Hindlimb laminar inflammatory response is similar to that present in forelimbs after carbohydrate overload in horses


email: belknap.16@osu.edu


Reasons for performing study: A significant proinflammatory response is known to occur in the forelimb lamina after carbohydrate administration. As the hindlimbs are often less affected by laminitis compared with the forelimbs, we assessed hindlimb inflammatory response in the early stages of carbohydrate-induced laminitis to determine whether differences in the response existed.

Objective: To determine whether a similar proinflammatory response occurs in the hindlimb laminae to that previously reported for the forelimb.

Methods: Archived laminar samples from 12 horses administered 17.6 g of starch (85% corn starch, 15% wood flour)/kg bwt via nasogastric tube that were anaesthetised either after developing a temperature >38.9°C (DEV; n = 6) or at the onset of Obel grade 1 lameness (OG1; n = 6) were used in addition to 6 control horses (CON) that were anaesthetised 24 h after administration of water. Real-time quantitative polymerase chain reaction for selected proinflammatory mediators and MAC387 immunohistochemistry were performed. The data were analysed nonparametrically to compare groups.

Results: Increases in laminar MAC387-positive leucocytes and laminar messenger ribonucleic acid (mRNA) concentrations (P<0.05) for interleukin-1β, interleukin-6, cyclo-oxygenase-2, chemokine (C-X-C motif)ligand (CXCL)1 and CXCL8 were present in both fore- and hindlimb laminae from horses with OG1 lameness. Both CXCL1 and CXCL8 were also increased in forelimb and hindlimb laminae in the DEV horses.

Conclusions: Administration of carbohydrate resulted in a similar inflammatory response in the hindlimb laminae to that previously reported for the forelimb laminae. These findings suggest that other factors, such as weightbearing, may play an important role in the development of laminitis after a systemic inflammatory condition develops.

Potential relevance: Evidence of inflammation in the hindlimb laminae suggests that the hindfeet should be addressed in the septic horse at risk for laminitis; however, laminitis is often less severe in the hindlimbs due to other factors, such as weightbearing and hoof angle.


Complemenary deoxyribonucleic acid








Chemokine (C-X-C motif)ligand


Deep dermal lamina






Messenger ribonucleic acid


Obel grade 1


Primary dermal lamina


Real-time quantitative polymerase chain reaction


The equine veterinarian is commonly confronted with laminitis as a sequela to sepsis associated with conditions such as strangulating intestinal lesions, enteritis, colitis, metritis and pleuropneumonia [1–4]. Except in cases of support limb laminitis, where the contralateral limb is at risk of laminitis regardless of whether it is a fore- or hindlimb [5], it is well known that clinical laminitis occurs much more frequently in the forelimb than in the hindlimb [2]. However, the reason for the much higher incidence or increased severity of laminitis in the forelimb digits is not known. One compelling factor commonly discussed is that the horse carries more weight on the forelimbs, resulting in more physical stress on the forelimb laminae. Although the forelimbs have been reported to bear 58% and the hindlimbs 42% of the horse's weight [6], it remains to be determined whether or not this difference is enough to result in a much higher incidence of laminitis in the forelimbs vs. the hindlimbs.

A complete understanding of the pathophysiology of laminitis remains to be elucidated; however, numerous pathological events have been reported to take place and are likely to play a role in laminar failure, including vascular alterations [7–10], enzymatic and metabolic dysfunctions [8,11–13] and inflammatory injury [14–16]. Interestingly, all of these studies concentrated on the forelimb laminae, leaving a paucity of information regarding laminar events in the hindfoot. Inflammatory signalling is reported to occur in most types of injuries purported to result in laminitis, including ischaemia-reperfusion [17–19], metabolic syndrome [20,21] and endotoxaemia/sepsis [22–26]. The presence of inflammation in the lamina of the forefeet during the development of laminitis has been well documented in both the black walnut extract and carbohydrate (CHO) overload experimental models of laminitis [15,16,27–33]. Recently, researchers in our laboratory have found increases in the mRNA concentrations of cytokines (i.e. interleukin [IL]-1β, IL-6 and IL-12p35) [33], chemokines (i.e. CXCL1 and CXCL8) [34], cyclo-oxygenase (COX)-2 [33] and endothelial adhesion molecules (i.e. intercellular adhesion molecule-1, E-selectin) [33] at the onset of lameness in the front laminae from horses given CHO. Leucocyte emigration, evidenced by immunostaining of calprotectin in neutrophils and macrophages, has also been reported at the onset of lameness in the dermal lamina of horses administered black walnut extract and CHO [35].

