Impacts of rumen-encapsulated methionine and lysine supplementation and low dietary protein on nitrogen efficiency and lactation performance of dairy cows

Low crude protein (CP) diets might be fed to dairy cows without impacting productivity if the balance of absorbed AA were improved, which would decrease the environmental impact of dairy farms. The aim of this study was to investigate the impacts of supplementing ruminally-protected Lys (RPL) and Met (RPM) at 2 levels of dietary CP on nutrient intake, milk production, milk composition, milk N efficiency (MNE), and plasma concentrations of AA in lactating Holstein cows and to evaluate these impacts against the predictions of the new NASEM (2021) model. Fifteen multiparous cows were used in a replicated 3 × 3 Latin square design with 21-d periods. The 3 treatments were (1) a high-protein (HP) basal diet containing 16.4% CP [MP balance of – 130 g/d; 95% of target values], (2) a medium-protein diet containing 15% CP plus RPL (60 g/cow per day) and RPM (25 g/cow per day) (MPLM) [MP balance of – 314 g/d; 87% of target values], and (3) a low-protein diet containing 13.6% CP plus RPL (60 g/cow per day) and RPM (25 g/cow per day) (LPLM) [MP balance of – 479 g/d; 80% of target values]. Dry matter intake was less for cows fed MPLM and LPLM diets compared with those fed the HP diet. Compared with the HP diet, the intake of crude protein, neutral detergent fiber, acid detergent fiber, and organic matter, but not starch, was lower for cows fed MPLM and LPLM diets. Milk production and composition were not affected by MPLM or LPLM diets relative to the HP diet. Milk urea N concentrations were reduced for the MPLM and LPLM diets compared with the HP diet, indicating that providing a low-protein diet supplemented with ruminal-protected AA led to greater N efficiency. There was no significant impact of treatment on plasma AA concentrations except for proline, which significantly increased for the MPLM treatment compared with the other 2 treatments. Overall, the results supported the concept that milk performance might be maintained when feeding lactating dairy cows with low CP diets if the absorbed AA balance is maintained through RPL and RPM feeding. Further investigations are needed to evaluate responses over a longer time period with consideration of all AA rather than on the more aggregated MP and the ratio between Lys and Met.


INTRODUCTION
High-yielding dairy cows require the indispensable amino acids (IAA) during lactation to support high levels of milk production.Historically, due to a lack of knowledge of individual IAA needs, N feeding has been metabolizable protein-based (NRC, 2001).With this approach, cows are probably overfed several IAA as well as some dispensable AA to ensure meeting the IAA needs across a wide range of diets, resulting in poor N efficiency.Using this approach, approximately 25% of the N fed to dairy cows is transformed into milk protein, and the remainder is lost (Chase et al., 2012;Apelo et al., 2014;Zhao et al., 2019).Providing highprotein diets to meet IAA needs is the source of low efficiency resulting in excess feed costs and excess N excretion in manure, which contributes to environmental pollution (Carder and Weiss, 2017;Lee et al., 2019).Therefore, reduced N losses and improved N utilization in dairy cattle are essential for biological, economic, and environmental reasons.One strategy to enhance N efficiency is to feed dairy cows lower-protein diets (LP) (Kalscheur et al., 2006;Broderick et al., 2008;Apelo et al., 2014); however, this strategy is typically associated with lower productivity (milk production, milk protein production, or both), and is economically unfavorable (Apelo et al., 2014;Giallongo et al., 2016).The loss in production is probably linked to the insufficiency of 1 or more of the IAA, and ruminal-protected AA (RP-AA) supplementation could compensate for insufficient supply of IAA.Adding a blend of RP-AA to low-protein diets could improve N efficiency by increasing milk protein production and decreasing N excretion into the environment which likely would add economic value for producers.Research has shown that Met and Lys often limit milk production in a variety of dairy cow diets, and milk protein responses to absorbed Met and Lys are almost linear within the observed biological range (NASEM, 2021).Many studies (Lee et al., 2015;Giallongo et al., 2016;Lee et al., 2019) have been conducted to investigate the impacts of ruminal-protected Lys and Met sources (RPL and RPM) under MP sufficient or mildly deficient dietary conditions (Robinson et al., 1998;Patton, 2010;Apelo et al., 2014).However, very few studies have been conducted using MP-deficient diets with the exception of some work during the transition period (Osorio et al., 2013;Girma et al., 2019;Lee et al., 2019) or using RP-AA with 1 level of low CP (Lee et al., 2012;Apelo et al., 2014;Lee et al., 2015).A new concept has been introduced by NASEM (2021) based on considering 5 IAA rather than the more aggregated MP and the ratio between Lys and Met.Therefore, the aim of the current study was to evaluate the impacts of adding the combination of RPL and RPM to 2 levels of low CP diets on DMI, milk yield and composition, milk N efficiency (MNE; milk N/N intake), and plasma concentrations of AA in lactating dairy cows, and to evaluate the observed responses against the predictions of the NASEM (2021) model.We hypothesized that the combined supplementation of RPL and RPM to a low CP diet would sustain milk yield and composition comparable to a high CP diet resulting in improved milk N efficiency and that these changes would be predicted by the NASEM (2021) model.

