Phospholipase D-dependent and -independent mechanisms are involved in milk protein secretion in rabbit mammary epithelial cells

Phospholipase D has been implicated in membrane traffic in the secretory pathway of yeast and of some mammalian cell lines. Here we investigated the involvement of phospholipase D in protein transport at various steps of the secretory pathway of mammary epithelial cells. Treatment of rabbit mammary explants with butanol, which blocks the formation of phosphatidic acid, decreased the secretion of caseins and to a lesser extent that of whey acidic protein. Butanol interfered with both the endoplasmic reticulum to Golgi complex transport of the caseins and secretory vesicle formation from the trans-Golgi network. In contrast, the transport of whey acidic protein to the Golgi was less affected. Activation of protein kinase C enhanced the overall secretion of both markers and interestingly, this stimulation of secretion was maintained for whey acidic protein in the presence of butanol. Transphosphatidylation assays demonstrated the existence of a constitutive phospholipase D activity which was stimulated by the activation of protein kinase C. We conclude that phospholipase D plays a role in casein transport from the endoplasmic reticulum to the Golgi and in the secretory vesicle formation from the trans-Golgi network. Moreover, our results suggest a differential requirement for phospholipase D in the secretion of caseins and that of whey acidic protein.     

Arachidonic acid metabolism and casein secretion in lactating rabbit mammary epithelial cells: effects of inhibitors of prostaglandins and leukotrienes synthesis

The metabolism of arachidonic acid (AA) in fragments of lactating rabbit mammary glands in vitro was studied by considering the distribution of 13-[14C]AA in the cells, and the effects of inhibitors of cyclooxygenase and lipoxygenase pathway on the basal and prolactin (PRL)-stimulated casein secretion. 13-[14C]AA was incorporated in all classes of lipids and PRL increased transiently the percentage of free fatty acid after 1 and 5 min. Ten microM ETYA (5,8,11,14-Eicosatetraynoic acid), a tetrayne analogue of AA inhibited prostaglandins F2 alpha (PGF2 alpha) production but not leukotrienes B4 and C4 (LTB4 and LTC4) production and increased basal casein secretion. 10(-4) M DCHA (Docosahexaenoic acid) a competitive inhibitor of prostaglandin-synthetase inhibited PGF2 alpha production but did not affect basal nor PRL-stimulated casein secretion. Fourteen microM indomethacin inhibited PGF2 alpha and LTC4 production and PRL-stimulated casein secretion. Ten microM NdgA (nordihydroguaiaretic acid) an inhibitor of lipoxygenase pathway, inhibited LTB4 and LTC4 production, increased basal level of casein secretion and inhibited PRL-stimulated casein secretion. Hundred microM caffeic acid, an inhibitor of glutathione-S-transferase (GST), a class of enzymes implied in the transformation of LTA4 into LTC4, had the same effect that NDGA on basal and PRL-stimulated casein secretion. These findings show that inhibitors of AA metabolites alter casein secretion.     

Suppressive effects of curcumin on milk production without inflammatory responses in lactating mammary epithelial cells

BackgroundCurcumin is a naturally occurring polyphenol found in Curcuma longa with multiple therapeutic properties, such as anti-inflammatory, wound healing and anti-cancer effects. Curcuma longa is also used as a galactagogue to improve milk production during lactation.PurposeTo assess curcumin could have therapeutic potential for breastfeeding mothers, we investigated whether and how curcumin influences milk production in lactating mammary epithelial cells (MECs) at the cellular and molecular levels.MethodsWe prepared a lactating MEC culture model that produced milk components and formed less-permeable tight junctions (TJs) to investigate the molecular mechanism of curcumin on milk production, TJs, and inflammation in vitro.ResultsCurcumin downregulated milk production in lactation MECs concurrently with inactivation of lactogenesis-relating signaling (STAT5 and glucocorticoid receptor). The maintenance of a less-permeable TJ barrier was also confirmed, although the TJ protein claudin-4 increased. Curcumin inactivated NFκB and STAT3 signaling, which are closely involved in inflammatory responses in weaning and mastitis mammary glands. The expression levels of IL-1β and TNF-α were also decreased by curcumin treatment. Furthermore, curcumin blocked activation of inflammatory signaling by lipopolysaccharide treatment in MECs, similar to those in MECs that were treated with diclofenac sodium. The drastic phosphorylation of ERK was induced by curcumin treatment in the absence of EGF. U0126, an inhibitor of ERK phosphorylation, attenuated the adverse effects of curcumin on lactating MECs.ConclusionThe results of the present study suggests that curcumin downregulates milk production via inactivation of STAT5 and GR signaling with concurrent suppression of inflammatory responses via STAT3 and NFκB signaling in MECs. These findings provide new insights into the role of curcumin as a mild suppressor of milk production without inflammatory damages in breastfeeding mothers.     

