Effects of non-digestible oligosaccharides on inflammation, lung health and performance of calves

Our objective was to determine the effects of non-digestible oligosaccharides (NDOs) on lung health and performance. Three hundred male Holstein-Friesian calves aged 18.0 ± 3.6 d received 1 of 6 treatments which they received for 8.5 weeks (period 1). Treatments included a negative control (CON), galacto-oligosaccharides (GOS) administered as a spray via the nose once daily (SPRAY), GOS administered via the milk replacer (MR) at 1% (GOS-L) and 2% (GOS-H), fructo-oligosaccharides (FOS) administered via the MR at 0.25% (FOS) and a combination of GOS and FOS administered via the MR at 1% and 0.25%, respectively (GOS-FOS). MR was fed twice daily. Feeding levels were equal between calves and increased progressively in time. Body weight was measured every 4 weeks and clinical health was scored weekly. Blood and broncho-alveolar lavage fluid (BALF) samples were collected bi-weekly from a sub-set of calves (n = 120). After period 1, all calves received the same control MR for 18 weeks until slaughter (period 2), during which general performance and clinical health were measured. Generally, infection pressure was high, with clinical scores and BALF pro-inflammatory TNFα concentrations increasing with time in period 1, which resulted in a high number of required group antimicrobial treatments (6 group antimicrobial treatments in 13 weeks, supplied to all calves). Average daily gain (ADG) adjusted to equal solid feed intake was increased for GOS-L (+61 g/d) compared with CON calves from experimental wk 1 to 5. Plasma white blood cell concentration tended to be lowered by GOS-L, plasma IL-8 concentration was reduced by all orally supplemented NDOs, plasma IL-6 was reduced by all NDO treatments except GOS-FOS and plasma IL-1β was reduced by all NDO treatments compared with CON, although this differed per time point for SPRAY. The


INTRODUCTION
Bovine respiratory disease (BRD) is a major health issue in dairy and veal calves.Clinical symptoms of BRD include coughing, nasal discharge, increased respiratory rate and rectal temperature, loss of appetite, apathy and reduced growth.Mortality due to BRD varies between 1.3% and 3.1% (Brscic et al., 2012, Pardon et al., 2012b, Pardon et al., 2013, Lava et al., 2016) Effects of non-digestible oligosaccharides on inflammation, lung health and performance of calves M. S. Gilbert,1 * † Y. Cai,23 † G. Folkerts,2 S. Braber, 2 and W. J. J. Gerrits 1  and BRD accounted for 56% of all morbidity cases in veal calves (Pardon et al., 2012b).Furthermore, BRD in early life in Holstein calves, as assessed by lung consolidation during the first 8 weeks of life, was associated with a reduced milk production during the first lactation (Dunn et al., 2018).
Bovine respiratory disease is a multifactorial disease, with different pathogens involved, such as Pasteurella multocida, Mannheimia haemolytica, Histophilus somni, Mycoplasma bovis, bovine viral diarrhea virus, bovine coronavirus, bovine respiratory syncytial virus, parainfluenzavirus type 3, bovine herpesvirus type 1, and bovine adenovirus (Autio et al., 2007, Pardon et al., 2011, Pardon et al., 2020).Due to the involvement of pathogenic bacteria in BRD, antimicrobial treatments are applied to combat or prevent BRD, and respiratory disease is the main cause for antimicrobial treatment supply in veal calves (Pardon et al., 2012a).Many pathogens involved in BRD are resistant to one or more antimicrobial agents (Catry et al., 2005, Pardon et al., 2011, Rérat et al., 2012), which is a major concern for both animal (Bengtsson and Greko, 2014) and human health (Ma et al., 2021).Therefore, strategies that prevent or reduce BRD are required.
One strategy is the use of oligosaccharides, such as bovine milk oligosaccharides or synthetic oligosaccharides, including oligofructose (fructo-oligosaccharides; FOS), galacto-oligosaccharides (GOS) and mannanoligosaccharides (MOS).In general, oligosaccharides are not digested by the host digestive enzymes and are fermented by the microbiota, resulting in changes in the microbial composition, such as promoting the growth of bifidobacteria and lactobacilli (Smiricky-Tjardes et al., 2003, Alizadeh et al., 2016, Castro et al., 2016, Boudry et al., 2017).Furthermore, non-digestible oligosaccharide (NDO) supplementation increases the concentration of fermentation products such as shortchain fatty acids and can increase intestinal barrier function (Alizadeh et al., 2016, Boudry et al., 2017).Besides the well-known prebiotic effects of NDOs on gut health, effects of NDOs on lung health have also been demonstrated.A reduction in lung inflammation and mortality was found in C57BL/6 mice fed pectinderived acidic oligosaccharides after a Pseudomonas aeruginosa-induced lung infection (Bernard et al., 2015).Supplementation of a mixture of GOS, long-chain FOS and low viscosity pectin prevented the development of lung emphysema following nasal lipopolysaccharide administration in BALB/c mice (Janbazacyabar et al., 2019).In pigs, supplementation with MOS reduced the increase in rectal temperature following infection with porcine reproductive and respiratory syndrome virus compared with pigs fed the control diet, although MOS supplementation did not reduce lung lesions caused by the infection (Che et al., 2011).Furthermore, GOS and alginate-oligosaccharides inhibited directly the growth of Escherichia coli in vitro (Asadpoor et al., 2021).Therefore, in addition to oral NDO supplementation, direct intranasal NDO supplementation provides an interesting strategy to reduce BRD as well.Although the exact mechanisms by which NDOs can affect lung health are not entirely known, the existence of a bidirectional gut-lung-axis seems evident (Anand and Mande, 2018) providing a target for influencing lung health.
Veal calves provide an interesting model to study the effects of NDOs on lung health.Veal calves originating from many different dairy farms are transported and mixed at a young age, resulting in a high infection pressure after arrival, with peak incidences of BRD between 2 and 6 weeks after arrival (Pardon et al., 2012b).In the current study, this natural exposure of veal calves to BRD pathogens was used to evaluate the effects of supplementation of NDOs.The objective of this study was to determine the effects of NDO supplementation, either via the milk replacer or via intranasal application, on lung health and performance of calves.We hypothesized that NDOs would reduce lung inflammation and thereby improve performance of calves.

