mass spectrometric data in both positive and negative ion modes. The raw data were first converted into the. mzXML format using ProteoWizard, and then peak alignment, retention time correction, and peak area extraction were performed with the XCMS software (The Scripps Research Institute, La Jolla, CA, USA). Subsequently, metabolite structure identification and data preprocessing were carried out. Finally, the total peak intensity data were normalized. The principal component analysis (PCA) was used to determine the global metabolic differences among the four groups. Statistical significance among species was declared at variable importance in projection (VIP) value > 1 and P-value <0.05 (Kong et al., 2025).

2.6. Statistical analysis

All data are expressed as mean and standard error of the mean (SEM). One-way analysis of variance (ANOVA) followed by post-hoc Duncan's test was performed to compare the differences between all experimental groups, using a SPSS 23.0 software (IBM Corporation, Armonk, NY, USA). The Kruskal–Wallis test was employed to assess differences in microbial composition at the phylum, genus, and species levels. A value of P < 0.05 was regarded as statistically significant. To explore the interaction between rumen fluid microbiota and metabolic biomarkers, spearman's correlation analysis was carried out. To examine the effects of varying dietary K₂CO₃ levels, both linear and quadratic regression analyses were employed.

The linear regression model is as follows:

$$Y={\beta }_0+{\beta }_{1}X+ϵ\text{;}$$

The quadratic regression model is as follows:

$$Y={\beta }_0+{\beta }_{1}X+{\beta }_2{X}^2+ϵ\text{.}$$

where Y represents the observation of the dependent variable; β₀ represents constant/intercept; X represents the independent variable; β₁ and β₂ represent regression coefficients; ϵ represents random error.

3. Results

3.1. Dry matter intake and milk composition

Table 2 lists the effects of supplementing K₂CO₃ on DMI and milk composition. The results showed that even though K₂CO₃ did not alter the DMI and milk production (P > 0.05), the milk composition was significantly affected. The milk protein yield was increased in MD compared with CON while the HD was significantly decreased compared with LD and MD cows (P = 0.014). In addition, the percentage of milk fat and milk protein have increased compared to the CON group (P < 0.05).

Table 2.

Dry matter intake and milk production responses of lactating cows.

Items	Groups1				SEM	P-value
	CON	LD	MD	HD		ANOVA	Linear	Quadratic
DMI, kg/d	22.94	23.52	22.77	21.93	0.378	0.529	0.259	0.336
Rectal temperature, °C	38.80a	38.14ab	38.63ab	37.99b	0.119	0.038	0.061	0.178
Respiratory rate, time/min	53.90	54.60	50.89	46.00	1.92	0.365	0.086	0.171
Milk yield, kg/d	37.48	37.98	36.99	36.41	0.589	0.814	0.429	0.661
3.5% FCM, kg/d2	38.12	40.79	40.30	39.05	1.596	0.339	0.656	0.206
ECM, kg/d3	37.98	40.66	40.52	38.55	1.479	0.178	0.744	0.086
Feed efficiency4	1.65	1.73	1.78	1.77	0.089	0.481	0.151	0.288
Milk composition yield, kg/d
Fat	1.35	1.50	1.50	1.44	0.023	0.065	0.205	0.022
Protein	1.18bc	1.26ab	1.29a	1.15c	0.018	0.014	0.753	0.004
Lactose	1.69	1.80	1.79	1.69	0.030	0.361	0.997	0.164
Milk composition, %
Fat	3.60b	3.95a	4.06a	3.95a	0.034	0.004	0.006	0.002
Protein	3.15b	3.32ab	3.49a	3.16b	0.104	0.006	0.316	0.002
Lactose	4.98	5.04	5.05	5.07	0.020	0.476	0.138	0.300
MUN, mg/dL	12.74	12.81	13.01	14.05	0.21	0.108	0.030	0.050
SCC, × 103/mL	171.40	100.53	121.54	112.08	16.23	0.410	0.270	0.333

DMI = dry matter intake; FCM = fat-corrected milk; ECM = energy -corrected milk; MUN = milk urea nitrogen; SCC = somatic cell count.

Superscripts differing across the same row denote significant differences (P < 0.05).

¹CON, the control group (basal diet); LD, the low dose group (basal diet + 250 g/d K₂CO₃·1.5H₂O per head); MD, the medium dose group (basal diet + 500 g/d K₂CO₃·1.5H₂O per head); HD, the high dose group (basal diet + 750 g/d K₂CO₃·1.5H₂O per head).

²3.5% FCM = 0.432 × Milk yield +16.216 × Fat yield.

³ECM = 12.96 × Fat yield +7.04 × Protein yield +0.3246 × Milk yield.

⁴Feed efficiency = ECM/DMI.

