Improved measurements of intracellular nitric oxide in intact microvessels using 4,5-diaminofluorescein diacetate

4,5-Diaminofluorescein diacetate (DAF-2 DA) has been widely used for the measurement of nitric oxide (NO) in living cells and tissues. We previously established a method that demonstrated platelet activating factor (PAF)-induced endothelial NO production in intact venules using DAF-2 DA. In previous applications, the loading dye was removed from the extracellular space before NO measurements. However, in high permeability vessels, endothelial cells quickly released the accumulated intracellular DAF-2 after the washout, which compromises the NO measurement. The objective of this study was to investigate if the presence of DAF-2 DA during NO measurements could overcome the dye retention problem and enhance the sensitivity of NO detection. Experiments were conducted in individually perfused rat venules, and endothelial NO was measured using fluorescence imaging under basal and stimulated conditions with continuous perfusion of DAF-2 DA. Continuous dye perfusion was found to promote a relatively constant endothelial dye concentration in both normal and high permeability vessels throughout the experiment. With the use of this method, the basal and stimulated NO was quantified after endothelial DAF-2 concentrations reached a steady state. Our results showed enhanced sensitivity of detecting PAF-stimulated NO compared with a previous method. We also found that the hydrolyzed intracellular DAF-2, the precursor of DAF-2 triazole, contributed significantly to the measured fluorescence and that an appropriate subtraction of non-NO-dependent intracellular DAF-2 fluorescence is critical for the assessment of NO in living tissues. This method overcame the dye leakage problem, enhanced the sensitivity of NO detection, and improved NO quantification, demonstrating significant advantages over existing methodologies using DAF-2.     

Detection of intracellular nitric oxide using a combination of aldehyde fixatives with 4,5-diaminofluorescein diacetate

Using 4,5-diaminofluorescein diacetate (DAF-2DA), which was recently developed for the detection of intracellular nitric oxide (NO) in living cells, we examined the sensitivity of intracellular NO in cells treated with some fixatives. Cultured human umbilical vein endothelial cells loaded with DAF-2DA in the presence of 10(-6) M acetylcholine showed intense fluorescence when fixed in paraformaldehyde or glutaraldehyde, but no fluorescence could be detected after fixation in ethanol or acetone. Fluorescence generation depended on the combination of each aldehyde fixative with DAF-2, which is produced enzymatically from DAF-2DA within the cells. Subtracting the fluorescence intensity of non-activated controls from that of cells activated by acetylcholine indicated the NO produced in the stimulated cells, since the control cells that took up DAF-2DA also generated fluorescence when treated with aldehyde fixatives. Thus, detection of intracellular NO by combining aldehyde fixatives with DAF-2DA is useful for examining the functions of NO in cells both in situ and in vivo.     

Determinants of shear stress-stimulated endothelial nitric oxide production assessed in real-time by 4,5-diaminofluorescein fluorescence

The extremely short biological half-life of endothelial-derived nitric oxide (NO) has impeded real-time measurements of NO synthesis. We used the membrane-permeable fluorescent probe 4,5-diaminofluorescein diacetate (DAF-2 DA) to study determinants of NO synthesis in bovine aortic endothelial cells (BAECs). A step increase in shear stress (SS) from 0.3 to 3.4 dyne/cm(2) triggered an increase in DAF-2 fluorescence starting 3.0 +/- 0.5 min after the flow rise and peaking at 44.7 +/- 7.2 min. This was abolished by intracellular Ca(2+) chelation, but was unaffected by blocking extracellular Ca(2+) influx or by inhibiting SS-related changes in intracellular pH. The increase in DAF-2 fluorescence occurred significantly earlier in BAECs transfected with either superoxide dismutase (SOD) or catalase (CAT), indicating concomitant reactive oxygen species (ROS) generation by SS and "competition" between ROS- and DAF-2-NO interactions. These data provide novel insights into several NO signaling determinants and reveal that DAF-2 can assess real-time SS-stimulated NO synthesis in endothelial cells. This should facilitate the analysis of NO-signaling pathways.     

