105

Comparable Growth of Virulent and Avirulent Mycobacterium tuberculosis in Human Macrophages In Vitro Simon Paul, Pairote Laochumroonvorapong, and Gilla Kaplan

Laboratory of Cellular Physiology and Immunology. Rockefeller University, New York, New York

The relative virulence and avirulence of Mycobacterium tuberculosis strains H37Rv and H37Ra were previously defined using animal infection models. To investigate host species' specificity of mycobacterial virulence, growth of the 2 M. tuberculosis strains in human monocyte-derived macrophages in vitro was studied. Mycobacterial growth was evaluated by acid-fast staining, electron microscopy, and colony-forming units (cfu) assay. As expected, the 2 strains demonstrated significantly different growth rates in mouse macrophages in vitro (53 h for H37Rv, 370 h for H37Ra). In marked contrast, in human macrophages the average division times of the strains were nearly equal (80 h for H37Rv and 76 h for H37Ra by cfu measurement, and 96 h for H37Rv and 104 h for H37Ra by acid-fast staining). These findings indicate that observations of mycobacterial virulence in murine systems may not necessarily translate to the human system, in which different mechanisms to control mycobacterial growth may be expressed. The virulence of mycobacteria, that is, the ability of mycobacteria to establish infection in vivo, is often host speciesspecific [1]. For example, while humans and guinea pigs are highly susceptible to both human and bovine species of mycobacteria, other animals (cat, horse, cattle) are susceptible only to bovine mycobacteria; the mouse has a low to moderate susceptibility to a broader range of mycobacteria, including human, bovine, and avian species [1]. Virulence factors identified as important in animal models mayor may not be similarly important in the human host. It has been suggested that in the mouse, the key to virulence may be the prevention of induction of reactive nitrogen intermediates in the macrophage [2]. However, even in the mouse system, there is evidence for other mechanisms for the killing or control of growth of mycobacteria. A role for reactive oxygen intermediates has been postulated [3]. In contrast, some experiments have demonstrated that bacterial growth is not reduced by inhibitors of generation of either reactive nitrogen intermediates or reactive oxygen intermediates by macrophages, suggesting that yet other mechanisms may be involved [4, 5]. The growth-regulatory mechanisms operative in the human macrophage infected with Mycobacterium tuberculosis are as yet unknown. M. tuberculosis H37Rv, a virulent human isolate, was originally characterized in 1905 [6]. In 1920, investigators at the Trudeau Institute established a line of H37Rv which retained

Received 20 November 1995; revised 29 February 1996. Informed consent was obtained from all patients before blood drawing. Guidelines for animal experimentation of the Laboratory Animal Research Center at Rockefeller University were followed in all animal studies. Grant support: NIH (AI-OI07l and AI-226l6). Reprints or correspondence: Dr. Simon Paul, Laboratory of Cellular Physiology and Immunology, Rockefeller University, Box 176, 1230 York Ave., New York, NY 10021. The Journal oflnfectious Diseases 1996;174:105-12 © 1996 by The University of Chicago. All rights reserved. 0022-1899/96/7401-0014$01.00

virulence by growth in synthetic Proskauer-Beck medium. In 1922, after passage of the original H3 7Rv strain through a modified enriched culture medium, the avirulent strain H37Ra was isolated. More recently (1995), the in vivo virulence of the Trudeau Institute M. tuberculosis H3 7Rv has been confirmed in comparison with a current clinical isolate of M. tuberculosis [7]: H37Rv was demonstrated to have equal or greater virulence in the mouse. That the virulence ofH37Rv has been maintained without passage in vivo is important, as mycobacteria are very host species-specific, and animal passage could conceivably select for a strain with virulence for a specific animal and not for the human host. M. tuberculosis H37Rv and the avirulent descendant, H37Ra, are often used for comparison studies [8]. In the mouse, the 2 strains are distinguished by their growth capabilities: Survival of the avirulent strain is controlled both in the infected host and in vitro in mouse phagocytes, while the virulent strain causes a progressive lethal infection [8, 9]. It is essential to determine whether the virulence characteristics identified in the mouse are relevant to human disease. Therefore, a direct comparison of the virulence of these 2 mycobacterial strains in a human model of mycobacterial infection is critically important. In the present study, we have investigated the growth of these virulent and avirulent mycobacterial strains in human monocyte-derived macrophages in vitro to determine whether the strains differ in their virulence in human macrophages and whether the mechanisms controlling growth of mycobacteria in the human macrophage are similar to those observed in mouse cells.

