Seven clinically healthy lactating (on average 92 days from parturition, range 30–123 days) primiparous cows (three Finnish Ayrshire and four Holstein-Friesian) were used as experimental animals. Experimental Escherichia coli mastitis was induced in one quarter of each cow twice at an interval of 14 days. The cows were housed in tie stalls and accustomed to the environment and handling for two weeks before the experiment. The cows were fed with good quality hay, silage and concentrated grain according to their energy requirements. Water was available ad libitum. The cows were milked twice a day, at 8 am and 4 pm. The milk composite somatic cell count (SCC) of the cows was less than 100 000 cells/ml and no bacteria were isolated from the milk before the challenges. Mean SCC in the milk of the test quarter before the first challenge was 15 200 cells/ml (range 3 000–57 000 cells/ml) and before the second challenge 14 300 cells/ml (range 5 000–25 000 cells/ml), respectively. Milk yield of the inoculated quarter before the first challenge was on average 3.8 kg (range 3–5.1 kg) and before the second 3.8 kg (range 3–5.3 kg). Mean total daily milk yield was 24.2 kg before the first challenge (range 18.6–31.5 kg) and 22.9 kg before the second challenge (range 15.7–33.0 kg). All cows were treated with flunixin meglumine (dose 2.2 mg/kg) once at 12 hours post challenge (PC), when the first clinical signs were observed, to comply with animal welfare requirements. The Ethics Committee of the Faculty of Veterinary Medicine, Helsinki, Finland approved the study protocol.

The Escherichia coli strain, FT238, isolated from clinical mastitis and used previously, was selected for experimental inductions 2728. The inoculation dose was prepared as described before 2928. One udder quarter of each cow was infused via the teat canal with an average dose of 1500 cfu of E. coli (range 1400–1600 cfu) and the inoculation was repeated after 14 days in another udder quarter. The quarters were infused four hours after the evening milking.

Blood samples were collected from the jugular vein of each cow before challenge and 12, 16, 20, 24, 36, 44, 60, 68 and 156 hours post challenge (PC). Serum was separated and kept frozen at -70°C for later determinations of SAA, Hp and LBP. EDTA blood was collected for leukocyte count (WBC) determination. Aseptic milk samples were collected from the experimental and contralateral quarter before the challenge and 12, 20, 36, 44, 60, 68, 84,108, 132 and 156 hours PC for bacteriology, SCC, N-asetyl-β-D-glucosaminidase (NAGase) activity, SAA, Hp and LBP determinations.

Systemic and local signs were monitored throughout the experimental period of 6 days: during the first day every 4 hours and thereafter twice a day at the time of milking. Heart rate, rectal temperature, rumen motility, appetite and general attitude were evaluated. The systemic signs were scored on a three point scale, 1 = no signs to 3 = severe signs; half values were also used 26. The udder was palpated for soreness, swelling, hardness and temperature, and appearance of milk assessed visually for clots, colour changes and changes in consistency. The local signs were scored on the same scale as systemic signs: milk 1 = normal to 3 = serous or clotty milk and udder 1 = no changes to 3 = severe swelling and soreness in the quarter. Cows with scores ≤1.5 were recorded as having mild mastitis, those with scores >1.5 but ≤2.5 as having moderate mastitis and those with scores from >2.5 to 3 as having severe mastitis. The milk yield from the experimental quarter and the total milk yield were measured before challenge and thereafter until the end of the experimental period.

Bacterial counts in the milk were determined by preparation of 10-fold dilution series of milk in sterile saline. Bacteria were cultured on blood agar at 37°C for 24 hours using serial dilutions and counted using a routine plate count method. Milk SCC was measured in Valio Ltd. Laboratories, Finland using a fluoro-optical method (Fossomatic-instrument, Foss Electric, Hillerød, Denmark). SCC values over 30 × 10⁶ cells/ml were recorded as 30 × 10⁶ cells/ml. Milk NAGase activity was measured using the fluorogenic method of Kitchen and co-workers (1978) 30 with a microplate modification developed by Mattila 31. Inter-assay and intra-assay CVs for NAGase activity were for the high control <5% and for the low control 7%. Values over 2.5 pmol/min/ml were expressed as >2.5 pmol/min/ml.