Therefore, due to the fact that tissue inflammation is a central component of pathological mechanisms reported to be involved in laminitis, the purpose of this study was to determine whether an inflammatory response occurs in the laminae of the hindlimb as previously reported in the forelimb laminae and if there is a difference in the magnitude of this response. Laminar tissue samples that were previously collected from control horses and horses administered CHO were used in order to assess mRNA concentrations of proinflammatory mediators by real-time quantitative PCR (RT-qPCR) and to assess the presence of leucocyte emigration by immunohistochemical techniques.

Materials and methods


The experimental protocol was approved by the Institutional Animal Care and Use Committee. Archived samples from 18 adult horses with a median body weight of 421 kg (ranging from 341 to 524 kg) and a median age of 5 years (ranging from 3 to 12 years old) were used in this study. Each horse was determined to be healthy and free of digital pathology determined by physical and lameness examinations and radiographic evaluation of the distal phalanx. Samples were obtained from 3 groups: a control group (CON; n = 6), a developmental period group (DEV; n = 6; determined by the onset of fever [>38.9°C], which was used as an indication of the presence of systemic inflammation) and an Obel grade 1 lameness group (OG1; n = 6; defined by the horse lifting its feet incessantly and/or having a short, stilted gait at trot). Prior to the experiment, the horses were quarantined for 2 weeks at the University of Missouri, housed in stalls and fed a regular ration of free choice hay. Laminar tissue from the forelimbs of these horses has been used in previous studies to assess inflammatory responses to carbohydrate overload [33–35].

Carbohydrate overload model

A carbohydrate gruel consisting of 85% cornstarch and 15% wood flour (17.6 g/kg bwt) was administered to each horse in the DEV and OG1 groups by nasogastric tube as previously described [33,36,37]. Six litres of deionised water were administered via nasogastric tube to each horse in the control group. All horses received complete physical examinations, consisting of rectal temperature, heart rate, respiratory rate, abdominal sounds, digital pulses, evaluation with hoof testers and gait evaluation immediately prior to nasogastric tube passage and at 2 h intervals following administration of either carbohydrate or water. Anaesthesia was induced in the DEV horses within 2 h of developing a temperature greater than 38.9°C (occurring between 12 and 22 h) and in the OG1 horses at the onset of lameness (occurring between 20 and 48 h). Control horses were anaesthetised 24 h after administration of water. Anaesthesia was induced with a combination of xylazine (1 mg/kg bwt i.v.) and ketamine (2.2 mg/kg bwt i.v.). Horses were placed in lateral recumbency and anaesthesia was maintained with isoflurane. While under general anaesthesia, the feet of all 4 limbs were rapidly removed (after placement of a tourniquet) by disarticulation of the metacarpophalangeal joint, and 1 cm thick sagittal sections of the digit were cut with a band saw. The laminae were rapidly harvested and immediately snap frozen in liquid nitrogen and later transferred to -80°C for storage until used for isolation of total RNA. Additional laminar samples were cut (approx. 1 cm × 1 cm × 3 mm) and immediately immersed in neutral-buffered 10% formalin for 24 h, followed by placement in 70% ethanol until paraffin embedding occurred. For each horse, 3 samples were embedded together in the same paraffin block. Once the tissues were collected, each horse was subjected to euthanasia with pentobarbital sodium and phenytoin sodium (20 mg/kg bwt i.v.).

Isolation of RNA and complementary DNA synthesis

Total RNA was extracted from 3 separate sections of dorsal lamina of each horse in the CON, DEV and OG1 groups using a kit (Absolutely RNA Miniprep)a which includes a DNase treatment to remove genomic DNA contamination. PolyA mRNA was then isolated using an mRNA extraction kitb followed by the production of complementary DNA (cDNA) via reverse transcription (Retroscript)c. A total of 400 ng of mRNA was used to make cDNA for each sample. The cDNA was frozen at -20°C and stored until used for RT-qPCR analysis.