MATERIALS AND METHODS
This research was carried out at Sino Farm, a commercial dairy farm in Beijing, China.All procedures for this research were approved by the Institute of Animal Science (IAS)'s Animal Care and Use Committee, Chinese Academy of Agricultural Sciences (CAAS), Beijing, China (protocol no. IAS2022-91).

Cows, Experimental Design, and Diets
A statistical power analysis was carried out with an α = 0.05 and power = 0.90, and the sample size required with 0.4 effect size was 15 cows, as estimated using G power 3.1 software (Faul et al., 2009).The variables used to determine sample size were milk yield and DMI.An effect size (ƒ) of 0.4 was used with variance explained by effect = 1.0, variance within group = 5.25, and partial η 2 = 0.16.The 15 multiparous lactating Holstein cows were selected and blocked to 1 of 5 blocks according to parity [(2.9 ± 0.52) (mean ± SD)], DIM (68.9 ± 2.92), milk production (45.0 ± 3.61 kg), and BW (671 ± 16 kg) at the beginning of the trial.Cows were randomly assigned within block to squares and treatment sequences within square.Each experimental period was 21 d in length (14 d for adaptation and 7 d for sample collection).Cows were housed in a single pen of a free stall barn fitted with 15 Insentec gates (RIC-Management by INSENTEC, Marknesse, Netherlands), with each cow allowed access to a single feeder.No other cows were allowed access to the pen.Cows had free access to water and were milked 4 times per day at equal intervals (0600, 1200, 1800, and 2400 h).
The treatments consisted of 3 diets: (1) high CP (HP) [CON; 16.4% CP, MP balance of -130 g/d] formulated based on a target milk yield of 38 kg/d which according to NASEM (2021) met 95% of the target values for MP; (2) medium CP (MPLM) (15% CP, MP balance of -314 g/d) which met 87% of MP target values supplemented with RPM [(25 g/cow per day of Meta-Smart dry (MS) (Adisseo Inc., Commentry, France), and RPL 60 g/cow per day of LysiPEARL (Kemin Industries, Inc., USA)]; and (3) low CP (LPLM) (13.6% CP, MP balance of -479 g/d) which met 80% of MP target values plus [25 g of RPM/cow per day and 60 g/cow per day of RPL].The 3 experimental diets were formulated using the NASEM (2021) model and software (Eighth revised editionV8 R2022.01.18;National Academies of Sciences, Engineering and Medicine, NASEM Dairy 8) for a cow producing 38 kg of milk at 3.10% true protein, and 3.80% milk fat and weighing 680 kg.The TMR was mixed and fed ad libitum 2 times per day at (07:30 (45%), and 15:00 (55%) h) using a TMR mixer with a target of 5-10% refusals.The ingredients and chemical composition of the experimental diets are presented in Tables 1 and 2, respectively.The RPM and RPL were fed top-dressed once per day (at the morning feeding) mixed with a small proportion of TMR.According to the manufacturer rumen-protected methionine (Meta-Smart dry, MS) contained 57% HMBi, 78% Methionine equivalent, and 50% bioavailability; so, each 1 g of MS was expected to deliver 0.22 g of metabolizable methionine.The ruminal-protected lysine (LysiPEARL) contained 47.5% L-Lysine mono-hydrochloride (3.2.3) with a ruminal degradation rate of 5.5%/hr.and 70% bioavailability, and, so, each 1 g of LysiPEARL was expected to deliver 0.33 g of metabolizable lysine-HCL.Predicted supply of MP and AA by NASEM (2021) are presented in (Table 3).

