Use of Aminoglycosides in Treatment of Infections Due to Intracellular Bacteria

Soon after antibiotics were introduced into medical practice, authors pointed out that microorganisms capable of surviving within phagocytic cells may be protected from the killing actions of antibiotics with a poor ability to cross the eukaryotic cell membrane, leading to therapy failures and disease relapses (86, 108). Following the pioneering work by Bonventre et al. (21) that showed that mouse peritoneal macrophages were impermeable to streptomycin in vitro, the aminoglycosides have been included in such a category. Failure of streptomycin and other aminoglycosides to treat infections due to strict or facultative intracellular pathogens (e.g., rickettsial diseases, chlamydial diseases, and typhoid fever) have reinforced this point of view. Also, in vitro infected cell systems have been developed to test the activities of antibiotics against intracellular pathogens, and in many instances they have shown the limited intracellular activity of aminoglycosides compared to their strong bactericidal potential in extracellular medium (52, 92).

However, at the same time, streptomycin was successfully used to treat infections due to facultative intracellular pathogens, such as tuberculosis (especially tuberculous meningitis) (120), plague (133), brucellosis (160), or tularemia (55). As for tuberculosis, clinical data were concordant with the demonstration by Mackaness et al. (106) that streptomycin could inhibit growth of Mycobacterium tuberculosis within macrophages. Both these clinical and experimental data have been overlooked for several decades. Later experiments by Tulkens et al. (22, 179) have confirmed that aminoglycosides are readily incorporated into phagocytes when these cells are in contact with the antibiotic for prolonged periods (i.e., more than 24 h).

The present review deals with the potential role of aminoglycosides, and especially streptomycin, in the antibiotic therapy of infections due to intracellular pathogens. Our goal is to highlight the concept that categorization of aminoglycosides as strictly ineffective against intracellular pathogens is in some ways too simplistic and does not correspond to clinical experience, since aminoglycosides remain in fact a “gold standard” therapy in a number of infections due to intracellularly surviving pathogens.

INTRACELLULAR SURVIVAL OF MICROORGANISMS

Intracellular microorganisms are referred to as strictly intracellular when they only multiply within eukaryotic cells and as facultatively intracellular when they can also be grown in axenic media (Table 1). However, this may not always reflect the situation in the infected human host, in which facultative intracellular bacteria may develop predominantly within eukaryotic cells. Examples include M. tuberculosis (especially for tuberculous meningitis), Francisella tularensis (the agent of tularemia), and Yersinia pestis (the agent of plague). Different strategies may allow intracellularly surviving pathogens to prevent their digestion by professional phagocytic cells (Table 1), i.e., polymorphonuclear cells (PMNs), monocytes, and macrophages (MPs), via the lysosomal pathway. Some of them infect lysosome-free cells such as erythrocytes, e.g., Plasmodium spp. and Babesia spp. among protozoa and the bacterial species Bartonella bacilliformis (the agent of Carrion's disease). Microorganisms may also infect nonprofessional phagocytic cells, including endothelial cells (e.g., Rickettsia spp., Bartonella spp.) and fibroblasts and epithelial cells (e.g., Chlamydia spp.), which display lower microbicidal potentials than PMNs and MPs. Few microorganisms are capable of multiplying inside PMNs because of the strong microbicidal properties (81), e.g., the agent of human granulocytic ehrlichiosis (75). Monocytes and MPs offer a more favorable niche for intracellular survival and multiplication, with both lower microbicidal properties (2) (especially a less efficient oxidative burst) and a prolonged lifetime compared to PMNs.

TABLE 1.

Intracellular microorganisms surviving within phagocytic cells

Site of multiplication and pathogen	Intracellular behavior	Target cells	pHa	Reference(s)
Cytosol without PLFb
Rickettsia spp.	Strict	Endothelial cells	∼6.5	44, 163
Shigella spp.	Facultative	Various	∼6.5	128, 147
Cytosol after PLF
L. monocytogenes	Facultative	Various	∼6.5	94, 100
Unfused vacuole
L. pneumophila	Facultative	MPs	5.7–6.5	87
Toxoplasma gondii	Facultative	MPs	6.8–7.0	155
Selectively fused vacuole
Chlamydia spp.	Strict	Epithelial cells	>6.0	82
M. tuberculosis	Facultative	MPs	NDc	37, 168
Mycobacterium avium	Facultative	MPs	6.3–6.5	37, 168
Mycobacterium marinum	Facultative	MPs	6.3–6.5	15
Salmonella spp.	Facultative	MPs	<7.0	138
Uncharacterized endosome
Ehrlichia chaffeensis	Strict	Monocytes	ND	16
HGE agent	Strict	PMNs	ND	75
Brucella spp.	Facultative	MPs	ND	66
Bartonella spp.	Facultative	Endothelial cells	ND	26, 42, 72, 73
Afipia felis	Facultative	ND	ND	25
Yersinia enterocolitica	Facultative	MPs	ND	175
Mycobacterium leprae	Strict	MPs	ND	65, 154
Nocardia asteroides	Facultative	MPs	ND	19
Acidic phagolysosome
C. burnetii	Strict	MPs	4.7	4, 78, 113
Leishmania donovani	Facultative	MPs	5.2	7
S. aureus	Facultative	MPs, PMNs	4.8	102, 117
Alkalinized phagolysosome
Histoplasma capsulatum	Facultative	MPs	∼7.0	53

^apH of the intracellular multiplication site. 

