The Comparative Pathology of the Glycosidase Inhibitors Swainsonine,Castanospermine,and Calystegines A3,B2,and C1 in Mice

Abstract

To study various polyhydroxy-alkaloid glycosidase inhibitors, 16 groups of 3 mice were dosed using osmotic minipumps with swainsonine (0, 0.1, 1, and 10 mg/kg/day), castanospermine, and calystegines A₃, B₂, and C₁ (1, 10, and 100 mg/kg/day). After 28 days, the mice were euthanized, necropsied, and examined using light and electron microscopy. The high-dose swainsonine–treated mice developed neurologic disease with neuro-visceral vacuolation typical of locoweed poisoning. Castanospermine- and calystegines-treated mice were clinically normal; however, high-dose castanospermine–treated mice had thyroid, renal, hepatic, and skeletal myocyte vacuolation. Histochemically, swainsonine- and castanospermine-induced vacuoles contained mannose-rich oligosaccharides. High-dose calystegine A₃–treated mice had increased numbers of granulated cells in the hepatic sinusoids. Electron microscopy, lectin histochemistry, and immunohistochemistry suggest these are pit cells (specialized NK cells). Histochemically, the granules contain glycoproteins or oligosaccharides with abundant terminal N-acetylglucosamine residues. Other calystegine-treated mice were histologically normal. These findings indicate that swainsonine produced lesions similar to locoweed, castanospermine caused vacuolar changes with minor changes in glycogen metabolism, and only calystegine A₃ produced minimal hepatic changes. These also suggest that in mice calystegines and castanospermine are less toxic than swainsonine, and as rodents are relatively resistant to disease, they are poor models to study such induced storage diseases.

Introduction

Several plants poisonous to livestock have been identified that contain polyhydroxy-alkaloids that inhibit specific cellular glycosidases. For example, swainsonine is found in many locoweeds (certain Astragalus and Oxytropis species) and is a potent inhibitor of lysosomal α-mannosidase and Golgi mannosidase II (Table 1). Purified swainsonine has been shown to reproduce identical disease and lesions to those produced by locoweed in various animal models (James et al., 1991; Stegelmeier et al., 1995). Recently swainsonine has also been identified with mixtures of other glycosidase-inhibiting polyhydroxy-alkaloids in toxic species of Ipomoea, Sida, Solanum, Physalis, and Convolvulus (Asano et al., 1995; Haraguchi et al., 2003; Molyneux et al., 1995). As many of the diseases induced by these plants differ from the lesions and clinical signs produced by swainsonine alone, more information is needed to determine the contribution of swainsonine and co-occurring alkaloids to these intoxications.

Castanospermine, an indolizidine alkaloid, is structurally and functionally similar to swainsonine (Table 1). It was first isolated from the seeds of the Moreton Bay chestnut tree (Castanospermum australe) that causes fatal gastroenteritis with myocardial degeneration and nephrosis in cattle and horses (Molyneux et al., 1990). Castanospermine inhibits α-glucosidase and was included in this work for comparison as the calystegines inhibit cellular β-glucosidases.