3.2. Complete blood count indexes and serum biochemical parameters

The content of white blood cell count and neutrophils of LD cows were higher compared to the CON group though not significantly (P > 0.05) (Table 3). Interestingly, the MD and HD groups had significantly lower white blood cell count (P = 0.002) and neutrophils levels (P = 0.040) compared to the LD group and also lower than the control group (P > 0.05). Table 3 also showed that supplementing K₂CO₃ affects serum biochemical parameters. The content of serum urea nitrogen was reduced (P = 0.029) in the MD group compared with CON, LD, and HD groups. The NEFA level in the serum showed no significant differences among the CON, LD, and MD groups (P > 0.05); however, the HD group markedly increased the NEFA content in comparison with those in the other three groups (P = 0.032). In addition, the TNF-α content in the serum in the MD group was significantly lower than the CON group (P = 0.031).

Table 3.

Effects of K₂CO₃ on complete blood count indexes and serum biochemical parameters of lactating cows.

Items	Groups1				SEM	P-value
	CON	LD	MD	HD		ANOVA	Linear	Quadratic
Complete blood count indexes
RBC, × 1012/L	5.91	6.17	5.95	6.18	0.085	0.561	0.463	0.763
WBC, × 109/L	8.83ab	10.22a	7.40b	7.51b	0.319	0.002	0.016	0.032
HGB, g/L	94.80	95.50	94.30	96.67	1.108	0.900	0.683	0.863
PLT, × 109/L	371.60	398.30	305.20	351.56	18.925	0.356	0.360	0.643
Neutrophils, × 109/L	4.03ab	4.79a	3.37b	3.11b	0.230	0.040	0.044	0.068
Lymphocytes, × 109/L	2.77	3.18	2.45	2.57	0.113	0.100	0.196	0.347
Monocytes, × 109/L	1.82	1.82	1.36	1.61	0.079	0.108	0.117	0.221
Eosinophils, × 109/L	0.14	0.33	0.16	0.15	0.028	0.052	0.658	0.198
Basophils, × 109/L	0.08	0.10	0.06	0.07	0.005	0.064	0.064	0.177
Serum biochemical parameters
GLU, mmol/L	3.73	3.81	3.88	3.70	0.056	0.671	0.951	0.513
TC, mmol/L	6.60	7.22	6.56	7.48	0.349	0.759	0.540	0.815
TG, mmol/L	0.12	0.13	0.12	0.13	0.006	0.870	0.460	0.765
ALT, U/L	21.51	23.24	19.51	20.30	0.775	0.368	0.299	0.565
AST, U/L	74.34	62.66	58.60	58.70	4.417	0.577	0.204	0.364
TP, g/L	71.60	67.89	73.57	71.20	0.823	0.095	0.555	0.782
UN, mmol/L	4.05a	4.39a	3.38b	4.39a	0.144	0.029	0.988	0.539
NEFA, μmol/mL	0.38b	0.38b	0.36b	0.46a	0.014	0.032	0.060	0.022
BHBA, mmol/L	0.86	0.91	0.85	0.89	0.020	0.789	0.894	0.982
IgG, g/L	11.49	11.63	12.06	11.75	0.149	0.612	0.380	0.528
IgA, g/L	1.13	1.11	1.19	1.16	0.016	0.359	0.228	0.490
IgM, g/L	3.04	3.08	3.19	3.18	0.049	0.652	0.229	0.478
TNF-α, ng/L	297.38a	256.06ab	215.93b	266.00ab	10.225	0.031	0.007	0.028

RBC = red blood cell count; WBC = white blood cell count; HGB = hemoglobin; PLT = platelet; GLU = glucose; TC, =total cholesterol; TG = triglycerides; ALT = alanine aminotransferase; AST = aspartate aminotransferase; TP = total protein; UN = urea nitrogen; NEFA = non-esterified fatty acids; BHBA = beta-hydroxybutyrate; IgG = immunoglobulin G; IgA = immunoglobulin A; IgM = immunoglobulin M; TNF-α = tumor necrosis factor-alpha.

Superscripts differing across the same row denote significant differences (P < 0.05).

¹CON, the control group (basal diet); LD, the low dose group (basal diet + 250 g/d K₂CO₃·1.5H₂O per head); MD, the medium dose group (basal diet + 500 g/d K₂CO₃·1.5H₂O per head); HD, the high dose group (basal diet + 750 g/d K₂CO₃·1.5H₂O per head).

3.3. Rumen fermentation parameters

Table 4 shows treatments had no effect (P > 0.05) on rumen pH value and total VFA concentration. However, the acetate: propionate ratio was significantly increased (P < 0.001) in LD, MD, and HD groups, due to the increased content of the acetate (P = 0.008), compared with CON. Similarly, the HD group significantly increased the concentration of butyrate (P = 0.007), and the content of isovalerate in the HD and MD groups was also significantly higher than in the CON and LD groups (P < 0.001).

Table 4.

Effects of K₂CO₃ on rumen fermentation parameters of lactating cows.

