Vol. 59, No. 4
INFECTION AND IMMUNITY, Apr. 1991, p. 1378-1386 0019-9567/91/041378-09$02.00/0 Copyright C) 1991, American Society for Microbiology
Ammonium Decreases Human Polymorphonuclear Leukocyte Cytoskeletal Actin Department
B. BRUNKHORST* AND R. NIEDERMAN Biology, Forsyth Research Institute, 140 The Fenway, Boston, Massachusetts 02115
Received 21 August 1990/Accepted 17 January 1991
Ammonium, a weak base produced as a metabolic by-product of urea metabolism by bacterial pathogens, inhibits a variety of motile polymorphonuclear leukocyte (PMN) functions. It was initially assumed that the mechanism of leukocyte inhibition was due to cytoplasmic alkalinization. However, while it is clear that ammonium can effect cytoplasmic alkalinization, current data indicate that alterations in chemotaxis, degranulation, and receptor recycling occur independently of cytoplasmic alkalinization. Since these are motility-related events, we examined the possibility that alterations in cytoskeletal actin may account for the effects of ammonium on PMN function. The results indicate that ammonium can inhibit degranulation, decrease cytoskeletal actin, and increase actin depolymerization rates. These findings are supported by five lines of evidence. First, formylmethionyl-leucyl-phenylalanine (fMLP)-induced elastase release was inhibited 3% in the presence of ammonium, and ammonium by itself did not stimulate elastase release. by 85% Second, ammonium treatment of resting PMNs caused a rapid 38% 6% decrease in cytoskeletal actin. Third, ammonium treatment accelerated the fMLP-induced depolymerization phase of the cytoskeletal actin transient by 150% ± 12%. Fourth, in resting PMNs treated with cytochalasin B or D, ammonium induced a 21% 4% and a 25% 5% decrease in cytoskeletal actin, respectively. Conversely, ammonium did not affect the ability of the cytochalasins to inhibit an fMLP-induced cytoskeletal actin transient. Fifth, pertussis toxin treatment of neutrophils did not affect the ammonium-stimulated decrease in cytoskeletal actin. These results suggest that ammonium can inhibit neutrophil function by altering cytoskeletal actin and therefore provide new information regarding potential pathogenic mechanisms for bacterial pathogens. ±
some-lysosome fusion, while others inhibit it (14, 15, 20). Findings such as this make it unlikely that ammonium's inhibitory activity operates strictly through interference with pH homeostasis. However, the mechanism(s) of ammonium's inhibitory activity on polymorphonuclear leukocytes (PMNs) is unknown. Because ammonium affects all forms of motility and because alterations in the actin cytoskeleton affect cell motility, we examined the hypothesis that ammonium's central effect is on the actin cytoskeleton.
Weak bases have been used for many years to experimentally alter the cytoplasmic pH and function of eukaryotic cells (9). Similarly, weak bases in general and ammonium in particular can alter the cytoplasmic pH of leukocytes (46) and can alter leukocyte function (17). The clinical relevance of this occurrence is particularly important in periodontitis. Urea and ureolytic bacteria are significant components of gingival crevicular fluid and pathogenic plaque, respectively (19). These bacteria can utilize urea to generate ammonium concentrations as high as 50 mM (32, 33, 50). Thus, in both cross-sectional and longitudinal studies, periodontal inflammation varies directly with the crevicular pH (4, 5, 7). The effects of weak base-mediated leukocyte alterations occur at three levels: (i) alteration of receptor-mediated activation (35, 61), including receptor-ligand binding (13) and receptor recycling (40); (ii) alteration of cytoplasmic regulation mediated by calcium (17, 56), cytoplasmic pH (55), and cytoskeletal actin (28, 34); and (iii) alteration of leukocyte function, including chemotaxis (40, 47), phagocytosis (11, 36, 54), degranulation (8, 11, 27), and oxygen-dependent and -independent killing mechanisms (11, 46, 54-56). It is important to recognize that the effects of weak bases are more complex than simple cytoplasmic alkalinization. For example, two weak bases, ammonium and chloroquine, both alkalinize the cytoplasm and inhibit oxidative metabolism (55) but have opposite effects on bacterial adhesion to leukocytes (54), phagosome-lysosome fusion (20), receptormediated endocytosis (35), and intracellular transport of myeloperoxidase (53). Similar divergent effects have been observed for other weak bases; some can accelerate phago-
