Growth Rate Paradox of Salmonella typhimurium within

Transcription

1 JOURNAL OF BACTERIOLOGY, June 1993, p /93/ $02.00/0 Copyright X 1993, American Society for Microbiology Vol. 175, No. 12 Growth Rate Paradox of Salmonella typhimurium within Host Macrophages KELLY ZAIGER ABSHIREt AND FREDERICK C. NEIDHARDT* Department of Microbiology and Immunology, The University of Michigan Medical School, Ann Arbor, Michigan Received 10 December 1992/Accepted 13 April 1993 The growth rate of Salmonella typhimurium U937 within host macrophages was estimated by two independent methods. The extent to which ribosomal protein L12 is acetylated to produce ribosomal protein L7 changes markedly with the growth rate. By this measure, the intracellular bacteria appeared to be growing rapidly. Measurements of viable bacteria, however, indicated that the bacteria were growing slowly. A solution of this apparent growth rate paradox was sought by treating U937 cells infected with S. typhimurium X3306 with ampicillin or chloramphenicol to help determine the number of bacteria that were actively growing and dividing in the intracellular condition. Use of these antibiotics showed that by 2 h after invasion, the intracellular bacteria consisted of at least two populations, one static and the other rapidly dividing. This finding implies that previously described changes in the gene expression of S. typhimurium are important for the survival and/or multiplication of the bacteria within the macrophage. Salmonella typhimurium is a facultative intracellular bacterium with the ability to survive and multiply within macrophages (4, 5). Within macrophages, S. typhimurium cells are generally found in membrane-bound structures called phagolysosomes, in which the bacteria encounter both oxygen-dependent and oxygen-independent bactericidal processes (6, 9). It has long been believed that S. typhimurium survives the microbicidal mechanisms of phagocytes by mounting adaptive responses that enable it to become an intracellular parasite of macrophages (4). We and others (1, 2) have previously shown that S. typhimurium alters its pattern of gene expression in response to conditions found within host macrophages. These changes are reflected by the induction and repression of protein spots (approximately 40 and 100, respectively) resolved on two-dimensional polyacrylamide gels. Although bacterial gene expression is now known to be altered within macrophage host cells relative to extracellular environments, very little is known about bacterial growth or growth rate in the intracellular environment. On the other hand, much is known about bacterial growth rate in extracellular environments. Within a certain range, the levels of various translation factors and ribosomal proteins within a bacterium have been shown to vary directly with the growth rate (11). This group of proteins includes several that are easily measured by two-dimensional gel electrophoresis, especially the ribosomal proteins L7/L12, S6, and S1 and the elongation factors Tu (EF-Tu), G (EF-G), and Ts (EF-Ts) (11). Ribosomal protein L7/L12 is especially useful in estimating growth rate. Protein L7 is the N-terminal acetylated form of protein L12. The sum of the two forms varies with growth rate, behavior which is similar to that of all other ribosomal proteins, but the rate of acetylation of L12 to form L7 does not keep pace with the increasing growth rate, leading to larger changes in the level of protein L12 than in those of other translation-related proteins. * Corresponding author. t Present address: Department of Infectious Diseases, Parke- Davis Research Division, Warner-Lambert Company, Ann Arbor, MI In experiments in which S. typhimurium was allowed to invade and grow within macrophage-like U937 cells, bacterial growth rate within the macrophages was estimated by plating for viable bacteria. At the same time, an independent estimate was provided by examining the proteins expressed by S. typhimunum while they were inside these macrophages. The two methods yielded apparently contradictory results: viable counts increased slowly, but the pattern of protein synthesis was characteristic of more rapidly growing bacteria. Evidence that resolves this paradox is presented here. MATERLiLS AND METHODS Bacterial strains