Calystegines were first identified as hydroxylated nortropane alkaloids, varying in both number and substitution pattern of the hydroxy groups, from Calystegia sepium (Goldmann et al., 1990). Subsequently, similar and additional calystegines were identified in Atropa belladonna, Convolvulus arvensis, and various other Solanum, Scopolia, Hyoscyamus, Morus, Datura, Physalis, and Ipomoea species (Asano et al., 1994; Asano et al., 1996; Casteel et al., 1989; Drager et al., 1994; Molyneux et al., 1993; Molyneux et al., 1994). Common calystegines include the trihydroxylated calystegines that are classified as group A, tetrahydroxylated as group B, and pentahydroxylated as group C. As shown in Table 1, some calystegines are potent glycosidase inhibitors with high in vitro affinities. Subsequently, it has been suggested that in vivo they are likely to disrupt intestinal glycosidases, lysosomal function, and glycoprotein processing. However, this has not been demonstrated in an animal model. As some calystegine-containing plants contain mixtures of other toxins, including swainsonine, it has been difficult to definitively prove the role that calystegines play in poisoning. For example, Ipomoea carnea occurs in tropical regions throughout the world. Poisoning has been reported in Brazil, Sudan, India, and Mozambique (Armien et al., 2007; De Blaogh et al., 1999; Idris et al., 1973; Tirkey et al., 1987). Additionally poisoning has been experimentally reproduced by feeding I. carnea to goats, sheep, and cattle (Adam et al., 1973; Armien et al., 2007; Damir et al., 1987; Idris et al., 1973). Mixtures of swainsonine (0.0029%) and calystegines B₁, B₂, B₃, and C₁ (0.0045%) have been isolated from Brazilian I. carnea (Armien et al., 2007; Haraguchi et al., 2003). However, the contribution that the calystegines play in Ipomoea toxicity is unknown. Recent studies suggest that the clinical signs and distribution of lesions differ from those reported for swainsonine alone (Armien et al., 2007; Hueza et al., 2005; James et al., 1991; Stegelmeier et al., 1995). Additional work is needed to define calystegine toxicity and determine their role in Ipomoea toxicity.

The purpose of this study was to develop a small animal model so that animals could be treated with relatively small amounts of purified alkaloids; to characterize, in a dose response fashion, the lesions of swainsonine, castanospermine, and calystegines A3, B2, and C1; and to compare the reported lesions produced by the whole Ipomoea plant with those produced by these respective purified glycosidase inhibitors.

Materials and Methods

Glycosidase Inhibitors

The individual alkaloids were purified from ground plant material using published techniques (Haraguchi et al., 2003). The purity of the samples was verified at better than 99% purity using GC-MS analysis (Haraguchi et al., 2003; Molyneux et al., 1993). The biological activities of the purified alkaloids were also verified demonstrating similar glycosidase inhibition as those reported using various in vitro enzyme inhibition assays (Haraguchi et al., 2003; Molyneux et al., 1991; Stegelmeier et al., 1995).

Animals

Forty-eight mature Swiss Webster male mice (Simonsen Laboratories Inc., Gilroy, CA) were randomly divided into 16 groups of 3 animals each. On study days 0 and 14, all mice were anesthetized with halothane and osmotic minipumps (Alzet 2002 with a delivery of 0.5 μL/h for 14 days) (Alzet Corp., Palo Alto, CA) were aseptically implanted in the subcutis over the right or left scapulae. On the day 14 surgery, the originally implanted pumps were removed and a second was installed over the other shoulder. The pumps were loaded with swainsonine, castanospermine, calystegine A₃, calystegine B₂, calystegine C₁, and saline to obtain target doses of 0.1, 1.0, and 10.0 mg/kg/day swainsonine, and 1.0, 10.0, and 100.0 mg/kg/day castanospermine, calystegine A₃, calystegine B₂, and calystegine C₁. Equal volumes of saline were loaded in the pumps implanted in the last group as a negative control. After 28 days of dosing, all the mice were humanely killed, blood was collected, and the animals were perfused via cardiac cannulation with Karnowsky’s fixative for 45 min. After fixation, the mice were necropsied and the tissues were prepared for histologic and ultrastructural examination. At necropsy, and when surgically removed, the pumps were salvaged; the remaining dose was removed and analyzed using GC/MS to determine the actual delivered dose.