Items	Groups1				SEM	P-value
	CON	LD	MD	HD		ANOVA	Linear	Quadratic
pH	6.36	6.49	6.53	6.54	0.029	0.094	0.021	0.041
Acetate, mmol/L	54.96b	60.52a	61.97a	63.22a	0.990	0.008	0.001	0.003
Propionate, mmol/L	33.60	34.17	35.97	33.28	0.467	0.172	0.035	0.080
Isobutyrate, mmol/L	1.56	1.33	1.48	1.38	0.071	0.697	0.552	0.764
Butyrate, mmol/L	14.06b	14.55b	15.24b	18.87a	0.589	0.007	0.002	0.002
Isovalerate, mmol/L	1.45b	1.66b	2.27a	2.37a	0.105	<0.001	<0.001	<0.001
Valerate, mmol/L	1.65	1.83	1.83	1.91	0.077	0.692	0.255	0.504
Total VFA, mmol/L	109.96	114.04	116.39	121.02	1.549	0.072	0.007	0.029
Acetate:propionate	1.53b	1.77a	1.85a	1.91a	0.038	<0.001	<0.001	<0.001

VFA = volatile fatty acids.

Superscripts differing across the same row denote significant differences (P < 0.05).

¹CON, the control group (basal diet); LD, the low dose group (basal diet + 250 g/d K₂CO₃·1.5H₂O per head); MD, the medium dose group (basal diet + 500 g/d K₂CO₃·1.5H₂O per head); HD, the high dose group (basal diet + 750 g/d K₂CO₃·1.5H₂O per head).

3.4. Metabolic profile in the rumen of cows in different groups

Next, the non-targeted metabolomic analysis on the six rumen fluid samples of cows that underwent metagenomic sequencing. A total of 808 and 537 metabolites were detected under positive and negative ion mode conditions, respectively (Fig. 1A). The principal coordinate analysis showed the obvious clustering of metabolite composition among the groups (Fig. 1B). To preliminarily visualize the difference in the distribution of metabolites among the CON, LD, MD, and HD groups, metabolites with P-value <0.05, and VIP >1.0 were defined as differential. Compared to the CON group, 14, 38, and 35 differential metabolites were screened in the LD, MD, and HD groups, respectively (Fig. 1C). Furthermore, the comparative analysis of the control group and the other three groups was shown in Fig. 2. The results indicated that eight common differential metabolites were identified among the three groups, they belong to organoheterocyclic compounds, lipids and lipid-like molecules, shikimates and phenylpropanoids, carbohydrates, nucleosides, nucleotides, and analogues, and benzenoids. Additionally, the number of unique differential metabolites between the LD, MD, and HD groups compared with the CON group were 0, 20, and 19, respectively. Notably, the MD group exhibited a marked increase in ruminal lipid metabolites, such as PC(16:0/16:0) and PC(18:2(9Z,12Z)/P-16:0).

Fig. 1.

Effects of K₂CO₃ on rumen metabolome. (A) The number of identified metabolites in the positive- and negative-ion modes. (B) Comparison of rumen metabolome was visualized using the principal component analysis (PCA). (C) The number of different metabolites in treatment groups LD, MD, and HD compared to the CON group. CON, the control group (basal diet); LD, the low dose group (basal diet + 250 g/d K₂CO₃·1.5H₂O per head); MD, the medium dose group (basal diet + 500 g/d K₂CO₃·1.5H₂O per head); HD, the high dose group (basal diet + 750 g/d K₂CO₃·1.5H₂O per head).

Fig. 2.

Effects of K₂CO₃ on ruminal metabolites. The alteration in particular rumen metabolites among groups. FC = fold change. CON, the control group (basal diet); LD, the low dose group (basal diet + 250 g/d K₂CO₃·1.5H₂O per head); MD, the medium dose group (basal diet + 500 g/d K₂CO₃·1.5H₂O per head); HD, the high dose group (basal diet + 750 g/d K₂CO₃·1.5H₂O per head).

3.5. Compositional profiles of the rumen microbiome and taxonomic differences among cows in four groups

The rumen microbiome of the four groups were compared in terms of microbial domains. First, the alpha diversity among groups showed no differences (P > 0.05) in these measures (Fig. 3A). The principal coordinates analysis (PCoA) plot results also didn't show a significant differentiation (Fig. 3B). The dominant bacterial phyla were Bacteroidota, Firmicutes A, and Firmicutes. Notably, the abundance of Bacteroidota exhibited significant differences, being higher in the rumen of the HD group compared to the CON group (P = 0.038; Fig. 3C). The dominant bacterial genus was Prevotella, followed by Ruminococcus E, Cryptobacteroides, and Sodaliphilus. At the genus level, the abundance of UBA2862 was significantly higher in the rumen of MD cows than in CON, LD, and HD cows (P = 0.046), and the abundance of UBA4372 was also significantly higher in the rumen of the MD and HD cows than CON cows (P = 0.011). The Cryptobacteroides was more abundant in the HD cows than in CON cows (P = 0.040) (Fig. 4A; Table S1). At the species level, three species exhibited significant differences, including Prevotella sp902794405 (P = 0.004), Prevotella sp900318795 (P = 0.011), and Prevotella sp902778265 (P = 0.006). For Prevotella sp902794405, the LD, MD, and HD groups were significantly higher than the CON group (P = 0.004). For Prevotella sp900318795, the HD group was substantially higher than the CON and LD groups (P = 0.011). Prevotella sp902778265 showed the highest abundance in the MD group and was markedly higher than both the CON and HD groups (P = 0.006) (Fig. 5A; Table S1). The most dominant archaeal phylum was Methanobacteriota (Fig. 3D), and the dominant archaeal genera included Methanobrevibacter A and Methanobrevibacter (Fig. 4B). For the differential abundance comparison analysis of archaea, only at the species level did the relative abundance of Methanobrevibacter A sp900318035 show significant differences (P = 0.008), the MD group was significantly higher than both the CON and LD groups, and the HD group was remarkably higher than the CON group (Fig. 5B; Table S1).

