Reasons for performing study: The timing of lamellar basement membrane (BM) changes occurring during laminitis development is incompletely understood.
Objectives: To determine the temporal progression of lamellar BM changes and whether laminin-332 (Ln-332) γ2 cleavage products are generated during laminitis development.
Methods: Eight clinically normal Standardbred horses were allocated into treatment (n = 5) or sham (n = 3) groups. The treatment group received, via nasogastric intubation, an oligofructose (OF) bolus (10 g/kg bwt) while the sham group was given water. Laminitis induction proceeded for 48 h followed by euthanasia. Lamellar biopsies were obtained prior to dosing and at intervals during the treatment period for analysis (at 12, 18, 24, 30 and 36 h and at 48 h following euthanasia).
Results: Changes in lamellar collagen type IV and Ln-332 were first observed at 12 h post dosing. A unique pattern of reactivity for the Ln-332 γ2 antibody D4B5 occurred, in which reactivity was observed only in lamellar tissue affected by laminitis. No bioactive Ln-332 γ2 proteolytic fragments were detected in lamellar samples.
Conclusions: Basement membrane changes occurred early during the laminitis process. Direct Ln-332 γ2 cleavage to release biologically active products did not appear to occur. Thus loss of stability or protein interaction of the BM is probably responsible for the γ2 specific reactivity observed.
Potential relevance: Basement membrane changes may a first step in lamellar failure occurring prior to detection with conventional methods. Thus, more sensitive detection methods of BM changes are required to study laminitis development.
Basement membranes (BM) are thin proteinaceous layers separating epithelial, endothelial, nerve and muscle cells from their underlying extracellular matrix (ECM). More than just a barrier, the BM is in fact a complex scaffold of proteins providing barrier and structural functions as well as regulating cell activities such as growth and migration through cell signalling. Basement membranes from different tissues have a common structure and are composed of collagens, laminins and proteoglycans (Aumailley 1995; Timpl 1996; LeBleu et al. 2007). Ultrastructuraly, the epithelial BM can be divided into 3 zones. Adjacent to the epithelial cell membrane is the lamina lucida, traversed by fine anchoring filaments that link hemidesmosomes to the lamina densa. The lamina densa is a network of intertwined, rigid, collagen type IV and flexible laminin polymers. These 2 layers are stabilised by proteoglycans (perlecan) and glycoproteins (nidogen) (Ghohestani et al. 2001; Quondamatteo 2002). Incorporated into the lamina densa are loops of type VII collagen (anchoring fibrils) that project into the ECM forming the sub basal lamina, that interacts with the interstitial collagen type I connecting the basal cell to the dermis (McMillan et al. 2003; Villone et al. 2008).
Lysis of the hoof lamellar BM and loss of its attachment to the lamellar epidermal basal cell (EBC) characterises the histopathology of laminitis (Pollitt 1996). Histochemical staining of the BM, in the acute phase of disease, shows that the BM disappears from between secondary epidermal lamellae (SEL) bases, close to the keratinised axis of primary epidermal lamellae (PEL). The BM also detaches from the SEL tips (Pollitt 1996). Subsequent immunohistochemical studies showed that loss of type IV collagen, type VII collagen and laminin from the lamellar BM occurs during the acute phase of the disease and mirrors the histochemical changes (Pollitt and Daradka 1998). Loss and cleavage of the anchoring filament protein Ln-332 (French and Pollitt 2004a) as well as a decreased number of hemidesmosomes (French and Pollitt 2004b; Nourian et al. 2007) are also reported to be features of the laminitis lesion. Ultrastructural analysis confirmed separation of the BM and loss of hemidesmosomes in laminitis as early as 24 h after dosing with oligofructose (OF) (French and Pollitt 2004b; Nourian et al. 2007). Seeking better understanding of laminitis pathology we used a novel lamellar biopsy technique to study laminitis development at earlier timepoints. We also attempted to confirm cleavage of Ln-332 by analysing lamellar tissue samples for the presence of cleavage products.