^bPLF, phagosome-lysosome fusion. 

^cND, not determined. 

Microorganism trafficking through endocytic organelles is a complex phenomenon, leading to the successive formation of early endosomes, late endosomes, and then true phagolysosomes, in which most microorganisms are digested because of the presence of a local acidity (i.e., pH ≤5) and enzymes such as acidic hydrolases (72, 77, 94, 125, 136). Microorganisms surviving within phagocytic cells may be currently placed into one of four categories (94) (Table 1): (i) intracytoplasmic, after exit from the endosomal compartment, with or without fusion of the phagosomal vacuole with lysosomes; (ii) residing in nonfused phagosomes, when the early phagosome does not fuse with any endocytic organelle of the host cell and does not acidify; (iii) residing in selectively fused phagosomes, when the early phagosome has been modified by a selective fusion event; and (iv) residing in lysosomes, when the endocytic cascade has been completed. In fact, the actual situation may be even more complex, with each intracellular pathogen being responsible for the formation of endocytic vacuoles with specific features.

Physicochemical conditions prevailing within each parasitophorous vacuole have been defined for only a few intracellular residing pathogens and only partially. Among such physicochemical factors, the influence of intracellular pH is one of the easiest to consider, and acidity is known to be deleterious for activity of most antibiotics and specifically of aminoglycosides (1, 49). The pH values in Chinese hamster ovary cells were found to be about 6.0 in early endosomes, 5.0 to 6.0 in late endosomes, and 5.0 to 5.5 in lysosomes (125). Similar values were obtained in fibroblasts cells (125). We found a lysosomal pH of 4.7 for P388D1 MP-like cells (113). The pH prevailing within parasitophorous vacuoles in infected cells has been defined for a limited number of intracellular microorganisms (Table 1).

INTRACELLULAR PHARMACOKINETIC AND PHARMACODYNAMIC PROPERTIES OF AMINOGLYCOSIDES

Several techniques have been used to assess the intracellular penetration and concentration of antibiotics within eukaryotic cells. All techniques share a common step, the in vitro exposure of a cell monolayer to antibiotic-containing medium for various incubation times. Then, the more problematic step of the procedure is in measuring the intracellular concentration of the antibiotic without interference with its extracellular counterparts. Washing cell monolayers to remove the extracellular antibiotic (180) potentially underestimates antibiotic cell uptake, because most antibiotics actually reemerge from the cell very rapidly during the washing procedure. A velocity gradient centrifugation technique (i.e., cells centrifuged rapidly through a lipid bilayer [180]) has been proposed as a more reliable method (183). Harvested cells are lysed, and the antibiotic concentration within cell lysates is quantified either by autoradiography (i.e., previously radiolabeled antibiotic), a fluorescence technique (i.e., fluorescent antibiotic such as a tetracycline or fluoroquinolone), or a chemical technique (e.g., high-performance liquid chromatography), whereas a microbiological technique of antibiotic dosage is used to assess the residual antimicrobial activity of the drug recovered. Results are expressed as C/E ratios, i.e., the ratio of the intracellular concentration of the antibiotic tested compared to the antibiotic concentration used in the incubation medium. Cell uptake methods may be coupled with a cell fractionation study, using a sucrose gradient centrifugation technique, for determination of the subcellular distribution of the antibiotic (180). The antibiotic is measured in the different cell fractions concomitantly with enzymes used as specific markers for the cytosol (e.g., lactic dehydrogenase) or lysosomes (e.g., N-acetyl-β-hexosaminidase or cathepsin D) (178, 180). This technique only allows determination of subcellular distribution of the antibiotic in these two cell compartments. Antibiotic penetration within phagosomes, i.e., in infected eukaryotic cells, remains totally undetermined. This is a particular shortcoming of these models since many intracellularly residing pathogens actually multiply in nonfused endosomal compartments (Table 1).