Reliable in vitro measurement of nitric oxide released from endothelial cells using low concentrations of the fluorescent probe 4,5-diaminofluorescein

4,5-Diaminofluorescein (DAF-2) and its membrane-permeable derivate DAF-2 diacetate are fluorescent probes that have been developed to perform real-time biological detection of nitric oxide (NO). Their use for intracellular imaging, however, has recently been seriously questioned and data using DAF-2 for extracellular NO detection at low levels, as for example released from endothelial cells, are rare. Here we show that a reliable detection of low levels of NO in biological systems by DAF-2 is possible (a) by using low DAF-2 concentrations (0.1 microM) and (b) by subtracting the DAF-2 auto-fluorescence from the measured total fluorescence. The described method allows easy real-time detection of endothelial NO formation.     

Interfering with nitric oxide measurements. 4,5-diaminofluorescein reacts with dehydroascorbic acid and ascorbic acid

4,5-Diaminofluorescein (DAF-2) is widely used for detection and imaging of NO based on its sensitivity, noncytotoxicity, and specificity. In the presence of oxygen, NO and NO-related reactive nitrogen species nitrosate 4,5-diaminofluorescein to yield the highly fluorescent DAF-2 triazole (DAF-2T). However, as reported here, the DAF-2 reaction to form a fluorescent product is not specific to NO because it reacts with dehydroascorbic acid (DHA) and ascorbic acid (AA) to generate new compounds that have fluorescence emission profiles similar to that of DAF-2T. When DHA is present, the formation of DAF-2T is attenuated because the DHA competes for DAF-2, whereas AA decreases the nitrosation of DAF-2 to a larger extent, possibly because of additional reducing activity that affects the amount of available N(2)O(3) from the NO. The reaction products of DAF-2 with DHA and AA have been characterized using capillary electrophoresis with laser-induced fluorescence detection and electrospray mass spectrometry. The reactions of DAF-2 with DHA and AA are particularly significant because DHA and AA often colocalize with nitric-oxide synthase in the central nervous, cardiovascular, and immune systems, indicating the importance of understanding this chemistry.     

Pitfalls and limitations in using 4,5-diaminofluorescein for evaluating the influence of polyphenols on nitric oxide release from endothelial cells

The reagent 4,5-diaminofluorescein (DAF-2) is a widely utilized and sensitive fluorescent probe for real-time assessment of nitric oxide (NO) production. In this study we investigated the feasibility of using DAF-2 for detection of NO release from EA.hy 926 human endothelial cells stimulated with plant polyphenols. Flavonoids have recently gained much interest because of reported beneficial effects on vasodilatation, which have been ascribed to stimulation of endothelial NO production. DAF-2 shows moderate fluorescence, and because certain phenolic compounds quench fluorescence or fluoresce themselves, we utilized liquid chromatography to avoid interference. Our investigations with (+)-catechin and trans-resveratrol as test phenolic compounds revealed various previously undescribed principal methodologic pitfalls and limitations. Under assay conditions (+)-catechin displayed a highly significant increase in fluorescence intensity so that a control of test compound stability is advisable. Moreover, DAF-2 was subject to conversion to triazolofluorescein (DAF-2T) under certain assay and storage conditions; thus control of spontaneous reagent conversion is advisable. Finally, formation of DAF-2T was dose-dependently inhibited by polyphenols to a degree consistent with their free radical scavenging activity. The inhibition of DAF-2T generation seems to contradict previous reports on enhanced NO release from endothelial cells by (+)-catechin and resveratrol. Therefore, the planning of experiments involving NO measurement in biological systems and interpretation of results requires substantial scrutiny.     

Oral atorvastatin therapy increases nitric oxide-dependent cutaneous vasodilation in humans by decreasing ascorbate-sensitive oxidants