Materials and methods
Tissue collection and preparation
Laminitis was induced in 5 normal Standardbred horses by alimentary carbohydrate overload using oligofructose (OF) (10 g/kg bwt) (van Eps and Pollitt 2006). Lamellar biopsies(approximately 10 x 10 mm) were collected from the right and left dorsal hoof wall of the forelimbs at 0, 12, 18, 24, 30 and 36 h with the horses standing, sedated with the appropriate digital nerve blocked with local anaesthetic (Croser and Pollitt 2006). At 48 h post OF dosing, following euthanasia by overdose with barbiturate, additional samples were obtained from tissue located directly beneath the central and lateral biopsy sites from both forefeet. A control group of horses (n = 3) were sham treated with water and biopsied in the same way. The locations and temporal patterns of the biopsy sites as well as the 48 h sampling sites are shown in Figure S1. In initial studies, to analyse the localisation of individual Ln-332 subunits, OF laminitis was also induced in a second group of horses (n = 4) subjected to euthanasia at 48 h along with sham treated control animals (n = 4) that received water. Mid wall, dorsal lamellar tissue samples were collected immediately after euthanasia. The project was approved by a University of Queensland Animal Ethics Committee (AEC–PCA) monitoring compliance with the Animal Welfare Act (2001) and The Code of Practice for the care and use of animals for scientific purposes (current edn). All animals were continuously monitored by the investigators and the horses were inspected by the Consultant Veterinary Officer to the Animal Welfare Unit at the University of Queensland at the request of the AEC. If pain could not be alleviated promptly, early euthanasia, even if this was prior to the planned conclusion of the experiment, was a proviso of the protocol.
Lamellar tissue samples were either fixed in 10% formalin for 24 h and embedded in paraffin or embedded in Tissue-Tek OCT and flash frozen. Specimens were sectioned at 5 µm and mounted on slides. For protein extraction, tissue samples were flash frozen in liquid nitrogen.
Paraffin sections were heated at 37°C for 30 min, followed by deparrafinisation using xylene (3 x 3 min) and rehydrated using a series of graded ethanol steps (100%, 2 x 5 min; 90%, 5 min; 80%, 5 min; 75% 5 min) to tap water. Antigen retrieval was performed using 0.05% Protease XXIV1 for 15 min at 37°C followed by washing with 50 mmol/l Tris-HCl pH 7.6. Tissue sections were then blocked with 2% normal goat serum2 for 30 min at room temperature followed by incubation in primary antibody diluted in antibody dilution solution with background reducing components3 for 18 h at 4°C. Primary antibodies used were the polyclonal antibody ab145094 detecting all 3 human Ln-332 subunits, the monoclonal D4B55 and the polyclonal Pab266 directed to the human γ2 subunit domain III (LEb). Antibodies detecting the human α3 subunit (K140)7 and the human β3 subunit (BM165)7 were also used, as were the antibody CIV228 that recognises the α1 and α2 chains of human collagen type IV. As a negative control, normal rabbit serum or mouse IgG was used in place of primary antibody. Following washing, sections were incubated in a biotin conjugated secondary antibody3 prepared in 2% normal goat serum for 30 min at room temperature then incubated in 3% H2O2 for 20 min to block endogenous peroxidase activity followed by washing. Strepavidin-horseradish peroxidase (HRP)9 was applied for 1 h at room temperature followed by washing. Development was performed using the substrate 3-amino-9-ethylcarbazole (AEC+ high sensitivity substrate)3 followed by nuclear staining with Mayer's haematoxylin. Slides were mounted with aqueous medium3.
Tissue sections were viewed with an Olympus BX-50 microscope equipped with a BX-FLA reflected light fluorescence attachment10 or by brightfield microscopy. Images were captured with a CoolSNAP-Pro CF monochrome digital camera using Image-Pro Plus software11.
A grading system was developed to estimate the amount of BM stained by each antibody. For lamellar tissue sections, 5 fields of view from different primary epidermal lamellae (PELs) from the base, middle and tip regions of each PEL were scored (1 = 100% stained, 2 = <75% stained, 3 = <50% stained, 4 = <25% stained) to give a value estimating the percentage of BM along the length of the SELs stained by each antibody. Generally, a higher number indicates less BM stained with the specific antibody. Combined scores for all 15 views from the 3 lamellar regions were averaged to give one score for each horse. Slides were randomly coded and the viewer was blinded to their identity. Scores for each time point compared to time 0 samples, as well as between laminitis and control groups at each time point were analysed using the Mann-Whitney test with significance set at P<0.05.
