Novel inflammatory cell infiltration scoring system to investigate healthy and footrot affected ovine interdigital skin

Introduction

Lameness in sheep, caused by interdigital dermatitis (ID) and footrot, is widespread in the UK, having a major welfare and economic impact. Footrot is caused by Dichelobacter nodosus which occurs in both virulent and benign forms. Footrot is defined by separation of the hoof from the underlying structures and an accumulation of necrotic material, with varying degrees of severity (Beveridge, 1941; Thomas, 1962), this damage is believed to be mediated by immune pathology rather than by bacterial enzymes and toxins (Egerton, Roberts & Parsonso, 1969).

Footrot development is characterised by invasion of neutrophils and lymphocytes into the dermis and epidermis in response to bacterial invasion of the epidermis (Davenport et al., 2014; Egerton, Roberts & Parsonso, 1969). Early histological studies described footrot as a degenerative condition of the stratum granulosum and spinosum which results in cellular degeneration (cell ballooning), formation of micro-abscesses and vacuoles which coalesce and progress to cavities (Beveridge, 1941; Deane & Jensen, 1955; Thomas, 1962). A recent histological study of clinically healthy and affected feet showed a progressive increase in lymphocyte and neutrophil infiltration into the dermis and epidermis between healthy, ID and footrot samples (Davenport et al., 2014). In footrot samples, purulence was seen in areas of epidermal degeneration, necrosis and epidermal-dermal clefts. As in previous studies, cytoplasmic ballooning and nuclear condensation were observed in the stratum spinosum of the epithelium (Thomas, 1962), as well as areas of fibrosis indicating a chronic reaction to tissue damage (Egerton, Roberts & Parsonso, 1969).

Recently, we have shown that IL-1β and CXCL-8, but not IL-6 and IL-17 mRNA expression levels correlate with D. nodosus load in footrot samples (Maboni et al., 2017a). In addition, stimulation of ex vivo organ explant of ovine interdigital skin infected with D. nodosus elicited IL-1β release (Maboni et al., 2017b).

The aim of this study was to perform a qualitative and quantitative analysis of histopathological features of the ovine interdigital skin comparing healthy and footrot affected feet. In order to evaluate inflammation in the epidermis and dermis a novel scoring system was developed and correlation of those scores with the expression of the pro-inflammatory cytokine IL-1β and D. nodosus load was investigated. In addition the depth of within tissue colonisation of eubacteria, D. nodosus and Fusobacterium necrophorum were examined in the context of hair follicle depth.

Materials and Methods

Tissue and cell scoring

For each tissue sample, 15 photomicrographs were scored for inflammation. The maximum score of five non-overlapping images from each of three slides per block was determined as the score for that tissue sample. A score of 0 represented no leukocytes in photomicrograph, score 1 occasional infiltration of single leukocytes visible, score 2 focal infiltration of leukocytes, score 3 coalescing leukocytes—individual loci could not be distinguished, and score 4 diffuse infiltration of leukocytes throughout the field of view. Examples of the epidermal and dermal scoring system are shown in Figs. 1 and 2, respectively. The scoring system was validated using two independent scorers blinded to sample identification.

Parakeratosis was defined as the retention of nuclear remnants in the stratum corneum (Brady, 2004). Grading in the interdigital skin stratum corneum was based on a three ordinal scale criteria as: the absence of parakeratosis (score 0), the presence of focal parakeratosis (score 1) and diffuse parakeratosis (score 2) (Figs. 3A–3C).

Micro-abscesses were defined as the aggregation of inflammatory cells and cellular debris, confined by fibrotic tissue walling. Grading was according to skin layer (intra-corneal, sub-corneal and dermal (Figs. 3D–3F)) and based on a nominal grading scale: presence/absence from five images throughout each tissue section.

In order to determine the area of ballooned cells per cross sectional area in the epidermis, photomicrographs were taken, saved in a Tagged Image File Format and uploaded into the analysis software Image Pro 6.3 (Media Cybernetics, Rockville, MD, USA). Each image was calibrated for the measurement of ballooned cells area in absolute values (µm²). Ballooned cells were identified and manually selected using the software to trace around each ballooned cell or group of cells. Percentage area of ballooned cells per photomicrograph calculated (total area of ballooned cell(s)/total area of epidermal cross section × 100, for example see Fig. 4A).