Real-time quantitative PCR procedure

Real-time quantitative PCR was performed using a thermocyclerb and quantified with external standards with the fluorescent format for SYBR Green I dye as previously described [27,28]. Previously reported primers were designed from equine-specific sequences for IL -1β, IL-6, COX-1, COX-2, CXCL1 and CXCL8 and the housekeeping genes (β-actin and glyceraldehyde-3-phosphate dehydrogenase) [27,28,34]. Amplified cDNA fragments of each gene were ligated into a vector (TOPO 010 Escherichia coli)d and the vectors linearised with HindIII restriction enzymed for the use of templates to generate a standard curve for the RT-qPCR [27,28].

Each cDNA sample was diluted 1 : 5 and 1 : 500 with 1x Tris EDTA buffer to be used in the cytokine and housekeeping PCRs, respectively. The PCRs were performed in glass capillaries containing 5 µl of diluted sample and 15 µl of PCR master mixture. Master mixture included the following: one unit of Taq polymerased, 0.2 units of uracil-N-glycosylaseb, 1 : 10,000 dilution of SYBR Green stock solution, forward and reverse primers, PCR nucleotide plusb and PCR buffer. The PCR buffer (20 mmol/l Tris-HCl) contained 0.05% each of Tween 20 and nonionic detergent. Primers for CXCL8 were used at a concentration of 2.5 µmol/l, and all remaining primers were used at a concentration of 5 µmol/l.

Uracil-N-glycosylase activation, to prevent PCR product carryover, was conducted at 50°C for 2 min and was followed by denaturation at 95°C for 2 min. Amplification occurred for 40–45 cycles, with the annealing temperature set at 1–5°C below the melting temperature for each specific set of primers; extension was set at 72°C for 5 s, and fluorescence acquisition for 10 s in the SYBR Green format. Single fluorescence acquisition in each cycle was set at either 80 or 82°C, depending on the melting temperature of the cDNA product of interest as previously described [27,28]. After amplification, melting curves of the PCR product were acquired through a stepwise increase in temperature from 65 to 95°C.

Standard and target samples were prepared in separate capillaries. Standard curves and water for negative control were performed for each gene of interest and amplified with each series of reactions. Standards were made up of 10-fold serial dilutions of linearised plasmids containing the different gene-specific cDNA inserts. All samples were run in duplicate. To further assess interassay variation between the fore- and hindlimb laminae, IL-6, CXCL8 and COX-2 RT-qPCRs were also performed using the control and OG1 samples from both the fore- and hindfeet of the same horses.


Laminar sections from control horses and sections from both the front and hind feet of the OG1 horses were used to determine leucocyte migration through the immunostaining of calprotectin, which will stain both activated macrophages and neutrophils, as well as stressed laminar epithelium [30]. The paraffin-embedded samples were sectioned at 5 µm thickness, deparaffinised, and treated with a protease solution at 0.5 protease units/ml for 4 min (Proteinase K)d. After blocking, sections were incubated with a mouse monoclonal anti-human calprotectin antibody using a 1 : 250 dilution at 37°C for 90 min (MAC387)e. A biotinylated anti-mouse antibody (VECTASTAIN ABC kit)f– was then applied using a 1 : 100 dilution at room temperature for 60 min. The immunoreactivity was observed using a peroxidase (DAB kit)f following the instructions of the manufacturer and Harris haematoxyling was used for the counterstain. The front lamina served as a positive control [35].

Image analysis

All slides were examined by light microscopy in order to identify the presence and location of stained cells, and counting of cells for each image was blinded to remove bias from data collection. Whole-slide digital images were obtained from stained sections using an automated scanning robot (T2 Scan Scope) with ×40 magnification at a spatial sampling period of 0.2 µm per pixel. Whole-slide digital images were assessed using a software program (Aperio Image Scope)h to capture images randomly from each section, which were then transformed into jpg files. Ten random images were captured for each of the 2 areas of interest, namely the deep dermal lamina (DDL; the area between the base of the laminae and the parietal surface of the distal phalanx, including the deep neurovascular structures) and the primary dermal lamina (PDL) for every horse. The number of immunoperoxidase-positive cells per high-power (×40) field of each image was determined manually. The average numbers of immunoperoxidase-positive cells for each horse in the DDL and PDL were calculated and used for statistical analysis.

Data analysis

Average copy number from each sample was determined for each gene (housekeeping and inflammatory marker). As previously reported, RT-qPCR data from the housekeeping genes for each sample were evaluated by the computer software program geNormi, and a normalisation factor was determined [38]. β-Actin and glyceraldehyde-3-phosphate dehydrogenase were selected and then used by the geNorm software to create a normalisation factor for each sample. To determine the corrected copy number value for each sample, the amplification data obtained by RT-qPCR for each gene were divided by the normalisation factor of the selected housekeeping gene for the same sample. After normalisation, the fold change from the average control value was calculated for each sample. Real-time quantitative PCR and calprotectin immunohistochemistry data were analysed nonparametrically using the Kruskal-Wallis and Dunn's multiple comparisons test to compare groups. Statistical significance was set at P<0.05 for all tests.