Lamellar tissue was pulversied under dry ice using a mortar and pestle followed by homogenisation on ice using a power homogeniser in RIPA buffer (20 mmol/l Tris-HCl pH 7.5, 150 mmol/l NaCl, 1% Triton-X 100, 0.5% sodium deoxycholate, 0.1% SDS containing Complete protease inhibitors)12, 1 ml /100 mg tissue, 5 x 20 s. Samples were centrifuged 10,000 x g, 10 min at 4°C and protein concentration of the supernatants was determined using the BCA protein assay9. Proteins were separated using SDS-PAGE mini gels13 under reducing conditions followed by transfer onto poly-vinyl-D-fluoride (PVDF) Hybond-P membrane14. Membranes were blocked with 5% milk prior to incubation in primary antibody overnight at 4°C (Pab26), followed by incubation with an HRP-conjugated secondary antibody15 for 1 h at room temperature. Immune complexes were detected by chemiluminescence using the Super Signal substrate9. Membranes were reprobed with a β-actin4 antibody to confirm equal protein loading.
Ln-332 γ2 chain specific changes occur in acute laminitis
Normal hoof tissue sections demonstrated immunostaining of the lamellar BM with a Ln-332 polyclonal antibody as well as with antibodies to both the α3 and β3 subunits (Fig S2). No reactivity in normal hoof lamellae was observed with the monoclonal γ2 antibody D4B5 (Fig 1a); however, lamellar BM staining was observed when the polyclonal γ2 antibody Pab26 was used (Fig 1b). In lamellar sections from horses subjected to euthanasia at 48 h after OF dosing and thus in the acute phase of laminitis, discontinuous loss and gaps in staining was present in the lamellar BM of all horses for Ln-332, the α3, β3 and the γ2 subunit detected with the antibody Pab26 (Fig S2, Fig 1d). Interestingly, immunostaining of the γ2 subunit using the antibody D4B5, was now clearly visible in the lamellar BM affected by laminitis (Fig 1c).
In the lamellar biopsy group of horses, initial loss of both Ln-332 and collagen type IV BM immunostaining occurred 12 h after OF dosing in all horses. Thinning and focal loss of collagen type IV and Ln-332 immunoreactivity was present between the bases of adjacent SELs (Fig 2 arrows). At later time points, loss of these proteins progressed along the perimeter of SELs. Cytoplasmic staining for Ln-332 in the basal cells of some lamellae occurred in 3 of 5 horses 48 h post OF dosing (data not shown). In some lamellae, SELs had completely separated from the secondary dermal lamellae (SDLs) with collagen type IV and Ln-332 staining remaining with the BM attached to the separated dermal tissue (data not shown).
Appearance of reactivity for the γ2 subunit detected by the antibody D4B5 initially occurred at 12–24 h post OF dosing in most horses. In Horse 2, the appearance of D4B5 was detected at 18–24 h post OF dosing and then disappeared. In some cases, D4B5 expression was initially detected only between the bases of SELs, while, at later time points, staining was present along the length of the SEL or only at the middle to tip (Fig 2 arrowheads). Where separation of the epithelial basal cell from the BM occurred, D4B5 staining remained with the BM of the detached SDL tissue (Fig 1e).
Lamellar biopsies from sham treated control horses (n = 3) showed strong BM reactivity along the entire SEL perimeter for Ln-332 (Fig 2d) and collagen type IV (data not shown) at all time points. Occasional faint foci of reactivity with the antibody D4B5 were observed throughout the temporal series in control animals; however, this was less than that observed in laminitis induction samples (data not shown). A minimal number of leucocytes detected by calprotectin immunostaining were present in sham treated tissue only after the 36 h timepoint, primarily in tissue areas surrounding biopsy sites (data not shown).
Overall, the BM changes observed within the biopsy group increased with the passage of time compared to the preinduction biopsy (0 h). The average Ln-332 immunostaining score was increased at each time point with significance (P<0.05) at 18, 36 and 48 h. Conversely, the average γ2 D4B5 immunostaining score decreased at each time point (P<0.05) except 24 h. Collagen type IV immunostaining was increased at each time point (P<0.05) except 30 h (Fig 2e). Similarly, the immunostaining scores for the laminitis group compared to the control group were also significantly different. The Ln-332 scores for the induction group were significantly more than control group at all time points except 24 and 30 h (P<0.05), while the D4B5 scores were less at all points except 24 h (P<0.05). Likewise, collagen type IV immunostaining of the laminitis group was more than the control group at all times except 36 h (P<0.05) (Fig 2e).