Congestion was defined as presence of dilated blood capillaries with visible red blood cells (erythrocytes) within the dermis and haemorrhage was defined as observed blood cells outside capillaries in tissues (Figs. 4B–4C). For both features, 15 photomicrographs were analysed per tissue with a nominal grading scale: presence/absence.

Basement membrane integrity was assessed using PAS stained sections scored for basement membrane disruptions at the dermal-epidermal junction as follows: score 0, no basement membrane disruption identified; score 1, focal basement membrane disruption identified; score 2, multiple basement membrane disruption identified (Figs. 4D–4F).

The IL-1β expression and D. nodosus quantification data have been published previously (Maboni et al., 2017a).

Results

To investigate the levels of inflammation within the interdigital skin, inflammation and occurrence of micro-abscesses were scored separately in the epidermis and dermis. In addition, in the epidermis, the area of ballooning was measured and the occurrence of parakeratosis was determined, and in the dermis, occurrence of haemorrhages and congestion were determined. The basement membranes were also assessed for disruptions.

In the epidermis, the same level of inflammation was observed in healthy (n = 55) and footrot (n = 30) samples with a median inflammation score 2 (minimum 1, maximum 4, Fig. 5A). In addition, no differences between healthy and footrot samples were seen with regards to the area of ballooned cells (Fig. 5B; n = 55 and 30 respectively). The proportion of samples with intra-corneal (healthy 43.64%, 24/55 (95% CI [59–90.4%]) and footrot 33.33% 10/30 (95% CI [27.2–73%])) or sub-corneal (healthy 25.45%, 14/55 (95% CI [20–50%])) and footrot 23.33%, 7/30 (95% CI [13.2–53%]) micro-abscesses were also not significantly different (Figs. 5C–5D). Parakeratosis was also observed to a very similar extent in both, healthy (72.73%, 40/55 (95% CI [59–84%])) and footrot (76.67%, 23/30 (95% CI [58–90%])) samples (Fig. 5E). However, while in footrot these were evenly split into samples with diffuse and focal parakeratosis, two thirds of the healthy samples with parakeratosis showed a diffuse pattern (Fig. 5E).

Basal membrane disruption was observed in 74.55% (41/55 (95% CI [61–85.3%])) healthy and 56.67% (17/30 (95% CI [37.4–74.5%])) footrot affected samples, with only a small proportion of samples with multiple disruptions (Fig. 5F).

In the dermis, a higher maximum inflammation score of 3 (minimum 2, maximum 4) was seen in healthy and footrot affected tissues (Fig. 6A; n = 54 and 29 respectively). There was also no difference between healthy and footrot affected tissues with regards to the proportion of samples with dermal micro-abscesses, haemorrhages or congestion (Figs. 6B–6D). Dermal micro-abscesses were observed in 8.5% of the tissues, with four out of 50 healthy tissues (7.41% (95% CI [2.2–19.2%])) and three out of 26 footrot affected tissues (10.34% (95% CI [2.4–30.1%]); Fig. 6B). Haemorrhages were observed in approximately one third of samples of each disease state (Fig. 6C; healthy = 30.91%, 17/50 (95% CI [28.6–62%]), footrot = 30.00%, 9/30 (95% CI [22–66%])). A similar level of congestion in blood vessels was found in clinically healthy (49.00%, 27/55 (95% CI [81.6–99.9%])) and footrot samples (40.00%, 12/30 (95% CI [41–86.6%])) (Fig. 6D).

We have shown previously that mRNA expression of IL-1β, similar to inflammation scores, were comparable in healthy and footrot affected tissues in parallel samples from the same feet (Maboni et al., 2017a). Here we analysed inflammatory scores, parakeratosis and presence of micro-abscesses in the context of IL-1β mRNA expression. IL-1β expression was significantly higher in the presence of diffuse parakeratosis in healthy tissues with no difference in footrot affected tissues (Figs. 7A–7B; n = 41 healthy and 23 footrot, p < 0.01). When comparing relative IL-1β expression with the inflammation scores they mirrored to an extent (Fig. 8).