Cell uptake of aminoglycosides has been evaluated in vitro using mouse or guinea pig peritoneal MPs (21, 22), rat embryo fibroblasts (178), cell lines such as J774 murine MP-like cells (180) and baby hamster kidney (BHK-21) cells (22), and human PMNs and human monocyte-derived MPs (135) (Table 2). Bonventre et al. (21) first reported that streptomycin was unable to penetrate within mouse peritoneal MPs. However, the same authors subsequently demonstrated that prolonged incubation (i.e., more than 24 h) resulted in a linear accumulation of the drug within cells for the first 6 days and then reaching a steady state, with C/E ratios greater than 3 (22). These observations were overlooked by Prokesh et al. (135) and Johnson et al. (93), who respectively demonstrated that aminoglycosides did not accumulate within human PMNs and rabbit alveolar MPs when incubated for 2 h. More recent investigations have confirmed that aminoglycosides do penetrate eukaryotic cells, with C/E ratios ranging from ≈2 to 5 after 2 to 4 days of incubation (178–181, 183). Current knowledge indicates that aminoglycosides cannot diffuse passively through the eukaryotic cell membrane because of their large size and negative charge (180). Cell uptake of aminoglycosides corresponds to an active mechanism of pinocytosis by the eukaryotic cell, which explains the slow intracellular accumulation of these drugs and is only detectable after 48 to 72 h of antibiotic exposure (180). This mode of penetration probably also explains why MPs which display high pinocytosis activity more efficiently concentrate aminoglycosides than do PMNs (165). Also, increased uptake of streptomycin by PMNs or MPs during phagocytosis has been reported (131, 166). After the aminoglycosides reach the eukaryotic cell membrane by pinocytosis, they accumulate quite exclusively (via the lysosomal pathway) within the acidic lysosomes, where they are retained because of their weakly basic nature (180). The lysosomal compartment corresponds to approximately 3 to 5% of the cell volume (177, 178). Thus, it has been estimated that aminoglycosides may reach concentrations within this cell compartment of up to 100- to 250-fold that in the extracellular medium (177, 178, 180). It may be hypothesized that cell infection influences the subcellular distribution of antibiotics, but this has not been investigated.

TABLE 2.

Aminoglycoside cell uptake

Antibiotic and cell type	Incubation time	Ea (μg/ml)	C/Eb	Reference
Streptomycin
Kupffer cells	30 min	1	<1	130
Mouse peritoneal MPs	1 h	5	<0.1	21
Mouse peritoneal MPs	2–4 h	2	<1	22
Mouse peritoneal MPs	20 h	30	<0.1	21
Mouse peritoneal MPs	24 h	2	∼2	22
Mouse peritoneal MPs	48 h	2	∼3.6	22
Guinea pig peritoneal MPs	48 h	2	>3	22
BHK-21	48 h	2	>3	22
Rat embryo fibroblasts	3 days	100	2.4	177
Rat embryo fibroblasts	4 days	500	2	179
Gentamicin
Rabbit alveolar MPs	1 min	18	0.27	93
Human PMNs	15 min	18	1.03	135
Rabbit alveolar MPs	15 min	18	0.48	93
Human PMNs	30 min	18	0.73	135
Rabbit alveolar MPs	30 min	18	0.39	93
Human PMNs	60 min	18	0.83	135
Rabbit alveolar MPs	60 min	18	0.56	93
Human PMNs	120 min	18	0.84	135
Rabbit alveolar MPs	120 min	18	0.56	93
Rat embryo fibroblasts	4 days	100	6	178
Rat embryo fibroblasts	4 days	500	5.2	179
Kanamycin
Rat embryo fibroblasts	4 days	100	2.6	178
Rat embryo fibroblasts	4 days	500	2.4	179
Amikacin
Rat embryo fibroblasts	4 days	500	3.6	179

^aE, antibiotic concentration used in the cell incubation medium. 

^bC/E, ratio of the intracellular concentration of the antibiotic to the extracellular one. 

Experimental data on intracellular pharmacokinetic properties of aminoglycosides obviously do not correlate well with their poor activity against many intracellular pathogens, especially those residing within lysosomes, such as Staphylococcus aureus (102, 117) or Coxiella burnetii (137). It has been hypothesized that, among antibiotics capable of intracellular penetration and concentration, a reduced activity may result from the deleterious action of the intracellular milieu (183). In particular, a low pH is a well-recognized deleterious condition for many antibiotics (Table 3). The activity of aminoglycosides is greatly influenced by pH, and these antibiotics are nearly devoid of activity at the acidic pH of lysosomes (i.e., ≈4.5 to 5) (1, 49). We have highlighted the influence of pH on intracellular activity of aminoglycosides within phagolysosomes in an in vitro model of S. aureus infection (117), as discussed below.

TABLE 3.

Influence of pH on activity of antibiotics

Antibiotic(s)	Activity at pHa
	<5	6	7
β-Lactams	−	++	+++
Aminoglycosides	−	+	+++
Chloramphenicol	−	+	+++
Tetracyclines	+	+++	++
Rifampin	+++	++	++
Erythromycin	−	+	+++
Fluoroquinolones	−	+	+++

^a−, no activity; +, low activity; ++, moderate activity; +++, maximum activity. 