Effect of nortriptyline on Ca2+ handling in SIRC rabbit corneal epithelial cells

To explore the effect of nortriptyline, a tricyclic antidepressant, on cytosolic free Ca2+ concentrations ([Ca2+]i) in corneal epithelial cells, [Ca2+]i levels in suspended SIRC rabbit corneal epithelial cells were measured by using fura-2 as a Ca2+-sensitive fluorescent dye. Nortriptyline at concentrations between 20-200 microM increased [Ca2+]i in a concentration-dependent manner. The Ca2+ signal was reduced partly by removing extracellular Ca2+. Nortriptyline-induced Ca2+ influx was inhibited by the store-operated Ca2+ channel blockers econazole and SK&F96365, the phospholipase A2 inhibitor aristolochic acid, and alteration of activity of protein kinase C. In Ca2+-free medium, 200 microM nortriptyline pretreatment greatly inhibited the rise of [Ca2+]i induced by the endoplasmic reticulum Ca2+ pump inhibitor thapsigargin. Conversely, pretreatment with thapsigargin or 2,5-di-tert-butylhydroquinone (BHQ; another endoplasmic reticulum Ca2+ pump inhibitor) nearly abolished nortriptyline-induced [Ca2+]i rise. Inhibition of phospholipase C with U73122 decreased nortriptyline-induced [Ca2+]i rise by 75%. Taken together, nortriptyline induced [Ca2+]i rises in SIRC cells by causing phospholipase C-dependent Ca2+ release from the endoplasmic reticulum and Ca2+ influx via store-operated Ca2+ channels.     

Store-operated Ca2+ current in prostate cancer epithelial cells. Role of endogenous Ca2+ transporter type 1

Ca(2+) influx via store-operated channels (SOCs) following stimulation of the plasma membrane receptors is the key event controlling numerous processes in nonexcitable cells. The human transient receptor potential vanilloid type 6 channel, originally termed Ca(2+) transporter type 1 (CaT1) protein, is one of the promising candidates for the role of endogenous SOC, although investigations of its functions have generated considerable controversy. In order to assess the role of CaT1 in generating endogenous store-operated Ca(2+) current (I(SOC)) in the lymph node carcinoma of the prostate (LNCaP) human prostate cancer epithelial cell line, we manipulated its endogenous levels by means of antisense hybrid depletion or pharmacological up-regulation (antiandrogen treatment) combined with functional evaluation of I(SOC). Antisense hybrid depletion of CaT1 decreased I(SOC) in LNCaP cells by approximately 50%, whereas enhancement of CaT1 levels by 60% in response to Casodex treatment potentiated I(SOC) by 30%. The functional characteristics of I(SOC) in LNCaP cells were similar in many respects to those reported for heterologously expressed CaT1, although 2-aminoethoxydiphenyl borate sensitivity and lack of constitutive current highlighted notable departures. Our results suggest that CaT1 is definitely involved in I(SOC), but it may constitute only a part of the endogenous SOC, which in general may be a heteromultimeric channel composed of homologous CaT1 and other transient receptor potential subunits.     

Suppression of EGF-induced cell proliferation by the blockade of Ca2+ mobilization and capacitative Ca2+ entry in mouse mammary epithelial cells