Fig. 3.

Effects of K₂CO₃ on rumen microbial community structure. (A) Alpha diversity indices. (B) Principal coordinates analysis (PCoA) plot based on the Bray–Curtis metric. The distribution of major bacterial phyla (C) and archaeal phyla (D) based on relative abundance, with significant Kruskal–Wallis test results. The asterisks mean statistically significant difference: ∗P-value <0.05. CON, the control group (basal diet); LD, the low dose group (basal diet + 250 g/d K₂CO₃·1.5H₂O per head); MD, the medium dose group (basal diet + 500 g/d K₂CO₃·1.5H₂O per head); HD, the high dose group (basal diet + 750 g/d K₂CO₃·1.5H₂O per head).

Fig. 4.

The relative abundance at the bacterial genus level (A) and archaeal genus level (B). The asterisks mean statistically significant difference: ∗P-value <0.05. CON, the control group (basal diet); LD, the low dose group (basal diet + 250 g/d K₂CO₃·1.5H₂O per head); MD, the medium dose group (basal diet + 500 g/d K₂CO₃·1.5H₂O per head); HD, the high dose group (basal diet + 750 g/d K₂CO₃·1.5H₂O per head).

Fig. 5.

The relative abundance at the bacterial species level (A) and archaeal species level (B). The asterisks mean statistically significant difference: ∗P-value <0.05. CON, the control group (basal diet); LD, the low dose group (basal diet + 250 g/d K₂CO₃·1.5H₂O per head); MD, the medium dose group (basal diet + 500 g/d K₂CO₃·1.5H₂O per head); HD, the high dose group (basal diet + 750 g/d K₂CO₃·1.5H₂O per head).

3.6. Correlation between rumen differential microorganisms, milk fat percentage, and rumen differential metabolites

The correlation between the relative abundance of the selected rumen differential genera and milk fat percentage showed that all genera, except Prevotella sp900318795 (Fig. 6A), exhibited significant correlations with milk fat percentage (P < 0.05; Fig. 6B, C, and 6D). This suggests that changes in the abundance of these three bacterial species in the rumen may play a critical role in increasing milk fat percentage. Additionally, the correlation between the shared rumen differential metabolites across the three groups and the differential bacterial species revealed that N-(tert-butoxycarbonyl) tryptophan, beta-doradecin, scopoletin, 1-methyladenosine, and N6-methyladenosine were significantly correlated with Prevotella sp902794405. This finding further indicates that Prevotella sp902794405 may be a key rumen microorganism mediating the regulatory effects of K₂CO₃ on the rumen microecosystem (Fig. 6E).

Fig. 6.

Relationship between milk fat, differential metabolites in rumen and key rumen taxa. (A﹣­­D) Shows the links between milk fat and key rumen taxa Prevotella sp900318795 (A), Prevotella sp902778265 (B), Prevotella sp902794405 (C), and MethanobrevibacterA sp900318035 (D), respectively. (E) The correlation network shows links between microbiome species relative abundance and different metabolites in rumen.