IN VITRO ACTIVITIES OF AMINOGLYCOSIDES AGAINST INTRACELLULAR BACTERIA

In vitro infected cell systems have been developed to test the activities of antibiotics against strict intracellular pathogens. These cell systems are also needed for facultative intracellular pathogens, because of the poor ability of MICs that have been determined in axenic medium to predict the in vivo efficacies of antibiotics. This is especially the case for aminoglycosides, which in many cases are poorly active against intracellular microorganisms despite high extracellular bactericidal activity. In these systems, cell cultures grown in vitro are exposed to a microbial inoculum for sufficient time (i.e., usually 1 to 2 h) to allow intracellular penetration of microorganisms and their establishment within their subcellular site of multiplication. After removal of nonphagocytized microorganisms from the extracellular medium to ascertain that only the intracellular activity of antibiotics is being evaluated, infected cell monolayers are exposed to the tested antibiotic. The mode of evaluation of antibiotic activity may vary according to the intracellular pathogen considered (118), including (i) a quantitative determination of residual viable bacteria (CFU counts, PFU counts, etc.); (ii) evaluation of the percentage of infected cells for microorganisms with complex intracellular cycles, such as Ehrlichia spp. or Chlamydia spp.; or (iii) other more sophisticated methods (i.e., flow cytometry for Chlamydia sp. [43] or the luciferase technique for mycobacteria [8, 90]).

Aminoglycosides are considered poorly active or not active against strictly intracellular bacteria, including Rickettsia spp. (142), Chlamydia spp. (162, 184, 185), Ehrlichia spp. (23, 24), and C. burnetii (137). As for facultative intracellular bacteria, aminoglycosides are usually considered poorly effective compared to their activities in axenic medium, such as for Salmonella spp. (98), Legionella pneumophila (58), and S. aureus (102, 117). A more significant intracellular activity of aminoglycosides was shown against Mycobacterium spp. (36, 39, 106), Brucella spp. (143), F. tularensis (115), Bartonella spp. (126), and Listeria monocytogenes (48). However, the poor activity of aminoglycosides against these pathogens in cellular models has prompted authors to hypothesize that the in vivo activities of these antibiotics (e.g., in brucellosis or tularemia) may be related to their ability to kill microorganisms in the extracellular milieu (rather than a true intracellular activity), preventing their spread from infected cells to uninfected ones.

It should be emphasized, however, that cell models currently used to test the activity of antibiotics against intracellular pathogens present a number of technical limitations that appear most critical when assessing the activities of aminoglycosides. First, removal of nonphagocytized microorganisms before evaluation of the activity of antibiotics is a critical stage of the methodology, and no methodology currently available allows accurate elimination of residual extracellular microorganisms without altering intracellular ones. Lysostaphin is used to remove extracellular staphylococci (50, 166). As for other intracellular pathogens, aminoglycosides are most often used for this purpose (e.g., L. monocytogenes, Brucella spp., Bartonella spp.); these antibiotics poorly penetrate the eukaryotic cell membrane when added in the cell incubation medium for short periods (usually less then 4 h), and they display a high extracellular bactericidal activity against most microorganisms. This procedure may induce overvaluation of the intracellular activity of the antibiotic tested because of interference of residual aminoglycoside with intracellular multiplication of microorganisms. This seems particularly troublesome when assessing the intracellular activity of an aminoglycoside and is still in practice only because no better alternative method is currently available. Conversely, we want to emphasize that the intracellular activities of aminoglycosides may have been underestimated in most in vitro studies, because the activity was evaluated after short periods of incubation of infected cells in the presence of the antibiotic, without considering the slow intracellular accumulation of aminoglycosides. In the following sections, specific examples have been developed to tentatively correlate currently available experimental data on activities of antibiotics against intracellular pathogens with clinical data for corresponding diseases.

SELECTED EXAMPLES OF CLINICAL ACTIVITY OF AMINOGLYCOSIDES

M. tuberculosis and tuberculous meningitis.

Tuberculous meningitis is still prevalent in developing countries where tuberculosis is endemic, but it is also occasionally diagnosed in industrialized countries (88, 96). Although effective antibiotic therapy is now available, tuberculous meningitis remains a serious health threat, with death in approximately one-third of involved patients, definite sequels in one-third, and total recovery in one-third (88, 96). Because the disease is clinically nonspecific and because of difficulties in culturing M. tuberculosis from the cerebrospinal fluid, a specific diagnosis and thus initiation of the appropriate antibiotic therapy are often delayed, which partly explains the poor prognosis.