MATERIALS AND METHODS
A project license was granted by the Central Committee for Animal Experimentation (The Hague, the Netherlands) after approval by the Animal Care and Use Committee of Wageningen University (AVD1040020185828, Wageningen, The Netherlands).This experiment was approved by the Animal Welfare Body of Wageningen University (2017.W-0017.001) and was conducted at the research facilities of the VanDrie Group (Scherpenzeel, the Netherlands).The study described in this paper was part of a larger project.Effects of NDOs, using calf primary bronchial epithelial cells, combined with selected parameters obtained in vivo as a reference, have been presented elsewhere (Cai et al., 2021, 2022a, Cai et al., 2022b).

Experimental Design
The experiment consisted of 2 periods.Period 1 started when calves arrived at the experimental facilities and lasted 8.5 weeks (experimental wk 1 to 9), in which NDO treatments were applied daily.Measurements were performed to evaluate effects of NDO treatments on (lung) health and performance.In period 2, no NDO treatments were applied and all calves received the same dietary treatment.Period 2 lasted from

Animals, Housing and Feeding
Calves were housed in a mechanically ventilated barn.Light was provided by daylight and artificial light was switched on between 0600 and 1800h.Calves were housed in pens (9 m 2 ) containing wooden-slatted floors.In the first 6 weeks after arrival, individual housing was applied (1.2 m 2 /calf) by placing stainless steel fences within the pens, allowing for individual monitoring of fecal consistency.After 6 weeks, the individual fencing was removed and calves were housed in groups of 5.In the first 3 weeks, a minimum temperature of 15°C was maintained and heat canons were used when needed.Thereafter, ventilation rates were adjusted to maintain a maximum ∆T of 5°C compared with the outside temperature and a maximum relative humidity of 80%.
A total of 300 male Holstein-Friesian calves of German origin were included in the study.This sample size was estimated based on Pardon et al. (2015), who calculated that in a similar setting of natural exposure to BRD pathogens, 56 calves per groups would be needed to demonstrate a reduction in BRD incidence by 25% (2 sided, 95% confidence, power = 0.8) in calves receiving either adequate or inadequate colostrum intake.We calculated the standard deviation of the work of Pardon et al. (2015) and used this in a power analysis.We estimated that 46 calves per group were needed based on a 25% reduction in BRD incidence and a standard deviation of 33% (one sided, 95% confidence, power = 0.8).We rounded the minimal group size to 50, because calves are housed in groups of 5.
Calves arrived at the facilities at 18.0 ± 3.6 d of age and 43.3 ± 3.2 kg of BW (mean ± SD) and were randomly assigned to pens.Pens were evenly distributed throughout the barn and the 6 treatments were blocked over pens to have an equal distribution of treatments throughout the barn.Calves received electrolytes solution upon arrival in the evening, and treatments started the next day.Treatments (n = 50 calves per treatment) included a negative control (CON) and 5 NDO supplementations; GOS administered as a spray via the nose (SPRAY), GOS administered via the milk replacer (MR) at 1% (GOS-L) and 2% inclusion (GOS-H), FOS administered via the MR at 0.25% inclusion (FOS) and a combination of GOS and FOS administered via the MR at 1% and 0.25%, respectively (GOS-FOS).NDOs of the oral NDO treatments were included in all MR feedings, which was twice daily.The SPRAY was administered once daily.The FOS used in the experiment was Frutalose ® OFP chicory oligofructose (Sensus, Roosendaal, the Netherlands) containing 97.6% DM and 92% FOS on DM basis.The GOS used was Vivinal GOS Syrup (FrieslandCampina DOMO, Borculo, the Netherlands) containing 74% DM and 59% GOS on DM basis (see Logtenberg et al., (2020) for a typical composition of the GOS in Vivinal GOS).These inclusion levels were based on our previous ex vivo studies, in which lower levels of FOS compared with GOS were required to reduce the Mannheimia haemolytica-induced release of cytokines and chemokines from calf primary bronchial epithelial cells (Cai et al., 2021(Cai et al., , 2022a)).For the SPRAY treatment, Vivinal GOS Syrup was diluted with saline to reduce the viscosity and 10 mL warm GOS solution (providing 0.99 g GOS DM) was sprayed into the nasopharynx once per day with an intranasal applicator (Rispoval, Zoetis B.V., The Netherlands) attached to a syringe.The researchers, except for the main researcher, were blinded to the treatment allocation during the study, except to the SPRAY treatment, as all researchers were involved in the application of the SPRAY.
The ingredient and nutrient composition of the MR in period 1 and 2 are in Table 1.Non-digestible oligosaccharides administered via the MR were included at the expense of lactose, corrected for the purity and DM of the NDO products used.Solid feed was composed of chopped wheat straw and concentrates.Solid feed was supplied from 1 week after arrival onwards.Straw and concentrates were supplied at a ratio of 16:84 for 2 weeks and was thereafter supplied at a ratio of 13:87.Concentrates comprised corn (412 g/kg), corn flakes (200 g/kg), lupines (195 g/kg), barley (102 g/kg), molasses (50 g/kg), vitamin-mineral mix (25 g/kg), urea (6 g/kg), palm oil (5 g/kg) and sodium-bicarbonate (5 g/kg) and were provided as flakes.Analyzed crude protein content of the concentrates was 140 g/kg DM.Feeding levels of both MR and solid feed were equal between calves and were based on a practical feeding scheme.Milk replacer increased progressively from 426 g/d to 1300 g/d in period 1 and from 1300 g/d to 2711 g/d in period 2. The MR was mixed with warm water (66°C) and supplied a temperature of approximately 42°C.In period 1, the MR concentration was 125 g/L and in period 2 it increased progressively from 125 g/L to 182 g/L.Solid feed supply increased progressively from 174 to 984 g/d in period 1 and from 1019 to 2865 g/d in period 2.
Calves were fed 2 equal MR meals at 0600 and 1600 h and were allowed access for at least 15 min, after which MR refusals were quantified.In case of MR refusals, calves were supplied with water to prevent dehydration.After wk 6 onwards, calves were restrained in the headlock for a maximum of 60 min.to enable individual MR feeding.Solid feed was provided after the morning MR meal.Solid feed refusals were removed twice a week and