and culture medium. The S. typhimunum SR-11 strain, X3306 (gyra1816), carrying the 100-kb virulence plasmid pstsr100 was chosen because it is highly virulent in mice (8). It was obtained from R. Curtiss III, Washington University, St. Louis, Mo. Several formulations of liquid MOPS (morpholinepropanesulfonic acid) medium were used for growing bacterial cultures (10). Glucose minimal MOPS medium contained glucose at 0.4% (excess) or 0.04% (limiting), and glucose rich MOPS medium without methionine contained 0.4% glucose, purine, and pyrimidine bases, five vitamins, and all amino acids except methionine (13). Sodium pyruvate (0.4%) and sodium acetate (0.4%) were used as alternative carbon sources in minimal MOPS medium. Cell cultures. For the maintenance of human macrophagelike U937 cell cultures, Dulbecco's modified Eagle medium (DMEM) containing 10% fetal bovine serum (FBS), 25 mm HEPES (N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid) (ph 7), 100 U of penicillin per ml, and 100 pg of streptomycin per ml was used. (DMEM and FBS were purchased from GIBCO BRL, Grand Island, N.Y.; HEPES and antibiotics were purchased from Sigma Chemical Co., St. Louis, Mo.). During infection experiments, DMEM-10% FBS-25 mm HEPES (ph 7) or DMEM without methionine- 10% FBS-25 mm HEPES (ph 7) was used. The addition of HEPES maintained the correct ph in cell cultures in tightly closed flasks, preventing radioactive contamination of the

2 VOL. 175, 1993 SALMONELLA TYPHIMURIUM GROWTH WITHIN MACROPHAGES 3745 incubator. HEPES also eliminated the need for a source of CO2 to maintain a constant ph. Culture and preparation of host macrophages. Macrophage-like U937 cells (12) were the host cells used for infection experiments. These cells were obtained from the laboratory of B. Eisenstein, University of Michigan, Ann Arbor. A stock culture of these cells was maintained in monocyte-like, nonadherent form in tissue culture medium (described above) and was passaged every 3 to 4 days for up to 12 to 13 passages. Before they were used in infection experiments, it was necessary to induce the monocyte-like U937 cells to differentiate into adherent, macrophage-like cells. Cells from the stock culture were diluted to 3 x 105 cells per ml in tissue culture medium, and 3 x 106 cells were placed in 50-mlvolume tissue culture flasks. Phorbol 12-myristate 13-acetate (Sigma) in dimethyl sulfoxide was added at a final concentration of 200 ng/ml. These cultures were then incubated for 48 to 72 h at 370C. Infection experiments. Cultures of macrophage-like U937 cells were infected with S. typhimurium and the bacteria were radioactively labeled while they were within the macrophages. All experiments were carried out at 370C, and bacteria and U937 cell cultures were kept at this temperature throughout each experiment to avoid temperature-induced cell responses. Overnight bacterial cultures were grown in glucose-limiting minimal MOPS medium. The overnight culture was diluted to an optical density at 420 nm (OD420) of approximately 0.05 in tissue culture medium without methionine. This culture was grown for several generations (to an OD420 of 0.5) to adapt the bacteria to the contents of this medium and then diluted to 3 x 107 cells per ml in fresh tissue culture medium without methionine, prewarmed to 370C. Samples (2 ml) of the diluted culture were placed in each tissue culture flask containing the U937 cells, resulting in a ratio of approximately 10 bacteria per macrophage. The flasks were incubated for 1 h to allow the bacteria to invade the host cells. The infected monolayers were then washed twice with warm Hanks balanced salt solution (GIBCO BRL) to remove bacteria that did not invade. Fresh tissue culture medium (2 ml) without methionine, containing 50 pg of gentamicin sulfate (Sigma) per ml, was added to each flask to kill any remaining extracellular bacteria. This was considered the beginning of the experiment (t = 0). At the beginning of each labeling period, cycloheximide (200,g/ml) was added to the flasks being labeled and then [35Slmethionine (150,uCi/ml) (ICN Radiochemicals, Irvine, Calif.) was added. Labeling was generally done from 0 to 4, 4 to 8, and 21 to 25 h after the invasion period. At the end of each labeling period, the U937 cells were lysed with a