3.7. Functional profiles of the rumen microbiome, differential functions among cows in four groups

Next, to preliminarily identify the functions of the cows’ rumen microbiome and visualize the differences among the CON, LD, MD, and HD groups, a Random Forest classifier was constructed and tested to identify distinguished microorganisms. The top 20 most important genes and pathways were enriched in the K₂CO₃ supplementation groups compared to the CON group. Among these, K08848 (RIPK4) was identified as the most significant gene, while ko03320 (peroxisome proliferator-activated receptor [PPAR] signaling pathway) was the most enriched pathway (Fig. 7A and B). Additionally, statistical analysis of metagenomic profiles (STAMP) analysis was performed to identify differentially abundant pathways among the three treatment groups compared to the CON group. More KEGG pathways were enriched in the LD, MD, and HD groups, with the MD group showing the most significant enrichment (Fig. 7C, D, and E). Finally, to further explore how K₂CO₃ supplementation influences genes encoding CAZymes in the rumen, a PCoA plot revealed significant differentiation among the groups (P = 0.047) (Fig. 8A). Furthermore, linear discriminant analysis effect size (LEfSe) of CAZymes, based on a linear discriminant analysis (LDA) threshold of 2, demonstrated clear differences across the four groups. Notably, the MD group exhibited the highest enrichment of genes encoding CAZymes (Fig. 8B). In the present study, a total of six CAZymes classes, namely auxiliary activities (AA), carbohydrate-binding modules (CBM), carbohydrate esterases (CE), glycoside hydrolases (GH), glycosyl transferases (GT), and polysaccharide lyases (PL) were identified. The GH constituted the largest fraction of CAZymes among the classes, while the GT was the second most abundant class. The minimum abundance was assigned to the AA CAZymes class. Results from the study indicated a significant difference in the abundances of CAZymes affiliated with the PL classes (Fig. 8C). The abundance of PL CAZymes significantly increased (P = 0.017) in the MD group. Consistent with the LEfSe analysis results, the heatmap-based abundance analysis further confirmed that the LD, MD, and HD groups exhibited higher enrichment of genes encoding CAZymes. Notably, PL0, GH124, and GT84 were identified as the top three most important CAZymes based on the random forest analysis (Fig. 8D). Overall, these findings suggest that supplementing 500 g/d per head of K₂CO₃ (MD group) is the most effective strategy for improving milk quality while maintaining the health of dairy cows. Furthermore, in the MD group, the genes were significantly upregulated in the rumen compared to the control group (Fig. 9). The MD group exhibited greater enrichment in pathways related to the biosynthesis of amino acids and long-chain fatty acid synthesis, highlighting its potential benefits in metabolic regulation (Fig. 10).

Fig. 7.

Predicted microbiota function differences among groups. The heatmaps show the relative abundances and the bar plots present the corresponding random forest importance rankings of the top 20 Kyoto Encyclopedia of Genes and Genomes (KEGG) orthologs (A) and pathways (B) in the rumen microbiome. (C–E) The stamp analysis of the top 20 enriched KEGG pathways (P-value <0.05) in the rumen microbiota for the comparisons between the control group and the LD (C), MD (D), and HD (E) groups, respectively. ∗P < 0.05. CON, the control group (basal diet); LD, the low dose group (basal diet + 250 g/d K₂CO₃·1.5H₂O per head); MD, the medium dose group (basal diet + 500 g/d K₂CO₃·1.5H₂O per head); HD, the high dose group (basal diet + 750 g/d K₂CO₃·1.5H₂O per head).

Fig. 8.

Effects of K₂CO₃ on rumen carbohydrate-active enzymes (CAZys). (A) Principal coordinates analysis (PCoA) of the rumen microbiota related to CAZys. (B) Linear discriminant analysis effect size (LEfSe) analysis of CAZys in the rumen fluid microbiota identified biomarker taxa, with linear discriminant analysis (LDA) scores >2 and significance at P < 0.05 (Wilcoxon signed-rank test). (C) Differences in the rumen microbiome at the CAZy family levels among groups based on Kruskal–Waillis test. ∗P < 0.05; ∗∗P < 0.01. (D) The heatmaps show the relative abundances and the bar plots present the corresponding random forest importance rankings of the top 20 CAZys of the rumen microbiome. CON, the control group (basal diet); LD, the low dose group (basal diet + 250 g/d K₂CO₃·1.5H₂O per head); MD, the medium dose group (basal diet + 500 g/d K₂CO₃·1.5H₂O per head); HD, the high dose group (basal diet + 750 g/d K₂CO₃·1.5H₂O per head). TPM = transcripts per million; AA = auxiliary activities; CBM = carbohydrate-binding modules; CE = carbohydrate esterases; GH = glycoside hydrolases; GT = glycosyl transferases; PL = polysaccharide lyases.

Fig. 9.

The different rumen microbial Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways in the MD group compared with the CON group. The red text represents KEGG enzymes enriched in the rumen microbiome of MD group cows. CON, the control group (basal diet); MD, the medium dose group (basal diet + 500 g/d K₂CO₃·1.5H₂O per head). TCA = tricarboxylic acid cycle.

Fig. 10.

Supplementation with 500 g/d of K₂CO₃ in the diet of mid-lactation dairy cows alters rumen microbial metabolism, promoted acetate production in the rumen, and improved milk fat and protein percentage. Upward arrows (↑) indicate increased levels or enrichment compared with the control group. CON, the control group (basal diet); LD, the low dose group (basal diet + 250 g/d K₂CO₃·1.5H₂O per head); MD, the medium dose group (basal diet + 500 g/d K₂CO₃·1.5H₂O per head); HD, the high dose group (basal diet + 750 g/d K₂CO₃·1.5H₂O per head).