Elevated low-density lipoproteins (LDL) are associated with cutaneous microvascular dysfunction partially mediated by increased arginase activity, which is decreased following a systemic atorvastatin therapy. We hypothesized that increased ascorbate-sensitive oxidant stress, partially mediated through uncoupled nitric oxide synthase (NOS) induced by upregulated arginase, contributes to cutaneous microvascular dysfunction in hypercholesterolemic (HC) humans. Four microdialysis fibers were placed in the skin of nine HC (LDL = 177 ± 6 mg/dl) men and women before and after 3 mo of a systemic atorvastatin intervention and at baseline in nine normocholesterolemic (NC) (LDL = 95 ± 4 mg/dl) subjects. Sites served as control, NOS inhibited, L-ascorbate, and arginase-inhibited+L-ascorbate. Skin blood flow was measured while local skin heating (42°C) induced NO-dependent vasodilation. After the established plateau in all sites, 20 mM ?ngname? was infused to quantify NO-dependent vasodilation. Data were normalized to maximum cutaneous vascular conductance (CVC) (sodium nitroprusside + 43°C). The plateau in vasodilation during local heating (HC: 78 ± 4 vs. NC: 96 ± 2% CVC(max), P < 0.01) and NO-dependent vasodilation (HC: 40 ± 4 vs. NC: 54 ± 4% CVC(max), P < 0.01) was reduced in the HC group. Acute L-ascorbate alone (91 ± 5% CVC(max), P < 0.001) or combined with arginase inhibition (96 ± 3% CVC(max), P < 0.001) augmented the plateau in vasodilation in the HC group but not the NC group (ascorbate: 96 ± 2; combo: 93 ± 4% CVC(max), both P > 0.05). After the atorvastatin intervention NO-dependent vasodilation was augmented in the HC group (HC postatorvastatin: 64 ± 4% CVC(max), P < 0.01), and there was no further effect of ascorbate alone (58 ± 4% CVC(max,) P > 0.05) or combined with arginase inhibition (67 ± 4% CVC(max,) P > 0.05). Increased ascorbate-sensitive oxidants contribute to hypercholesteromic associated cutaneous microvascular dysfunction which is partially reversed with atorvastatin therapy.     

cGMP-independent inotropic effects of nitric oxide and peroxynitrite donors: potential role for nitrosylation

Nitric oxide (NO) has concentration-dependent biphasic myocardial contractile effects. We tested the hypothesis, in isolated rat hearts, that NO cardiostimulation is primarily non-cGMP dependent. Infusion of 3-morpholinosydnonimine (SIN-1, 10(-5) M), which may participate in S-nitrosylation (S-NO) via peroxynitrite formation, increased the rate of left ventricular pressure rise (+dP/dt; 19 +/- 4%, P < 0.001, n = 11) without increasing effluent cGMP or cAMP. Superoxide dismutase (SOD; 150 U/ml) blocked SIN-1 cardiostimulation and led to cGMP elaboration. Sodium nitroprusside (10(-10)-10(-7) M), an iron nitrosyl compound, did not augment +dP/dt but increased cGMP approximately eightfold (P < 0.001), whereas diethylamine/NO (DEA/NO; 10(-7) M), a spontaneous NO. donor, increased +dP/dt (5 +/- 2%, P < 0.05, n = 6) without augmenting cGMP. SIN-1 and DEA/NO +dP/dt increase persisted despite guanylyl cyclase inhibition with 1H-(1,2,4)oxadiazolo-(4,3,-a)quinoxalin-1-one (10(-5) M, P < 0.05 for both donors), suggesting a cGMP-independent mechanism. Glutathione (5 x 10(-4) M, n = 15) prevented SIN-1 cardiostimulation, suggesting S-NO formation. SIN-1 also produced SOD-inhibitable cardiostimulation in vivo in mice. Thus peroxynitrite and NO donors can stimulate myocardial contractility independently of guanylyl cyclase activation, suggesting a role for S-NO reactions in NO/peroxynitrite-positive inotropic effects in intact hearts.     

Inhibitory effects of catecholamines and anti-oxidants on the fluorescence reaction of 4,5-diaminofluorescein, DAF-2, a novel indicator of nitric oxide

4,5-Diaminofluorescein (DAF-2) is a newly developed indicator of nitric oxide (NO). Two amino groups of DAF-2 are oxidized by NO. We investigated the effects of reducers on the NO-induced oxidation of DAF-2. NOC-5 (0.1-10 microM), a NO-donor, concentration-dependently elicited fluorescence with 10 microM DAF-2. The rate of the fluorescence reaction was dependent on the width of the excitation band path. The presence of catecholamines (1 microM), but not tyrosine or phenylephrine, attenuated the fluorescence induced by NOC-5. Ascorbate and other reducers like dithiothreitol, 2-mercaptoethanol, or glutathione (all 1 mM) abolished the fluorescence. These results suggest that reducers attenuate the NO-induced fluorescence of DAF-2 mainly through an anti-oxidative action.     