Since footrot pathology is mediated by host inflammatory responses whilst the disease is initiated by D. nodosus, a virulent D. nodosus load against host inflammatory cells was compared in healthy (n = 30 for both epidermis and dermis) and affected tissues (n = 18 epidermis and 17 dermis). Inflammatory scores increased as virulent D. nodosus load increased (Fig. 9). In the epidermis, the virulent D. nodosus load was significantly higher (p ≤ 0.0001) in tissues with an inflammatory score of 3 compared to scores 1 and 2 respectively (Fig. 9B, Fig. S1).

Bacterial localisation in ovine interdigital skin

Eubacterial D. nodosus and F. necrophorum DNA was quantified across the entire skin depth. Eubacterial load in healthy samples (n = 5) was similar throughout the tissue depths but peaked at depth of 201–400 µm (Fig. 10). In footrot samples (n = 4) eubacterial load progressively decreased from the outermost skin surface to a depth of 1,000 µm, eubacterial load was then similar from 1,200–2,200 µm.

In footrot samples, D. nodosus was quantified across depths of 601–2,200 µm, whereas it was detected throughout all levels of the healthy samples. F. necrophorum in healthy samples was detected in deeper skin tissues (601–2,200 µm) but was present throughout every levels in footrot samples (Fig. 10). The distribution pattern of eubacterial load in ovine interdigital skin samples were investigated with respect to depth of skin sections in the context of depth of follicles in skin. A relatively low eubacterial load of range of 0.4–19.6 pg/section was observed across healthy samples. In comparison, footrot samples eubacterial load was higher in four out of five samples with peak values of 1,218 pg/section observed in the superficial skin depth (≤500 µm; Fig. 10). Localisation of eubacterial load did not extend beyond follicular depths (2,119–2,710 µm) in healthy samples while in footrot samples, eubacterial load extended beyond follicular depths (1,460–2,699 µm; Fig. 10, Fig. S2).

Discussion

In this study, histopathological features of the ovine interdigital skin were analysed and compared between healthy and footrot affected feet using a novel scoring system. In addition, these features were compared to levels of IL-1β and virulent D. nodosus. Of note in this study was the severity of inflammatory cell infiltration that was found to be similar between healthy and footrot affected. Inflammation in healthy samples may have been caused by unfavourable ground conditions of the pasture due to wet weather. Under those condition previous studies have also noted an increase in increase the prevalence of footrot and interdigital dermatitis (Beveridge, 1941; Emery, Stewart & Clark, 1984; Graham & Egerton, 1968; Wassink et al., 2003; Wassink et al., 2004). Alternatively, inflammatory cell infiltrations could be an indication of subclinical disease not yet progressed to visible signs of ID and footrot.

Previous studies have qualitatively described mild inflammatory cell infiltration in healthy ovine feet (Angell et al., 2015; Davenport et al., 2014). To the best of our knowledge, this is the first study to quantitatively examine inflammatory cells in the ovine feet using a new scoring system. Similar investigations in cattle also found inflammatory cell infiltration in apparently healthy hooves (Mendes et al., 2013; Tarlton et al., 2002). However, Tarlton and colleagues attributed the inflammatory response observed to the changes in collagen expression and keratinization of epidermal laminae caused by intensive breeding regimes during the peripartum period (Tarlton et al., 2002). Interestingly, bovine digital dermatitis, which is driven by inflammatory response in its pathogenesis, is also characterised by inflammatory cell infiltration (Mendes et al., 2013; Refaai et al., 2013).

Histological lesions including cell ballooning, parakeratosis and micro-abscesses were similarly observed in both conditions and in areas associated with pathology including inflammation in the skin (Refaai et al., 2013). Cellular ballooning in the epidermis is an indication of host cell response to stimuli and may be preceded by keratinocytes hyper-proliferating to remove pathogen infected areas (sloughing) (Edwards, Dymock & Jenkinson, 2003). Hyper-proliferation of keratinocytes without a corresponding rate of differentiation results in parakeratosis. This has been shown to occur in low grade inflammation caused by microbial presence such as F. necrophorum (Egerton, Roberts & Parsonso, 1969). Parakeratosis may trigger the release of pro-inflammatory cytokines such as IL-1β (Chang et al., 1992) which in-turn mediates the recruitment of inflammatory cells. This may explain the higher levels of IL-1β expression associated with diffuse parakeratosis in this study.