4. Discussion

Mid-lactation dairy cows face the challenge of maintaining both high milk yield and quality. In this period, DMI has reached its peak, coinciding with the lactation peak, making it the main stage of milk production (Garg et al., 2013). Moreover, when the risk of metabolic diseases increases, cows significantly mobilize body tissues to maintain lactation. For example, inflammatory processes trigger competition for nutrients, causing the body to prioritize immune system demands at the expense of production performance (Ingvartsen et al., 2003). Therefore, it is crucial to maintain the lactation peak for a longer duration and improve milk quality while ensuring cow health during this period. K₂CO₃ has been utilized to maintain potassium and electrolyte balance, particularly in helping cows mitigate heat stress during summer months (Staples, 2007). Despite these benefits, the specific mechanisms by which K₂CO₃ improves production performance and health in mid-lactation dairy cows remain incompletely understood. In this study, K₂CO₃ supplementation significantly improved the milk fat and protein ratio. Further investigation revealed that these positive effects were mediated through the regulation of rumen fluid microbiota and microbial metabolites. Specifically, K₂CO₃ affects the abundance of key rumen microbial groups such as Prevotella, and functional enrichment analysis shows it promotes metabolic pathways including ‘galactose metabolism’ and ‘fructose and mannose metabolism’, thereby providing more substrates for milk fat and milk protein biosynthesis. In addition, supplementing with K₂CO₃ increases the production of acetate, a precursor for milk fat synthesis, and significantly enriches rumen microbial functions in fatty acid biosynthesis, fatty acid elongation pathways, and amino acid metabolism pathways, which may be important reasons for the increased milk fat and protein ratio. These findings not only highlight the functional properties of K₂CO₃ in enhancing lactation performance but also elucidate its underlying mechanisms at the microbial and metabolic levels. By broadening our understanding of role of K₂CO₃ in rumen microbiota and metabolite regulation, this study paves the way for its potential application in improving the productivity of mid-lactation dairy cows.

Milk fat is one of the primary components of milk, serving not only as the main source of milk flavor compounds but also as a carrier for essential fat-soluble nutrients (Ding et al., 2020). Therefore, it is considered one of the key indicators for assessing milk quality. Davis and Brown described “low-fat milk syndrome” as a result of a high ratio of readily digestible carbohydrates to fiber in the diet, which creates unfavorable conditions in the rumen (Davis and Brown, 1970). Based on these observations, different minerals were proposed to help stabilize rumen pH, thereby reducing the incidence of milk fat depression (Phillipson and Ash, 1970). K₂CO₃ is widely recognized as an effective rumen buffer, playing a crucial role in maintaining normal ruminal acid-base balance and improving rumen metabolism (West et al., 1986). This study demonstrated that increasing the proportion of K₂CO₃ in the diet can enhance milk fat concentration, aligning with findings from previous research (Alfonso-Avila et al., 2017; Harrison et al., 2012). However, the specific mechanisms by which K₂CO₃ supplementation mitigates the effects of high-concentrate diets on milk fat depression remain unclear. Then, this study observed that K₂CO₃ supplementation increased the rumen's acetate-to-propionate ratio, as well as the concentrations of acetate. In cows, about one-half of fatty acids come from de novo fatty acids synthesis, with acetate and β-hydroxybutyrate as the primary precursors (Bauman and Griinari, 2003). The other half of milk fatty acids are preformed fatty acids taken up from circulation by the mammary gland. Acetate is the main short-chain fatty acids, and it dose-dependently stimulates milk fat synthesis by increasing de novo fatty acids synthesis in lactating dairy cows (Urrutia and Harvatine, 2017) and regulating milk fat synthesis and lipogenic gene expression in the bovine mammary gland (Maxin et al., 2011; Jacobs et al., 2013). Furthermore, dose-dependent alterations in rumen metabolism were observed, and the number of differential metabolites between the MD group (moderate-dose K₂CO₃ supplementation) and the CON group was higher than that observed in the LD and HD groups. This suggests that excessive K₂CO₃ supplementation may exert inhibitory effects on rumen microbial metabolic activity. Notably, compared to the CON group, the MD group exhibited a marked increase in ruminal lipid metabolites such as PC(16:0/16:0) and PC(18:2(9Z,12Z)/P-16:0). These phosphatidylcholine species are structurally characterized by saturated (palmitic acid, 16:0) and polyunsaturated (linoleic acid, 18:2) fatty acid chains, which are known precursors for milk fat synthesis and membrane lipid assembly (Potts et al., 2023). Previous studies have demonstrated that elevated levels of such phospholipids are positively correlated with enhanced lactation performance in dairy cows, likely through improving lipid bioavailability and metabolic efficiency in the mammary gland. Specifically, reported that dietary interventions modulating ruminal lipid profiles, particularly through phosphatidylcholine enrichment, can significantly boost milk yield and fat content, aligning with the findings in this study (Lima et al., 2024). This may be attributed to the ability of K₂CO₃ to improve the production of VFAs in the rumen and optimize the proportions among various fatty acids (Sharma et al., 2018), thereby exerting a positive regulatory effect on the ruminal environment. This, in turn, this alleviates the inhibitory impact on lipid-metabolizing microorganisms, potentially promoting the breakdown of fats as well as the release and metabolism of fatty acids. This may be a result of changes in the structure and function of the rumen microbiota (Zhao et al., 2023). Therefore, the metagenomic sequencing of the rumen microbiome was conducted to further investigate these changes.