MATERIALS AND METHODS
Reagents. Nitrobenzo-oxadiazole phallacidin (NBD-phallacidin) was obtained from Molecular Probes, Inc. (Eugene, Oreg.). Cytochalasin B (CB), dimethyl sulfoxide (DMSO), ethylene glycol-bis(,-aminoethyl ether)-N-N-N' -N' -tetraacetic acid (EGTA), formylmethionyl-leucyl-phenylalanine (fMLP), N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid (HEPES), MgCl2, piperazine-N,N'-bis(2-ethanesulfonic acid) (PIPES), and Triton X-165 were from Sigma Chemical Co. (St. Louis, Mo.). Ficoll-Hypaque was from Flow Laboratories (McLean, Va.). Hanks balanced salt solution (HBSS) with and without Ca2+ and Mg2+ was from GIBCO-BRL (New York). Methyl-succinyl-alanyl-prolyl-valyl-methylcoumnarin amide (MCA) was obtained from Peninsula Laboratories (Belmont, Calif.). NH4Cl was from Fisher Scientific (Pittsburgh, Pa.). Calibrated fluorescent polystyrene beads were from Flow Cytometry Standards Corp. (Research Triangle Park, N.C.). Pertussis toxin (PT) was from List Biologicals (Campbell, Calif.). Human neutrophil elastase was from Calbiochem (San Diego, Calif.). CB and MCA were dissolved in DMSO and diluted in HBSS plus 10 mM
Corresponding author. 1378
VOL. 59, 1991
HEPES (pH 7.4) (HBSS + HEPES). fMLP was dissolved in DMSO and diluted in HBSS + HEPES. NH4Cl and human neutrophil elastase were diluted in HBSS + HEPES. The addition of stimuli diluted in DMSO never exceeded 0.5% of the cell suspension volume. Isolation of neutrophils. Whole blood was drawn from healthy donors by venipuncture into Vacutainers containing 1,500 U of heparin per ml. Leukocytes were separated by centrifugation through a discontinuous Ficoll-Hypaque gradient at 700 x g for 40 min at 25°C (26). The granulocyte layer was removed, 20 ml of Ca2+- and Mg2+-free (CMF) HBSS + HEPES was added, and the cell suspension was centrifuged at 200 x g for 10 min at 4°C. Contaminating erythrocytes were removed by addition of 12 ml of doubledistilled H20 to the pellet for 40 s, and isotonicity was restored with 3 ml of 5x CMF HBSS + HEPES. This suspension was centrifuged at 200 x g for 10 min at 4°C. The cell pellets were resuspended in HBSS + HEPES. PMNs isolated in this manner were 95% pure as determined by visual examination and 95% viable as determined by trypan blue exclusion. Measurement of elastase release. Elastase activity was measured by the increase in fluorescence attributed to the elastase-specific cleavage product of the MCA substrate (49). Cell suspensions of 106 PMN per ml in HBSS + HEPES were warmed to 37°C for 5 min with 5 jig of CB or 2 FLM CD in the presence or absence of 30 mM NH4Cl while being stirred in a SLM-5000 spectrofluorimeter (SLM Instruments, Urbana, Ill.). We selected 30 mM ammonium on the basis of concentrations released by pathogenic bacteria in vitro and in vivo (21, 27, 57). In addition, ammonium neither inhibited the cleavage of the elastase-specific substrate by purified human neutrophil elastase nor affected the fluorescence of the elastase-specific fluorophore (data not shown). Initially, CB was used to allow for the comparison of our results with previous studies. However, since CB has a pleiotropic effect on PMNs, we repeated all studies with CD. MCA (5 ,uM final concentration) was added to the stirred cell suspension at 4 min. Fluorescence of PMN suspensions plus MCA was measured by continuously measuring 460-nm emissions following excitation at 380 nm. After completion of the 5-min preincubation period, some cell suspensions were challenged with 10-6 or 108 M fMLP. A standard curve was created to correlate the change per minute in MCA fluorescence to micrograms of HNE. This was done by determining the MCA fluorescence in stirred suspensions of MCA plus various concentrations of HNE. The addition of CB or CD, fMLP, or NH4Cl did not affect the fluorescence of the cleaved substrate or the ability of HNE to cleave the substrate. NBD-phallacidin staining of neutrophils. Actin polymerization was measured by using the NBD-phallacidin method essentially as previously described (22). Neutrophils (106/ml) were incubated at 37°C in HBSS + HEPES for 5 min in the presence of buffer, 30 mM NH4Cl, 5 ,ug of CB per ml, and 2 F.M CD or a mixture of 5 pLg of CB per ml or 2 puM CD and 30 mM NH4Cl before addition of 10-6 or 108 M fMLP. At timed intervals under the conditions described above, 200 p.1 of the cell suspension was removed and simultaneously fixed and extracted at 4°C in a solution containing 2.5% paraformaldehyde, 2.5% Triton X-165, and PHEM buffer (300 mM PIPES-124 mM HEPES-50 mM EGTA-10 mM MgCl2 [pH 6.9]) (42). This and all subsequent preparative procedures were carried out at 4°C. Each cell sample was incubated in the extraction-fixation buffer for 30 min, centrifuged at 200 x g for 10 min, and finally stained with NBD-phallacidin in
AMMONIUM DECREASES PMN F-ACTIN
paraformaldehyde-PHEM buffer (0.2 p.M final NBD concentration) for at least 30 min at 4°C. The cell samples were washed once in paraformaldehyde-PHEM buffer and finally resuspended in this same buffer for cytometric analysis. Once stained, the cells were kept covered by foil to prevent photobleaching. Cytometric analysis of NBD-phallacidin-stained cells. Fluorescence was measured on a Coulter EPICS C flow cytometer equipped with a Coulter MDADS data acquisition system. Day to day fluorescence data were standardized with calibrated fluorescent polystyrene beads. These standard beads were also used to create a standard curve that correlated log fluorescence to linear values (43). PMNs were identified by measurement of forward and right-angle light scatter. NBD-stained neutrophils were excited at 488 nm, and emission was determined at 510 nm. The mean log fluorescence of 5,000 cells was determined and converted into linear fluorescence values by using the standard curve. Relative cytoskeletal F-actin was determined by dividing the linear fluorescence value (LFV) at a given time by the initial linear fluorescence value (22, 28) (see Fig. 2 to 6). PT treatment of neutrophils. A stock solution of 100 ng of PT per ml was made in phosphate buffer consisting of 0.11 M NaPO4 and 0.5 M NaCl (pH 7.0). PMNs (106/ml) in CMF HBSS + HEPES were incubated with 500 ng of PT per ml for 120 min while being rocked at 37°C. Sham treatment of cells was performed under the same conditions with buffer only. Cells were washed free of PT with a 5 x volume excess of CMF HBSS + HEPES and centrifuged for 10 min at 200 x g at 4°C. The cell pellets were diluted to 106 PMN per ml in HBSS + HEPES, and NBD staining for cytoskeletal F-actin was performed as described above. All results are expressed as the percentage of sham cell responses. We determined that maximal inhibition of fMLP-stimulated responses occurred when cells were incubated with 500 ng of PT per ml. Therefore, 500 ng of PT per ml was used for the experiments described here. Statistics. All data are reported as means + standard errors of the mean (SEM). Statistical significance between the means was assessed by the Student's t test. RESULTS Effects of ammonium on degranulation. Klempner and Styrt (27) initially demonstrated that ammonium chloride inhibited fMLP-stimulated degranulation, as determined by lysozyme release in CB-treated neutrophils. We therefore initially examined the effects of ammonium on fMLP-stimulated degranulation. However, we made two alterations in the experimental approach. First, we measured elastase release, and second, we used both CB and CD. fMLP induced an immediate release of elastase at a constant rate in CB-treated cells (Fig. 1). The rate of elastase release was dependent on the fMLP concentration; 10-6 M fMLP induced 3.26 + 0.18 p.g/min/ml, 10-7 M fMLP induced 2.7 ± 0.07 p.g/min/ml, and 108 M fMLP induced 0.73 ± 0.01 p,g/min/ml. Ammonium pretreatment inhibited elastase release, and the inhibition varied inversely with the fMLP dose (Fig. 1); 10-6 M fMLP was inhibited by 38% ± 4%, 10-7 M fMLP was inhibited by 46% + 3%, and 10-8 M fMLP was inhibited by 85% ± 3%. Ammonium by itself did not elicit elastase release. The replacement of CB by CD had no demonstrable effects on the results. For example, in CDtreated cells ammonium caused a 92% ± 8% inhibition in 108 M fMLP-stimulated elastase release. Taken together,
BRUNKHORST AND NIEDERMAN
%foa a a 0 i S
c u a
. . . . . . . . . . . . . .