Materials and Methods Mycobacteria. Mycobacterial strains were obtained from the Trudeau Mycobacterial Collection (Saranac Lake, NY). M. tuberculosis H37Rv (TMC #102) and H37Ra (TMC #201) were used. A 1.O-mL vial of each strain was grown for 7 days in 7H9 liquid

106

Paul et al.

medium (Difco, Detroit) containing 0.05% Tween 80 (Sigma, St. Louis) at 37°C with daily agitation, and aliquots were frozen at - 70°C. For each experiment, an aliquot was thawed and grown in 10.0 mL of 7H9 medium with 0.05% Tween 80 (Sigma) to mid-exponential growth phase (5 days). For infection of macrophages, the culture was then sonicated for 30 s in a bath sonicator, and the bacteria were washed twice in RPMI 1640 (Life Technologies GIBCO BRL, Gaithersburg, MD) with 2% human serum (pooled AB+; Biocell, Carson, CA) and 0.05% Tween 80. A singlecell suspension was obtained after removal of bacterial clumps with a 10-min 1OOO-g centrifugation. The top 2 mL of the supernatant was collected, and the bacteria were counted in a PetroffHauser Chamber and used immediately for infection of monocytes. About 67% of both mycobacterial strains counted visually were capable of giving rise to a colony on solid medium. Human monocytes. Peripheral blood mononuclear cells (PBMC) were isolated from human buffy coat blood preparations (New York Blood Center, New York) by centrifugation on ficollhypaque (Pharmacia, Uppsa1a, Sweden) as described [10]. Briefly, buffy coat preparations were diluted 1:1 with RPMI 1640 culture medium, layered on ficoll-hypaque, and centrifuged at 500 g for 30 min at 25°C. PBMC were then washed three times with RPMI 1640 with 1% human serum (R 1), including a low-speed 100-g 10-min spin to remove platelets. The cells were resuspended at 5 6/mL, X 10 and 0.5-mL aliquots were plated in wells of a 24-well Falcon tissue culture dish (Becton Dickinson Labware, Lincoln Park, NJ). After 1 h, nonadherent cells were removed with three washes of warm R 1, and the cells were reincubated in RPMI 1640 with 10% human serum (RIO) at 37°C with 5% CO 2 • Monocytes were incubated for up to 9 days without change of medium before infection. The number of adherent cells per well was determined by counting nuclei after lysis of the monolayer with 0.008% digitonin (Sigma) in PBS. Infection ofhuman monocytes. To determine the effect of differentiation of monocytes on the growth rate of mycobacteria, monocytes were infected with mycobacteria on day 0, 5, and 8 after isolation. To ensure that identical monocyte numbers and populations were infected, monocytes were not washed after infection and the medium was not changed. The MOl in this case was based on the number of PBMC used for adherence and the percentage of monocytes measured by flow cytometry. An estimated 2 X 105 monocytes per well were infected with an equal number of colony-forming units (cfu). Growth of mycobacteria in monocytes. To compare growth rates of different strains of mycobacteria, monocytes were infected on day 6 after isolation at an MOl of 1:1. After 6 h, the bacterial suspension was removed, the monolayer was washed three times with warm Rl to remove nonadherent bacteria, and the monolayer was then reincubated in RIO. The effectiveness of washing at each time point was assured by cytospin preparations (Shandon, Pittsburgh) of the medium overlying the macrophage monolayer (cytospins were done onto glass slides at 2000 rpm for 10 min). The cytospin preparations demonstrated that bacteria in the medium were associated with the few (often more heavily infected) cells that detached over the 6-day growth period. Antibiotics were not used at any point, and extracellular growth of mycobacteria in RIO culture medium was not seen (data not shown). Growth of mycobacteria in mouse peritoneal macrophages. Pathogen-free C57BL/6 mice were obtained from Jackson Laboratories (Bar Harbor, ME) and maintained under barrier conditions.

lID 1996; 174 (July)