solution of 0.1% sodium deoxycholate in phosphate-buffered saline to free the intracellular bacteria. A sample of the U937 cell lysate was diluted and plated for viable bacterial counts. The remainder of the U937 cell lysate was chilled in ice water, placed in a centrifuge tube, and pelleted at 10,000 rpm in a Sorvall RC-2B centrifuge for 10 min. The bacteria were then washed with lx MOPS buffer (10), transferred to Eppendorf tubes, and pelleted for 10 min in an Eppendorf centrifuge. The bacterial pellets were stored at -70 C until extracts were made. A series of infection experiments tested the effect of the antibiotics ampicillin and chloramphenicol on viable cell counts. These experiments were carried out as described above except that no radioactive labeling was done and no extracts were made. To test the effect of ampicillin, the antibiotic was added at a final concentration of 100 pg/ml for three periods: 0 to 2, 2 to 4, and 4 to 8 h after invasion. U937 cell monolayers were lysed as described above, and the bacteria were plated for viable counts. Chloramphenicol (also 100 pg/ml) was employed in exactly the same way. Steady-state labeling of S. typhimurium. All experiments in which bacteria were labeled under extracellular conditions were conducted aerobically in a shaking water bath. Overnight bacterial cultures were grown in glucose-limiting minimal MOPS medium, glucose minimal MOPS medium, or glucose rich MOPS medium without methionine. The overnight cultures were diluted to an OD420 of approximately 0.05 and were allowed to grow for a minimum of four generations before label was added to the cultures. In all cases, growth was monitored by observing the OD420 for the duration of the experiment. All labeling was done with [35S]methionine at 50 pci/ml. All labeled cultures were rapidly chilled to 00C in ice water following the labeling period, transferred to Eppendorf tubes, pelleted, and stored at -70'C. Cultures were steady-state labeled in several media: glucose rich MOPS, glucose minimal MOPS, DMEM, pyruvate minimal MOPS, and acetate minimal MOPS. In each case, the medium contained 0.02 mm cold methionine. This concentration of methionine is 1/10 the normal rich MOPS medium concentration and is added to ensure uniform labeling of the bacterial cultures over the course of the experiment. Steadystate labeling was carried out over two to three generations, and the OD420 was always kept below 0.5 to ensure that the cold methionine was not exhausted from the medium. Cellular extracts and two-dimensional gels. Extracts were prepared and analyzed by two-dimensional gel electrophoresis as already described (1). Measurement of protein spots on two-dimensional gels. The levels of various protein spots on two-dimensional gels were measured with the QUEST package for computer-aided image analysis (7) as previously described (1). RESULTS The growth rate of S. typhimurium within macrophage-like U937 cells determined with ribosomal protein L7/L12 appeared to be rapid. The two forms of ribosomal protein L7/L12 are well resolved on two-dimensional gels, making comparison of their levels easy. Figure 1 shows the levels of these proteins in S. typhimurium X3306 growing with three different doubling times. Doubling times were varied by growing the bacteria in media of various richnesses. In glucose rich MOPS medium, the bacteria doubled every 30 min and the radioactivity of protein L12 was greater than that of L7. In DMEM, the bacteria doubled every 41 min and the radioactivity of protein L12 appeared to be approximately equal to that of L7. In glucose minimal MOPS medium, the bacteria doubled every 50 min and the amount of radioactive protein L12 was very low compared with that of L7. Figure 1 also shows L7 and L12 made by S. typhimunum X3306 within the macrophage-like U937 cells during three labeling periods (0 to 4, 4 to 8, and 21 to 25 h). In all three labeling periods, the rates of formation of proteins L7 and L12 appeared to be approximately equal. This suggested that the intracellular bacteria were doubling at the fairly rapid rate of 40 min or so. The results obtained by visual inspection of the levels of ribosomal protein L7/L12 were verified by quantitation of protein levels with computer-aided image analysis and PDQuest software. The percentage of protein L7/L12 in the L12 form was calculated for steady-state growth in five different growth media. In glucose rich MOPS medium, the