Measurements
General health of the calves was monitored daily.Calves were weighed at arrival (wk 1) and in wk 5, 8, 13, 17, 21 and 25 with a digital body weight scale.All calves were weighed on the same day.Hemoglobin concentration was determined in jugular blood at arrival and in experimental wk 10, 14 and 18.A minimum hemoglobin concentration of 5.5 mmol/L was targeted.Iron (iron dextran, 20%) was injected intramuscularly when required, preceded by supplying acetylsalicylic acid (Preventiron, Dopharma, Raamsdonksveer, the Netherlands) with the MR the day before iron injections to prevent iron shock.Medical treatment was applied when required based on clinical signs of illness.Individual antimicrobial treatments included trimethoprim with sulphamethoxazol or paromomycin against diarrhea, oxytetracycline against leg problems and depocillin, florfenicol, ampicillin, amoxicillin or marbofloxacine against respiratory problems.Group medical treatment to all calves was applied when 10% of the calves had been treated within 5 d, or 5% of the calves had become ill within 24 h or when the situation required group treatment in the expert judgement of a veterinarian.Group antimicrobial treatments included doxycycline or tilmicosin against respiratory problems, and ampicillin or tylosin against serositis.All individual and group medical treatments were recorded.Calves that died were sent to the Animal Health Service (Deventer, the Netherlands) for autopsy.
Period 1. Measurements were performed for all calves or for a sub-set of calves.This sub-set of calves included 2 calves per pen (n = 20 per treatment), selected on BW at arrival closest to the average BW of all calves at arrival.Blood samples were collected by venipuncture in the jugular vein at arrival before the first MR feeding from all calves, and in experimental wk 3, 5 and 7 from the sub-set of calves.Blood was collected in 9 mL and 4 mL K 2 -EDTA tubes and were kept on ice for collection of plasma or kept at room temperature for analysis of hematological parameters, respectively.Plasma was collected after centrifugation at 2000 x g and 4°C for 20 min and was stored at −20°C pending cytokine and chemokine analyses.
experimental wk 7, fecal samples were collected directly from the rectum of the sub-set of calves and pH was measured in fresh fecal samples (Mettler Toledo, Se-venGo SG23).Broncho-alveolar lavage fluid (BALF) samples were collected from the sub-set of calves in experimental wk 2, 4, 6 and 8. To collect a BALF sample, the calf was restrained in the headlock.The nostrils were cleaned with cotton wool and 70% ethanol.Thereafter, a sterile flexible silicon tube was inserted via the nasal cavity into the trachea until the tube was fixed in a bronchus.Placement of the silicon tube in the lungs was verified before BALF sampling by a coughing reflex and/or sticking out of the tongue.In case the tube was inserted into the esophagus (i.e., no reflexes), the tube was retrieved, passing the epiglottis and a second attempt was performed.If the tube was swallowed and inserted into rumen, the tube was retrieved completely, cleaned and flushed with saline before the tube was inserted again.After appropriate fixation in the lungs, saline (30 mL, ~37°C) was inserted completely and immediately retrieved.BALF samples were stored on ice in 30 mL tubes until further processing in the lab the same day.
In experimental wk 9, 1 calf per pen was dissected from all NDO treatments supplemented with the MR and the control treatment (n = 50 in total), with the objective to evaluate degradation kinetics of oligosaccharides along the intestinal tract (not part of this manuscript).At dissection, lungs were scored for lesions, with scores ranging from 0 (no pneumonia) to 3 (severe pneumonia), according to Leruste et al. (2012).
Period 2. Clinical scoring was performed weekly as in period 1, except that rectal temperature was not measured.In addition, general performance (BW, MR and solid feed refusals, medical treatments, Hb) was measured as mentioned previously.In experimental wk 28, calves were slaughtered, carcasses were weighed and lungs were scored for lesions, with scores ranging from 0 (no pneumonia) to 5 (severe pneumonia).
BALF samples were filtered using a 70 µm cell strainer (Corning, NY).Subsequently, the BALF suspension was centrifuged at 400 x g and 4°C for 5 min.Supernatant was collected and stored at −80°C pending cytokine and chemokine analyses.The cell pellets were resuspended with 1mL cold fetal bovine serum.Cells were counted using a Cellometer Bright Field cell counter (Nexcelom Bioscience, Lawrence, MA).After cell counting, 0.5 × 10 6 cells were used for making cytospins, which were stained with Diff-Quick (Medion Diagnostics, Medion Diagnostics International Inc., Miami, FL) according to manufacturer's description.For differential BALF cell counts, a minimum of 400 cells was counted.
BALF supernatants were analyzed for TNFα at all time points and for IL-1β, IL-6 and IL-8 concentrations in wk 6 using the same ELISA kits as used for the plasma samples.