Mice were injected with 1.0 mL of 5% casein (Sigma) in H20 intraperitoneally. After 3 days, the peritoneal cavity was washed out with 5 mL of RPMI 1640, the cells obtained were washed once with RPMI 1640, and 0.5 mL of 2 X 106 cells/mL in RPMI 1640 with 10% heat-inactivated fetal calf serum (Hyclone, Logan, UT) was plated per well of a 24-well tissue culture plate. Cultures were incubated for 18 h at 37°C with 5% CO 2 and infected in an identical manner from the same bacterial suspension as described for infection of human monocytes. Antibiotics were not used at any time. Acid-fast staining. The medium overlying the infected monocyte monolayer was aspirated and cytospun. The mono layers of infected monocytes (without washing) and the cytospin preparations were fixed for 30 min in 10% formalin (Fisher Scientific, Pittsburgh), dried, and stained with auramine/rhodamine (TB fluorescent stain, set T; Fisher) according to the manufacturer's instructions. After drying and mounting, bacteria were observed using a Nikon BG-12 excitation and OG-1 barrier filter for epifluorescence. Duplicate monolayers were fixed at each time point. To evaluate mycobacterial growth, all coverslips from one experiment were stained together, and counting was blinded for date and strain of mycobacteria. Electron microscopy. Monolayers of monocytes on coverslips were fixed in 2% glutaraldehyde/O.1 M sucrose/O.l M cacodylate (Poly sciences, Warrington, PA) buffer for 1 h at 4°C and washed twice in 0.1 M cacodylate buffer. The mono layers were postfixed in OS04 (Polysciences) for 6 h at 4°C, stained for 2 h with 0.25% uranyl acetate, dehydrated in increments with alcohol, and embedded in epoxy resin blocks. Sections were stained with uranyl acetate and lead citrate and examined with a Jeol JEM 100CX transmission electron microscope. Photographs were taken on Kodak electron imaging film. cfu assay. At each time point, triplicate wells were assayed for cfu. Culture medium was removed and the monolayer was lysed with PBS/0.05% Tween 80/0.008% digitonin. Medium and lysate were bath sonicated for 30 s to disperse bacilli. Serial dilutions ofthe bacterial suspensions were plated (6 replicates/dilution) on Middlebrook 7HIO agar plates. The cfu data are presented as the total cfu per well, the sum of that in the medium and the macrophage lysate.

Results MOL To measure intracellular growth of mycobacteria, it is essential to obtain an even infection with only a few bacilli per cell. Suspensions of bacteria were centrifuged, therefore, to remove clumps before infection of monocytes. Cord formation was minimized by the use of a single-cell suspension to ensure the accuracy of both cfu measurement and counting of acid-fast bacilli (AFB). When an MOl of 1: 1 was used for a 4-h period of infection with either H37Rv or H37Ra, 18% ± 2.6% and 14% ± 1.7% of human monocyte-derived macrophages were infected, respectively (table 1). An average of 2.5 (H37Rv) and 2.3 (H37Ra) bacilli per infected cell was observed (table 1). A slightly greater percentage of uptake of H37Rv was seen compared with H37Ra. However, the percentage of cells infected and the average number of bacilli per cell were not significantly different between strains. Neither M. tubercu-

JID 1996; 174 (July)

Virulent and Avirulent Mycobacterial Growth

Table 1. Comparison of colony-forming unit (cfu) assay and acidfast staining to determine bacilli at baseline in macrophages infected with M. tuberculosis H37Rv or H37Ra. Method, measurement Acid-fast staining AFB added/well (X 104 ) % of macrophages infected AFB/infected macrophage Calculated AFB/well (X 104)* % uptake of AFB cfu assay Human macrophages cfu added (X 104 ) cfu after washing (X 104) % uptake of cfu Mouse macrophages % uptake of cfu

H37Rv

H37Ra

0.4 2.6 0.2 1.2 12

lO±l.l 14 ± 1.7 2.3 ± 0.2 3.2 ± 1.0 32 ± 10

5.7 ± 1 3.0 ± 0.8 53 ± 24

7.0 ± 0.5 1.9 ± 0.3 27 ± 6

43

37

10 18 2.5 4.5 45

± ± ± ± ±

NOTE. After 6 h of infection, triplicate wells were lysed for cfu assay and duplicate coverslips were stained for acid-fast bacilli (AFB). All data for human macrophages are averages of results obtained from 3 separate experiments. Mouse macrophage data are from 1 experiment. For AFB enumeration, 100 infected cells were counted on each of 2 coverslips for each strain for each experiment. cfu assay was also done for initial inoculum in each experiment. Average viability of infecting inoculum for both strains was 67%. * Calculated from no. of AFB/macrophage, % of macrophages infected, and no. of macrophages/well. No. of macrophages/well (12 ± 2 X 104 ) was determined on day 6 of infection by lysis of the monolayer with PBS with 0.008% digitonin and counting of nuclei by phase-contrast microscopy.