Sampling, Analysis, and Calculations
The TMR offered and orts were recorded daily for each cow and used to calculate DMI.Weekly samples of TMR and the main individual feed ingredients (i.e., alfalfa hay, corn silage, flaked corn, and soybean meal) were collected and analyzed for nutrient concentrations.Samples of TMR and orts from each cow were collected on d 15, 18, and 21 of each period and stored at -20°C for later analysis.The feed and orts samples were pooled by treatment and period.Diet samples and orts were dried for 48 h at 65°C to measure dry matter.Using a Cyclotec 1093 Mill (Tecator AB, Höganäs, Sweden), the dry TMR samples were ground through a 1-mm screen before analysis.To measure the absolute DM, diet samples were further dried at 105°C for 4 h.
Crude protein (CP, N × 6.25) was determined using a macro- Kjeldahl N test (method no. 976.05;AOAC, 2000) with a Kjeltec digester 20 and a Kjeltec System 1026 distilling unit (Tecator AB).The content of ADF and NDF were analyzed using the procedure of Van Soest et al. (1991) using α-amylase and with the addition of sodium sulfite.Ether extract content was determined using a Soxhlet HT6 apparatus (Tecator AB) according to method no.920.39 (AOAC, 2000).Incineration at 550°C was used to determine the ash content and calculate the OM content by subtracting ash from 100 according to method no.942.05 (AOAC, 2000).According to the National Research Council (NRC, 2001), Non-fiber carbohydrates (NFC) were calculated as NFC = 100 -(% NDF + % CP + % EE + % Ash).
Metabolizable Protein (MP), energy balances (EBAL), and energy density were estimated based on observed DMI, milk yield, and ingredient composition for each treatment group using the NASEM (2021) software.
Net energy intake was calculated according to the following equation (NASEM, 2021): Net energy intake (NE I ) = daily DMI × NE L density of the diet.
The NE requirements were calculated using the following equation (NASEM, 2021):

Blood Samples
Approximately 10 mL of blood was sampled from a coccygeal vessel of each cow after 1 h of the morning and afternoon feedings on d 15, 18, and 21 of each experimental period.The blood was collected into hepa-   rinized vacuum tubes and centrifuged at 3000 × g for 20 min at 4°C.Separated plasma samples were stored at −20°C until later analysis.The plasma samples were pooled by period and cow upon analysis.Concentrations of free AA were measured by an automatic AA analyzer (type 1290 Infinity II, Agilent Technologies) using the method described by Pereira et al. (2020).

Milk Samples
In each experimental period, milk production was recorded at each milking time.Duplicate milk samples from each cow were collected at each milking time on d 19, 20, and 21 of each experimental period.To make a representative milk sample, daily milk samples were pooled from 4 milking sessions and mixed according to the average milk yield at each milking session [(morning, afternoon, evening, and night; volume ratio 20%:20%:30%:30%; these ratios represented the mean observed production ratios for cows from each milking session)].One aliquot of milk sample was preserved with bronopol-B2 (800 Broad Spectrum Microtabs II; D&F Control Systems, Inc., MA, USA) and stored at 4°C before being tested for milk composition (fat, protein, lactose, MUN) using a near-infrared reflectance spectroscopy analyzer (Foss Electric, Hillerød, Denmark), and SCC using a Fossomatic 5000 apparatus (Foss Electric, Hillerød, Denmark).Fat-corrected milk (4% FCM), and ECM were calculated according to (Tyrrell and Reid, 1965)

Body Condition Score (BCS)
Body condition scores were assessed for each animal at d 15, 18, and 21 during each experimental period by 2 qualified experts using physical palpation combined with optical evaluation.The 5-point scale used to evaluate BCS ranged from 1 for extremely thin to 5 for obese cows (Ferguson et al., 1994).