M. tuberculosis is a facultative intracellular bacterium which multiplies in monocytes and MPs (37, 168). In axenic medium, M. tuberculosis is highly susceptible to streptomycin (28), which displays a bactericidal activity, and to other aminoglycosides, including gentamicin (146) and amikacin (69, 85, 145). Assessment of the activity of streptomycin against M. tuberculosis isolates grown in MPs has shown that this compound is far less active than in axenic medium, which has forced the idea that this compound is only active against extracellularly growing M. tuberculosis. However, early in vitro experiments from Mackaness (106) showed that streptomycin could inhibit growth of M. tuberculosis within mouse MPs. Such observations were later confirmed by others using animal-derived MPs or cell lines (152, 169, 170), but also by Crowle et al. (36, 39), using human monocyte-derived MPs. Abraham and Duthie have hypothesized that the intracellular activity of streptomycin may be impaired by acidic pH within the M. tuberculosis-containing phagosomes (1) rather than a poor ability to penetrate within the eukaryotic cell. Crowle and May (38) also demonstrated that alkalinization of M. tuberculosis-containing vacuoles by using the lysosomotropic agent chloroquine resulted in an expected increased activity of streptomycin. Using color indicators, Sprick found that the pH in M. tuberculosis-containing vacuoles in infected mouse MPs was between 4.5 and 5.0 (164). However, other investigators have suggested that M. tuberculosis may reside in less acidic intracellular environments, as previously supposed (37). More recent experiments have confirmed that M. tuberculosis multiplies in selectively fused phagosomes which lack membrane proton ATPases (94), thus preventing their drastic acidification (37, 168).

The antibiotic therapy of tuberculous meningitis has been determined empirically and is currently based upon a few clinical studies (45, 46, 88, 96, 120). Most current recommendations, in fact, take into account recent advances in antibiotic therapy of pulmonary tuberculosis, which may not always apply to the situation of meningitis. Before the antibiotic era, tuberculous meningitis was considered always fatal. Introduction of streptomycin allowed a 25 to 36% survival rate, which is an undeniable argument for its clinical usefulness, although many patients relapse on antibiotic withdrawal (45, 88, 96, 120). Although the prognosis of tuberculous meningitis further improved with the introduction of antituberculous agents with better pharmacokinetic properties within the cerebrospinal fluid, such as isoniazid, ethambutol, and pyrazinamide (54), streptomycin is still used in combination with other antituberculous agents (45, 88).

Y. pestis and plague.

Plague has caused several world pandemics in the past, and it is still prevalent in some parts of Africa, Asia, and South America (68, 173). Rats are the primary reservoir of Y. pestis in nature from which fleas become contaminated. The enlarged lymph nodes of bubonic plague, the most frequent presentation of the disease, occur following cutaneous inoculation of Y. pestis by an infective flea. The most severe form of the disease is pneumonic plague, which may lead to a near 100% mortality rate and which also may be responsible for rapid spread of the disease via infectious aerosol inhalation, possibly leading to epidemics and even pandemics.

Y. pestis, the etiological agent of plague, is a gram-negative, facultatively intracellular bacterium (127, 167). In axenic medium, Y. pestis is highly susceptible to many antibiotics (63, 156), including aminoglycosides, although multidrug resistance in this species has been reported recently in one strain (68). The in vitro activities of antibiotics against intracellular Y. pestis have never been assessed. In the experimental pneumonic plague model in mice, streptomycin is highly effective (20, 29).

Streptomycin is considered highly effective for treatment of plague, whereas gentamicin and kanamycin have been proposed as alternative aminoglycosides (30, 133). Chloramphenicol has been advocated for treating plague meningitis (18) because of its more favorable pharmacokinetic properties within the cerebrospinal fluid compared to that of aminoglycosides. Tetracyclines or sulfonamides are used for prophylactic purposes (68, 133). Although fluoroquinolones are active against Y. pestis in in vitro and animal models, data in humans are not available (89). Thus, aminoglycosides and especially streptomycin remain a first-line antibiotic therapy of plague, which is well correlated with experiments conducted in animal models. In vitro cell systems would be useful to better define the type of action of aminoglycosides against the plague bacillus.

Brucella spp. and brucellosis.

Many animals are potential reservoirs for Brucella species (32, 187). Milk and dairy products are the main sources for human contamination. Four serotypes are responsible for human brucellosis: Brucella melitensis, B. abortus, B. suis, and B. canis. Brucellosis most often manifests as an acute bacteremic disease, and it may be complicated by secondary hematogenous localization in almost every organ (32, 187). Chronic brucellosis is defined as a disease evolving for more than a year, without evident secondary localization.