losis H37Rv nor H37Ra replicated in cell-free RIO culture medium (data not shown). The numbers of bacilli in macrophage cultures at baseline were compared by two methods (table 1). The number of intracellular bacteria per culture calculated from light microscopic counting of AFB was 4.5 X 104 for H37Rv and 3.2 X 104 for H37Ra. By comparison, the number ofH37Rv and H37Ra cfu measured after lysis of the macrophage monolayer at this time point was 3.0 X 104 and 1.9 X 104 , respectively. The difference observed between AFB and cfu bacillary numbers is accounted for by the 67% viability of the bacilli in the infecting inoculum. Effect of monocyte differentiation on mycobacterial growth. Continuous culture of monocytes has been reported to lead to an increase in the ability of the cells to inhibit growth of M. tuberculosis [11]. We therefore compared the growth rates of the avirulent strain H37Ra in monocytes infected on days 0, 5, and 8 after introduction into culture (figure 1). The bacterial generation time (measured by the cfu assay) in monocytes infected on day 0 in culture was 44 h; in monocytes infected on day 5 in culture, the generation time was 103 h, and in monocytes infected on day 8, 189 h. Thus, the longer the monocytes were differentiated in culture before infection, the slower the growth rate of H37Ra. We chose a 6-day period of differentiation of mononuclear phagocytes for all subsequent experiments for two reasons. First, day 6 monocytes are intermediate in their control of H37Ra growth. As the monocytes at this point are neither maximally permissive nor maximally inhibitory, any factors

107

that enhance or reduce the growth rate of the mycobacteria should be detectable. Second, during the further 6 days of culture necessary to allow mycobacterial growth, the monocytes remained adherent and appeared healthy by both light and electron microscopic evaluation. Morphologic analysis of the infected monocytes. For each experiment, PBMC from a single donor were infected in parallel with either H37Rv or H37Ra. In all cultures examined by electron microscopy during the first 6 days of infection, the bacilli were found inside phagocytic vacuoles and not free in the cytoplasm (figure 2A-C). Most organisms were found either singly or dividing in tight vacuoles. Time points beyond 6 days were not included for measuring growth rates, as electron micrographs revealed that infected macrophages were dying and then being phagocytosed by neighboring cells (figure 2D). Phagocytosis of dying infected macrophages results in a change in the intracellular location of the mycobacteria from a single tight vacuole to a large debris-filled phagosome with multiple bacilli. Interestingly, there appeared to be a resequestration of mycobacteria from these large phagosomes into single vacuoles (figure 2D). Thus, enclosure in a tightly apposed vacuole appears to be the preferred location for M. tuberculosis within the human macrophage. Growth rates of H37Rv and H37Ra measured by cfu assay. In the growth curves measured by cfu assay, there appeared to be a 1- to 2-day lag period before exponential growth of both strains of mycobacteria (figure 3). The apparent lag in bacillary growth observed by cfu assay (which was not seen by acidfast staining) might occur if the complete physical separation of the bacteria required for the formation of 2 separate cfu did

400

,;)

0 ~

~

$

'c

:::> 0>

!

100

~!~-!--!

c

'E c

0 (5

y!!~

o

20

! 4

8

10

12

14

16

18

Day of Macrophage Culture

Figure 1. Effect of duration in culture of monocytes on growth of M. tuberculosis H37Ra. Monocytes on day 0 (_), day 5 (e), or day 8 (.&) after isolation and introduction into culture were infected with 2 X 105 cfu ofH37Ra. To allow infection of identical cell populations at each time point, mono layers were not washed before or after infection (adherence varies with duration in culture). Colony counts were determined for combined monolayer lysate and overlying media at each time point shown. Results are mean of triplicates for each point.

108

Paul et al.