Special Stains and Immunochemistry

Tissues with histologic vacuolation were stained with periodic acid Schiff (PAS) and Luxol fast blue following previously described techniques (Morgolis and Pickett, 1956; Selby et al., 1971). Lectin histochemistry was performed using modifications of previously reported techniques (Alroy et al., 1985; Castagnaro, 1990; Stegelmeier et al., 1995). Briefly tissues on substrated slides were deparaffinized, rehydrated, and incubated in methanolic, 3% hydrogen peroxide to block the endogenous peroxidases. After blocking, the slides were washed 3 times in deionized water (2.5 min) and once in phosphate buffered saline (PBS, 10 min, 37°C). The tissue was then digested in a warm (37°C) trypsin bath (0.25 g calcium chloride, 0.25 g trypsin, and 250 mL deionized water, pH = 7.7) for 25 min. After digestion the slides were washed once in water and 3 times in PBS (5 min each). Nonspecific binding was blocked by incubating the slides for 60 min in 2% equine serum and 0.2% bovine serum albumin in PBS. The slides were then incubated with the following concentrations of biotinylated lectins (Sigma Chemical Co., St. Louis, MO). Arachis hypogea (PNA) 0.05 mg/mL, Concanavalia ensiformis (Con-A) 0.01 mg/mL, Glycine X 3max (SBA) 0.05 mg/mL, Phytolacca americana (PWM) 0.05 mg/mL, Triticum vulgaris (WGA) 0.05 mg/mL, and Ulex europaeus (UEA-1) 0.02 mg/mL. Specifically bound lectins were then identified in the tissues using a peroxide-linked avidin/biotin complex (Vectastain, Vector Laboratories Inc., Burlingame, CA) following the manufacturer’s suggested procedure. As negative controls, binding specificity was confirmed either by omitting the lectin or by specifically blocking tissue binding using lactose (PNA), α-D-methyl-mannose (ConA), α-D-N-acetyl-galactosamine (SBA), N-acetyl-glucosamine (PWM), acetyl-neuraminic acid (WGA), and α-L-fucose (UEA-1). The tissues were counter-stained with hematoxylin, dehydrated, and coverslipped.

Immunohistochemistry was done using minor modification of published techniques (Pinard et al., 2006). Briefly, paraffin-embedded tissue sections were deparaffinized, rehydrated, and incubated in antigen retrieval solution (Dako Inc., Carpinteria, CA) for 20 min. Endogenous peroxidase was blocked by incubating in 3% hydrogen peroxide for 15 min, and nonspecific binding was blocked using protein-blocking agent (Dako) washes of 10 min before application of the primary antibody (anti-VWF, 1:1000; anti-alpha 1-antitrypsin, 1:100; pan cytokeratins, 1:100; chromogranin A, 1:1000; chromgranin B, 1:100; PGP9.5, 1:1000; synapthophypin, 1:1000; CD3, 1:1000; CD45, 1:100; Factor 8, 1:1000; vimentin, 1:10; antilysozyme antibody, 1:1000). The primary antibodies (rabbit origin antibodies from Dako and R&D Systems Minneapolis, MN) were allowed to react with the tissue for 30 min at room temperature. Specific binding was detected using an autostainer (Ventana Medical Systems Inc., Tucson, AZ). The label, streptavidin-biotin-immunoperoxidase (Dako), was used for detection. 3,3′-diaminobenzidine was used as substrate (Dako), and the sections were counterstained with Mayer’s hematoxylin. Positive immunohistochemical controls included liver from control mice to which the appropriate antisera were added and negative controls in which the primary antibodies were replaced with homologous nonimmune sera.

Tissues were processed for both light and electron microscopy following commonly used techniques. Serial sections of all neurologic sections were also stained with a modified Bodian silver stain and luxol fast blue/periodic acid Schiff stains (Gato, 1987; Yamamoto and Hirano, 1986).

Statistical Analysis

Dose and daily weight gains were compared between groups by analysis of variance using SAS statistical software and the general linear model procedures (SAS Statistical Software 1986, SAS Institute Inc., Cary, NC). Regressions were also done to determine if there was a relationship between dose and daily weight gain.

Results

The actual dose each group received was higher than the projected dose for all of the alkaloid groups (Table 2). However, all the differences in the dose were consistent between groups with all animals receiving about 40% more than the projected doses.

There were no significant differences between the weight gains of the different castanospermine or calystegine groups. However, the high-dose swainsonine mice did gain less than controls and all the other groups (P = 0.03); these mice also ate less. The high-dose castanospermine and calystegine groups all tended to have higher daily food consumption (up from 4.7–5.0 g/mouse/day for the control and low-dose groups to 5.3–5.8 g/mouse/day for the high-dose groups). No statistical comparisons were possible as the animals were penned as groups and food consumption is reported as average food consumed or wasted per mouse per day (Table 2). All of the castanospermine, the calystegine, and the low- and medium-dose swainsonine groups appeared to be clinically unaffected by the dose. However, the high-dose swainsonine–treated mice were clinically anxious, easily excited, and slight intention tremors were apparent upon movement. No significant gross lesions were found in mice in any of the treatment groups.