Horses in the DEV group developed fevers, with an average high temperature of 39.8°C (range 39.6–40.6°C) occurring between 12 and 22 h post-CHO administration (median 17 h). Horses in the OG1 group developed signs of Obel grade 1 lameness (lifting feet incessantly and demonstrating a short, stilted gait at trot) occurring between 20 and 48 h post-CHO administration (median 25 h). Laminar mRNA concentrations of CXCL1 and CXCL8 were significantly increased in the DEV (P<0.05) and in the OG1 (P<0.01) horses when compared with control animals (Fig 1 and Table 1). Laminar mRNA concentrations of IL-1β, IL-6 (P<0.001) and COX-2 (P<0.01) were significantly increased only in the OG1 horses when compared with control animals (Fig 1 and Table 1). No change in laminar mRNA concentrations of COX-1 was present in either the DEV or the OG1 group (Fig 1 and Table 1). Both fore- and hindfeet laminae at the OG1 time point demonstrated increases (P<0.05) in laminar IL-6, CXCL8, and COX-2 mRNA concentrations when compared with their respective controls (Fig 2). However, no significant difference between the fore- and hindfeet laminae were noted in the OG1 horses (Fig 2). Comparisons between previously reported fold-change values for inflammatory mediators of the forefeet laminae [33,34] and the present results for hindfeet laminae are demonstrated in Table 1. Leucocyte emigration occurred in the OG1 horses (fore- and hindfeet laminae) in both the DDL and the PDL, evidenced by a significantly increased number of calprotectin-positive cells per high-power field when compared with controls (Fig 3). However, no significant differences in the numbers of calprotectin-positive cells between the fore- and hindfeet were noted in the OG1 horses. Increased calprotectin staining was also present in the epithelial cells from the OG1 horses (Fig 4), similar to that previously reported for the forelimb laminae [35].

Figure 1.

Median fold changes in hindlimb laminar inflammatory mediator mRNA concentrations after CHO administration in horses (DEV and OG1). Significant increases in inflammatory mediators when compared with control (CON) horses are indicated as *P<0.05, **P<0.01 and ***P<0.001.

Table 1. Fold changes* in lamellar inflammatory mediator mRNA concentrations after carbohydrate overload administration
GeneDevelopmental stage (DEV)Onset of lameness (OG1)
  1. Values displayed are significantly different from control values (P<0.05); NS, nonsignificant fold change from control values. *Expressed as median fold change from respective controls (fore vs. hind). **Values from data of previous reports by Leise et al. [33] for IL-1β, IL-6, COX-1 and COX-2 and by Faleiros et al. [34] for CXCL1 and CXCL8.

IL-1β NSNS11.2 ↑27 ↑
IL-6 NSNS2089 ↑2749 ↑
CXCL1 4.5 ↑9.3 ↑9.8 ↑20 ↑
CXCL8 45.4 ↑40 ↑25.6 ↑129 ↑
COX-2 NSNS31.2 ↑51.4 ↑
Figure 2.

Control (CON) and OG1 samples from both the fore (F) and hind (H) laminae of the same horses were performed in the same RT-qPCR for IL-6, CXCL8 and COX-2 to compare the response of the forelimb and hindlimb laminae directly. Median fold changes in laminar (both fore and hind) IL-6, CXCL8 and COX-2 mRNA concentrations after CHO administration in OG1 horses were significantly increased when compared with CON horses (*P<0.05). There was no significant difference between fore- and hindlimb laminae of the OG1 horses.

Figure 3.

Average number of calprotectin (MAC387)-positive cells per high-power (×40) field in the deep dermal lamina (DDL) and primary dermal lamina (PDL). Leucocyte emigration was significantly increased (*P<0.05) in the OG1 horses (both fore and hind) when compared with control animals. There was no difference between the fore- and hindlimb laminae. Images above graphs (Bars: 100 μm) demonstrate the DDL and PDL regions from the hindlimb laminae of OG1 horses.

Figure 4.