Detection of Ln-332 proteolytic fragments during laminitis development
A schematic representation of the normal Ln-332 heterotrimer and known processing sites for fragment production is shown in Figure S3. The full length 150 kDa γ2 and the processed 105 kDa γ2' subunits were present in lamellar tissue extracts from controls and all time points of laminitis development. However, no proteolytic fragments equivalent to γ2x, LEb or other fragments were present in biopsy samples (Fig 3).
Lamellar BM changes were present 12 h after OF dosing to induce laminitis. There was loss of immunostaining for collagen type IV and Ln-332 particularly between SEL bases in all 5 horses. Conventional periodic acid Schiff (PAS) staining of sections from the same 12 h samples detected diminished BM staining in only 2 of the 5 horses (Croser and Pollitt 2006). Thus, our serial biopsy immunohistochemical staining technique was more sensitive than histochemical staining and detected the earliest BM lesion so far described for OF induced laminitis. The possibility that lesions occur even earlier is being investigated in our laboratory.
Interestingly, while the γ2 subunit antibody Pab26 reacted strongly with the normal lamellar BM, the antibody D4B5 did not. However, clear D4B5 reactivity appeared in the BM of horses developing laminitis. Thus, laminitis appeared to have induced changes in the lamellar BM that allowed de novo recognition of the D4B5 epitope. Initially, we hypothesised that loss of Ln-332 immunoreactivity was due to direct Ln-332 γ2 subunit cleavage (French and Pollitt 2004a) by increased expression and activity of the matrix metalloproteases (MMP), MMP-2 and membrane type 1 MMP (MT1-MMP) (Johnson et al. 1998; Pollitt et al. 1998; Kyaw-Tanner and Pollitt 2004; Kyaw-Tanner et al. 2008). Both MMP-2 and MT1-MMP cleave Ln-332, resulting in the production of biologically active proteolytic fragments. Ln-332 γ2 processed fragments corresponding to the ∼80 kDa γ2x form have been detected in tissue undergoing normal remodelling (Giannelli et al. 1997; Pirila et al. 2001) as well as in the pathogenic remodelling associated with tumour invasion (Giannelli et al. 1997). Furthermore, the γ2 LEb domain is freely released from the γ2 subunit by either MMP-2 or MT1-MMP (Schenk et al. 2003; Koshikawa et al. 2005) and displays biological functions that include triggering cell migration (Schenk et al. 2003; Koshikawa et al. 2005) and increased MMP-2 transcription (Schenk et al. 2003). Direct processing the of lamellar Ln-332 γ2 subunit to the γ2x form could destabilise the lamellar BM structure, while free LEb could increase MMP activation during laminitis. However, using sensitive chemiluminescent western blotting techniques with detection limits in the low femtogram range, no proteolytic fragments corresponding to Ln-332 γ2x or the free LEb domain were detected in lamellar tissue during the development of laminitis, even though clear changes in the BM were observed in these horses. Thus, the changes occurring in the lamellar BM that precipitated recognition of the D4B5 epitope were probably not the result of direct γ2 subunit cleavage and production of biological active fragments. Alternatively, the equine Ln-332 γ2 subunit may be resistant to cleavage by the MMPs responsible for the production of the expected fragments. The cleavage site in the LEb domain, where conversion to the γ2' form occurs is well conserved among species (Vailly et al. 1994; Amano et al. 2000; Koshikawa et al. 2000) and exists in the horse. However, variations in the cleavage sites for conversion to the γ2x form exist between species (Vailly et al. 1994; Giannelli et al. 1997; Koshikawa et al. 2005).