Statistics
Data from the last 7 d of each experimental period were analyzed using the PROC MIXED procedure of SAS (version 9.4; SAS Institute Inc., Cary, NC).The following statistical model was used to assess the dietary treatment effects: where Y = the dependent variable; µ = the overall mean; S i = the fixed effect of square (I = 1 to 5); T i = the fixed effect of the treatment (i = 1 to 3); P j = the random effect of period (j = 1 to 3); D k = the random effect of day within period (k = 15 to 21 or d 15, 18, and 21 for items sampled on 3 d); C l = the random effect of cow within square (l = 1 to 15); and e ijkl = the random residual error.The error term was assumed to be normally distributed with mean = 0 and constant variance.Data on milk composition, feed efficiency (ECM / DMI), MNE, BCS, and BW was only from 3 d of the last 7 d of each experimental period and were analyzed as repeated measures assuming an autoregressive (1) covariance structure with treatment, period, cow, sampling time, and the interaction between treatment, period, and sampling time in the model.All values are provided as least squares means.Interactions were removed from the model when they were not significant (P > 0.05), but all main effects were retained regardless of significance.The Tukey-Kramer option of PROC MIXED was used to separate means when the main treatment effect was significant.Significance was declared for all analyses at P ≤ 0.05 and trends toward significance at 0.05 < P ≤ 0.10.

RESULTS
The experimental diets and the chemical compositions of nutrients are presented in Tables 1 and 2, respectively.According to NASEM (2021) model estimations using the observed DMI, milk yield, milk composition, and BCS throughout the experiment, the HP diet provided MP less than target values by approximately 5%, whereas MPLM and LPLM diets were 13% and 20% below MP targets, respectively.It should be noted that these predictions were based on the model-calculated MP target values, and such predictions are known to have a significant slope bias (NASEM, 2021;Ch. 20, pp. 414 -469).The balance of digestible Lys for the HP diet was -11% below the NASEM (2021) target values, while the MPLM diet was only -2% below the target value, and the LPLM diet was -11% below the target value.The MPLM diet was balanced for digest- ible Met, while HP and LPLM diets were around 17% and 5% deficient, respectively.The MPLM and LPLM diets were about 6% and 22% deficient in RDP relative to recommendations.The DM consumed by all cows was lower than predicted by the NASEM model, at around 1 and 1.5 kg for the HP diet; and 3 and 3.25 kg for both MPLM and LPLM diets using the predictions based on animal and on animal and feed factors, respectively (Table 4).The equation based on animal and fiber is assumed to be more accurate and regarded as the best prediction of DMI responses to dietary changes although it does not consider protein effects.

Predicted AA
The amino acids (AA) supplied by the diets and metabolizable AA efficiencies as estimated by the NASEM (2021) model using the observed DMI, milk yield and composition, and BCS are presented in Table 3. Estimates for absorbed Met and Lys for HP treatment were 2.14 and 7.12% of MP, respectively, while for MPLM and LPLM treatments were 2.81 and 2.89% of MP for Met and 8.57 and 8.50% for Lys, respectively.The efficiency of predicted metabolizable Met and Lys for the HP treatment indicated a relative deficiency as they were greater than 0.80 and 0.75, respectively, as compared with the recommended values of 0.73 and 0.72, respectively (NASEM, 2021), Whereas the values were slightly lower for MPLM (0.65 and 0.66) and LPLM (0.67 and 0.71) than the target values, indicating Met and Lys supplementation raised the duodenal supply of Lys and Met to levels that were nearer to recommendations by NASEM (2021).Regarding other IAA, HP and MPLM treatments resulted in predicted metabolizable AA efficiencies for Ile, Leu, Phe, Thr, Trp, and Val that were at or below target values; however His efficiencies were above targets (0.82 vs 0.75 and 0.89 vs 0.75, respectively).In contrast, the LPLM treatment resulted in predicted metabolizable AA efficiencies from all IAA that were above target values.