Statistical Analyses
Calves that died were removed from the data set.Statistical analyses were performed with SAS 9.4 (SAS Institute Inc., Cary, NC).Except for feed intake and adjusted ADG data (see below), all observations were performed on individual calves.Calves were, however, housed in groups of 5 per pen, and a random effect of pen was included in all models.In each model, treatment differences were evaluated using pre-set contrasts (CON vs. SPRAY, CON vs. GOS-L, CON vs. GOS-H, CON vs. FOS, CON vs. GOS-FOS, FOS vs. GOS-FOS, GOS-L vs. GOS-FOS and GOS-L vs. GOS-H), using the LSMESTIMATE statement including Bonferroni adjustment.Studentized residuals of each model were checked visually for homogeneity of variance and data transformations were applied to obtain homogeneity of variance when needed based on visual inspection of Q-Q plots.
For performance data, calves that died in period 1 (n = 9) or in the first 3 weeks of period 2 (n = 6) and 1 calf with excessive MR refusals (>40L, i.e., 9.5% of supply), were removed from the data set for period 1.For period 2, 2 calves with excessive milk refusals (>100L, i.e 5.8% of supply) were removed from the performance data set, in addition to the calves that died in period 2 (n = 9).Refusals of solid feed were recorded per pen.To account for this, ADG was adjusted to the average realized level of solid feed intake for each period.To obtain this adjusted ADG, first the residuals of the GLM model ADG = solid feed intake were obtained.These residuals represent the difference between the observed and predicted ADG based on the model.Subsequently, these residuals were added to the mean ADG of the specific period.This adjusted ADG is the ADG at equal solid feed intake.Performance data per period, blood Hb and WBC concentration at arrival, and fecal pH (wk 7) were analyzed for the treatment contrasts as described above using the MIXED procedure.ADG and carcass weight data were squared to obtain homogeneity of variance.For BW data and ADG in period 2, BW at arrival was included as a covariate in the model.
Continuous parameters measured over time (blood, BALF, rectal temperature) were analyzed using the MIXED procedure with treatment, time and treatment × time interaction as fixed effects, and included time as a repeated statement with calf as unit.Treatment contrasts were evaluated as described above, but if the interaction effect of the main model was significant, these treatment contrasts were evaluated per time point separately.The covariance structure per model was selected based on fit statistics (lowest AIC and BIC).For blood parameters, the values at d 1, before the first feeding, were included as a covariates.Blood concentrations of IL-6 were inverse-transformed and blood concentrations of IL-8 and white blood cells (WBCs) were log-transformed.Blood percentages of neutrophils were arc-sin transformed, percentages of lymphocytes and eosinophils were squared, and percentages of monocytes and basophils were angular-transformed.BALF cell concentrations were square-root transformed and BALF macrophage and neutrophil percentages were arc-sin transformed.Percentages of BALF lymphocytes were very low with many calves having 0% lymphocytes and these data were analyzed for treatment effects per week separately, using the Kruskal-Wallis test including the Dwass-Steel-Critchlow-Fligner method for evaluating treatment comparisons.Concentrations of TNFα and IL-6 in BALF were square-root transformed and BALF IL-8 and IL-1β concentrations were log-transformed.Rectal temperature data were inverse-transformed.
Discrete parameters measured over time (clinical and fecal scores) were analyzed for treatment contrasts per week using the GLIMMIX procedure with a Poisson distribution.Fecal scoring was performed daily for the first 9 d, and the median fecal score for wk 1 and 2 was obtained and used in the model.Lung lesion scores at dissection and slaughter were analyzed for treatment contrasts using the GLIMMIX procedure with Poisson distribution as well.Individual antimicrobial treatment supply was analyzed for treatment contrasts for period 1 and 2 separately, using the GLIMMIX procedure with a negative binomial distribution, including all calves.Mortality was analyzed for treatment contrasts using the GLIMMIX procedure with a binary distribution.Treatment contrasts were always evaluated using Bonferroni adjustment.
Principal component analysis (PCA) was performed on blood (wk 7) and BALF (wk 6) inflammatory parameters to evaluate interrelations.The FACTOR procedure with the principal axis method was used and prior communality estimates were set at 1.The extracted principal components (PC) were subjected to varimax rotation, and PCs with an eigenvalue > 1.5 were retained.Variables with a loading ≥ 0.40 or ≤ −0.40 in the rotated factor pattern, were considered to load on that PC.
Differences were considered significant when P < 0.05 and considered a trend when P < 0.10.Results are expressed as non-transformed means and their pooled SEM P-values of treatment contrasts represent Bonferroni-adjusted P-values.