The rumen contains a microbial community, including bacteria, fungi, and protozoa, that plays a pivotal role in lipid metabolism in dairy cows (Conte et al., 2022). Different microorganisms have varying capacities for lipid metabolism (Brown et al., 2023). Previous studies have primarily focused on the effects of K₂CO₃ on the rumen's bacterial population, which dominates in numbers (Kittelmann et al., 2015). However, given that archaea also be essential for nutrient degradation and have complex interactions with bacteria, this study explored not only the effects of K₂CO₃ on the bacterial community but also its influence on archaea using metagenomics. The results showed that K₂CO₃ primarily affected the bacterial community in the rumen, promoting the growth of beneficial microorganisms, such as Bacteroidetes (e.g., Prevotella species). These microbes are crucial in ruminal lipid metabolism. Members of Bacteroidetes, along with certain fiber-degrading bacteria, are more efficient at utilizing complex dietary lipids, breaking down both saturated and unsaturated fatty acids, thereby enhancing lipid digestion (Flint et al., 2012). This is particularly important for the production of acetate, which is essential for milk fat synthesis (Zhao et al., 2021). The elevated acetate levels, along with an increased acetate-to-propionate ratio, could explain this finding. Interestingly, K₂CO₃ significantly increased the abundance of MethanobrevibacterA sp900318035, a member of the Methanobrevibacter genus. It has been previously reported that Methanobrevibacter, the dominant methanogen in the rumen (Jin et al., 2017), belongs to the hydrogenotrophic methanogens, which produce methane (CH₄) by reducing hydrogen (H₂) (Su et al., 2018). The increased abundance of Methanobrevibacter A sp900318035 may be due to the large amounts of hydrogen produced as a byproduct of acetate formation. Since methanogens in the rumen, such as Methanobrevibacter, use hydrogen as a substrate, an increase in acetate production leads to more hydrogen availability, potentially enhancing methanogen activity. Finally, a correlation analysis was performed between the identified differential bacterial strains and milk fat percentage. The results revealed significant positive correlations between milk fat percentage and Prevotella sp902778265, Prevotella sp902794405, and Methanobrevibacter A sp900318035, which was consistent with the above findings. To further assess the potential impact of these bacterial species on rumen metabolism and to identify key species involved, an additional correlation analysis was conducted between the shared metabolites (identified from comparisons between the three K₂CO₃ treatment groups and the CON group) and the relative abundance of the four bacterial species. Based on these associations, Prevotella sp902794405 was suggested as a potential primary bacterial strain responsive to K₂CO₃ supplementation. However, these associations did not constitute direct evidence. Isolation experiments and in vitro monoculture studies are required to validate this hypothesis.

In the functional analysis of the rumen metagenome, the rumen microbiome of K₂CO₃-supplemented dairy cows exhibited an enrichment of KEGG pathways related to carbohydrate degradation, including “galactose metabolism” and “fructose and mannose metabolism”. This suggests that the microbiome in these cows may have a greater capacity for carbohydrate breakdown, leading to enhanced substrate availability for milk fat and milk protein biosynthesis. Additionally, the enrichment of genes encoding CAZymes, particularly PLs, further supports the enhanced ability of the rumen microbiome in K₂CO₃-supplemented cows to degrade complex substrates. Furthermore, the LEfSe analysis revealed a significant enrichment of CAZymes in the MD group. The elevated concentrations of major VFAs in the rumen of MD group cows suggest that their ruminal microbiota exhibits enhanced efficiency in converting hydrolytic products into VFAs, consequently improving energy metabolism and overall metabolic efficiency.

Notably, the K₂CO₃ treatment groups—particularly the moderate-dose (MD) group—exhibited significant enrichment in fatty acid biosynthesis and fatty acid elongation pathways compared to other groups. This observation may be attributed to the substantial increase in Prevotella abundance in the rumen, as this genus has been demonstrated to efficiently utilize complex dietary lipids by degrading both saturated and unsaturated fatty acids, thereby enhancing lipid digestion (Flint et al., 2012). This hypothesis is further supported by the metabolomics data, which showed significant elevations in rumen lipid metabolites such as phosphatidylcholines (PC(16:0/16:0) and PC(18:2(9Z,12Z)/P-16:0)) in the K₂CO₃-treated groups. Collectively, these findings suggest that K₂CO₃ upregulates fatty acid biosynthesis and fatty acid elongation pathways by enriching Prevotella in the rumen, subsequently increasing the availability of lipid precursors (e.g., PC(16:0/16:0), PC(18:2(9Z,12Z)/P-16:0), and acetate) essential for milk fat synthesis. Interestingly, the metagenomic analysis revealed that the PPAR signaling pathway was not only enriched in K₂CO₃-supplemented groups but also identified as the most significant metabolic pathway through a random forest analysis. This finding aligns with previous reports establishing PPAR's crucial role in fatty acid catabolism (Burdick et al., 2006; Liu et al., 2022). Specifically, PPAR regulates key genes involved in mitochondrial and peroxisomal β-oxidation of long-chain fatty acids, thereby maintaining tight regulation of intracellular long-chain fatty acid levels (Loor et al., 2007). Based on these findings, it is proposed that activation of the PPAR signaling pathway may represent a key mechanism through which K₂CO₃ enhances lipid biosynthesis in ruminants.