Dry Matter and Nutrient Intake
Cows consuming the HP diet had greater DMI (P < 0.05; Table 4) than those consuming the other 2 treatments (MPLM and LPLM).Changes in daily OM consumption were consistent with the changes in DMI.Consumption of NDF and CP was less for MPLM and LPLM treatments compared with the HP diet (P < 0.001).Dietary CP consumption was 0.55 and 0.91 kg/d less for the MPLM and LPLM diets, respectively, compared with the HP diet.Daily ADF consumption for cows in MPLM and LPLM treatments was less than those in the HP treatment (P < 0.001); however, there was no difference between the MPLM and LPLM treatments.Daily starch consumption was greater for the LPLM treatment as compared with the MPLM and HP diets (P < 0.001).Body weight and BCS values were not different among treatments (P > 0.05).

Milk Performance and Feed and N Efficiencies
Milk production did not differ among treatments (P > 0.05; Table 5), and there was generally no effect of treatments on milk component production and content with the exception of MUN, which was reduced for the low-and medium protein treatments (P = 0.001; Table 5).Feed efficiency values calculated as milk production/ DMI for MPLM and LPLM diets were increased compared with the HP diet (P = 0.03), while values tended to be higher when calculated as ECM / DMI (P = 0.06).Milk N efficiency was increased for the LPLM diet compared with HP and MPLM diets (P = 0.02).

Plasma AA Concentrations
Plasma concentrations of AA are presented in Table 6.Plasma concentrations of dispensable AA (DAA) were mostly uninfluenced by treatments except for Pro which was reduced (P = 0.04) by the LPLM diet compared with HP and MPLM diets.Supplementing low-and medium protein diets with RPL and RPM did not influence the plasma concentrations of Lys and Met and other IAA (P > 0.05).

Feed Intake and Experimental Diets
The objective of the current study was to investigate whether dietary MP for lactating dairy cows could be reduced below previous recommendations without sacrificing milk production and composition when combined with the addition of selected RP-AA.A second objective was to evaluate the observed data against the predictions of the new NASEM (2021) model.Because the addition of a nutrient to a diet must occur at the expense of one or more other nutrients, it is always challenging to assign cause and effect.In the current work, the reduction in dietary MP and addition of RPAA was associated with a slight reduction in ADF and substantial increases in starch, thus the results could also reflect contributions from starch and reduced ADF.Increased dietary starch content has been shown to stimulate milk and milk protein production (Boerman et al., 2015).Providing MP below the recommended target values of dairy cows may not sustain the desired levels of milk production for the long-term and may decrease milk protein yield and content; however, this effect was not observed consistently across the literature and largely depended on DMI responses (Huhtanen and Hristov, 2009;Lee et al., 2012;Lee et al., 2015).In the present study, the observed reduction in DMI for the  MPLM and LPLM diets that were deficient in MP was consistent with the observations of Lee et al. (2011) and Lee et al. (2012) where DMI was reduced with a low MP diet when fed for 10 weeks.On the other hand, Lee et al. (2015) did not observe an effect when a diet 10% deficient in MP was fed over a shorter period.A similar response was observed by Giallongo et al. (2015) with a diet 5% deficient in MP.In previous studies (Dhiman and Satter, 1993;Weigel et al., 1997;Allen, 2000), the reduction in DMI may have been caused by a deficiency in RDP which may decrease microbial growth in the rumen and reduce ruminal fiber digestion.However, in the current experiment, the diets were not deficient in RDP; thus, this is not the expected cause.Liu and Van-deHaar (2020) also observed a reduction in DMI for LP diets despite being adequate in RDP.This indicates the reduction is due to low MP in general and not to RDP specifically.In addition, diet NEL concentrations did not differ across treatments, and thus this was not the apparent cause of the DMI depression.Other factors may also affect DMI in dairy cows, such as changes in metabolism, AA balance, hormone concentrations, and neural regulation.In non-ruminants, dietary AA concentrations have been observed to affect food intake.Previous studies in rats showed that the brain regulates food intake response to AA-imbalanced diets (Gietzen et al., 1986;Gietzen et al., 2007).In the former study, the authors observed a decrease in food intake for rats fed Thr-deficient diets.In the present study, the NASEM 2021 evaluations of the MPLM and LPLM diets predicted Thr efficiencies that were slightly greater than the target efficiency indicating a deficiency.The same was observed for Trp and Val.Thus, we might conclude that feeding dairy cows diets deficient in 1 or more IAA may negatively affect the intake center which lies between the hypothalamus and the lower part of the brainstem leading to DMI depression.Mechanisms driving DMI depression with low CP diets await further characterization (Sinclair et al., 2014).