3 ai..;..i^45 :g: ''i 3746 ABSHIRE AND NEIDHARDT ~~~i.,i'icd X -*,0 O-.i4*A ~4-hr; si7. 7. i. U o ( a, 0 co J. BACT1ERIOL. 107 FIG. 1. Comparison of levels of L12 and L7 forms of ribosomal protein L7/L12 at growth rates of 30, 45, and 55 min and during intracellular growth at 0 to 4, 4 to 8, and 21 to 25 h postinvasion. bacterial doubling time was 30 min; in DMEM, 41 min; in glucose minimal MOPS medium, 50 min; in pyruvate minimal MOPS medium, 75 min; and in acetate minimal MOPS medium, 245 min. Growth rate expressed as k, the growth rate constant (k = In2/doubling time in hours), was plotted against the percentage of this ribosomal protein in the unacetylated (L12) form to obtain a calibration curve (Fig. 2). The percentage of L12 for the labeling periods of 0 to 4, 4 to 8, and 21 to 25 h was also calculated, and the calibration curve was used to determine the apparent bacterial growth rate in the intracellular condition. For 0 to 4, 4 to 8, and 21 to 25 h, the apparent values of k were 1.26, 1.08, and C,. 30 CD _j ) k (1 /hr) FIG. 2. Relationship between growth rate (k) and percentage of ribosomal protein L7/L12 found in the L12 form. Shown next to each closed square is the growth rate, expressed as doubling time in minutes, measured for S. typhimurium during growth in five different media. The open circles indicate where the data obtained for proteins L7 and L12 from cells labeled in the intracellular condition fall on this calibration curve FIG. 3. Actual intracellular growth of S. typhimurium expressed as viable counts (-) compared with viable counts that would be expected with a 40-min doubling time (5). (equivalent to doubling times of 33, 39, and 37 min), respectively. Therefore, quantitation of the levels of ribosomal protein L7/L12 supported the results of visual analysis of gels: the protein pattern of the intracellular bacteria was characteristic of rapidly growing bacteria. The growth rate of S. typhimurium within U937 macrophage-like cells determined by bacterial viable counts appeared to be slow. Viable counts were obtained in the following way. Cells of S. typhimurium X3306 were allowed to invade U937 macrophages. At various time points, the U937 cells were lysed with a mild detergent to release the intracellular bacteria. The lysate was diluted and plated, and colonies were allowed to develop. The intracellular growth curve (Fig. 3, solid squares) shows that the total number of viable bacteria did not increase much in the first 4 h and increased only about threefold over the next 4 h of growth. Figure 3 compares the growth curve obtained from viable counts with the growth curve that would be expected if the bacteria were doubling every 40 min (open squares). This figure shows that the viable cell number should have increased 60-fold in 4 h and 3,800-fold in 8 h if the cells were growing with a 40-min doubling time instead of the threefold increase that was seen with viable counts. Ampicillin and chloramphenicol were used to assist in explaining this growth rate paradox. There are several possible explanations for this growth rate paradox. First, it was possible that the bacteria were growing rapidly but were being killed by their hosts at a rate that was at first equal to and later slightly less than their growth rate, resulting in a very small net increase in cell number. Second, it was possible that a small percentage of the bacteria were multiplying rapidly in the intracellular environment while the remainder were viable but not growing and that ribosomal proteins from only the rapidly growing fraction were observed through labeling. Third, there was a possibility that the bacteria were growing as slowly as the viable counts indicated and for some unknown (but interesting) reason the ribosomal proteins were not adjusted by the normal growth rate mechanisms. Since the third possibility is not directly testable, the first