2 3 4 Time (min)
I -300-200-100 0
100 200 300
FIG. 1. Effects of NH4C1 on elastase release. Time course of elastase release by PMN incubated for 5 min with 5 p.g of CB per ml and stimulated with 30 mM NH4C1 (--), 10-8 M fMLP (...), or 10-8 M fMLP following preincubation with CB plus 30 mM NH4C1 (-) as described in Materials and Methods. Stimuli were added at time zero. Data represent averaged data of three separate experiments.
FIG. 2. Effects of NH4C1 on cytoskeletal F-actin. Time course of relative F-actin of PMN incubated for 5 min in the presence (A and O) or absence (A) of 30 mM NH4C1. fMLP (106 M) was added at time zero (U and O). At indicated times, the cells were fixed, lysed, stained, and analyzed on the flow cytometer for relative F-actin content. Data represent the mean ± SEM of more than three separate experiments.
these results confirm and expand the results of Klempner and Styrt (27). Effects of ammonium on the F-actin cytoskeleton. Previous studies (27, 55) suggested that ammonium's primary mechanism of action is lysosome alkalinization, which prevents lysosome-membrane fusion. However, recent work from this and other laboratories indicate that ammonium may affect the receptor Kd, fMLP receptor recycling, and the actin cytoskeleton (13, 28, 40). We therefore examined the effects of ammonium on the actin cytoskeleton more closely. Figure 2 displays the effects of fMLP and ammonium on the actin cytoskeleton. fMLP induced a transient increase and then a decrease in cytoskeletal F-actin, as initially described by Howard and coworkers (22, 23). Cytoskeletal F-actin reached a maximum within 10 s, depolymerized slightly, and repolymerized, exhibiting a second maximum at 80 s before finally depolymerizing toward baseline. In contrast to fMLP, ammonium alone caused an immediate decrease in cytoskeletal F-actin, which reached its lowest level within 10 s and remained below baseline throughout the 5-min assay period (Fig. 2). When fMLP was added to ammonium-treated cells, there was an immediate increase in cytoskeletal F-actin. Statistical analysis indicated that the fMLP-stimulated maximal increase in cytoskeletal F-actin was not significantly different in ammonium-treated and untreated cells (Table 1). Furthermore, the ammonium pretreatment did not affect the fMLPinduced actin polymerization rate or final cytoskeletal F-actin level (Table 2). However, ammonium pretreatment did affect the resting actin level as well as actin depolymerization induced by fMLP. Actin depolymerization was affected in three ways (Table 2). First, depolymerization started significantly earlier in ammonium-treated cells (10 s after stimulation) (Fig. 2). Second, ammonium doubled the fMLPstimulated depolymerization rate (Table 2). Third, ammo-
nium abrogated the second peak in relative cytoskeletal F-actin induced by 10-6 M fMLP (Fig. 2; Table 1). Effects of ammonium and cytochalasin on the F-actin cytoskeleton. Cytochalasin is used to maximize chemoattractant-stimulated degranulation and can bind to the barbed end of actin filaments to alter actin polymerization (12). Since cytochalasin was used in previous and current degranulation studies, we examined the effects of ammonium and cytochalasin on the F-actin cytoskeleton. Our initial degranulation studies used CB in order to compare our results with previous studies (27). However, because CB has