RESULTS
At arrival, calves weighed 43.3 ± 3.2 kg of BW (mean ± SD) and this did not differ among treatments (Table 3).Blood Hb and WBC concentration at arrival did not differ among treatments (P > 0.10 for all treatment contrasts), and averaged 5.5 ± 0.08 mmol/L and 8.1 ± 0.17 × 10 9 /L, respectively.

Performance and Clinical Health
Adjusted ADG in the first 4 weeks after arrival was higher for the GOS-L (+ 61 g/d; P = 0.001) compared with the CON calves (662 ± 11 g/d; Table 3).Body weight at arrival explained variation in body weight at later time points (P < 0.001).GOS-L calves (65.3 ± 0.6 kg; P = 0.002) were heavier and GOS-H calves (64.0 ± 0.6 kg; P = 0.06) tended to be heavier compared with CON calves (62.6 ± 0.7) at experimental wk 5. GOS-L calves (86.7 ± 0.8 kg) were also heavier than CON calves (82.4 ± 0.9 kg) in wk 8 (P = 0.03).Adjusted ADG in period 2 (1437 ± 18.3 g/d on average) and body weight at the end of period 2 (254 ± 2.4 kg on average) did not differ among treatments.
Fecal consistency scores did not differ among treatments (P > 0.10), but did change over time (P < 0.001), which was generally reflected by higher fecal scores in experimental wk 1 to 3, compared with fecal scores in wk 4 to 6.In wk 1, fecal consistency score averaged 3.8 ± 0.10, whereas in wk 6, fecal consistency averaged 2.6 ± 0.10.When categorizing fecal consistency score 4 and 5 as diarrhea, 59% of the calves had diarrhea in wk 1 and 8% of the calves had diarrhea in wk 6. Fecal pH (wk 7) averaged 6.5 ± 0.12 and did not differ among treatments (P > 0.10).
In period 1, 66 individual antimicrobial treatments were administered, of which 78.8% related to respiratory problems, 13.6% related to diarrhea, 1.5% related to (poly)serositis and 6.1% related to other problems (e.g., leg problems).Individual antimicrobial treatments did not differ among treatments in period 1 (P > 0.10).In this period, 4 antimicrobial group treatments were applied.Three of these group treatments were related to respiratory infections and the last one to (poly) serositis.In the same period, 7 calves died and 2 were euthanized (2 from the SPRAY, 5 from the GOS-L, 1 from the GOS-H and 1 from the FOS treatment) and 2 calves removed from the trial (due to excessive milk refusals and poor health; 1 from the SPRAY and 1 from the FOS treatment).In period 2, 88 individual antimicrobial treatments were administered, of which 48.9% related to respiratory problems, 1.1% related to diarrhea, 14.8% related to (poly)serositis and 35.2% related to other problems.Individual antimicrobial treatments did not differ among treatments in period 2 (P > 0.10).Two antimicrobial group treatments were applied related to (poly)serositis in the first 3 weeks of period 2. In period 2, 9 calves died (4 from the CON, 1 from the SPRAY, 1 from the GOS-L, 2 from the GOS-H and 1 from the GOS-FOS treatment), of which 6 calves died in the first 3 weeks of period 2. Mortality did not differ among treatments (P > 0.10).For 12 out of 18 calves that died, the cause of death was likely (poly)serositis, related to infection with Mannheimia haemolytica.

Inflammatory Parameters in Blood and BALF
Blood inflammatory parameters are presented in Table 4. White blood cell concentration changed over time (P < 0.001), with the highest concentration in wk 7 (9.3 ± 0.56 × 10 9 cells/L) and WBC values at arrival explained variation in WBC concentration later in time (P < 0.001).Overall, WBC concentration of GOS-L calves tended to be lower compared with CON calves (P = 0.06).Differential cell counts were not affected by treatments but changed over time (P < 0.001) and values upon arrival explained variation in neutrophil (P = 0.03), monocyte (P < 0.001) and basophil percentages (P = 0.09) later in time.
Plasma TNFα was not affected by treatment, but increased with 43% from wk 5 to 7 (P < 0.001).Plasma IL-8 was 66 times higher in wk 7 than in wk 5 (P < 0.001) and orally supplemented NDOs (P < 0.001 for all) lowered plasma IL-8 compared with the CON treatment, but SPRAY treatment did not (P = 1.00).Feeding GOS-L separately resulted in a lower plasma IL-8 concentration compared with feeding GOS-L and FOS together (P < 0.001).Feeding FOS separately tended to lower plasma IL-8 concentration compared with feeding GOS-L and FOS together in wk 5 (P = 0.08), but not in wk 7 (P = 0.39), which resulted in an treatment × time effect (P = 0.01).Plasma IL-6 decreased 12% from wk 5 to 7 (P < 0.001) and was lower for all NDO treatments compared with the CON treatment.Feeding GOS-L (P = 0.002) or FOS (P = 0.07) separately resulted in a lower plasma IL-6 concentration compared with feeding them together.Plasma IL-1β decreased 80% from wk 5 to 7 (P < 0.001) and generally all NDO treatments lowered plasma IL-1β concentration compared with CON.The treatment × time effect (P = 0.01) was caused by a lower IL-1β concentration for  2 Body weight at arrival was included as a co-variable in the model (P < 0.001).
3 Body weight at arrival was included as a co-variable in the model (P = 0.004).