Milk Production, Composition, and N Efficiency
The impacts of the concentration of CP in diets on the performance of dairy cows have been thoroughly documented.Some research (Leonardi et al., 2003;Colmenero and Broderick, 2006) observed no impact on the production performance with diets that were comparatively low in CP content 13.2 to 15.1% CP, while others (Cressman et al., 1980;Wu and Satter, 2000;Lee et al., 2011) observed a reduction in the yields of milk or milk protein content.In the current study, milk production and composition (fat, protein, and lactose) did not differ among treatments except for ECM which tended to be higher for HP treatment compared with LPLM treatment (P = 0.06), despite milk production being numerically reduced by approximately 1 kg for MPLM and LPLM treatments compared with the HP treatment.The increase in DMI for the HP treatment in the current study elevated the NEL intake, and it was previously demonstrated that promoting the energy supply enhances the synthesis of milk protein (Hanigan et al., 1998a); as a result, the higher DMI of the HP diet may have been assumed to enhance the yield of milk protein, which it did not.It seems unlikely that the lack of a response was due to the experimental period length, given that the impact of dietary energy on milk production is observable within 10 d of altering the energy content of the diet (Aston et al., 1995).Zhao et al. (2019) observed that RPM supplementation partly recovered the reduction in milk protein yield due to feeding LP diets; but, the reduction was not improved when adding RP-Thr.Also, the reduction was mitigated by adding the 4 RPAA (Met, Thr, Leu, and Ile), indicating that Thr may not be needed for full recovery of milk protein production when all selected AA were supplied, but this needs more investigation.Therefore, the lack of change in plasma concentration of IAA in the current study indicates all treatments provided an AA supply that was similarly matched with body use; thus this could explain the lack of differences in milk protein among treatments.In a Latin square study with 4-week periods, Colmenero and Broderick (2006) found no impact on milk output and DMI when dietary CP was varied from 13.5% to 19.5% while observing a linear decline in FCM and a numerical trend for decreasing milk yield with the 13.5% CP diet.This might indicate that depending only on CP as a measure of protein supply may be inaccurate.At the same trend, Cabrita et al. (2011) observed a reduction in the production of milk and protein for cows that consumed diets containing 14.3% CP compared with a diet containing 15.7% CP despite balancing the AA profile of the diet.In contrast, the MPLM and LPLM diets in the present study maintained the milk composition at approximately the same level as the milk composition of the HP diet, where the percentages of milk protein, fat, and lactose were similar in all treatments.These observations are in agreement with a previous 4 × 4 Latin square experiment (Broderick et al., 2008) that supplemented 4 different CP diets with varied Met levels, where the performance of LP diets supplemented with Met was better than predicted relative to CON treatment.Lee et al. (2012)  plasma and a general trend of around 17% reduction in plasma IAA was observed, while in the present study, low MP diets did not affect plasma IAA, including His.In addition, all treatments met NASEM (2021) targets for IAA efficiency.Provision of metabolizable AA in concert with no changes in plasma IAA concentrations across treatment diets is consistent with the lack of change in milk yield and milk composition (milk fat, protein, and lactose), but surprising given the variation in DMI.It is possible the cows were near or at maximum production, and the LPLM diet plus some potential energy release from body stores (e.g., mobilize fatty acids from adipose tissues) was adequate to support all needs, thus a response could not be elicited by the HP and MPLM diets.It also seems unlikely that the lack of a response was due to the experimental period length, given that the impact of dietary protein on milk and milk protein is fully manifested within 1 week of the diet change regardless of whether cows were shifting from high protein to low or the reverse (Liu and VandeHaar, 2020).
Milk Urea Nitrogen (MUN) is considered an effective indicator of N excretion and N utilization efficiency in dairy cows (Jonker et al., 1998) as well as it is easy and convenient to measure (Guo et al., 2004).Previous experiments reported that MUN levels are elevated due to excessive protein feeding (DePeters and Ferguson, 1992;Broderick and Clayton, 1997).Decreasing MUN levels due to feeding MPLM and LPLM diets in the current study reflects greater N efficiency (Broderick et al., 2008).
Our findings regarding MNE are in agreement with the previous studies that found a significant improvement in N efficiency of about (8.3%) when reduced CP in the diet from 17.1% to 12.3% (Kalscheur et al., 2006) and an improvement of about (8%) when dietary CP was decreased from 18.7% to 14.8% (Ipharraguerre and Clark, 2005).This effect will only occur if the decline in milk N output is less than the decline in N intake; it is the balance of N that matters, not just the level of CP in the diet, and that explains why Apelo et al. ( 2014) observed no differences among treatments in values of N efficiency when reduced CP from 17% to 15%.These findings have economic and environmental impacts since they involve lowering the availability of one of the most costly dietary components (St-Pierre, 2012).

NASEM Model Evaluations
The NASEM (2021) model overestimated DMI by 1.5 kg/d for the HP diet and 3.25 kg for both MPLM and LPLM treatments by using the 2nd equation (based on animal/fiber); the feed factors considered in that equation are ADF, NDF, ADF/NDF, fNDF (forage content in the diet), and fNDFD (digestibility of forage NDF measured in vitro or in situ); however, the model did not capture the differences, so, one may conclude that the reduction in intake is not due to high starch and low ADF, it is due to CP.Moreover, there is no protein factor in the DMI equations, so one might guess that the equations would fail to capture the effect of protein.Furthermore, the data set included data from North American Holstein cows, and cows that were used in the present study are of New Zealand genetic stock and thus one would expect them to be somewhat different than North American cows.Dry matter intake is influenced by a variety of factors, including feed composition, dairy cow physiological state, environment, and management factors, and because of an incomplete understanding and variability of data to define the interactions among all of these factors, make it is challenging to achieve the constant accurate prediction of DMI in ruminants (NASEM, 2021).
The NASEM (2021) model also overestimated milk production by 2 kg for cows fed the HP diet and 1 kg for cows consumed both MPLM and LPLM compared with the observed milk yield for the 3 treatments (see Table 2).Lee et al. (2012) reported that NRC (2001) underestimated milk yield by 7 kg for cows consuming an MP deficient diet relative to those fed an adequate MP diet.NASEM (2021) demonstrated that the 2001 model dramatically overestimated MP allowable production due to an overestimate of the efficiency of conversion of MP to milk protein thus explaining the latter observations.The Nutrient Allowable milk prediction by NASEM (2021) is driven from milk protein and fat predictions, both of which were also underpredicted thus explaining the milk production bias.The milk protein prediction is primarily driven by the supply of 5 IAA and digestible energy intake (DEI).It is also scaled to an estimate of the maximum production achievable by the cows in the prevailing environment.If set too high by the user, the model will over-predict production of milk protein, and thus this may be the cause of the bias in protein.However, the lack of representation of some physiological mechanisms such as urea recycling (Huhtanen and Hristov, 2009) likely contributed.
The milk fat production equation is much more empirical, and thus the bias may simply represent differences in genetic potential and management between the cows used herein and those represented in the literature.