pleiotropic effects on the neutrophil, all studies were also done with CD, which specifically affects cytoskeletal actin (12). As in the degranulation studies, the effects of ammonium were similar in the presence of either CB or CD. Figure 3 displays the effects of CB and ammonium on PMNs. Contrary to results of prior studies (12, 58, 59), CB and CD alone caused a significant increase in cytoskeletal F-actin. From a baseline of 1.0, CB and CD stimulated a relative increase in cytoskeletal actin to 1.6 + 0.11 and 1.66 ± 0.14, respectively. In contrast to cytochalasin alone, when cytochalasin and ammonium were added simultaneously, there was an immediate decrease in cytoskeletal F-actin. Ammonium plus CB or CD resulted in an immediate decrease to 0.79 + 0.03 or 0.84 + 0.03 relative F-actin, respectively. This decrease in relative F-actin was less than that induced by ammonium alone (0.62 + 0.06). Furthermore, unlike ammonium alone, in the presence of ammonium and cytochalasin the cytoskeletal F-actin increased toward baseline by 5 min (Fig. 3) after the initial decline, and the initial depolymerization rate appeared to be slower than that induced by ammonium alone (Fig. 3; see Table 4). However, these two rates were not significantly different. Thus, while ammonium caused a
VOL. 59, 1991
AMMONIUM DECREASES PMN F-ACTIN
TABLE 1. Effects of NH4Cl on fMLP-induced actin cytoskeletal response Relative F-actin"
30 mM NH4CJ 10-6 M fMLP 10-8 M fMLP 30 mM NH4C1, then 10-6 M fMLP 30 mM NH4CI, then 10-8 M fMLP
% Inhibition of maximum change'
End effect (5 min)
1.00 1.00 1.00 0.76 ± 0.1 0.81 ± 0.1
(0.62 ± 0.06) 2.51 ± 0.2
-0.38 ± 0.06 1.51 ± 0.2
2.32 ± 0.26 2.24 ± 0.19 1.89 ± 0.15
1.32 ± 0.3 1.48 ± 0.19 1.08 ± 0.21
NA NA NA 2 ± 7 19 ± 10
0.76 + 0.1 1.49 + 0.2 1.58 ± 0.18 1.58 ± 0.2 1.38 ± 0.2
a All values given as mean ± SEM. b Maximum change is peak relative F-actin - baseline F-actin. P - 0.05 for all values. C Values are figured by the following equation: 100 - [(maximum change in relative F-actin of ammonium-treated cells/maximum change in relative F-actin of control cells) x 100]. NA, Not applicable.
decrease in cytoskeletal F-actin, cytochalasin appeared to reverse this effect. Effects of ammonium and cytochalasin on the fMLP-stimulated F-actin cytoskeleton. As indicated previously, fMLP induced a substantial actin transient. In cells treated with CB or CD this fMLP-induced actin transient was inhibited by 80% ± 6% and 85% ± 2%, respectively (Table 3). Other workers have noted similar inhibition of fMLP-induced actin transients by CB treatment (12, 59). Our degranulation experiments utilized ammonium and cytochalasin simultaneously. Therefore, we examined their combined effects on fMLP-stimulated cytoskeletal F-actin transients. Again, the effects of either CB or CD were similar. Ammonium and CB, added simultaneously, inhibited the fMLP-stimulated actin transient to the same extent as CB alone (74% ± 6% and 80% ± 6%, respectively) (Tables 3 and 4). The same was true in cells incubated with CD. Ammonium and CD, added simultaneously, inhibited the fMLPstimulated actin transient to the same extent as CD alone (85% ± 3%, and 85% ± 4% respectively). However, ammonium, even in the presence of cytochalasins, was able to increase the residual fMLP-stimulated actin depolymerization rate by sixfold and to decrease the final cytoskeletal F-actin level by one-half (Tables 3 and 4). Taken together, the results indicate that ammonium can induce a lower resting level in cytoskeletal F-actin and an increase in actin depolymerization rate. This effect occurs in cells treated with ammonium alone, ammonium plus CB or CD, ammonium plus fMLP, and ammonium plus CB or CD and then fMLP. Furthermore, the presence of ammonium