Bacteria belonging to the genus Brucella are gram negative and facultatively intracellular. Monocytes and MPs are considered the target cells in humans (14). Since Brucella species are fastidious bacteria, their antibiotic susceptibilities in axenic medium have been usually determined either in Mueller-Hinton, tryptic soy, or brucella medium, supplemented with hemoglobin and vitamins. Although tetracycline compounds are the most effective antibiotics in axenic media, Brucella spp. are also susceptible to aminoglycosides, including streptomycin (3, 73, 97, 124, 153, 174), gentamicin (80, 124, 153), tobramycin (124), and amikacin (124). Using the killing curve technique, Rubinstein et al. (143) demonstrated a high in vitro bactericidal activity of streptomycin against B. melitensis (143). The activities of antibiotics against intracellular Brucella species have been poorly evaluated, and results obtained with aminoglycosides are contradictory. Freeman et al. (64) reported that streptomycin inhibited intracellular growth of Brucella spp. at a 10 μg/ml extracellular concentration. In contrast, Richardson et al. (139) reported that the same antibiotic, used at extracellular concentrations ranging from 2 to 50 μg/ml, was not effective in preventing growth of a B. abortus strain within bovine uterine mucosal cells despite up to 6 days of antibiotic exposure. Goldenbaum et al. (74) also reported that streptomycin at 5 μg/ml could not inhibit growth of a B. abortus strain within guinea pig peritoneal MPs, although the addition of chlorpromazine resulted in a significant intracellular killing effect after 24 and 48 h of incubation of cultures. More recently, Fountain et al. (62) found that streptomycin at an extracellular concentration of 100 μg/ml did not affect intracellular growth of B. abortus and B. canis strains within murine or guinea pig peritoneal MPs, whereas liposome-encapsulated streptomycin was highly bactericidal. However, the activity of streptomycin was evaluated after only 24 h of incubation of infected cultures with the antibiotic, which may not have been sufficient to allow its intracellular accumulation. Similar results have been obtained with gentamicin (57).

Streptomycin, doxycycline, and rifampin have become the mainstay in antibiotic therapy of brucellosis (32, 79, 160, 187). However, it has also become obvious that none of these antibiotics can definitely cure Brucella spp. infection in all patients when used as monotherapy, with relapse rates of over 30% of cases (32, 79, 160, 187). Thus, combination therapy has been advocated, including the two gold standard regimens currently recommended by the WHO, i.e., the combination of doxycycline with either streptomycin or rifampin for at least 6 weeks (160). More recently, gentamicin and netilmicin have been used as alternatives to streptomycin, with promising results (158, 159). The combination of doxycycline plus streptomycin has been found to be superior to that of doxycycline plus rifampin (10, 32, 160, 161), with 0 to 8% relapses with the former regimen versus 5 to 17% with the latter (79, 160). The difference between these two regimens is more pronounced in cases of osteoarticular secondary localization (10, 32, 160, 161). Thus, streptomycin remains a first-line antibiotic for therapy of brucellosis, although its use in combination with a second antibiotic (usually doxycycline) is a necessity to prevent relapses on antibiotic withdrawal. Further experiments are needed to reassess the activity of aminoglycosides against intracellular Brucella spp., since available data are contradictory and not well correlated to the current clinical experience, which demonstrates the usefulness of these antibiotics to treat brucellosis.

F. tularensis and tularemia.

F. tularensis biovar tularensis and F. tularensis biovar palearctica are the etiological agents of tularemia (56, 90). Tularemia may have different clinical presentations, the more suggestive being the combination of a chronic lymphadenopathy in the draining territory of a cutaneous eschar in patients with a medical history of tick bite or contact with a wild animal (especially rabbits or hares) (56, 90, 91, 132). Other clinical manifestations include pharyngitis, conjunctivitis, pneumonitis, or a pseudotyphoidal syndrome.

Francisella species are gram-negative, facultative intracellular bacteria, with monocytes and MPs being natural target cells in infected humans (59, 60, 67). Only a few in vitro studies have determined the antibiotic susceptibilities of F. tularensis isolates (13, 111). All strains tested have been susceptible to most antibiotics, including aminoglycosides. However, MICs determined in axenic medium do not correlate well with the clinical experience in treating human tularemia. Nutter et al. (129) reported early that streptomycin but not penicillin G could inhibit growth of F. tularensis within rabbit alveolar MPs grown in vitro. We have recently determined the activities of several antibiotics against an F. tularensis biovar palearctica strain grown in P388D1 MP-like cells (115). In our system, aminoglycosides (i.e., gentamicin, streptomycin, and amikacin) and fluoroquinolones were highly bactericidal. Interestingly, the activities of aminoglycosides against intracellular F. tularensis increased progressively with prolonged incubation time, to a maximum after 72 h incubation of cultures. This is well correlated with the reported slow accumulation of aminoglycosides within eukaryotic cells (116, 178, 179).