The concentration of SAA in serum and milk was determined by using a commercial ELISA test (Tridelta Development, Wicklow, Ireland). Serum and milk samples were initially diluted 1:500 and 1:50, respectively. For very high SAA values, samples were diluted as necessary up to 1:5000 for serum samples and up to 1:15000 for milk (maximum concentrations 750 mg/l and 2250 mg/l, respectively). The inter-assay and intra-assay coefficients of variation (CV) for SAA analysis were <10% and <7%. Milk and serum Hp concentrations were determined using the method based on the ability of Hp to bind to haemoglobin 32 and using tetramethylbenzidine as the substrate 33. The assay is aimed for determining of Hp in serum but was here adapted to be used for milk 34. Optical densities of the formed complex were measured using a spectrophotometer at 450 nm (Multiskan MS, Labsystems, Vantaa, Finland). Lyophilized bovine acute phase serum was used as a standard and calibration was according to the European Union concerted action on standardization of animal APPs (number QLK5-CT-1999-0153). The inter-assay and intra-assay CVs for Hp analysis were <10% and <12%.

LBP concentrations in serum and milk were determined with a commercially available LBP ELISA kit, cross-reacting with bovine LBP (LBP ELISA for various species, Hycult Biotechnology, Uden, The Netherlands). Milk and serum samples were initially diluted 1:500 and 1:1000 respectively, and assayed following the instructions of the manufacturer. For high concentrations, milk was diluted up to 1:5000 and serum up to 1:2000. The optical density at 450 nm and a correction wavelength of 550 nm were measured on a spectrophotometer (Multiskan MS, Labsystems). The LBP concentration was determined by extrapolation using linear regression from a standard curve of known human LBP concentrations. The inter-assay and intra-assay CVs for LBP analysis were <13% and <9%.

Leukocyte count (WBC) was determined 24 hours after sampling using an automated multiparameter analyzer with software for animal samples (Cell-Dyn 3700 System, Abbot Diagnostic Division, Abbot Park, IL, USA).

Linear random-intercept models were used to explore time trend differences between challenge times in milk production data, milk SCC, milk NAGase, WBC and all APP measurements. Bacterial counts in milk and local and systemic sign differences between challenges were tested using generalized linear mixed models in which a Poisson distribution was used for response variables. The cow was included as a random factor. Polynomials for time in increasing order and their interactions with challenge occasion were fixed factors and were added until significant, for modeling changes in time at both challenges. Overall time trend differences between challenges were tested with an F-test. As there were different intervals between sampling, isotropic spatial exponential correlation structures were used for modeling serial correlations of repeated measurements within cows. Logarithmic transformation of milk SCC, NAGase and APPs in milk and serum was used. The nlme-package 35 with statistical software R 2.5.0 36 was used for fitting linear random-intercept models and generalized linear mixed models were fitted using the GLIMMIX procedure 37 software with the SAS/STAT 9.1 (SAS Institute Inc., Cary, NC, USA).

After both challenges all cows became infected and developed clinical mastitis within 12 hours after inoculation. One cow was excluded from the experiment because of acute spontaneous coliform mastitis after the first challenge. All cows showed systemic and local inflammatory response after both challenges. Systemic response began within 12 hours, being moderate in all cows at 12 hours PC based on the clinical severity scoring system. Systemic signs disappeared in cows after both challenges until 36 hours PC. Local signs were still recorded at the end of the experimental period of 6 d after the first challenge, but disappeared by 60 hours PC after the second challenge. In both challenges, cows developed a similar systemic response, but their local responses varied more. After the second challenge, local clinical signs were significantly milder (p < 0.05) but no statistically significant differences were noted in systemic signs (Figure 1).

The daily milk yield was at its lowest 36 hours PC after both challenges, being on average 16 kg after the first challenge and 17.1 kg after the second. After 6 days PC the total milk yields in both groups returned to pre-challenged levels. The total daily milk yield during the experimental period was significantly higher for the second challenge (p < 0.05). The milk yield of the infected quarter was lowest at 36 hours PC, being 1.1 kg (range 0 – 2.5 kg) after the first challenge and 1.4 kg (range 0.8 – 2.5 kg) after the second. The milk yield from infected quarters was significantly higher after the second challenge (p < 0.05; Figure 2).

Bacterial counts in the milk of the challenged quarters peaked at 12 hours PC at both challenge times, being on average 18.1 × 10⁶ cfu/ml in the first challenge and 6800 cfu/ml in the second challenge. Bacteria were still isolated in low numbers from one cow (80 cfu/ml) 6 days PC after the first challenge, but after the second challenge were eliminated totally in all cows within 68 hours. Overall bacterial counts were lower at the second challenge (p < 0.05; Figure 3).