The role of intracellular Ca2+ stores and capacitative Ca2+ entry on EGF-induced cell proliferation was investigated in mouse mammary epithelial cells. We have previously demonstrated that EGF enhances Ca2+ mobilization (release of Ca2+ from intracellular Ca2+ stores) and capacitative Ca2+ entry correlated with cell proliferation in mouse mammary epithelial cells. To confirm their role on EGF-induced cell cycle progression, we studied the effects of 2,5-di-tert-butylhydroquinone (DBHQ), a reversible inhibitor of the Ca2+ pump of intracellular Ca2+ stores, and SK&F 96365, a blocker of capacitative Ca2+ entry, on mitotic activity induced by EGF. Mitotic activity was examined using an antibody to PCNA for immunocytochemistry. SK&F 96365 inhibited capacitative Ca2+ entry in a dose-dependent manner (I50: 1-5 microM). SK&F 96365 also inhibited EGF-induced cell proliferation in the same range of concentration (I50: 1-5 microM). DBHQ suppressed [Ca2+]i response to UTP and thus depleted completely Ca2+ stores at 5 microM. DBHQ also inhibited EGF-induced cell proliferation at an I50 value of approximately 10 microM. The removal of these inhibitors from the culture medium increased the reduced mitotic activity reversibly. Using a fluorescent assay of DNA binding of ethidium bromide, no dead cells were detected in any of the cultures. These results indicate that the inhibitory effects of SK&F 96365 and DBHQ on cell proliferation were due to the inhibition of capacitative Ca2+ entry and Ca2+ mobilization suggesting the importance of capacitative Ca2+ entry and Ca2+ mobilization in the control of EGF-induced cell cycle progression in mouse mammary epithelial cells.     

Role of the Ca2+ -activated Cl- channels bestrophin and anoctamin in epithelial cells

Two families of proteins, the bestrophins (Best) and the recently cloned TMEM16 proteins (anoctamin, Ano), recapitulate properties of Ca(2+)-activated Cl(-) currents. Best1 is strongly expressed in the retinal pigment epithelium and could have a function as a Ca(2+)-activated Cl(-) channel as well as a regulator of Ca(2+) signaling. It is also present at much lower levels in other cell types including epithelial cells, where it regulates plasma membrane localized Cl(-) channels by controlling intracellular Ca(2+) levels. Best1 interacts with important Ca(2+)-signaling proteins such as STIM1 and can interact directly with other Ca(2+)-activated Cl(-) channels such as TMEM16A. Best1 is detected in the endoplasmic reticulum (ER) where it shapes the dynamic ER structure and regulates cell proliferation, which could be important for renal cystogenesis. Ca(2+)-activated Cl(-) channels of the anoctamin family (TMEM16A) show biophysical and pharmacological properties that are typical for endogenous Ca(2+)-dependent Cl(-) channels. TMEM16 proteins are abundantly expressed and many reports demonstrate their physiological importance in epithelial as well as non-epithelial cells. These channels are also activated by cell swelling and can therefore control cell volume, proliferation and apoptosis. To fully understand the function and regulation of Ca(2+)-activated Cl(-) currents, it is necessary to appreciate that Best1 and TMEM16A are embedded in a protein network and that they probably operate in functional microdomains.     

Ca2+ transport and Ca2+-dependent ATP hydrolysis by Golgi vesicles from lactating rat mammary glands.

Ca2+ transport across mammary-gland Golgi membranes was measured after centrifugation of the membrane vesicles through silicone oil. In the presence of 2.3 microM free Ca2+ the vesicles accumulated 5.8 nmol of Ca2+/mg of protein without added ATP, and this uptake was complete within 0.5 min. In the presence of 1 mM-ATP, Ca2+ was accumulated at a linear rate for 10 min after the precipitation of intravesicular Ca2+ with 10 mM-potassium oxalate. ATP-dependent Ca2+ uptake exhibited a Km of 0.14 microM for Ca2+ and a Vmax. of 3.1 nmol of Ca2+/min per mg of protein. Ca2+-dependent ATP hydrolysis exhibited a Km of 0.16 microM for Ca2+ and a Vmax. of 10.1 nmol of Pi/min per mg of protein. The stoichiometry between ATP-dependent Ca2+ uptake and Ca2+-stimulated ATPase varied between 0.3 and 0.7 over the range 0.03-8.6 microM-Ca2+. Both Ca2+ uptake and Ca2+-stimulated ATPase were strongly inhibited by orthovanadate, which suggests that the major mechanism by which Golgi vesicles accumulate Ca2+ is through the action of the Ca2+-stimulated ATPase. However, Ca2+ uptake was also decreased by the protonophore CCCP (carbonyl cyanide m-chlorophenylhydrazone), indicating that it may occur by other mechanisms too. The effect of CCCP may be related to the existence of transmembrane pH gradients (delta pH) in these vesicles: the addition of 30 microM-CCCP reduced delta pH from a control value of 1.06 to 0.73 pH unit. Golgi vesicles also possess a Ca2+-efflux pathway which operated at an initial rate of 0.5-0.57 nmol/min per mg of protein.     