Comparison of calprotectin immunostaining of the hindlimb laminae between a CON horse (top) and an OG1 horse (middle). Note there is little to no staining of leucocytes in the CON horse vs. the OG1 (inset shows leucocyte staining in the primary dermal lamina surrounding blood vessels). In addition, there is increased staining present in the stressed laminar epithelium in the OG1 horse compared with CON. Bars: 80 μm.


Laminar failure secondary to a septic process or systemic inflammatory condition, occurring most commonly in forelimbs, is believed to be a result of multiple factors, including both physiological factors (e.g. inflammation, metabolic dysfunctions, enzymatic dysfunctions and vascular changes) and mechanical factors, such as weightbearing. Obtaining a better understanding of which factors play a significant role in the development of laminar failure is essential to finding successful preventative and treatment regimes. The presence of inflammation in the lamina during the developmental and initial stages of lameness has been well documented in the forefeet [15,16,27,28,30–33,38] and is believed to play an important role in laminar failure. However, to our knowledge, the majority of pathological events documented to occur in the forelimb laminae in laminitis, including inflammation, have not been assessed in the hindlimb laminae. The results of this study show that inflammation does indeed occur in the hindfeet and appears to be of a similar magnitude to that that reported for the forefeet [33,34]. Increases in hindlimb laminar gene expression of several proinflammatory mediators were present after the administration of CHO, including the chemokines CXCL1 and CXCL8 in both the DEV and OG1 horses, and IL-1β, IL-6 and COX-2 in the OG1 horses. No difference in laminar mRNA concentration was found between the OG1 hindlimb laminae and OG1 forelimb laminae for the 3 genes (IL-6, CXCL8 and COX-2) in which forelimb and hindlimb samples were assessed simultaneously in the same RT-qPCR, showing that the degree of laminar inflammation was the same for both the forelimb and hindlimb. Additionally, comparisons between previously reported fold changes for the other genes of interest (IL-1β and CXCL1/Gro-α) and the results from this study demonstrate a similar magnitude of increase in inflammatory gene expression [33]. Moreover, the magnitude of increase in leucocyte emigration into the laminar dermal tissue at the OG1 time after CHO administration was the same for the forelimb and hindlimb laminae, as no significant differences were noted between the feet in the face of marked increases between CON and OG1 horses in both fore- and hindlimbs. Thus, the similar increases in laminar inflammatory mediator expression and leucocyte emigration indicate that comparable inflammatory processes occur in the forelimb and hindlimb laminae during the early stages of laminitis. Regardless of whether laminar inflammation occurs due to laminar vascular compromise, laminar cellular activation/injury from circulating pathogen products and mediators, local laminar enzymatic or metabolic dysregulation, or a combination of these events, laminar inflammatory processes would be expected to be similar if the same severity of laminar dysregulation and injury were occurring in both the forelimb and hindlimb laminae. Thus, the present data suggest that the forelimb and hindlimb laminae undergo a similar degree of laminar injury/dysfunction initially, and that biomechanical forces are the primary factors that place the forelimb laminae at greater risk of structural failure in laminitis.

Although there are reports evaluating biomechanical differences between the fore- and hindlimb of the horse [39–41], there are limited studies of how the biomechanics of the hoof and lamina differ between fore- to hindfeet [6,42]. However, the most logical reason why the forelimbs are more affected in cases of laminitis is due to weightbearing loads. As discussed earlier, normal horses stand with the majority of their body weight (58%) on the forelimbs, with equal distributions between the right and left limbs [6]. In addition, vertical ground reaction forces of the forelimb have been reported to be greater at both the walk and the trot when compared with the hindlimb [40]. Although reports are conflicting, weight appears to play a role in the development of laminitis and outcome in some cases [1,43,44]. Horses that were reported to weigh more than 550 kg were twice as likely to develop laminitis post enteritis [1], and Baxter reported that horses with lower body weights were more likely to survive distal displacement of the third phalanx [44]. Although the forelimbs of horses are more commonly affected in laminitis, it is the authors' observation that involvement of all limbs most commonly occurs in severe cases of sepsis, with distal displacement of the third phalanx within the hoof capsule (indicative of complete laminar failure) occurring much more commonly than rotation. It is possible that these horses initially develop laminitis in their forefeet, become painful and subsequently shift more weight to the rear, resulting in increased stress on the hindlimb laminae. Although no significant difference in mean limb load was found after administration of CHO overload, increases in mean limb load in the hindlimbs has been found in chronic cases of laminitis [6]. Thus, it is possible that the combination of severe inflammation and increased weight on the hindfeet (in attempts to remove weight from the more painful forefeet) results in the condition where all 4 feet are affected. It is also possible that, due to the severity of the septic disease process in these animals, all laminae are so severely injured that they fail circumferentially with minimal stress, thus resulting in distal displacement of all 4 distal phalanges.