No significant histologic changes were found in any of the control animals or in any animals receiving 0.1 mg swainsonine/kg/day. All of the mice receiving 1.6 mg swainsonine/kg/day and 14.1 mg swainsonine/kg/day mice had vacuolization of the proximal convoluted tubules of the kidney, and the thyroid follicular epithelium (Figure 1a). Lectin stains of these vacuoles stained positive with WGA, Con-A, and PWM (suggestive of increased residues or oligosaccharides containing free α-D-mannose, α-glucose, β-(1–4)-N-acetyl-glucosamine, and N-acetyl-neuramic acid). Glycine max lectin (SBA) consistently stained macrophages and other cells of the reticulo-endothelial system in both control and treated animals (suggestive of normal amounts of α-D-acetyl-glucosamine and α-D-mannose). No vacuolar-related staining was detected with the other lectins. Many of the vacuoles were also diastase resistant, PAS positive.

The high swainsonine dose mice also had vacuolation of many of the dendritic cells and macrophages in the spleen, lymph node, thymus, liver, heart, and lung. They also had extensive vacuolation of hepatocytes, pancreatic acinar cells, parietal cells of the stomach, and cells of the salivary gland and prostate gland. Vacuolation was less severe in various neurons and glial cells, especially those of the peripheral ganglia, nuclei in the pons and medulla, and the Purkinje cells (Figure 2).

Only the high-dose castanospermine-treated mice had mild vacuolar changes in the proximal convoluted tubule epithelium and the thyroid follicular epithelium (Figure 1b). These mice also had minimal vacuolation of the periportal hepatocytes. Affected hepatocytes also had increased amounts of diastase-resistant and diastase-sensitive PAS-positive material. Some skeletal myocytes were also vacuolated with increased amounts of PAS-positive material, most likely glycogen.

Mice treated with calystegine A₃ at 140 mg/kg/day had increased numbers of round to oval granulated cells within hepatic sinusoids (Figure 3a). The granules were large (0.1–0.2 μm) and eosinophilic, and they often seemed to displace the nucleus to the margins of the cell. Lectin histochemistry demonstrated that these granules were positive with WGA and PWM and lightly positive with ConA (Figure 3b). No altered or vacuolar-related staining was detected with the other lectins in any of the calystegine-treated mice. Ultrastructurally the A₃-induced granulated cells were flattened along the hepatic sinusoids beneath the endothelial cells and they were often adjacent to Kupffer cells and hepatocytes. Ultrastructurally their cytoplasm was markedly distended with numerous electron-dense granules. The granules were 400–600 A in diameter, homogeneously electro-dense, membrane-bound granules with small numbers of more dense condensates on the margins. Smaller numbers of multivesicular bodies and mitochondria were also present (Figure 3c). Immunohistochemically the granules and cell membranes were negative for antibodies against factor 8, VWF, several pan-cytokeratins, vimentin, chromogranin A and B, synapthophysin, PGP9.5, CD3, and CD45. The granule’s vacuoles were strongly PAS positive, and staining was diastase resistant (Figure 3d). Hepatocytes of these mice were swollen with minimal cytoplasmic vacuolation (Figure 3a). No additional histologic changes were detected in other tissues. No histologic, histochemical, or ultrastructural changes were detected in any of the calystegine B₂- or C₁-treated mice.