Identification of the cysteine nitrosylation sites in human endothelial nitric oxide synthase

S-nitrosylation, or the replacement of the hydrogen atom in the thiol group of cysteine residues by a -NO moiety, is a physiologically important posttranslational modification. In our previous work we have shown that S-nitrosylation is involved in the disruption of the endothelial nitric oxide synthase (eNOS) dimer and that this involves the disruption of the zinc (Zn) tetrathiolate cluster due to the S-nitrosylation of Cysteine 98. However, human eNOS contains 28 other cysteine residues whose potential to undergo S-nitrosylation has not been determined. Thus, the goal of this study was to identify the cysteine residues within eNOS that are susceptible to S-nitrosylation in vitro. To accomplish this, we utilized a modified biotin switch assay. Our modification included the tryptic digestion of the S-nitrosylated eNOS protein to allow the isolation of S-nitrosylated peptides for further identification by mass spectrometry. Our data indicate that multiple cysteine residues are capable of undergoing S-nitrosylation in the presence of an excess of a nitrosylating agent. All these cysteine residues identified were found to be located on the surface of the protein according to the available X-ray structure of the oxygenase domain of eNOS. Among those identified were Cys 93 and 98, the residues involved in the formation of the eNOS dimer through a Zn tetrathiolate cluster. In addition, cysteine residues within the reductase domain were identified as undergoing S-nitrosylation. We identified cysteines 660, 801, and 1113 as capable of undergoing S-nitrosylation. These cysteines are located within regions known to bind flavin mononucleotide (FMN), flavin adenine dinucleotide (FAD), and nicotinamide adenine dinucleotide (NADPH) although from our studies their functional significance is unclear. Finally we identified cysteines 852, 975/990, and 1047/1049 as being susceptible to S-nitrosylation. These cysteines are located in regions of eNOS that have not been implicated in any known biochemical functions and the significance of their S-nitrosylation is not clear from this study. Thus, our data indicate that the eNOS protein can be S-nitrosylated at multiple sites other than within the Zn tetrathiolate cluster, suggesting that S-nitrosylation may regulate eNOS function in ways other than simply by inducing dimer collapse.     

Photoactivation and calcium sensitivity of the fluorescent NO indicator 4,5-diaminofluorescein (DAF-2): implications for cellular NO imaging

The fluorescent indicator of nitric oxide (NO), 4,5-diaminofluorescein (DAF-2), and its membrane-permeable derivative (DAF-2 diacetate) have been recently developed to perform real-time biological imaging of NO. In this study, we show that DAF-2 is strongly influenced by factors other than the concentration of NO itself. Using measurements with a fluorimeter as well as fluorescence microscopy, we found that the divalent cation concentration in the medium, as well as the incident light, strongly affects the ability of DAF-2 to detect NO. Calcium, in particular, enhanced the signal detection of NO released by NO donors by up to 200 times. With multiple and longer exposures to light, no bleaching of the dye was observed but, instead, a potentiation of the fluorescence response could be measured. While these two properties will affect the use and interpretation of the hitherto acquired data with this fluorescent compound, they may also open up new possibilities for its application.     

Simultaneous measurement of intracellular Ca(2+) and nitric oxide: a new method

Different methods to measure the unstable radical nitric oxide (NO) have been established. We are going to present a new method to measure intracellular calcium and NO simultaneously in endothelial cells. A new fluorescent dye (DAF-2) has been developed recently which binds NO resulting in an enhanced fluorescence. We loaded porcine aortic endothelial cells with Fura-2, a fluorescent dye commonly used to measure intracellular calcium, and DAF-2 simultaneously (cell permeable dyes). Using excitation wavelengths of lambda 340 nm (Fura-2) and lambda 485 nm (DAF-2) we could show that thrombin induces an intracellular calcium increase and simultaneously a NO formation in endothelial cells which could be blocked by a NO synthase inhibitor. This new method of a simultaneous measurement of intracellular calcium and NO provides the possibility to follow intracellular calcium and NO distributions online, and is sensitive enough to monitor changes of NO formed by the constitutive endothelial NO-synthase.     

Reductive nitrosylation and S-nitrosation of hemoglobin in inhomogeneous nitric oxide solutions.