4
ADG adjusted for solid feed intake per pen of calves.
SPRAY compared with CON in wk 5 (P = 0.01), but not in wk 7 (P = 0.19), and because feeding GOS-L (P = 0.03) or FOS (P = 0.02) separately resulted in a lower plasma IL-1β concentration compared with feeding them together, but only in wk 7. None of the plasma cytokine or chemokine concentrations differed between the GOS-L and GOS-H treatment (P > 0.10).
Inflammatory parameters in BALF are presented in Table 5. Concentrations of total cells in BALF were not affected by treatment and were higher in wk 6 and 8 compared with wk 2 and 4 (P < 0.05).Differential cell counts in BALF changed over time (P < 0.001), with macrophages being the most prevalent cell type in wk 2 (78.0 ± 4.8%) and 4 (71.4 ± 5.5%), decreasing to 45.1 ± 5.6% and 46.6 ± 6.2% in wk 6 and 8, respectively.This coincided with a relative increase in neutrophils from 21.9 ± 4.8% and 28.5 ± 5.5% in wk 2 and 4, respectively, to 54.2 ± 5.7% and 52.9 ± 6.2% in wk 6 and 8, respectively.Lymphocytes represented a small fraction of the cells in BALF, with 0.1 ± 0.07% in wk 2 and 0.6 ± 0.21% in wk 6. Macrophage percentage (interaction effect P = 0.03; Figure 2) was higher for GOS-L than for CON (P = 0.005), but only in wk 6, and tended to be higher for GOS-FOS compared with FOS (P = 0.08), but only in wk 2. This coincided with a lower percentage of neutrophils (interaction effect P = 0.01; Figure 2) for GOS-L (P = 0.002) compared with CON calves in wk 6. Concentrations of TNFα in BALF increased in time (P < 0.001) and were highest in wk 6 and 8. Concentrations of TNFα were lower for GOS-H (P = 0.02) and tended to be lower for GOS-L (P = 0.06) compared with CON.In wk 6, IL-8 concentration was lower for GOS-L and GOS-H (P < 0.001 for both) compared with CON and it was lower for GOS-L than for GOS-FOS calves (P = 0.005).Concentrations of IL-6 in BALF were lowered by SPRAY (P = 0.02), GOS-L (P < 0.001) and GOS-H (P = 0.01) and tended to be lowered by GOS-FOS (P = 0.07) compared with CON.Concentrations of IL-1β were lowered by SPRAY (P = 0.01), GOS-L (P < 0.001), GOS-H (P = 0.005) and FOS (P = 0.02) compared with CON and it was lower for GOS-L than for GOS-FOS calves (P = 0.01).None of the BALF cytokine or chemokine concentrations differed between the GOS-L and GOS-H treatment (P > 0.10).

Interrelations between Blood and BALF Inflammatory Parameters
Loadings per PC extracted from PCA on BALF (wk 6) and blood parameters (wk 7) to evaluate interrelations are presented in Table 6.Four PCs with an eigenvalue >1.5 were retained, accounting for 54% of the total variance.Total cell concentration, neutrophil percentage and TNFα concentration in BALF loaded positively and percentage of macrophages and lymphocytes in BALF loaded negatively on PC1.Plasma WBC concentration and neutrophil percentage loaded positively and percentages of lymphocytes and basophils loaded negatively on PC2.Interleukin-1β and IL-8 in both BALF and plasma loaded positively on PC3.

DISCUSSION
In the current study, the effects of NDOs on lung health and performance of calves were studied.Veal calves provided a suitable model for this, because of the high natural exposure to BRD pathogens, generally resulting in peak incidences of BRD between 2 and 6 weeks after arrival (Pardon et al., 2012b).Similarly, in the current study, clinical scores, measured using a BRD-scoring system, increased in time, indicating the presence of BRD in veal calves.Pro-inflammatory TNFα concentrations in BALF also increased in time, which points to lung inflammation.Hence, 6 group antimicrobial treatments were applied in 13 weeks to prevent potential excessive morbidity and mortality.This was high compared with the number of group antimicrobial treatments typically required at this facility (i.e., 2-3 group antimicrobial treatments).Overall, this indicates that (respiratory) infection pressure was sufficiently high to evaluate NDO effects on lung health.
The inclusion levels of NDOs were based on our previous ex vivo observations using calf primary bronchial epithelial cells.Lower levels of FOS compared with GOS were required to reduce the Mannheimia haemolytica-induced release of cytokines and chemokines (Cai et al., 2021(Cai et al., , 2022a)).In addition, the inclusion levels were chosen to prevent NDO-induced diarrhea, because diarrhea has been observed when supplementing GOS at 3.35% in MR fed to Holstein calves (Castro et al., 2016).Fecal consistency was monitored closely during the first 9 d of the experiment to adjust NDO inclusion levels if needed.However, fecal consistency was not higher for any of the NDO treatments compared with the CON treatment, indicating that the current NDO inclusion levels did not induce diarrhea.Generally, fecal consistency decreased in time, which is in line with veal calf practice, where diarrhea mainly occurs in the first 3 weeks after arrival, as demonstrated by Pardon et al. (2012b).
Blood and BALF samples were used to evaluate the effects of NDOs on inflammatory responses.Principal component analysis was used to evaluate interrelations between inflammatory parameters in BALF and blood.The PCA retained 4 PCs with few cross-loadings.Interestingly, WBC concentrations and relative shifts in cell types loaded on separate PCs for BALF (PC1) and blood (PC2).These loading patterns within sample type represented well-known relations during inflammation, with increased total WBC concentrations being related to an increase in neutrophils.For instance, in mice with LPS-induced acute lung injury, the total cell increase 2 IL-8 tended to be lower for FOS than for GOS-FOS calves in wk 5 (P = 0.080), but did not differ in wk 7 (P = 0.39), resulting in a treatment × time effect (P = 0.010).
3 IL-1β was lower for SPRAY compared with CON in wk 5 (P < 0.001), but not in wk 7 (P = 0.19); IL-1β did not differ between FOS and GOS-FOS in wk 5 (P = 0.11), but was lower for FOS compared with GOS-FOS in wk 7 (P = 0.018); and IL-1β did not differ between GOS-L and GOS-FOS in wk 5 (P = 0.25), but was lower for GOS-L compared with GOS-FOS in wk 7 (P = 0.030); resulting in an interaction effect (P < 0.001).2014, Cai, 2021).In addition to blood and BALF parameters, clinical scoring was used in this study to evaluate the effects of NDOs.Clinical scoring can be insensitive to diagnosing (subclinical) lung infections (van Leenen et al., 2020) and these scores were done only once a week, indicating that disease can be missed.Principal components were obtained using BALF and blood parameters of wk 6 and 7, respectively, and interestingly, PC2 correlated with clinical scores in wk 7 (r = 0.34, P > 0.001) and PC3 correlated with clinical scores in wk 6 (r = 0.23, P = 0.01).