did not alter the inhibitory effect of CB or CD on the fMLP-induced actin polymerization transient. This last point suggests that ammonium and cytochalasins have entirely different mechanisms of action (see Discussion). Effects of cytochalasin and then ammonium on the F-actin cytoskeleton. In addition to increasing fMLP-induced degranulation, cytochalasin is thought to bind to the barbed end of the actin filament, to prevent actin monomer addition, and thus to prevent actin polymerization (6). Paradoxically, data presented here suggest that both CB and CD can stimulate an increase in cytoskeletal F-actin. In contrast, ammonium stimulates a decrease in cytoskeletal F-actin. The data therefore suggest that ammonium and cytochalasin have different effects on cytoskeletal actin. We therefore determined whether the ammonium-stimulated depolymerization of cytoskeletal actin would be affected by cytochalasin pretreatment. As previously noted, CB or CD caused an increase in cytoskeletal F-actin (Fig. 3). Addition of ammonium to CB- or CD-treated cells caused an immediate depolymerization of the actin cytoskeleton (Fig. 4). The 2.0
TABLE 2. Effects of NH4Cl on fMLP-induced actin cytoskeletal response
30 i a
Actin response rate" Stimulus
Depolymerization Peak 1
2.0 ± 0.45 0.3 ± 0.3 1.32 ± 0.43
30 mM NH4CI 10-6 M fMLP 10-8 M fMLP 30 mM NH4Cl + 10-6
NA 8.5 ± 1.6 5.2 ± 2.0 7.8 ± 0.9
0.62 ± 0.08
NA 0.24 + 0.06 NA NA
M fMLP 30 mM NH4Cl + 10-8 M fMLP
5.3 ± 0.9
1.12 ± 0.54
aAll values are given as the mean ± SEM. Polymerization and depolymerization rates are calculated as the change in relative F-actin per minute. NA, Not applicable.
100 200 300 -300 -200 -100 0 Tim(suec) FIG. 3. Effects of CB and NH4Cl on cytoskeletal F-actin. Time course of relative F-actin of PMN exposed to 5 ,ug of CB per ml (M) and then stimulated with 10-6 M fMLP (V) or a mixture of 5 ,ug of CB per ml plus 30 mm NH4Cl (O) and then stimulated with 10-6 M fMLP (V). Data represent the mean ± SEM of more than three separate experiments.
BRUNKHORST AND NIEDERMAN
TABLE 3. Effects of CB and NH4Cl on fMLP-induced actin cytoskeletal response Relative F-actin' Maximum changeb
fMLP CB CB + fMLP NH4Cl CB + NH4C1 CB + NH4Cl, then fMLP CB, then NH4C1
1.00 1.00 1.54 ± 0.16 1.00 1.00 0.89 ± 0.12
1.42 ± 0.16
2.51 1.61 1.84 0.62 0.79 1.28 1.03
1.51 0.61 0.30 -0.38 -0.21 0.39 -0.39
± 0.03" ± 0.06"
+ 0.03"'A' ± 0.0.31dh ± 0.14
± 0.12e ± 0.06d ± 0.09 ± 0.07
% Inhibition of maximum changeC NA NA 80 ± 6' NA NA 74 ± 6" NA
End effect (5 min)
1.49 1.54 1.75 0.76 0.89 0.96 0.89
± ± ± ± ± ± ±
0.2 0.16 0.24 0.07
" All values are given as the mean + SEM. b Maximum change is peak relative F-actin - baseline relative F-actin. ' Values are figured by the following equation: 100 - [(maximum change in relative F-actin/maximum change in relative F-actin induced by fMLP) x 100]. NA, Not applicable. d p 0.05 compared with baseline " P C 0.05 compared with fMLP. f P < 0.01 compared with maximum change in relative F-actin by fMLP. g P - 0.05 compared with CB. h p