Elucidating the reaction of nitric oxide (NO) with oxyhemoglobin [HbFe(II)O2] is critical to understanding the metabolic fate of NO in the vasculature. At low concentrations of NO, methemoglobin [HbFe(III)] is the only detectable product from this reaction; however, locally high concentrations of NO have been demonstrated to result in some iron-nitrosylhemoglobin [HbFe(II)NO] and S-nitrosohemoglobin (SNO-Hb) formation. Reductive nitrosylation through a HbFe(III) intermediate was proposed as a viable pathway under such conditions. Here, we explore another potential mechanism based on mixed valenced Hb tetramers. The oxidation of one or two heme Fe(II) in the R-state HbFe(II)O2 has been observed to lower the oxygen affinity of the remaining heme groups, thus creating the possibility of oxygen release and NO binding at the heme Fe(II) sites. This mixed valenced hypothesis requires an allosteric transition of the Hb tetramer. Hence, this hypothesis can account for HbFe(II)NO formation, but not SNO-Hb formation. Here, we demonstrate that cyanide attenuated the formation of SNO-Hb by 30-40% when a saturated NO bolus was added to 0.1-1.0 mM HbFe(II)O2 solutions. In addition, HbFe(II)NO formation under such inhomogeneous conditions does not require allostericity. Therefore, we concluded that the mixed valenced theory does not play a major role under these conditions, and reductive nitrosylation accounts for a significant fraction of the HbFe(II)NO formed and approximately 30-40% of SNO-Hb. The remaining SNO-Hb is likely formed from NO oxidation products.     

Hyaluronan in breast cancer: correlations with nitric oxide synthases and tyrosine nitrosylation.

Reactive oxygen species (ROS), including nitric oxide (NO(*)), are associated with all steps of carcinogenesis. Hyaluronan (HA), a high-molecular-mass glycosaminoglycan overexpressed in a variety of human malignancies also has ROS-scavenging properties. We histochemically studied the level of HA in breast carcinoma cells and their stroma and compared it with the expression of NO(*) synthases (NOSs), major antioxidant enzymes, and nitrotyrosine. We also assessed whether the level of HA correlates with traditional prognostic factors of breast cancer and survival. Stromal HA level was moderate or high in all the samples studied (n=185), and 84% of the lesions showed HA-positive carcinoma cells. Intense stromal HA signal was associated with high neuronal NOS expression (p=0.009), whereas tumor-cell associated HA was inversely correlated with nitrotyrosine expression (p=0.027). Of the traditional prognostic factors, tumor cell-associated HA was correlated with poor differentiation (p=0.011), and high stromal HA levels were associated with aggressive features of the carcinomas such as large primary tumor (p=0.002), poor differentiation (p=0.019), and estrogen (p=0.012) and progesterone receptor negativity (p=0.009). High stromal HA level also significantly predicted poorer survival. The strong positive correlation between neuronal NOS and stromal HA could reflect NO(*)-stimulated synthesis of HA, an extracellular matrix alteration that favors breast cancer progression. Furthermore, it is suggested that, while acting as a scavenger of NO(*)-derived radicals, cell-associated HA undergoes partial fragmentation, release from receptors, and further degradation in lysosomes, and thus becomes undetectable in histological sections.     

Sustained nitric oxide delivery delays nitric oxide-dependent apoptosis in macrophages: contribution to the physiological function of activated macrophages

Treatment of the macrophage cell line RAW 264.7 with the short-lived NO donor S-nitrosoglutathione triggers apoptosis through the release of mitochondrial mediators. However, continuous supply of NO by long-lived NO donors protected cells from apoptosis through mechanisms that involved the maintenance or an increase in the levels of the inhibitor of apoptosis proteins (IAPs) cIAP-1, cIAP-2, and xIAP and decreases in the accumulation of p53 and in the levels and targeting of Bax to the mitochondria. As a result of these changes, the activation of caspases 9 and 3 was notably delayed, expanding the time of viability of the macrophages. Moreover, inhibition of NO synthase 2 activity after 8 h of stimulation of RAW 264.7 cells with LPS and IFN-gamma accelerated apoptosis via an increase in the processing and activation of caspases. These data suggest that NO exerts an important role in the autoregulation of apoptosis in macrophages.     