Tularemia is a disease characterized by a tendency to relapse despite appropriate antibiotic therapy, especially in severely diseased patients. Only the aminoglycosides are considered to achieve a 100% cure (55). Streptomycin has been used for several decades and remains the gold standard antibiotic therapy of tularemia (61). Since streptomycin may be no longer available in clinical practice in some areas, gentamicin has been used as an alternative in a limited number of patients, with encouraging results (35, 112). Tetracyclines, especially doxycycline, are currently considered the most useful alternative to aminoglycoside therapy, but relapses are reported in up to 10% of tularemic patients treated with these drugs (56, 131, 144). Beta-lactams are ineffective (34). Fluoroquinolones have been more recently reported to be useful in a limited number of patients, but more clinical experience is needed (140, 148, 171). In conclusion, antibiotic susceptibility of F. tularensis as determined in an in vitro cell system correlated well with clinical experience in treating human tularemia. In particular, the observation that aminoglycosides display high in vitro bactericidal activity against intracellular F. tularensis is compatible with their ability to prevent relapses in tularemic patients (55, 112).

Bartonella species and verruga peruana.

B. bacilliformis is responsible for Carrion's disease, a disease endemic in the mountainous regions of Peru, Ecuador, and Colombia (5, 6, 76, 109). The acute clinical manifestations of Carrion's disease, referred to as Oroya fever, correspond to a potentially fatal B. bacilliformis bacteremia with acute hemolysis (76, 83, 109, 150, 151). The chronic form of the disease, called verruga peruana, corresponds to the emergence of benign cutaneous angiomatous tumors and is typically a relapsing illness (9, 109). Bartonella henselae and Bartonella quintana are more widespread species responsible for various diseases in humans, including bacillary angiomatosis (119), which shares many similarities with verruga peruana. The former disease mainly affects immunocompromised patients (including AIDS patients) and corresponds to angiomatous cutaneous lesions and/or angiomatous lesions that may involve almost every organ (119).

Bartonella species are gram-negative, facultatively intracellular bacteria. B. bacilliformis has been visualized in erythrocytes in patients suffering from Oroya fever (5, 76). Bartonella species can infect endothelial cells in vitro, leading to endothelial cell proliferation (26, 42, 70, 71), and are supposed to infect these cells in vivo, inducing the angiomatous lesions typical of verruga peruana (9) and bacillary angiomatosis (42). We have recently evaluated the in vitro susceptibilities of B. henselae, B. quintana, and B. bacilliformis in blood-enriched medium (114, 157). All strains were highly susceptible to most antibiotics tested. The activities of antibiotics against intracellular B. bacilliformis or intracellular B. quintana have not been evaluated. However, intracellular susceptibility determinations for B. henselae in an endothelial cell system showed that only aminoglycosides, especially gentamicin, display intracellular bactericidal activity (126).

Chloramphenicol is usually recommended as a first line antibiotic therapy of Oroya fever (40, 109, 182), whereas a macrolide has been advocated for bacillary angiomatosis (119). However, although a number of antibiotics may be effective to treat the acute stage of Bartonella infections, relapses remain frequent on antibiotic withdrawal (109). More strikingly, antibiotics are much less effective in treating patients suffering from chronic infection. Although Oroya fever and verruga peruana are caused by the same bacterium (i.e., B. bacilliformis), chloramphenicol is ineffective in treating the latter disease (109), and it also does not prevent the occurrence of verruga peruana (109) when administered in acutely infected patients. Streptomycin is considered the drug of choice in this chronic stage of Carrion's disease, allowing a near 100% cure (109). Thus, aminoglycosides and especially streptomycin represent the gold standard antibiotic therapy for verruga peruana, which is a chronic disease due to invasion of endothelial cells by B. bacilliformis. Because of similarities in the pathophysiologies of Carrion's disease and bacillary angiomatosis, experience in treating the former disease may help to better define the optimum therapy for the latter one.

L. monocytogenes and listeriosis.

Listeriosis is a food-borne disease of worldwide occurrence (84, 103). Although in the majority of cases ingestion of L. monocytogenes remains completely uneventful, it can cause diarrhea (11) and some patients may experience severe diseases, including bacteremia, meningitis, and encephalitis (103). Immunocompromised patients, especially those with impaired cell-mediated immunity and pregnant women and their fetus are particularly at risk for acquiring severe listeriosis (103).

L. monocytogenes is a facultative intracellular bacterium that can multiply within the cytoplasm of many cells, including MPs, epithelial cells, fibroblasts, and hepatocytes (94, 100). In axenic medium, L. monocytogenes is highly susceptible to aminoglycosides, especially gentamicin (84). Moreover, these compounds display a bactericidal activity (104, 176). On the other hand, true in vitro synergism has been demonstrated for the combination of a penicillin with gentamicin (12, 51, 99, 105, 123, 176). When the activities of amoxicillin and gentamicin against intracellular L. monocytogenes (in HeLa cells) were determined after 24 h incubation of infected cultures with each antibiotic, amoxicillin displayed a significant intracellular killing effect at concentrations of ≥1 μg/ml, whereas gentamicin was ineffective (121). More recently, however, Drevets et al. (48) showed that cell lysates obtained from a W1C3 MP cell line preincubated for 3 days in the presence of various concentrations of gentamicin displayed listericidal activity. L. monocytogenes resides within MPs inside phagosomes rather than in the cytosol (47, 134). Since lucifer yellow is slowly accumulated by pinocytosis within L. monocytogenes-containing vacuoles, investigators have suggested that gentamicin could also be slowly concentrated by pinocytosis within these cell compartments after the 72-h incubation period, with subsequent intracellular listericidal activity. The combination of an amino-penicillin with gentamicin has been shown to be superior to monotherapy with the former drug in animal models (51, 122, 149).