4 VOL. 175, 1993 SALMONELLA TYPHIMURIUM GROWTH WITHIN MACROPHAGES A co a FIG. 4. Intracellular growth of S. typhimurium in the presence (0) and absence (-) of 100 pg of chloramphenicol per ml. Dashed lines indicate treatment intervals. two possibilities were explored. Two antibiotics were employed in order to learn whether intracellular bacteria were dividing rapidly and/or were being killed by the host cells. Chloramphenicol or ampicillin was added to U937 cells infected with S. typhimurium X3306, and viable bacterial counts were obtained to determine the effects of these drugs on bacterial survival. Consider two possible intracellular situations. In the first, bacteria are growing rapidly and are being killed almost as rapidly. In the second, most bacteria are in growth stasis and are not being killed, while a small fraction of them are growing. A bacteriostatic drug such as chloramphenicol could be expected to lower viable counts in the first situation and to have little effect in the second. Likewise, a drug like ampicillin that kills growing cells could be expected to lower viable counts in the first situation and to have little effect in the second. The intracellular growth curves in Fig. 4 and 5 show the results of the antibiotic experiments. The results varied depending on the time interval measured. From 0 to 2 h after invasion, the intracellular bacteria were susceptible to both antibiotics; each antibiotic caused the viable counts to drop significantly below normal. The ampicillin result indicated that during this interval early in infection, many of the cells were dividing, and the chloramphenicol result indicated that the host cells were able to kill the bacteria. From 2 to 4 and 4 to 8 h after infection, neither ampicillin nor chloramphenicol had much effect on viable counts. The ampicillin result suggested that most of the bacteria were viable but not dividing during these later time intervals. The chloramphenicol result suggested that a large number of the intracellular bacteria were not susceptible to killing by the host cells even during chloramphenicol inhibition of growth. The 4- to 8-h results are consistent with the interpretation that most intracellular bacteria by that time are not growing and are resistant to killing and that there is a small, relatively rapidly growing population. DISCUSSION The results of treating S. typhimunum-infected macrophage-like U937 cells with ampicillin and chloramphenicol suggest a solution to the growth rate paradox. Intracellular ' FIG. 5. Intracellular growth of S. typhimurium in the presence (0) and absence (-) of 100 pg of ampicillin per ml. Dashed lines indicate treatment intervals. S. typhimurium cells apparently exist in two populations: a rapidly dividing population and a nondividing but viable (static) population. Bacteria that were labeled while within the macrophage were probably a mixture of the rapidly growing population and the static population, but with growing cells most likely incorporating the majority of the [35S]methionine. This would explain the growth rate determination with protein L7/L12 and its conflict with results obtained by viable counts. These results can also help explain a technical problem. In general, it was never possible to get large amounts of [35Sjmethionine incorporated into the intracellular bacteria. The fact that only a small proportion of the intracellular bacteria were actively growing could account for the low rate of labeling. The idea that there are two bacterial populations present in the intracellular environment fits in an interesting way with some recent results of Buchmeier and Heffron (3). By electron microscopy, these investigators were able to determine that approximately half of intracellular S. typhimurium bacteria reside in phagosomes that were not fused with lysosomes and that the other half of the intracellular bacteria were in fused phagolysosomes. They observed dividing bacteria within unfused phagosomes, but they never saw dividing bacteria within fused phagolysosomes. These observations provide a physical model for our findings: the two intracellular populations of S. typhimurium, one growing and one static, might correspond to their locations in unfused and fused phagosomes, respectively. A second aspect of our measurements is of interest. The macrophages appeared to kill the bacteria effectively early (0 to 2 h) but not later (2 to 4 and 4 to 8 h) in infection. When chloramphenicol was present early in the infection, inhibiting synthesis of the appropriate proteins, the bacteria were apparently left susceptible to killing by the host cells. Later in the infection, after the bacteria were allowed time to adapt and to synthesize the appropriate proteins, the addition of chloramphenicol did not enhance the killing of intracellular bacteria by the macrophages. This result supports the notion that the changes observed (1) in the pattern of gene expression of S. typhimuium in response to intracellular conditions are important for survival and/or multiplication of the bacteria