Plasma AA Concentration
The concentration of AA in plasma may provide an overall predictor of AA metabolic status in dairy cows (Morris and Kononoff, 2020).Several studies were conducted to evaluate the effects of individual IAA on the Seleem et al.: LOW PROTEIN DIET AND RUMEN-PROTECTED AMINO ACIDS plasma concentration of AA; however, the results were inconsistent.Some studies reported increased plasma AA concentration (Wang et al., 2010;Lee et al., 2015;Giallongo et al., 2016), but, in others no changes (Van den Bossche et al., 2023) or, in fact, decreased plasma AA concentrations (Swanepoel et al., 2010;Apelo et al., 2014).The reasons underpinning that inconsistency are still obscure and not fully understood.In the present study, plasma concentrations of IAA were not affected significantly by supplementation.In chicks, it was previously demonstrated that plasma free AA concentrations and the AA in dietary protein were highly correlated (Almquist, 1954); also, in ruminants, there is a correlation (Martineau et al., 2019;Letelier et al., 2022), although likely less than in non-ruminants, because of rumen activity.Thus, the lack of change in plasma concentration of IAA in the current study suggests all treatments provided the same AA supply, which is supported by the milk production observations (milk yield and milk composition), being equal among treatments.Lee et al. (2012) reported diets that were deficient in MP supplemented with RPL or RPM reduced IAA plasma concentration except for Met and Lys, indicating the supplies of the decreased AA (His, Thr, and Val) may have been insufficient with the MPdeficient diets.That is based on the assumption that the plasma AA profile is an indicator of AA supply to the mammary gland and that DAA have no impact on milk yield or milk protein synthesis (Doepel and Lapierre, 2010); and also suggesting that supplying RPL and RPM to the MP-deficient diets was, at least in part; successful in delivering the required digestible AA.
In the current study, plasma concentrations of DAA were mostly uninfluenced by treatments (see result section), except for increased Pro for MPLM.As mentioned above, it was previously reported that DAA has no impact on milk yield or milk protein synthesis (Doepel and Lapierre, 2010); however, NASEM (2021) reported significant effects of the combination of all DAA plus Arg, Phe, Thr, Trp, and Val on milk protein production, and thus the DAA effect cannot be ruled out.Evaluating the impacts of adding another IAA (His, Leu, and Ile) along with Met and Lys according to the NASEM (2021) concept is a topic worth further investigation.In light of the foregoing, since the plasma concentration of IAA did not differ among treatments along with adequate AA profile for the 3 treatments that were supported by production observations (milk yield and composition, BCS, and BW), we can say, under the conditions of the present study, low CP diets supplemented with RPL and RPM maintained the milk performance (milk yield and composition) to the same level of HP diet (16.4% CP).

CONCLUSIONS
Low MP diets (13% and 20% below target values) supplemented with RPL and RPM significantly reduced DMI but did not significantly reduce milk production and composition compared with a control diet at 95% of the target MP value for dairy cows producing 38 kg/d of milk.The supplemented low MP diets also had improved milk N efficiency and reduced MUN.The NASEM (2021) model underpredicted the production of milk protein; while overestimating milk yield by 2 kg for the HP diet and 1 kg for the supplemented diets; In addition, it failed to capture the observed reductions in DMI for the MPLM and LPLM treatments.Overall, rumen-protected Lys and Met supplementation maintained milk performance (milk yield and composition) of cows that were fed diets extremely deficient in MP to the same level of HP diet (16.4% CP), indicating the dietary IAA utilization efficiency will be improved with lowering the supply of MP-AA.Further investigations are needed under NASEM (2021) concept.
Seleem et al.: LOW PROTEIN DIET AND RUMEN-PROTECTED AMINO ACIDS

Table 2 .
Seleem et al.:LOW PROTEIN DIET AND RUMEN-PROTECTED AMINO ACIDS Chemical composition and evaluation of the experimental diets with or without RP-AA fed to multiparous Holstein lactating cows (% of DM unless otherwise noted)
supplemented a low MP diet (-338 g/d MP balance, 13.5% CP) with 30 g/d RPM and 100 g/d RPL which caused approximately the same FCM, milk fat, and lactose, and reduced milk protein yield compared with the MP-adequate diet.In that study, a 42% reduction in His concentration in Seleem et al.: LOW PROTEIN DIET AND RUMEN-PROTECTED AMINO ACIDS