Amino-penicillins remain the gold standard therapy for listeriosis (84), although a high dosage should be used (i.e., at least 6 g daily), especially in cases of meningitis or meningoencephalitis. Prolonged duration of the antibiotic therapy (i.e., at least 2 to 3 weeks) has been advocated to prevent relapses on antibiotic withdrawal (84). However, frequent clinical failures are still observed despite long-term therapy (84), which fits well with the demonstration that amino-penicillins are not able to eradicate L. monocytogenes from infected cell cultures in vitro (121). Based on experimental studies in animals (51, 122, 149), the combination of an amino-penicillin with gentamicin is now the preferred first-line therapy for listeriosis (84, 103), although the superiority of such a combination remains to be firmly established in humans. Thus, the role of an aminoglycoside in antibiotic therapy of listeriosis remains to be clearly defined but may represent another example of the potential intracellular activity of these compounds.

S. aureus and osteoarticular infections.

S. aureus, a ubiquitous pyogenic bacterium, has the ability to cause chronic infections, especially of bones and joints (41). S. aureus can survive within lysosomes of phagocytic cells (33, 95, 107, 141, 172), although the local acidic pH inhibits its intracellular multiplication. It was recognized early that its ability to survive within phagocytic cells, especially PMNs, may cause infection relapse despite appropriate antibiotic therapy (17, 101, 110).

Although aminoglycosides are highly bactericidal against S. aureus in vitro, these antibiotics are considered poorly effective or ineffective against intracellular staphylococci (27, 107). Lam and Mathison (102) first hypothesized that inhibition of S. aureus growth within the intracellular environment resulted in impaired activity of most antibiotics against such “resting” bacteria. We have evaluated in vitro the influence of slow cell uptake and acidity within the phagolysosome on activity of amikacin against intracellular S. aureus (117). P388 murine MP-like cells were preincubated in the presence of amikacin for 3 days before infection with S. aureus in order to allow slow accumulation of the drug within cells. Nonpreincubated cells served as controls. After infection of cells with an S. aureus inoculum, cultures were further incubated with amikacin alone or in combination with the lysosomotropic alkalinizing agents chloroquine and ammonium chloride. Measurements of phagolysosomal pH in S. aureus showed that lysosomotropic agents induced a significant vacuolar alkalization, from a pH of 4.8 up to 6.8. When using preloaded cells, amikacin induced a significant intracellular killing effect against S. aureus when combined with a lysosomotropic agent but not when used alone. No intracellular killing effect was demonstrated when using nonpreloaded cells.

These experiments clearly showed that amikacin displayed a bactericidal activity against intracellular S. aureus only when antibiotic-cell contact occurred for sufficient time to allow slow intracellular accumulation of the antibiotic and, additionally, when the S. aureus vacuolar pH was raised to a level which allowed restoration of its antimicrobial activity. Similar reasoning probably explains why rifampin, which both penetrates within eukaryotic cells (181, 183) and is more active at acidic pHs (186), displays a bactericidal activity against intracellular S. aureus (110). Rifampin is much safer than the combination of an aminoglycoside with a lysosomotropic agent for staphylococcus-related infections, although it should not be used alone because of the rapid emergence of resistant mutants. For example, this compound has been used successfully in combination with a fluoroquinolone in patients suffering from chronic S. aureus osteoarthritis (17, 31).

CONCLUSIONS

The current general feeling remains that aminoglycosides are poorly effective in treating infections due to intracellularly surviving pathogens, because of their poor ability to penetrate the eukaryotic cell membrane. In vitro intracellular pharmacokinetic studies have invalidated such a theory. Aminoglycosides actually concentrate within phagocytes. Their reduced activities against the intracellular forms of microorganisms may be related to their almost exclusive concentration within the lysosomal cell compartment and their inactivation because of local acidity, the activity of aminoglycosides being extremely pH susceptible. However, current in vitro models for the evaluation of the intracellular activities of aminoglycosides are poorly predictive of the clinical situation since, in many instances, these antibiotics are valuable therapeutic alternatives for infections due to intracellular pathogens, including the treatment of mycobacterial diseases, plague, brucellosis, tularemia, listeriosis, or bartonellosis. More reliable in vitro models should be elaborated to reassess the activities of aminoglycosides against intracellular microorganisms, in order to improve our knowledge of the mode of action of these antibiotics in infections due to these pathogens. These models should take into consideration the slow intracellular accumulation of aminoglycosides within eukaryotic cells.