Role of microtubules in milk secretion — Action of colchicine on microtubules and exocytosis of secretory vesicles in rat mammary epithelial cells

Summary Effect of colchicine on microtubules was studied in mammary epithelial cells treated both in vivo and in vitro with the alkaloid. Three hours after the intramammary infusion of colchicine, secretory activity of mammary epithelia ceased, milk constituents accumulated and were randomly distributed within the cytoplasm, sometimes leaking into the perialveolar connective tissue, and autophagic vacuoles were prevalent. It appeared that an accelerated involutionary process was occurring. No microtubules were observed after this treatment. In vitro treated cells appeared to be less affected by the alkaloid. Although numerous casein-containing secretory vesicles accumulated in the cytoplasm, lipid droplet accumulation was less, and fewer autophagic vacuoles were observed, although lysosomes were commonly observed. Occasionally, obliquely sectioned microtubules were found in cells treated with low concentrations of colchicine but were absent at higher colchicine concentrations; however, paracrystalline inclusions (tubulin aggregates) were observed in some cells at all concentrations of the drug. These observations provide evidence that drugs which interfere with microtubule integrity reduce the secretory activity in mammary epithelia. This evidence is consistent with the concept of an association of the microtubular system and the secretory process.     

Endocytic prolactin routes to the secretory pathway in lactating mammary epithelial cells

Plasma-borne prolactin is carried from blood to milk by transcytosis across the mammary epithelial cell through the endocytic and secretory pathways. To determine the precise route of prolactin endocytosis, intracellular transport of biotinylated prolactin was monitored, in parallel with endocytosis of fluorescein isothiocyanate-conjugated dextran and IgG, by using pulse-chase experiments in lactating mammary fragments and in enzymatically dissociated acini. Biotinylated prolactin was sorted to vesiculo-tubular organelles whereas dextran was very rapidly carried to the lumen and IgG remained accumulated in the basal region of cells. To determine whether prolactin uses routes into and across the Golgi and trans-Golgi network, localisation of biotinylated prolactin was combined with the immunofluorescence detection of caseins and, respectively, p58 and TGN38. Biotinylated prolactin strongly colocalised with caseins during a chase but not all or only very little with p58 and TGN38. To characterise the organelles involved in transcytosis, gold-labelled prolactin, experimentally accumulated in late endosomes and which recovered a normal transport, was localised by electron microscopy. In mammary fragments incubated at low temperature, and in mammary fragments from rats fed with a lipid-deprived diet, transport of gold-labelled prolactin was restored by increasing the temperature and by adding arachidonic acid, respectively. These data demonstrate that a sorting occurs very rapidly between prolactin, dextran and IgG. They suggest that prolactin may reach the biosynthetic pathway after direct fusion between multivesicular bodies and secretory vesicles.     

Endocytosis and intracellular transport of transferrin across the lactating rabbit mammary epithelial cell.

To study the transcytosis and segregation of ligand in the mammary epithelial cell, endocytosis and intracellular transit of human blood transferrin were followed in lactating rabbit mammary epithelial cells. Human transferrin labeled with biotin added to an incubation medium was bound to the basal membrane of mammary epithelial cells and carried across the cell to the lumen of the acini within 5-60 min. At the same time, biotinylated human transferrin accumulated at the apex of the cell. After incubation with human transferrin labeled with colloidal gold, label was detected inside endosome-like structures, vesicles and saccules of the Golgi apparatus, and inside the lumen within 2-5 min. A significant label accumulated at the apex of the cell after 30-60 min. Biotin labeling did not modify the time of transit of human transferrin, as attested by comparison with the time of transit of native transferrin. Human transferrin was never detected inside vesicles containing casein micelles. In contrast, rabbit milk transferrin was immunocytochemically detected inside vesicles containing casein micelles. These results indicate that transcytosis of human transferrin follows a pathway different from vesicles that carry casein micelles.     

A Ca2+-stimulated adenosine triphosphatase in Golgi-enriched membranes of lactating murine mammary tissue.