Milk SCC of the challenged quarters started to increase from the baseline values after both challenges within 12 hours and reached the maximum level at 20 hours PC, being over 25 × 10⁶ cells/ml after first challenge and 20.7 × 10⁶ cells/ml after the second. In both groups SCC gradually decreased after challenges. At the end of the experimental period of 6 d, SCC was on average 5.8 × 10⁶ cells/ml (range 541 000 – 18.1 × 10⁶ cells/ml) after the first challenge and 541 000 cells/ml (range 256 000–705 000 cells/ml) after the second challenge. The difference between the groups was not statistically significant (Figure 3).

NAGase activity of the milk after both challenges peaked at 20 hours PC, being on average 1.95 pmol/min/μl (range 0.65 – >2.5) after the first challenge and 1.90 pmol/min/μl after the second (range 0.63 – >2.5). After the first challenge NAGase activity remained elevated over the experimental period, but returned to the baseline value by this time after the second challenge. The difference between the challenges was not statistically significant (Figure 4). Milk SCC and NAGase activity in the contralateral control quarters remained at the pre-challenged levels in both groups after both challenges.

Before the first challenge, mean milk SAA concentrations were 7.1 mg/l ± 11.0 mg/l and before the second, 0.4 mg/l ± 0.4 mg/l. Milk SAA concentrations in both groups started to increase after 12 hours PC and reached the maximum (mean 1315.9 mg/l ± 947.3) at 60 hours PC after the first challenge and at 44 hours PC (mean 925.0 mg/l ± 609.1) after the second challenge. After the second challenge, SAA concentration decreased faster: mean concentration by the end of the experimental period was 16.6 ± 11.9 mg/l. Milk Hp started to increase after both challenges 12 hours PC and peaked at 44 hours at 0.60 g/l (± 0.49 g/l) after the first challenge and at 36 hours at 0.32 g/l (± 0.17 g/l) after second challenge. The Hp concentrations in milk returned to background levels within 156 hours after both challenges, faster after the second challenge. LBP concentrations in milk started to rise 12 hours PC and peaked at 36 hours PC, being on average 203.5 ± 44.3 mg/l after the first challenge and 169.0 ± 167.7 mg/l after the second. LBP was still increasing 6 d after the first challenge, but had reached the pre-challenge level by that time after the second challenge. Statistically significant differences between the two challenges were established for milk SAA (p < 0.05) and Hp (p < 0.05).

The concentrations of SAA in serum started to rise slowly after challenges until 24 hours PC, concentrations peaking after the first challenge by 60 hours PC (mean 447.9 mg/l ± 164.8) and after the second challenge by 44 hours PC (mean 307.1 mg/l ± 66.2). In both groups the SAA in serum subsequently decreased gradually, but had not reached the base levels by the end of the experimental period. However, there were no statistically significant differences between the two challenges.

The same trend was found for serum Hp concentrations, which started to rise after 24 hours and peaked at 60–68 hours after both challenges, reaching, on average, 1.70 g/l (± 0.68) in the first challenge and 1.13 g/l (± 0.08) in the second challenge. Haptoglobin concentrations in serum then decreased and were on average 0.61 g/l (± 0.54) by 6 days PC after the first challenge and 0.23 g/l (± 0.10) after the second challenge. Serum Hp concentrations were significantly lower in the later challenge (p < 0.001; Figure 5)

The basic concentrations of serum LBP before the challenges were on average 10.8 mg/l (± 7.7) after the first challenge and 10.0 mg/l (± 6.4) after the second. Serum LBP started to increase rapidly in both groups and peaked at 36 hours PC, being on average 148.6 mg/l (± 41.8) after the first challenge and 108.9 mg/l (± 31.6) after the second. No statistically significant difference was recorded between the challenges (Figure 5).

WBC started to decrease after both challenges, being at the lowest 12 h PC (on average 2.03 × 10⁹ cells/l at first and 2.97 × 10⁹ cells/l at second challenge), then starting to increase, being at its highest an average of 8.61 × 10⁹ cells/l (range 4.85–10.9 × 10⁹ cells/l) at 60 hours after the first challenge and at 24 hours 10.47 × 10⁹ cells/l (8.02–15.8 × 10⁹ cells/l) after the second. WBC levels were higher after the second challenge during the whole experiment. The difference in WBC levels was statistically significant (p < 0.05; Figure 5).