Table 4 (Continued
Feeding GOS improved the performance of calves, mainly in the first 4 weeks after arrival.Solid feed intake differed between pens of calves and, therefore, ADG was adjusted for solid feed intake.This adjusted ADG is the ADG at equal solid feed intake and is therefore a measure of feed efficiency.GOS supplemented at 1% via the MR increased adjusted ADG in the first weeks after arrival and resulted in a higher BW in wk 5 and 8. Similarly, in piglets, on-top supplementation of GOS also improved growth performance, associated with improved gut function (Tian et al., 2018) and reduced systemic inflammation (Xing et al., 2020).In addition to performance, GOS reduced systemic and respiratory inflammation.Orally supplemented GOS reduced IL-6, IL-8 and IL-1β concentrations in both plasma and BALF, and GOS-L also reduced the percentage of neutrophils in BALF in wk 6, a week with high clinical scores.Generally, the GOS dose (1% or 2%) had little impact on inflammatory parameters.Several mechanisms might underlie the effects of GOS.Galacto-oligosaccharides can change the composition of gut microbiota (Castro et al., 2016) and exert anti-inflammatory effects via gut-lung axis (Anand andMande, 2018, Cai et al., 2020).Although short-chain fatty acids have been identified as key mediators of the gut-lung axis (Dang and Marsland, 2019), it can be questioned whether they also played an important role in the current study.Calves received solid feed, resulting in fermentation in the rumen and fermentation of ruminal undegraded material in the hindgut.For comparison, in experimental wk 9, GOS-H calves received 25 g/d of GOS compared with 984 g/d of solid feed, which is composed mainly of carbohydrates, suggesting that NDO supplementation only marginally contributed to total short-chain fatty acid production.Microbiota-independent mechanisms could also underly the effects of GOS.Part of the GOS may be absorbed into the systemic circulation (Difilippo et al., 2015) and directly inhibit the activation of the pro-inflammatory signaling pathway caused by pathogens in the lungs, including the TLR4 pathway (Cai et al., 2022a).
Several NDOs exhibit antibacterial effects, either by inhibiting the growth of pathogens or by killing them directly (Asadpoor et al., 2021).We found that GOS can lower the viability of Mannheimia haemolytica in vitro (Cai et al., 2022b).Therefore, GOS were also locally administered via intranasal application in the current study, to evaluate the direct effects of GOS on lung health.The intranasal GOS dosage (0.99 g DM per day) was based on the dilution required to obtain a spray flow of the GOS syrup, which is viscous when undiluted, and on a maximum volume of 10 mL to minimize fixation time of calves to administer the spray.The amount of GOS provided via nasal spray was lower compared with GOS provided via the MR (e.g., GOS-spray provided 4.1 times and 12.2 times less GOS compared with the GOS-L treatment at the start and in wk 8 of the experiment, respectively).Likely, a lower dosage is required to enable effects in the lung directly compared with indirectly via the gut.Indeed, intranasal GOS application did reduce some of the systemic and respiratory inflammatory parameters in a comparable manner to orally supplemented GOS.Whether this GOS-spray effect is due to GOS reaching the lungs directly, or GOS being ingested after mucociliary clearance from the trachea or bronchus cannot be separated.A preliminary test (approved by the Animal Welfare Body of Wageningen University, 2017.W-0017.002)using 4 male Holstein-Friesian calves and spray containing Evans blue, revealed that the intranasal administration likely supplied GOS into the trachea and esophagus (unpublished data).Despite the reduction in lung and systemic inflammation, the GOSspray treatment tended to decrease ADG from wk 5 to 8. Likely, this reduction in clinical performance with GOS-spray is due to the stress of daily handling the calves during the spray administration.
Feeding 0.25% FOS in the MR reduced inflammatory cytokine and chemokine concentrations in plasma and reduced IL1β concentrations in BALF, but did not improve performance or clinical health.In contrast, supplementation of FOS (10 g/d) to Simmental × Red Holstein veal calves tended to improve BW gain between 10 and 13 weeks of age (Kaufhold et al., 2000).Supplementation of FOS was higher compared with the current study (where FOS supply increased from 1.1 to 3.2 g/d from experimental wk 1 to 9, respectively), FOS was supplied on-top and calves were not fed any solid feed, potentially explaining the difference with our study.Mechanisms underlying the effects of FOS, beside effects on the microbiota (Philippeau et al., 2010), could include the prevention of epithelial barrier dysfunction by pathogens.We showed that FOS prevented Mannheimia haemolytica-induced epithelial barrier dysfunction and inflammatory responses in vitro, which might be related to the inhibition of TLR5-mediated pro-inflammatory signaling (Cai et al., Gilbert et al.: OLIGOSACCHARIDES AND LUNG HEALTH IN CALVES 2021), demonstrating the different mechanisms of action between GOS (TLR4-mediated; Cai et al., 2022a) and FOS (TLR5-mediated; Cai et al., 2021).
The effect of a combination of GOS (1%) and FOS (0.25%) on lung health and performance was also evaluated in the current study.Feeding GOS-FOS reduced inflammatory cytokine and chemokine concentrations in plasma and tended to reduce IL6 concentrations in BALF.However, feeding the combination of GOS and FOS was not more effective than feeding GOS or FOS alone, e.g., calves fed GOS-FOS had higher plasma and BALF cytokine and chemokine concentrations compared with calves fed GOS-L, and had higher or tended to have higher plasma cytokine and chemokine concentrations compared with calves fed FOS.This suggests that even though the modes of action of GOS and FOS likely differ, this GOS-FOS combination does not exhibit synergistic effects in calves.Combinations of GOS and FOS are already included in infant formulas within Europe (Akkerman et al., 2019).Infants receiving formula with GOS and long-chain FOS (9:1) had increased numbers of fecal bifidobacteria and lactobacilli (Moro et al., 2002) and had reduced incidences of acute diarrhea and respiratory tract infections (Bruzzese et al., 2006).In mice, the combination of GOS and long-chain FOS decreased the ovalbumin-induced inflammation (Vos et al., 2007).In such studies, often long-chain FOS are used, typically with a 9:1 ratio between GOS and FOS, to resemble the molecular size distribution of human milk oligosaccharides (Knol et al., 2005).In the current study, shorter chain FOS (degree of polymerization 2-10) in a 4:1 GOS: FOS ratio were used, based on our previous ex vivo studies using calf primary bronchial epithelial cells (Cai et al., 2021(Cai et al., , 2022a)).In humans, immune responses differed depending on the chain length of FOS/inulin (Vogt et al., 2017).Whether different structures of FOS or different ratios between GOS: FOS are more suitable for calves requires further research.
Potential carry-over effects of feeding NDOs on clinical health and performance were also assessed.Infection pressure was lower in period 2, with clinical scores averaging 1.3 ± 0.01 in period 2 compared with 3.1 ± 0.18 in wk 6 of period 1 and clinical scores and performance in the 18 weeks following NDO supplementation were not affected.