A membrane fraction isolated from lactating murine mammary tissue and enriched for the Golgi membrane marker enzyme galactosyltransferase exhibited Ca2+-stimulated ATPase activity (Ca-ATPase) in 20 microM-free Mg2+ and 10 microM-MgATP, with an apparent Km for Ca2+ of 0.8 microM. Exogenous calmodulin did not enhance Ca2+ stimulation, nor could Ca-ATPase activities be detected in millimolar total Mg2+ and ATP. When assayed with micromolar Mg2+ and MgATP the Ca-ATPases of skeletal-muscle sarcoplasmic reticulum and of calmodulin-enriched red blood cell plasma membranes were half-maximally activated by 0.1 microM- and 0.6 microM-Ca2+ respectively. All three Ca-ATPases were inhibited by similar micromolar concentrations of trifluoperazine, but the Golgi activity was unaffected by quercetin in concentrations which completely inhibited both the sarcoplasmic-reticulum and red-blood-cell enzymes. The results are consistent with the hypothesis that the high-affinity Ca-ATPase is responsible for the ATP-dependent Ca2+ transport exhibited by Golgi-enriched vesicles derived from lactating mammary gland [Neville, Selker, Semple & Watters (1981) J. Membr. Biol. 61, 97-105; West (1981) Biochim. Biophys. Acta 673, 374-386].     

Freeze-fracture observations of the lactating rat mammary gland. Membrane events during milk fat secretion

Membrane events during milk fat secretion were analyzed by freeze-fracture of the rat mammary gland. Two modes of milk fat secretion were observed: extrusion of fat droplets surrounded by a portion of the apical plasma membrane of the alveolar epithelial cells and, less frequently, release into the alveolar lumen of fat droplets contained in intracytoplasmic vacuoles. The extrusion process consists of two asynchronous events: clearing of membrane particles (probably including integral membrane proteins) and bulging of the apical plasma membrane. Most fat droplets are extruded with a bilayer membrane envelope (milk fat globule membrane) partially devoid of particles. The segregation of membrane particles may represent the onset of a process of structural degradation of the milk fat globule membrane.     

Involvement of the Ca2+-dependent K+ channel activity in the hyperpolarizing response induced by epidermal growth factor in mammary epithelial cells

Epidermal growth factor (EGF) induces a hyperpolarizing response of 5-20 mV amplitude in mouse mammary epithelial cells in culture. The amplitude of the hyperpolarizing response was reduced by more than 60% within several minutes after addition of blockers of voltage and/or Ca2+-dependent K+ channels such as tetraethylammonium (7 mM) or quinine (0.29 mM). Both nifedipine (0.15 mM), a blocker of the Ca2+ channel, and ruthenium red (2 mM), an inhibitor of the Ca2+-binding site, also reduced the amplitude of the hyperpolarizing response by more than 60%. The Ca2+ ionophore, A23187 (3.8 microM), induced a large hyperpolarization, which was 25-40 mV and lasted about 3 min. These data suggest that activity of the Ca2+-dependent K+ channel was involved in the EGF-induced hyperpolarizing response of the mammary epithelial cells.     

Ca2+ dependence of inositol 1,4,5-trisphosphate-induced Ca2+ release in renal epithelial LLC-PK1 cells

We have studied arginine vasopressin (AVP)-, thapsigargin- and inositol 1,4,5-trisphosphate (InsP₃)-mediated Ca²⁺ release in renal epithelial LLC-PK₁ cells. AVP-induced changes in the intracellular free calcium concentration ([Ca²⁺]_i) were studied in indo-1 loaded single cells by confocal laser cytometry. AVP-mediated Ca²⁺ mobilization was also observed in the absence of extracellular Ca²⁺, but was completely abolished after depletion of the intracellular Ca²⁺ stores by 2 μM thapsigargin. Using ⁴⁵Ca²⁺ fluxes in saponin-permeabilized cell monolayers, we have analysed how InsP₃ affected the Ca²⁺ content of nonmitochondrial Ca²⁺ pools in different loading and release conditions. Less than 10% of the Ca²⁺ was taken up in a thapsigargin-insensitive pool when loading was performed in a medium containing 0.1 μM Ca²⁺. The thapsigargin-insensitive compartment amounted to 35% in the presence of 110 μM Ca²⁺, but Ca²⁺ sequestered in this pool could not be released by InsP₃. The thapsigargin-sensitive Ca²⁺ pool, in contrast, was nearly completely InsP₃ sensitive. A submaximal [InsP₃], however, released only a fraction of the sequestered Ca²⁺. This fraction was dependent on the cytosolic as well as on the luminal [Ca²⁺]. The cytosolic free [Ca²⁺] affected the InsP₃-induced Ca²⁺ release in a biphasic way. Maximal sensitivity toward InsP₃ was found at a free cytosolic [Ca²⁺] between 0.1 and 0.5 μM, whereas higher cytosolic [Ca²⁺] decreased the InsP₃ sensitivity. Other divalent cations or La³⁺ did not provoke similar inhibitory effects on InsP₃-induced Ca²⁺ release. The luminal free [Ca²⁺] was manipulated by varying the time of incubation of Ca²⁺ -loaded cells in an EGTA-containing medium. Reduction of the Ca²⁺ content to one-third of its initial value resulted in a fivefold decrease in the InsP₃ sensitivity of the Ca²⁺ release. © 1993 Wiley-Liss, Inc.     