Simultaneous measurement of intracellular nitric oxide and free calcium levels in chordate eggs demonstrates that nitric oxide has no role at fertilization

At fertilization in sea urchin, the free radical nitric oxide (NO) has recently been suggested to cause the intracellular Ca(2+) rise responsible for egg activation. The authors suggested that NO could be a universal activator of eggs and the present study was set up to test this hypothesis. Intracellular NO and Ca(2+) levels were monitored simultaneously in eggs of the mouse or the urochordate ascidian Ascidiella aspersa. Eggs were either fertilized or sperm extracts microinjected. Sperm-induced Ca(2+) rises were not associated with any global, or local, change in intracellular NO, although we were able to detect NO produced by the addition of a NO donor. Furthermore, the NO synthase inhibitor N(G)-nitro-L-arginine methyl ester had no effect on sperm-induced Ca(2+) release but did block completely ionomycin-induced NO synthase activation. Therefore, we suggest that the current data provide evidence that NO has no role in the fertilization of these two chordate eggs.     

Heme oxygenase-1 induction by endogenous nitric oxide: influence of intracellular glutathione

To investigate the influence of glutathione (GSH) on cellular effects of nitric oxide (NO) formation, human colon adenocarcinoma cells were transfected with a vector allowing controlled expression of inducible nitric oxide synthase (iNOS). Protein levels of oxidative stress-sensitive heme oxygenase-1 (HO-1) were analyzed in the presence or absence of GSH depletion using L-buthionine-[S,R]-sulfoximine and iNOS induction. While no effect was observed in the presence of iNOS activity alone, a synergistic effect on HO-1 expression was observed in the presence of iNOS expression and GSH depletion. This effect was prevented by addition of N-methyl-L-arginine. Therefore, targeting of endogenous NO may be modulated by intracellular GSH.     

Application of 4,5-diaminofluorescein to reliably measure nitric oxide released from endothelial cells<Emphasis Type="Italic">in vitro</Emphasis>

Here we describe in more depth the previously published application of the fluorescent probe 4,5-diaminofluorescein (DAF-2) in order to reliably measure low levels of nitric oxide (NO) as released from human endothelial cells invitro. The used approach is based on the following considerations a) use low concentrations of DAF-2 (0.1 μM) in order to reduce the contribution of DAF-2 auto-fluorescence to the measured total fluorescence, and b) subtract the DAF-2 auto-fluorescence from the measured total fluorescence. The advantage of this method is the reliable quantification of NO in a biological system in the nanomolar range once thoroughly validated. Here we focus in addition to the previous publication (Leikertet al.,FEBS Lett 2001, 506:131–134) on aspects of validation procedures as well as limitations and pitfalls of this method. Published: June 2, 2003     

TNF-alpha controls intracellular mycobacterial growth by both inducible nitric oxide synthase-dependent and inducible nitric oxide synthase-independent pathways

The role of TNF-alpha in the control of mycobacterial growth in murine macrophages was studied in vitro. Infection of macrophages from TNF-alpha gene disrupted (TNF-knockout (KO)) mice with recombinant Mycobacterium bovis bacillus Calmette Guérin (BCG) expressing the vector only (BCG-vector) resulted in logarithmic growth of the intracellular bacilli. Infection with BCG-secreting murine TNF-alpha (BCG-TNF) led to bacillary killing. Killing of BCG-TNF was associated with rapid accumulation of inducible NO synthase (iNOS) protein and the production of nitrite. The uncontrolled growth of BCG-vector was associated with low iNOS expression but no nitrite production. Thus, iNOS expression appears to be TNF-alpha independent but iNOS generation of NO requires TNF-alpha. In cultures of TNF-KO macrophages infected with BCG-TNF, inhibition of iNOS by aminoguanidine (AMG) abolished the killing of the bacilli. However, the growth of the organisms was still inhibited, suggesting an iNOS-independent TNF-alpha-mediated growth inhibition. To confirm this, macrophages from iNOS-KO mice were infected with either BCG-vector or BCG-TNF. As expected, no nitrite was detected in the culture medium. TNF-alpha was detected only when the cells were infected with BCG-TNF. In the iNOS-KO macrophages, the growth of BCG was inhibited only in the BCG-TNF infection. These results suggest that in the absence of iNOS activity, TNF-alpha stimulates macrophages to control the growth of intracellular BCG. Thus, there appears to be both a TNF-alpha-dependent-iNOS-dependent killing pathway as well as a TNF-alpha-dependent-iNOS-independent growth inhibitory pathway for the control of intracellular mycobacteria in murine macrophages.