CONCLUSIONS
All tested NDOs reduced systemic and lung inflammation in calves under high natural infection pressure.For GOS supplemented via the MR at 1%, this resulted in an increased performance during the first 4 weeks after arrival.There was no dose-dependent effect of GOS.
Combining GOS (1%) and FOS (0.25%) was generally less effective in reducing inflammation compared with feeding GOS (1%) or FOS (0.25%) alone, suggesting the absence of a synergistic effect.Galacto-oligosaccharides were also administered intranasally, and this also reduced inflammation, but tended to negatively impact performance, likely due to the stress of daily handling of the calves.Overall, this study demonstrates the potential of NDOs to alleviate systemic and respiratory inflammation in calves.
Gilbert et al.: OLIGOSACCHARIDES AND LUNG HEALTH IN CALVESexperimental wk 10 to slaughter at experimental wk 28 and included general (performance) measurements.
Gilbert et al.: OLIGOSACCHARIDES AND LUNG HEALTH IN CALVES Gilbert et al.: OLIGOSACCHARIDES AND LUNG HEALTH IN CALVES
of least squares means differences were estimated following Bonferroni adjustment.

Figure 2 .
Figure2.Percentage of macrophages (A) and neutrophils (B) in broncho-alveolar lavage fluid (BALF) of veal calves not supplemented with non-digestible oligosaccharides (CON) or supplemented with non-digestible oligosaccharides (SPRAY, galacto-oligosaccharides (GOS) applied with a spray into the nose; GOS-L, low (1%) supplementation of GOS via the milk replacer (MR); GOS-H, high (2%) supplementation of GOS via the MR; FOS, supplementation of fructo-oligosaccharides (FOS, 0.25%) via the MR and GOS-FOS, supplementation of GOS (1%) and FOS (0.25%) via the MR).Labels: treatments differ from each other, with P < 0.01 (**) or there is a trend for a difference at P < 0.10 ( †), where treatment contrasts are evaluated per week separately because of a treatment × week interaction (P = 0.03 for panel A and P = 0.01 for panel B).

Table 1 .
Gilbert et al.: OLIGOSACCHARIDES AND LUNG HEALTH IN CALVES Ingredient and nutrient composition of the experimental milk replacers

Table 2 .
Gilbert et al.: OLIGOSACCHARIDES AND LUNG HEALTH IN CALVES Clinical scoring used for assessing lung health in calves 1 Pardon (2012)Pardon (2012).

Table 6 .
Loadings per principal component (PC) extracted by principal component analysis with varimax rotation performed on inflammatory parameters in broncho-alveolar lavage fluid (BALF) and blood samples of veal calves 1