Visfatin is present in bovine mammary epithelial cells, lactating mammary gland and milk, and its expression is regulated by cAMP pathway

Visfatin was originally identified as a growth factor for immature B cells, and recently demonstrated to bind insulin receptor. Visfatin mRNA and protein were detected by RT-PCR and Western blot analysis in cloned bovine mammary epithelial cells, lactating bovine mammary gland and human breast cancer cell line, MCF-7. Immunocytochemical staining localized the visfatin protein in the cytosol and nucleus of both cells. Quantitative-RT-PCR analysis revealed that the expression of the visfatin mRNA was significantly elevated when treated with forskolin (500 microM), isopreterenol (1-10 microM) and dibutyric cyclic AMP (1 mM) for 24 h, and significantly reduced when treated with insulin (5-50 ng/ml) and dexsamethasone (0.5-250 nM) for 24 h. These results indicate that mammary epithelial cells express the visfatin protein and secrete them into the milk.     

Role of Ca2+ release from ryanodine stores in generation of NMDA-induced Ca2+ signal in rabbit hippocampus.

In vivo microdialysis combined with measurements of 45Ca efflux from pre-labelled rat hippocampus has been utilised in our laboratory to demonstrate NMDA-evoked 45Ca2+ release to dialysate, reflecting calcium-induced calcium release (CICR) via ryanodine receptors (RyR). In the present study we attempted to reproduce this phenomenon in the rabbit hippocampus. Application of 1 mM NMDA to dialysis medium induced a decrease in Ca2+ concentration in dialysate, as a result of extracellular Ca2+ influx to neurones. The release of 45Ca2+ was not observed, instead a decrease in 45Ca2+ efflux rate from the NMDA treated rabbit hippocampus was noted, along with release to dialysate of prostaglandin D2, taurine and phosphoethanolamine. All these effects, reflecting different steps of intracellular calcium signalling, were insensitive to 100 microM dantrolene and 50 microM ryanodine, RyR modulators known to interfere with NMDA-evoked 45Ca2+ release in the rat hippocampus. Thus, although the results of this study demonstrate the role of extracellular Ca2+ influx to neurones in NMDA-evoked generation of Ca2+ signal in the rabbit hippocampus, the activity of CICR was not detected.     

Role of Ca2+ in induction and secretion of dengue virus-induced cytokines

The present study was undertaken to investigate the role of calcium ions (Ca₂₊) in the induction and secretion of the dengue type 2 virus induced cytotoxic factor and the cytotoxin. This was done by using calcium channel blocking drugs such as verapamil, nifedipine or diltiazem hydrochloride. The production of cytotoxic factor was significantly reduced by treatment of dengue type 2 virus infected mice with verapamil. Similarly, a dosedependent inhibition of the secretion of cytotoxic factor was observed, when spleen cells of the virus-primed mice were treatedin vitro with the 3 calcium channel blockers. The production of cytotoxin by macrophages was abrogated by pretreatment with calcium channel blockers but had little effect on its secretion as shown by treatment of macrophages with verapamil at 1 h after the induction to later periods up to 18 h. The findings thus show that in the induction of both the cytokines Ca₂₊ plays a critical role; on the other hand it is required for the secretion of the cytotoxic factor but not for that of the cytotoxin.