In regions where the lamellar basal cells were separated from their dermal lamellar BM to form empty ‘bubbles’, reactivity for all Ln-332 subunits remained with the BM. This suggests that in these areas, a second type of pathology exists in which the BM appears to still be intact but separated from the cell. In these areas, the entire Ln-332 molecule may have detached from its hemidesmosme (HD) binding partner integrin α6β4 by cleavage of the C-terminal end allowing for its release. This is supported by EM studies in which separation at the level of the lamina densa resulted in an increased lamina lucida width and a decrease in the number of hemidesmosomes at both the 24 and 48 h post OF timepoints of laminitis development (French and Pollitt 2004b; Nourian et al. 2007). Plasmin is one protease known to cleave the intact Ln-332 α3 subunit to remove the LG4-5 domains (Goldfinger et al. 1998) along with removal of both a small fragment from the cell adhesive LG3 domain of the α3 subunit as well as LN domain of the β3 subunit involved in collagen type VII binding (Ogura et al. 2008). This is an alternative mechanism allowing for the entire BM to fall away form the basal cell. Events like these may also allow for either direct conformational changes in Ln-332 or result in changes in interaction with other BM components such as collagen type VII or other Ln forms in the lamina densa to allow for binding of the γ2 antibody D4B5.
A final possibility is that recognition by the antibody D4B5 during laminitis is the result of changes occurring in the surrounding BM. A recent model of Ln-332 orientation in human skin suggests that the N-terminal arm of the γ2 subunit is embedded deep within the lamina densa of the BM (McMillan et al. 2003) therefore the epitope recognised by the D4B5 antibody may be masked in the normal lamellar BM. As collagen type IV, the major component of the lamina densa, forms a mesh network where it acts to stabilise the BM, it is feasible that this network may prevent interaction with the antibody epitope. As laminitis develops, loss of collagen type IV and VII protein in the lamellar BM, observed in this study as well as by Pollitt and Daradka (1998), may reveal the LEb epitope and allow binding of the antibody D4B5 to the γ2 subunit by direct interaction. Additionally, processing and destabilisation of the proteoglycans and glycoproteins of the lamellar BM is a possible explanation. Significantly increased lamellar expression of the ADAMTS-4 gene occurs early in laminitis development (Budak et al. 2009; Coyne et al. 2009) and if increased protein translation and enzyme activity follow gene upregulation then ADAMTS-4 proteolytic processing of the lamellar BM could be responsible for revealing the D4B5 epitope. Although no reports of cleavage of the major BM proteoglycan perlecan exist to date, ADAMTS-4 is able to cleave small leucine rich proteoglycans in addition to hyaluronan binding proteoglycans, aggregcan and brevican (Tortorella et al. 1999; Kashiwagi et al. 2004).
Dorsal hoof wall lamellar biopsies were successfully used to study the histological and molecular events during laminitis development and accurately represented events occurring in the lamellae. The results obtained were similar to those from single time point sampling. The lamellar biopsy technique employed resulted in little local inflammation or tissue damage. Similar equally successful results have been recently reported (Hanly et al. 2009). Thus, lamellar biopsy techniques may be considered for experimental study as well as for clinical use.
In summary, we have established that notable changes occur in the lamellar BM early in OF induced laminitis, prior to clinical signs of lameness and much earlier than previously recognised. Use of a monoclonal antibody directed to the LEb domain of the γ2 subunit, provided evidence that an epitope-revealing conformational change in BM structure occurred specific to the γ2 subunit. Direct Ln-332 γ2 cleavage did not occur. Changes in the interaction or stability of attachment components within the lamellar BM undoubtedly occur but by a yet to be explained mechanism.
Funding was provided by the Rural Industries Research and Development Corporation of Australia and the Animal Health Foundation of St Louis, Missouri. M.B.V. was the recipient of an Endeavour International Postgraduate Research Scholarship from the University of Queensland. We thank Emma Croser, Christopher Owens, Katie Asplin, Ali Nourian and Marianne Keller for assistance with tissue collection and animal experiments.
1 Sigma St. Louis, Missouri, USA.
2 Vector Laboratories, Burlingame, California, USA.
3 Dako Carpinteria, California, USA.
4 Abcam, Cambridge, Massachusetts, USA.
5 Millipore Billerica, Massachusetts, USA.
6 Dr Karl Tryggvason, Karolinska Institute, Stockholm, Sweden.
7 Dr Peter Marinkovich, Stanford University, California, USA.
8 Thermo Scientific, Fremont, California, USA.
9 Pierce, Rockford, Illinois, USA.
10 Olympus Optical Company, Tokyo, Japan.
11 Media Cybernetics, Silver Spring, Maryland, USA.
12 Roche Basel, Switzerland.
13 NuSep, Frenchs Forest, New South Wales, Australia.