J1D 1996; 174 (July)

Figure 2. Morphology ofM. tuberculosis-infected human rnacrophages by transmission electron microscopy. Macrophages were infected on day 6 at I; I ratio with strain H37Rv or H37Ra. A, Day 5 after infection with 1I37Ra; low numbers of bacilli per phagocyte are seen, and single or occasional double bacilli are localized in vacuoles. B, Day 4 after infection with H37Ra ; dividing bacilli (small arrows) are found within single vacuole. C, Day 6 after infection with H37Rv: larger numbers of bacilli per macrophage are seen in isolated vacuoles . D, Day 9 after infection with H37Ra: macrophages containing large debris-filled vacuoles with many bacilli. Bacilli from large vacuole appear to be segregating into smaller tight vacuoles (arrows). N, nucleus.

Virulent and Avirulent Mycobacterial Growth

JID 1996; 174 (July)

109

H37Rv

H37Ra

.-...

CFU

C'?

a

T"""

x .....(J) .~

c

100

100

10

10

:J en

c E ~

0 LL

>C

0 0

0

Q)

ctl .J::

o,

·r-

AFB

en

10

0

~

o

ctl

~ \J Q)

+-'

o

Q)

C

•

AFB 10

~;~

.~•

~

Q)

o,

CO

LL

«

0

2

3

4

5

6

o

2

3

4

5

6

Days Post Infection Figure 3. Gro wth of 1l37Rv and 1l37R a in hu man macro phages measu red by colony-forming unit assay (top) and by count ing of acid-fas t bacilli (AF I3; bottom) . Data are presented for 3 experime nts; matching symbols repre sent same experiment using cells from single blood donor. For cfu assay. each point is averag e of data from 3 sep arate wells and is sum of cfu in monolayer and culture medium overly ing monolayer. For AFB co unting, duplicate coverslips were evaluated by counting bacilli in 100 infec ted macroph ages per covers lip. Data are average number of bacilli per infected phagocyte (% of ce lls infected did not change over 6-day study ).

not take place as rapidly as the appearance of 2 separate bacilli visualized by acid-fast sta ining (figure 2B). In three experiments, the doubling rates measured for H37Rv were 55, 48, and 187 h (table 2). In the same experiments, the corresponding doubling rates measured for H37Ra were 62, 81, and 168 h. Thus, while there is variability in growth rates between the experiments (possibly reflecting differences in donor or isolation conditions; buffy coat preparation is carried out by the New York Blood Center), within each experimen t, the relative growth rates of the virulent and avirulent strains were similar. Growth rate measured by acid-fast staining. Acid-fast staining and microscopic counting of bac illi in infected monocytes revealed exponential growth in all experiment s for both H37Rv and H37Ra (figure 3). After 6 days of mycobacterial

growth, 90% of infected cell s still contained < 32 bacteria per cell (figure 4). Doubling tim es measured (calculated as described for cfu data) were 79 ::+:: 5 h for H37Rv and 77 ::+:: 10 h for H37Ra. Thus by acid -fast staining, the 2 strains of M. tuberculosis had identical growth rates. To determine whether all the bacilli in the cells were replicating at the same rate, the frequenc y distribution of mycobacteria within the monocytes was compared at days 0 and 6 (figure 4). The results show a unimodal frequency distribution on both days 0 and 6, indic ating that the entire population of bacilli was replicating at a similar rate. The frequ ency distributions were essentially identi cal for H3 7Rv and H37Ra, indicating that the infecti on and grow th of these strains we re also identical at the single-cell level and not j ust in the total culture.

Paul et al.

110

Table 2. Generation times (in hours) of M. tuberculosis H37Rv and H37Ra measured by colony-forming unit (cfu) assay and by acid-fast staining.

Type of macrophage, method Human Acid-fast staining

Mean cfu assay

Mean Mouse cfu assay Mean

Relative generation time, H37Rv:H37Ra

Experiment

H37Rv

H37Ra

I 2 3

83 69 86 80* 186 54 48 96*

80 58 87 76 168 63 81 104

1: 1.0 1:0.8 I: 1.0

73 33 53

550 190 370

1:7.5 1:5.6

I 2 3

1 2

1:0.9 1:1.2 I: 1.7

NOTE. Data reflect growth from day 0 to day 6 after infection. At each time point for each strain, 2 coverslips were fixed and stained for acid-fast bacilli (AFB), and average no. of bacilli/infected macrophage was counted in 200 infected cells. Generation times were calculated from best-fit straight line through logarithmic plot of the cfu counts vs. time and of AFB counts vs. time for 3 experiments. For mouse macrophages, generation times are based on cfu measured in triplicte on day 0 and day 6. * Not significant (P :3 .05) vs. H37Ra.