The rumen microbiome functions regarding nitrogen metabolism contribute to differential host milk protein yield. The result showed that multiple nitrogen metabolism pathways were significantly enriched in MD group cows including “valine, leucine, and isoleucine biosynthesis”, “Phenylalanine, tyrosine and tryptophan biosynthesis”, and “Serine and lysine biosynthesis”. These amino acids play a crucial role in microbial protein synthesis (Allison et al., 1966), ultimately supporting the production of milk protein in the mammary gland of the MD group.

Previous studies have demonstrated that rumen microbiota in ruminants play a central biological role in vitamin B metabolism. Rumen microorganisms can synthesize various vitamin B compounds through their own metabolic pathways (Russell and Rychlik, 2001). These vitamins serve as essential enzyme cofactors, directly participating in metabolic processes related to milk components, such as fatty acid synthesis and amino acid catabolism (Chen et al., 2011). In this study, metabolic pathway analysis revealed a significant enrichment of the biotin metabolism pathway in the rumen microbiota of the MD group. Biotin, as a cofactor of carboxylases, plays a crucial role in carbon dioxide transfer reactions (Zempleni et al., 2009). This finding suggests that biotin may influence milk fat synthesis by regulating carboxylation reactions. Notably, exogenous vitamin B supplementation has been shown to enhance milk component yield in dairy cows (Kaur et al., 2019). Based on the results of this study, K₂CO₃ treatment may improve milk quality through the following mechanisms: (1) modulating the rumen environment to promote the proliferation of vitamin B-synthesizing microbiota; and (2) enhancing the activity of biotin-dependent carboxylases, thereby accelerating the conversion of precursor substances into milk components. These findings provide new theoretical insights into the molecular mechanisms by which K₂CO₃ regulates milk quality.

In summary, this study found that K₂CO₃ can significantly improve lipid metabolism and amino acid biosynthesis in the rumen of mid-lactation dairy cows by regulating the ruminal environment, promoting the production of volatile fatty acids, and optimizing the microbial community. These mechanisms ultimately increase milk fat and protein ratio, enhancing both production performance. While it was highlighted that dietary K₂CO₃ improved milk fat content by enhancing ruminal lipid metabolism in mid-lactation dairy cows, there were some limitations to this study. Firstly, although all experimental animals were at the same lactation stage, inherent individual differences among cows could still influence the experimental results. Additionally, ion concentrations and osmotic pressure in the rumen were not measured, and these physiological parameters may directly affect microbial metabolic enzyme activity. Furthermore, the study design only allowed for measurements at the end of the experimental period, which limits our understanding of the dynamic changes occurring during the intervention. Finally, although a mechanistic pathway linking K₂CO₃ supplementation to improved milk quality via modulation of the microbiota and metabolites was proposed, this chain of effects lacked direct functional validation. Future research should incorporate additional in vitro experiments to better isolate the effects of K₂CO₃ from potential host-mediated influences.

5. Conclusion

In conclusion, this study systematically investigated the effects of different K₂CO₃ supplementation levels on the physiological health, milk quality, and rumen metabolism of mid-lactation dairy cows. Focusing on changes in rumen microbial community structure, function, and metabolic profiles, the results showed that K₂CO₃ optimized the rumen metabolic environment and increased the production of acetate and phosphatidylcholines (PC(16:0/16:0) and PC(18:2(9Z,12Z)/P-16:0))—key precursors for milk fat synthesis. Functional analysis of rumen microbiota further revealed significant enrichment of fatty acid biosynthesis and elongation pathways in the MD group, which contributed to increased milk fat ratio. Simultaneously, the MD group also showed significant enrichment of amino acid biosynthesis pathways, ultimately enhancing milk protein ratio. Based on these findings, supplementing 500 g/d of K₂CO₃ per head (MD group) was suggested to be the most effective strategy for improving milk quality. These discoveries not only deepen our scientific understanding of K₂CO₃'s mechanisms of action and expand the theoretical foundation of this research field, but also provide valuable practical guidance for its application in dairy production.

Credit Author Statement

Xiaojing Liu: Writing – review & editing, Writing – original draft, Visualization, Methodology, Investigation. Haoqi Han: Validation, Investigation. Xinyue Zhang: Writing – review & editing, Methodology, Investigation. Fanlin Kong: Writing – review & editing, Methodology, Investigation. Dongwen Dai: Investigation, Methodology. Yangyi Hao: Writing – review & editing, Conceptualization, Supervision. Wei Wang: Supervision, Methodology, Conceptualization. Shengli Li: Writing – review & editing, Writing – original draft, Supervision, Methodology, Funding acquisition, Conceptualization.

Data availability statement

The raw reads of metagenome sequencing are available at NCBI project PRJNA1230829.

Declaration of competing interest

We declare that we have no financial and personal relationships with other people or organizations that can inappropriately influence our work, and there is no professional or other personal interest of any nature or kind in any product, service and/or company that could be construed as influencing the content of this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article.