As expected, higher IL-1β expression levels corresponded with high epidermal inflammatory score grade (score 4) when compared to lower score grade (score 1) in footrot but not in healthy feet. This suggests that IL-1β expression level may be directly proportional to the severity of inflammatory cell infiltrates in the epidermis. To investigate this further, future studies could include immunohistochemical visualization of IL-1β protein in relation to cell infiltration or other lesions. Similarly, significantly increased virulent D. nodosus loads tend to correspond with higher epidermal inflammatory score grades (scores 2 and 3) when compared to lower grade (score 1) in footrot but not healthy samples. This suggests that the severity of inflammatory response in footrot is dependent on the abundance of virulent D. nodosus. This also confirms previous studies based on severity and bacteria load (Maboni et al., 2017a; Maboni et al., 2016). The initiation of footrot is mostly associated with virulent D. nodosus due to its ability to degrade host extracellular matrix, a trait conferred by the presence of the acidic protease AprV2. It is different from a second benign phenotype which possess the acidic protease AprB2 (Kennan et al., 2010). Interestingly, not all cases of virulent D. nodosus correlate to severe clinical manifestations since it has been reported in apparently healthy feet without signs of ID or footrot (Maboni et al., 2016; Moore et al., 2005; Stauble et al., 2014). In the UK, virulent D. nodosus has been reported as the predominant strain (Maboni et al., 2016; Moore et al., 2005).

Studies using whole biopsies reported eubacteria presence in healthy ovine interdigital skin (Calvo-Bado et al., 2011; Maboni et al., 2016; Witcomb et al., 2015). Interestingly, we detected eubacterial DNA down to depths of 3 mm in healthy samples. A previous study of healthy human skin also demonstrated eubacterial DNA in deep facial and palm skin tissues (5 mm), with localisation including hair follicles and also dermal stroma (Nakatsuji et al., 2013).

As expected, eubacterial load was significantly higher in the outermost layer of footrot skin (≤200 µm) in comparison to the same depth in healthy samples. Other studies on a human skin model have also reported higher bacteria abundance in the superficial layers of human skin (Lange-Asschenfeldt et al., 2011; Roeckl & Mueller, 1959). Detection of bacteria DNA in deep layers of healthy skin may indicate that the skin is not an impervious barrier as previously thought (Nakatsuji et al., 2013).

Previous studies based on Giemsa stain and fluorescent in situ hybridisation (FISH) have suggested D. nodosus was primarily localised in superficial epidermis of ovine interdigital skin (Egerton, Roberts & Parsonso, 1969; Witcomb et al., 2015). However, Witcomb and colleagues reported a single cell of D. nodosus in the dermis of footrot infected ovine skin (Witcomb et al., 2015). In contrast to FISH, qPCR technique is able to amplify target DNA, thereby detecting low levels of D. nodosus.

Hair follicles have been reported to serve as a bacterial reservoir and hence, possible routes of bacterial entry into healthy skin (Montes & Wilborn, 1970). The analysis of sequential transverse sections in our study showed that eubacterial load corresponded to follicular depth in healthy samples but extended beyond follicular depth in footrot samples. This suggests that hair follicles may play a role in eubacteria localisation in intact healthy skin. In contrast, in a similar polymicrobial disease of bovine feet, digital dermatitis, no association was found between treponemes and hair follicles (Evans et al., 2009).

Conclusions

Data presented in this study showed for the first time that there are no consistent differences in the level and range of histological lesions examined between healthy and footrot affected feet. Interestingly, the novel inflammatory cell infiltration scoring system developed and validated in this study mirrored the pro-inflammatory cytokine IL-1β and confirmed an association between severity of inflammatory response and increased virulent D. nodosus load.

Supplemental Information