5 3748 ABSHIRE AND NEIDHARDT within the macrophages. It is possible that in a host animal, some of these proteins may be induced by various conditions encountered by the bacteria previous to their uptake by macrophages. This could result in the bacteria being more prepared for the intracellular conditions within macrophages than they are in a tissue culture model system of infection. The indication that there are two distinct populations of intracellular bacteria leads to questions about the relative importance of each group to overall survival within macrophages. One would usually assume that the much smaller, replicating population would be responsible for persistence of the bacteria within the host cells. In this case, the host cell would be responsible for the bacteria being in a static state, and over a longer period, these bacteria eventually would be killed. However, since a large proportion of the intracellular bacteria are apparently resistant to killing by the host cells and yet appear to be in a nondividing state, it is possible that at least one of the keys to intracellular survival is to differentiate into this nongrowing state. Since the pattern of proteins seen on two-dimensional gels of intracellular bacteria is likely to be that of a mixture of the rapidly growing population and the static population, some of the proteins induced during intracellular growth could be important to cells in the static state. An interesting point of speculation is that the ability to enter a nongrowing but resistant state during infection of macrophages may be a virulence factor for S. typhimurium to survive extreme changes in its environmental conditions. If and when conditions within the host improve, the bacteria may begin growing again. There may be many situations in which the importance of being static outweighs the importance of growing. REFERENCES 1. Abshire, K. Z., and F. C. Neidhardt Analysis of proteins synthesized by Salmonella typhimurium during growth within a host macrophage. J. Bacteriol. 175: J. BA=rRIOL. 2. Buchmeier, N. A., and F. Heffron Induction of Salmonella stress proteins upon infection of macrophages. Science 248: Buchmeier, N. A., and F. Heffron Inhibition of macrophage phagosome-lysosome fusion by Salmonella typhimurium. Infect. Immun. 59: Carroll, M. E. W., P. S. Jackett, V. R. Aber, and D. B. Lowrie Phagolysosome formation, cyclic adenosine 3':5'-monophosphate and the fate of Salmonella typhimunum within mouse peritoneal macrophages. J. Gen. Microbiol. 110: Collins, F. M Immunity to enteric infections in mice. Infect. Immun. 1: Elsbach, P., and J. Weiss Phagocytic cells: oxygenindependent antimicrobial systems, p In J. I. Gallin, I. M. Goldstein, and R. Snyderman (ed.), Inflammation: basic principles and clinical correlates. Raven Press, New York. 7. Garrels, J. I The QUEST system for quantitative analysis of two-dimensional gels. J. Biol. Chem. 264: Gulig, P. A., and R. Curtiss m Plasmid-associated virulence of Salmonella typhimurium. Infect. Immun. 55: Klebanoff, S. J Phagocytic cells: products of oxygen metabolism, p In J. I. Gallin, I. M. Goldstein, and R. Snyderman (ed.), Inflammation: basic principles and clinical correlates. Raven Press, New York. 10. Neidhardt, F. C., P. L. Bloch, and D. F. Smith Culture medium for enterobacteria. J. Bacteriol. 119: Pedersen, S., P. L. Bloch, S. Reeh, and F. C. Neidhardt Patterns of protein synthesis in E. coli: a catalog of the amount of 140 individual proteins at different growth rates. Cell 14: Sundstrom, C., and K. Nilsson Establishment and characterization of a human histiocytic lymphoma cell line (U-937). Int. J. Cancer 17: Wanner, B., R. Kodaira, and F. C. Neidhardt Physiological regulation of a decontrolled lac operon. J. Bacteriol. 130: