Holstein dairy cows aged 3–15 years were used in this study. To prevent cross-contamination, all animals were housed separately on-site according to positive or negative infection status with MAP. All experimental procedures were approved by the IACUC (National Animal Disease Center, Ames, IA). Housing facilities are accredited by the American Association for Accreditation of Laboratory Animal Care.

Cows were stratified into infection status groups utilizing diagnostic tests measuring serum MAP-specific antibody levels (Herdchek; IDEXX, Westbrook, ME), bovine IFN-γ plasma levels (Bovigam; Prionics, La Vista, NE), and fecal shedding detected by culture on Herrold's egg yolk medium (Becton Dickinson, Sparks, MD) as previously described (20). Clinical cows (n = 10) had an average age of 5 years and were ELISA positive for MAP serum antibody, with an average S/P ratio of 2.44. The average MAP-specific IFN-γ recall response for this group was OD₄₅₀ 0.50 ± 0.35 (Abs_450nmMPS–Abs_450nmNS). Most of these animals were also culture positive for MAP, having an average of ~138 CFU/g fecal matter. Subclinical cows (n = 10) had an average group age of 8 years and were mostly ELISA negative for MAP serum antibodies, with one subclinical cow weakly positive having an S/P of 0.91. The subclinical group had an average IFN-γ OD₄₅₀ of 0.16 ± 0.09. Two subclinical cows were fecal culture positive for MAP and had an average shedding value of 1.5 CFU/g fecal matter. Control cows (n = 11), averaging 8 years of age, were negative for all diagnostic tests.

Samples of mid-ileal tissue were collected at necropsy and rinsed with Dulbecco's phosphate buffered saline (D-PBS) at pH 7.4 (Sigma-Aldrich, St. Louis, MO). The luminal side was placed on top of a liver section with a thin layer of Tissue-Tek optimum cutting temperature (OCT) compound (Sakura Finetek, Torrance, CA) interface to protect microvilli structure from mechanical and environmental disruption. This also aided in determining tissue orientation upon cryosectioning. Tissue samples were then wrapped in aluminum foil and snap-frozen by placement in a tin cup of cold isopentane (Sigma-Aldrich) housed within a small cooler of dry ice and 95% ethanol slurry for a minimum of 5 min, then wrapped again in aluminum foil and transferred on dry ice to be stored at −80°C.

When cryosectioning was ready to be performed, tissue sections were removed from storage at −80°C and placed in a cryostat to acclimate to −20°C for at least 30 min. The entire tissue sample was embedded within Tissue-Tek optimal cutting temperature (O.C.T.; Sakura Finetek, Torrance, CA) and cut in 6 μm sections, then adhered to MAS slides (Matsunami Glass, Bellingham, WA). The tissue sections were dried overnight at room temperature and fixed in a 50% acetone/50% methanol solution for 5 min. Slides were stored at −80°C until ready for further processing.

Prior to beginning the tissue immunofluorescence (IF) protocol, CYP27B1 antibody (PA5-79128; Invitrogen, Carlsbad, CA) had its stock buffer exchanged to remove bovine serum albumin (BSA) using the Pierce Antibody Cleanup Kit (44600; Thermo Scientific, Rockford, IL). Buffer exchanged antibody was transferred to a 30 K molecular weight cutoff Pierce concentrator microcentrifuge column (88529; Thermo Scientific) and centrifuged at 12,000 × g in 30 s increments to a final volume of 100 μl. The 100 μl of antibody was conjugated using the AF594 Lightning-Link conjugation kit (ab269822; Abcam, Waltham, MA) according to the manufacturer's instructions. Additionally, Rab5 primary antibody was directly conjugated to AF594 using the same Lightning-Link conjugation kit.

Primary antibodies used in this study were split into multiple protocols due to our constrains of four lasers. All antibodies were diluted in 0.05M Tris buffer and incubated for 1 h each at room temperature in a humidified chamber protected from light. The tissue underwent 3 wash steps with 0.05 M Tris/0.2% Tween20/0.9% NaCl (Tris/Tween buffer) following incubation with each antibody.

Tissue sections were removed from storage at −80°C and allowed to equilibrate to room temperature. A thin hydrophobic layer was drawn around each tissue section using a PAP pen (Life Technologies, Carlsbad, CA) to keep liquid components localized. Tissue was rehydrated with 0.05 M Tris buffer for 10 minutes, followed by addition of 3,3′-diaminobenzidine (DAB) (Vector Labs, Burlingame, CA) for 10 min in a dark, humidified chamber to quench autofluorescence. Slides were then washed 3 times alternating with 0.05 M Tris buffer and Tris/Tween buffer. Non-specific labeling was blocked for 30 min using serum-free blocking buffer (X0909, Agilent Technologies, Santa Clara, CA). Primary and secondary antibodies were added following this step and are detailed in the individual protocols below.

CYP27B1 and CYP24A1 antibodies were paired together, and this protocol began with labeling of ileal tissue macrophages with CD68 primary antibody (1:300, M0718; Agilent Technologies), followed by labeling with an AF647 conjugated secondary antibody (1:750, 115-605-205; Jackson Labs). Next, the CYP24A1 (1:25, LS-C407760; LSBio, Seattle, WA) primary antibody was added, followed by AF488 secondary antibody (1:250, A11070; Invitrogen). Lastly, CYP27B1 was labeled in the tissue using a primary antibody directly conjugated to AF594 (1:40, PA5-79128; Invitrogen).

Detection of Rab5 began with addition of CD68 (1:300, Agilent Technologies) and MAP (1:1,000, #272 in-house, NADC) primary antibodies to the tissue concurrently in a cocktail. Secondary antibodies conjugated to AF647 (115-605-205; Jackson Labs) or AF488 (1:1,000, A11070; Invitrogen) were also added together and were used to detect each primary antibody, respectively. Rab5 (1:40, MBS612620; MyBioSource) directly conjugated to AF594 was added to the tissue last.

The Rab7 protocol also began with labeling of tissue macrophages and MAP through addition of CD68 (1:300, Agilent Technologies) and MAP (1:1,000, #272 in-house, NADC) primary antibody cocktail. Secondary antibodies were also added in a cocktail, with CD68 being detected by AF647 (1:1,000, 115-605-205; Jackson Labs) and MAP detected by AF488 (1:1,000, A11070; Invitrogen). Rab7 primary (1:100, ab50533; Abcam) was then incubated with the tissue, followed by an AF594 secondary antibody (1:250, A21145, Invitrogen).

The Vitamin D receptor (VDR) protocol began with the labeling of CD68 (1:300, Agilent Technologies) followed by addition of AF647 secondary antibody (1:750, 115-605-205; Jackson Labs). The VDR primary antibody (1:10, LS-C407668; LSBio) was then added, followed by labeling with an AF594 secondary antibody (1:250, A11072, Invitrogen).

Lastly, for all protocols counterstaining was achieved using DAPI diluted in 0.05 M Tris buffer to a concentration of 1μg/ml and incubated with the tissue for 10 min. Slides underwent one more wash step with 0.05 M Tris buffer prior to addition of VECTASHIELD Vibrance mounting media (Vector Labs) and coverslipping with 24 × 30 mm Richard-Allan Scientific Slip-Rite #1.5 (Thermo Scientific, Carlsbad, CA). Mounting media was allowed to cure overnight at room temperature in the dark prior to imaging.

Prior to histochemical staining, all slides were equilibrated to room temperature and rehydrated in deionized water for 5 min. Sequential slides corresponding with IF protocols were stained using Ziehl-Neelsen acid fast (AF) and Harris' hematoxylin and eosin (H&E) methods to detect acid-fast mycobacteria and assess tissue morphology, respectively.

Slides that underwent AF staining were incubated for 1 h in carbol fuchsin solution at room temperature. Stock acid alcohol was used to decolorize tissue for ~10 s, which resulted in the tissue obtaining a pale pink color. Slides were then washed in a running water bath for 10 min, followed by a 5 min counterstain with Harris' hematoxylin in a Leica autostainer (Leica Biosystems, Buffalo Grove, IL). Lastly, slides were rinsed again in a running water bath and coverslipped using xylene substitute mountant (Epredia, Portsmouth, NH).

H&E staining was also performed in a Leica autostainer beginning with incubation for 5 min in Harris' hematoxylin to label nuclei, followed by rinsing in deionized water for 5 min. Slides were then dipped in acid alcohol and rinsed in a deionized water bath for 5 min, then incubated for 1 min in 70% alcohol. To stain cytoplasm, slides were incubated with eosin for ~15 s. The slides were dehydrated by dipping twice in 95% alcohol and three times in 100% alcohol, each for 1 min. Lastly, slides were cleared of OCT by dipping three times in Pro-Par for 5 min each prior to coverslipping using xylene substitute mountant (Epredia, Portsmouth, NH).

A subset of cows from the CYP27B1/CYP24A1 protocol (Set 1) were used for the Rab5, Rab7, and VDR protocols (Set 2) due to tissue availability. New tissue sample cryostat chucks were made for this subset of cows, and concurrent AF and H&E slides were stained for pathology analysis.

Tissue lesion severity and presence of MAP was evaluated histologically by a board certified veterinary pathologist. Tissue slides were examined at 200 × and 20 fields were read each. MAP burden and granuloma severity were scored within a range of 0–5, with methods adopted from Palmer et al. (21) and Stabel et al. (22) and used previously in our lab (23–25). Briefly, a score of 0 indicated no MAP/granuloma presence and a score of 1 represented scarce MAP and granulomas within each tissue section. Increasing scores correlate with increasing MAP burden and granuloma severity, with a score of 5 representing a high amount of MAP present and severe granulomas disrupting most of the tissue architecture.

Images were acquired with a 60 × Nikon Plan Apochromat lambda objective with 1.4 numerical aperture using oil immersion and 6.2 s pixel dwell time on a Nikon A1 Resonance Plus confocal microscope using NIS-Elements Advanced Research software v5.21 (Nikon, Melville, NY). The instrument contains a four-laser gallium-arsenide-phosphide/normal photomultiplier tube (GaAsP PMT) fluorescence detector unit (A1-DU4) with two GaAsP PMTs (488 and 561 nm) and two normal PMTs (405 and 640). Fluorescent signal was detected sequentially using the following solid-state diode lasers and bandpass filters: 405 nm (450/50 nm), 488 nm (525/50 nm), 561 nm (600/50 nm), and 640 nm (685/70 nm). A minimum of 10 images were acquired per cow.

Following image acquisition, binary layers for each laser channel were created with thresholds established from control slides to exclude non-specific background. The average area (μm²) expressing fluorescence signal for each marker was measured per image field. All analyses were run on unaltered images and representative images chosen for this manuscript have had lookup tables applied within the Nikon NIS-Elements software to brighten signal and were post-processed in Adobe Photoshop (version 22.0; San Jose, CA) to further increase color brightness for printing purposes. All alterations were uniformly applied to the entire image.

Statistical analysis was performed using R Statistical Software (version 4.0.3, R Foundation for Statistical Computing, Vienna, Austria) and RStudio (version 1.3.1093, Boston, MA). Data measuring total amount of macrophages present and MAP burden were aggregated from separate IHC protocols performed and analyzed using the mixed model function “lme” from package “nlme” (26) to account for repeat observations. Statistical models for all other confocal data sets were analyzed by ANOVA using the linear model function “lm” (27) and all post-hoc tests were performed using the package “emmeans” (28) with a Tukey adjustment for multiple comparisons. Transformations of the data by log or log (1 + x) were performed where standardized residuals were not normally distributed. Pathology assessment scores were analyzed by using the Fisher's exact test with two-sided probability (Pr < 0.05) through function “fisher.test” from package “stats” in base R (27). Pearson correlations were performed using function “rcor.test” from package “ltm” and a multiple comparisons correction was applied using the Holm-Bonferroni method (29).

Immunofluorescent assessment of CD68+ macrophages in the mid-ileum tissue from each cow showed animals in the clinical stage of infection had significantly greater numbers compared to subclinical (Figure 1; P < 0.05) and JD- cows (P < 0.05). No significant difference in the number of macrophages were observed between JD- control and subclinical animals.

Histologically present granulomatous inflammation was scored for each JD status group (Table 1). Subclinically infected cows showed no granulomatous lesions; however, one subclinical cow had a small focus of epithelioid macrophage aggregates, but no AF bacteria were present. All clinical cows had granulomatous lesions present. A significant association between subclinical/clinical JD status and granulomatous inflammation score was shown through Fisher's Exact Test (P < 0.001; Table 1).

Total MAP burden and intracellular MAP within macrophages was also measured by IF. Cows in clinical stage disease had significantly greater (Figure 2A; P < 0.001) overall MAP burden compared to JD- control and subclinically infected animals. No significant difference in the overall number of MAP between JD- controls and subclinicals was observed; however, no MAP was detected in JD- control cows and ~66% of subclinical cow tissue samples had detectable MAP. Clinical cow tissue samples had a MAP detection rate of 100%. Additionally, clinical cows had greater intracellular MAP than subclinical cows and JD- controls as measured by MAP colocalization with CD68 (Figure 2B; P < 0.001).

Histochemical analysis of MAP burden was performed through AF staining. Aligning with the confocal IF data, clinical cows had the greatest amount of MAP present in the ileum. The presence of MAP was not detected by AF stain in tissues from subclinical or JD- control cows. Additionally, Fisher's Exact Test revealed higher AF scores were associated with the clinical group for both set 1 (P < 0.001; Table 1) and set 2 protocols (P < 0.01) when compared to the subclinical group. MAP was largely found in the lamina propria, and some animals observed bacilli extending from the lamina propria into the submucosa (Table 1). Furthermore, a strong, positive correlation was observed between AF and granulomatous inflammation scores (r = 0.935, P < 0.001). Positive correlations were also observed between the amount of macrophages detected by IF and both AF score (r = 0.679, P < 0.01) and granulomatous inflammation score (r = 0.602, P < 0.05).

Early and late endosomal markers Rab5 and Rab7 were investigated by IF and measured through confocal microscopy. Total expression of each endosomal marker was measured by its colocalization with CD68, with and without colocalization with MAP.

Expression of Rab5 within tissue macrophages was shown to be significantly reduced in clinical cows compared to both subclinicals (Figure 3A; P < 0.01) and JD- controls (P < 0.05). No significant differences were observed between subclinical and JD- control animals. In contrast, Rab5 colocalization with intracellular MAP was significantly higher for clinical cows compared to JD- control (Figure 3B; P < 0.05) and subclinical cows (P < 0.05). Fluorescent expression of Rab5, MAP, and macrophages within the bovine ileum from a representative healthy (Figures 4A–D), subclinical (Figures 4E–H), and clinical cow (Figures 4I–L) are shown in a composite confocal microscopy image. This figure also shows visual representation of the increased MAP burden between subclinical and clinical cows.

Rab7 expression within tissue macrophages was not different among infection status groups (Figure 5A); however, clinical cows had significantly greater colocalization of Rab7 with MAP within tissue macrophages when compared to subclinical cows and JD- controls (Figure 5B; P < 0.05). Notably, no Rab7 colocalization was detected with intracellular MAP from subclinical cows. Fluorescent expression of Rab7, MAP, and macrophages within the bovine mid-ileum is shown in a confocal microscopy image from a representative clinical cow (Figures 6A–D).

Expression levels of vitamin D hydroxylases CYP24A1 and CYP27B1 were measured within the bovine ileum for each group of cows. CYP24A1 showed no significant level of variation among infection status groups (Figure 7A). In contrast, CYP27B1 expression was significantly reduced in clinical cows compared to JD- controls (Figure 7B; P < 0.05) and subclinical cows (P < 0.05). A representative confocal microscopy image from a JD- control (Figures 8A–D) and clinical cow (Figures 8E–H) show a high degree of CYP24A1 labeling in the enterocytes of the intestinal villi. CYP27B1 expression is also expressed in the villi enterocytes albeit in lower amounts. Furthermore, CYP27B1 expression was observed to colocalize with CD68+ macrophages. A high level of expression was also observed in another undetermined cell type which appears to dimly label with CD68 or not at all.

Additionally, VDR expression in the mid-ileum was investigated which revealed no variation in expression among infection status groups (Figure 9). A fluorescent microscopy image from a representative clinical cow shows VDR labeling highly concentrated in the nuclei of enterocytes, along with lower levels of expression in CD68+ macrophages (Figures 10A–D).