Discussion

Ipomoea carnea toxicity has been primarily attributed to the toxic effects of swainsonine. However, many portions of the plant contain nearly as much calystegine B₂ and C₁, with smaller amounts of B₁ and B₃ (Armien et al., 2007; Haraguchi et al., 2003). The contribution these calystegines play in toxicity is unknown. The lesions of Ipomoea carnea poisoning are similar to locoweed (Astragalus and Oxytropis spp.) and swainsonine poisoning, but the progression and many of the clinical signs are more severe and progressive. Clinical locoweed poisoning is a chronic disease that requires weeks of ingestion for animals to progressively lose condition and develop weakness, trembling, proprioceptive deficits, altered gait, anxiety, changes in demeanor, and reluctance to stand or move. With extended poisoning, animals become emaciated, dehydrated, and recumbent. Most poisoned animals are euthanized, and related mortalities are commonly due to some accident or misadventure (Stegelmeier et al., 1998; Stegelmeier et al., 1999). Ipomoea poisoning progresses rapidly with high reported mortality. In some areas, Ipomoea poisoning is known as “mata cabra” or “goat killer.” Reports suggest poisoned animals have seizures and die suddenly. Although the exact cause of these Ipomoea-associated deaths has not been identified, it has been suggested that it is due to swainsonine with the added effect of calystegine poisoning.

Under these study conditions, we found that relatively high doses of calystegines B₂ and C₁ caused no detectable change in mice. These mice were treated with about 140 mg/kg/day with purified calystegine B₂ or C₁. This dose is more than 60 times higher than the doses associated with induced and clinical Ipomoea poisoning in goats (Armien et al., 2007). These findings suggest that in mice calystegines alone probably contribute little to Ipomoea toxicity directly and that additional toxins or toxin-synergism should be investigated.

Only relatively high doses of swainsonine produced neurologic histologic changes in mice similar to those seen in locoweed poisoning of livestock. The dose, histologic changes, and lectin histochemistry are similar to those previously reported in rats and hamsters (Stegelmeier et al., 1995; Stegelmeier et al., 1999; Stegelmeier et al., 2007). These findings support the hypothesis that rodents are relatively resistant to swainsonine toxicity. Most livestock and wildlife species develop significant clinical and histologic lesions at swainsonine doses of 0.4 mg/kg, with durations as short as several weeks (Stegelmeier et al., 1999; Stegelmeier et al., 2007). These doses are nearly 25 times lower than the swainsonine doses required to produce relatively subtle lesions in these mice. This has several implications. It indicates that rodents are poor models of swainsonine-induced neurologic disease. It also has implications for the safety studies of swainsonine, castanospermine, and the calystegines, as swainsonine has been evaluated as an antiviral, antimetastatic, and cancer therapy or therapeutic adjuvant (Goss et al., 1994; Oredipe et al., 2003b; Oredipe et al., 2003a). When the preclinical safety studies were done using rodent models, the risk analyses concerning safety of swainsonine may have been underestimated. Additional work is needed to identify why rodents are less susceptible and to determine if that mechanism can be used to alter toxicity in other species.

If rodents are similarly resistant to calystegine toxicity, our findings suggest that rodent infusion models are poor indicators of calystegine toxicity. Similar to our findings of few detectable calystegine-induced histologic or ultrastructural lesions in mice, Hueza et al. were also unable to produce lesions in rats dosed orally with purified calystegine _B1–3 and C₁ (Hueza et al., 2005). These rats were orally dosed with doses of 3.0 mg/Kg (B₁), 2.4 mg/Kg (B₂), 1.2 mg/Kg (B₃), and 1.8 mg/Kg (C₁) for 14 days. We had hypothesized that much higher doses, longer durations, and continuous infusion would produce lesions in mice. This seems not to be the case. Because calystegine isolation is difficult and expensive, our current efforts are to develop a sensitive animal model for calystegine toxicity in small goats or pigs.

As shown in Table 1, many calystegines are potent glycosidase inhibitors with glycosidase affinities similar to those of swainsonine and castanospermine. However, this study suggests that at these doses in mice, they cause minimal clinical or microscopic pathology. Theoretically the calystegines should inhibit cellular α- and β-galactosidase and β-glucosidase. Similar to mannosidase II that is inhibited by swainsonine, these enzymes are important steps in N-linked glycoprotein synthesis (Elbein, 1989; Goldmann et al., 1996). Other than the minimal hepatic changes in the high-dose A₃ group, we were unable to detect structural changes in these mice suggestive of altered glycoprotein synthesis or widespread accumulation of stored cellular metabolites. More work is needed to determine if there were undetected calystegine-induced functional, biochemical, or molecular changes. It may be that calystegines synergize with swainsonine to produce the more progressive and severe disease seen in Ipomoea poisoned animals. Alternatively, the ability of calystegines to inhibit enzymes that are important to digestive processes may alter the absorption and excretion of swainsonine, exacerbating its effect. Such combined or “multiple hit” theories are common in plant toxicities. For example, the synergistic effects of black sage and tetradymia toxins are required to produce liver disease and photosensitization (big head) in sheep (Johnson, 1974). Additionally, multiple hit inhibition of both mannosidase II and lysosomal α-mannosidase by swainsonine has been used to explain the difference in progression and lesion distribution between genetic and swainsonine-induced mannosidosis (Stegelmeier et al., 1995).