Other factors in relation to weight may be more important than total bodyweight alone. For example, hoof size and/or shape may be additional factors to consider. Weightbearing with greater contact area, secondary to footing (soft vs. hard ground) or hoof size, presumably reduces mechanical stress acting on the laminar cell at or near the epidermal–dermal interface [42]. The overall shape of the hindfoot varies from that of the forefoot, often having a shorter toe and steeper hoof to ground angle [45]. Altering the hoof size and shape by shorting the toe through trimming or bevelling of the ground surface of the shoes and increasing the hoof angle through the placement of heel wedges are common goals of treating the laminitic foot [46]. Varying hoof angles may also produce areas of greater hoof wall strain, thereby increasing the likelihood of laminar failure. However, conflicting results concerning dorsal laminar hoof wall strain have been reported, with some studies reporting decreased strain with increasing hoof angle through wedging of the heel [47,48], whereas other studies have found no difference in dorsal laminar hoof wall strain [49] or weightbearing in the toe region following application of a heel wedge [50]. Other studies have reported a change in distribution of forces on the hoof wall, with changes in hoof angles insinuating increased stress on the dorsal hoof wall with a lower angle, but did not include data documenting decreased strain [42,51]. Although many of the models are derived from ex vivo measurements [42,49,51], findings from these studies suggest, at least in part, that hoof shape and size may play a role in laminar function and failure.

Another possibility for the increased frequency of forelimb laminitis vs. hindlimb that has been considered is vascular compromise secondary to increased weightbearing. Several researchers have documented decreased blood flow and/or changes in blood pressures during periods of increased weightbearing and increased blood flow in nonweightbearing conditions [52–56]. Progressive increases in distal arterial resistance associated with the increased pressure exerted on the foot have been reported and are suspected to result from the compression of the soft tissue and vascular structure inside the foot, resulting in both an increase in peripheral arterial resistance and an increase in digital venous pressure [53]. Additionally, it has been determined that digital venous pressures increase with increasing loads on the digit and decrease with decreasing loads (which was purported to provide a hydraulic shock-absorbing function of the digital vasculature [54]). Although these data certainly provide support for the hypothesis that decreased digital blood flow due to greater weightbearing in the forelimbs may exacerbate the laminar injury occurring in laminitis and may therefore be the major factor putting the forelimbs at higher risk, the present data do not support this premise. Similar to the inflammatory response to sepsis, inflammation is a major component of the injury that occurs post ischaemia, evidenced by increases in cytokine expressions and increases in leucocyte emigration into affected tissue [57–60]. Thus, the similar magnitude of inflammatory signalling in the fore- and hindlimbs indicates that a vascular event occurring only in the forelimbs is an unlikely cause of the predisposition of the forelimbs to undergo structural failure in the laminitic horse.

Overall, this study provides the first evidence that inflammation does occur with equal magnitude in the hindlimbs as previously reported for the forelimbs in the early stages of CHO-induced laminitis. The presence of inflammation in this model and the infrequency of laminitis occurring in all 4 limbs secondary to grain overload or other systemic inflammatory diseases suggest that inflammatory injury plays only a partial role in laminar failure. Further research is warranted to determine the importance of other possible factors involved, including mechanical factors, such as weightbearing and the shape of the foot.

Authors' declaration of interests

The authors did not declare any conflict of interest.

Sources of funding

Supported by USDA NRI CSREES 2007-35204-18563.


The authors thank Dr Andrew Parks for his critical review of this manuscript and Mr Marc Hardman for his technical assistance involving graph and image development.

Manufacturers' addresses

a Stratagene, LaJolla, California, USA.

b Roche, Indianapolis, Indiana, USA.

c Ambion Inc., Austin, Texas, USA.

d Invitrogen, Carlsbad, California, USA.

e Abcam, Cambridge, Massachusetts, USA.

f Vector Laboratories, Burlingame, California, USA.

g Fisher Scientific, Kalamazoo, Michigan, USA.

h Aperio Image Scope, Vista, California, USA.

i Ghent University, Ghent, Belgium.