Mouse peritoneal macrophages. The relative virulence of the 2 mycobacterial strains in mouse peritoneal macrophages was also evaluated in vitro. The average generation time for virulent H37Rv was 53 h in the mouse macrophage (table 2). In contrast, the generation time for H37Ra in the mouse cells was 370 h; this is significantly greater than the generation time observed for H37Ra in any of the experiments with human macrophages. Thus, the difference in virulence between H37Rv and H37Ra was apparent only when the mycobacteria were cultured in mouse macrophages. This result in mouse macrophages confirms the virulent and avirulent nature of the actual bacterial inocula used in the human experiments. Furthermore, these results demonstrate that the equal growth rates seen in human macrophages are not an artifact of the mycobacterial culture methods before or after infection but are a property of the host (human) cell.

Discussion Our present understanding of mycobacterial infection relies heavily on the study of infection in the mouse model. The virulence of mycobacteria, however, is often host speciesspecific [1], and it is important, therefore, to compare results obtained in animals with results obtained in a human system. Here we report a striking inability of human macrophages, compared with mouse macrophages, to control the in vitro growth of an avirulent strain of M. tuberculosis. This unexpected difference between the mouse and human infection mod-

JID 1996; 174 (July)

els underscores the importance of studying mycobacterial infection in a human model. To investigate the growth of mycobacteria in human monocyte-derived macrophages in vitro required the optimization of several factors in the culture system, including differentiation of monocytes, MOl, and duration of infection. The importance of these factors was often made clear only by electron microscopy, which demonstrated effects that would be overlooked in cumulative data such as cfu and even light microscopic acid-fast staining. First, electron microscopy facilitated the determination of the time point at which infected monocytes began to die (day 9 after infection, figure 2D). This cell death resulted in phagocytosis of the dying macrophage by neighboring healthy macrophages and would not be detected by methods such as trypan blue exclusion or by light microscopic acid-fast staining. It is not known whether growth rates of the intracellular organisms might be modified by this process of host cell death and rephagocytosis. Therefore, we terminated the experiments before host cell death began to occur in order to evaluate mycobacterial growth within viable macrophages. It is interesting to note that the phagocytosed cell debris, which includes intact mycobacteria, appears to again segregate into different compartments (figure 2D). The mycobacteria were separated from the large phagocytic vacuole into tighter, discrete extensions which might pinch off as separate vacuoles containing a single mycobacterium. Similar morphology has been observed in vivo in mice infected via aerosol with virulent M tuberculosis (unpublished data). A second factor evaluated before the comparison of growth rates was the evenness of infection at the initial and subsequent time points. Analysis of the frequency distribution of bacilli per infected host cell demonstrated a unimodal distribution throughout the 6-day time course (figure 4). This unimodal distribution indicated that mycobacteria in all cells were replicating at similar rates regardless of the total number of organisms per phagocyte. In contrast, in preliminary studies with avirulent Mycobacterium smegmatis, a bimodal distribution developed, with mycobacterial growth occurring more rapidly in heavily infected cells (unpublished data). Finally, a third factor investigated to establish culture conditions to be used for measuring intracellular growth rates of mycobacteria was the differentiation of monocytes before infection. As has been shown previously with H37Rv [11], the period of monocyte differentiation can affect the subsequent growth rates measured. We measured the growth of H37Ra in different aged monocyte-derived macrophages (figure 1). For subsequent measuring of growth rates, we chose a period of differentiation in which an intermediate rate of mycobacterial growth was seen, so as to allow either an increase or decrease in growth to be detectable. On the basis of the above results, the following system of culture and infection of monocytes was used for the subsequent experiments. Six-day-old monocytes were infected with a single-cell bacterial suspension at a multiplicity of 1:1, resulting

Virulent and Avirulent Mycobacterial Growth

JID 1996; 174 (July)

200

r--

H37Rv

150

iO'

0

-

100

-

Q)

~

H37Ra

r--

o (/)

III

n..

50

..c

0..

o

U

n

0

ro

~

o

~

200

Q)

.D

E

:::J

150

Z