Only the high-dose calystegine A₃–treated mice had swollen and heavily granulated hepatic sinusoidal cells with mild hepatocellular swelling. The electron-dense granules morphologically resemble secretory granules, and the histologic and ultrastructural location and cellular features suggest these are pit cells. Pit cells are specialized NK cells that are immunophenotypically, morphologically, and functionally different from circulating NK cells. They are named pit cells because of their characteristic granules that resemble the pit in a grape (Luo et al., 2000). Most cells contain around 13 granules. Pit cell granules are generally ~0.3 μm, round, electron-dense, and membrane-bound structures. The dense inclusion in some granules may contain a less dense halo on one side, and there is often a progression of granules with less electron-dense material until they are similar to multivesicular bodies (Wisse et al., 1976). The granules in the calystegine A₃–treated mice were about 2 to 3 times larger (0.4–0.6 μm) than normal pit cell granules. They became the predominate ultrastructural feature as they are much larger than mitochondria, Golgi, and multivesicular bodies. The immunohistochemical studies failed to demonstrate immunoreactivity with antibodies to various other endothelial, lymphocyte, macrophage, or epithelial cell markers. There are several NK-lymphocyte-specific antibodies, but none are available for formalin-fixed/paraffin-embedded tissues (Luo et al., 2000). The pit cell vacuoles were diastase resistant, PAS positive. Lectin and chemical histochemistry suggests the vacuoles are filled with glycoproteins or oligosaccharides rich with terminal N-acetylglucosamine residues. Only weakly positive response with ConA (which binds high mannose glucosaccharides), these are different from the mannose-rich glycosaccharides seen in swainsonine-induced vacuoles. Additional studies are needed to better identify and define this cell type and to determine how calystegine A₃ treatment alters their function.

Calystegine A₃ has been isolated from potatoes (Solanum tuberosum), sweet potatoes (Ipomoea batatas), egg plant (Solanum melongena), and chili peppers (Capsicum frutescens) (Asano et al., 1997). It inhibits rat β-glucosidase and human α-galactosidase and β-xylosidase, but at higher IC₅₀ values than calystegines B₂ or C₁ (Table 1). Calystegines B₂ or C₁ are potent inhibitors of bovine, human, and rat β-glucosidase and α-galactosidase and human β-xylosidase at rates between 5 and 25 times higher than A₃. This suggests that the calystegine A₃–induced vacuoles are probably not due to inhibition of these enzymes. Certainly more work is needed to determine if calystegine A₃ alters some other enzyme or even if it is directly toxic to certain cell populations. The lack of clinical response suggests that this change may be an incidental finding as the altered sinusoidal cells did not appear to cause significant clinical or functional disease.

These findings indicate that swainsonine-treated mice developed clinical and histologic lesions similar to locoism of livestock, but only at relatively high doses. This suggests that mice are similar to other rodents as they are less susceptible to swainsonine intoxication than most livestock. Mice treated with high castanospermine doses developed mild renal, thyroid, hepatic, and skeletal muscle lesions similar to genetic glycogenosis. In contrast, calystegines B₂ and C₁ similar to those in Ipomea carnea did not induce significant histological lesions in these conditions; calystegine A₃ produced minimal hepatic change of unknown importance. These findings suggest that mice are a relatively poor model for calystegine toxicity. The difference in toxicity between swainsonine and the calystegines also suggests that most of the toxicity of plants containing mixtures of these glycoside-inhibiting alkaloids is due to swainsonine.