Combating Multidrug-Resistant Pathogens with Host-Directed Nonantibiotic Therapeutics

ABSTRACT

Earlier, we reported that three Food and Drug Administration-approved drugs, trifluoperazine (TFP; an antipsychotic), amoxapine (AXPN; an antidepressant), and doxapram (DXP; a breathing stimulant), identified from an in vitro murine macrophage cytotoxicity screen, provided mice with 40 to 60% protection against pneumonic plague when administered at the time of infection for 1 to 3 days. In the present study, the therapeutic potential of these drugs against pneumonic plague in mice was further evaluated when they were administered at up to 48 h postinfection. While the efficacy of TFP was somewhat diminished as treatment was delayed to 24 h, the protection of mice with AXPN and DXP increased as treatment was progressively delayed to 24 h. At 48 h postinfection, these drugs provided the animals with significant protection (up to 100%) against challenge with the agent of pneumonic or bubonic plague when they were administered in combination with levofloxacin. Likewise, when they were used in combination with vancomycin, all three drugs provided mice with 80 to 100% protection from fatal oral Clostridium difficile infection when they were administered at 24 h postinfection. Furthermore, AXPN provided 40 to 60% protection against respiratory infection with Klebsiella pneumoniae when it was administered at the time of infection or at 24 h postinfection. Using the same in vitro cytotoxicity assay, we identified an additional 76/780 nonantibiotic drugs effective against K. pneumoniae. For Acinetobacter baumannii, 121 nonantibiotic drugs were identified to inhibit bacterium-induced cytotoxicity in murine macrophages. Of these 121 drugs, 13 inhibited the macrophage cytotoxicity induced by two additional multiple-antibiotic-resistant strains. Six of these drugs decreased the intracellular survival of all three A. baumannii strains in macrophages. These results provided further evidence of the broad applicability and utilization of drug repurposing screening to identify new therapeutics to combat multidrug-resistant pathogens of public health concern.

INTRODUCTION

Multiple-antibiotic-resistant (MAR) bacteria represent a growing and pressing public health concern of the 21st century (1–3). Worldwide, it is estimated that MAR pathogens lead to 700,000 deaths every year; however, the death toll is predicted to reach 10 million by the year 2050 in the absence of new interventions (4). Although the natural selection process in microorganisms can explain the emergence of resistance, the human contribution through the over- and unregulated use of antibiotics in both the food and health care industries cannot be overlooked as a major factor in the emergence of MAR pathogens (1, 5, 6). Further compounding the threat of antimicrobial resistance, the discovery and/or availability of new classes of antimicrobial agents has stalled, with only a few new antibiotics being approved for use in clinics in the past 2 decades (1, 7, 8).

As traditional drug discovery is both a highly inefficient and a costly process, alternative approaches to the identification of antimicrobial drugs are being sought (9–11). Previously, through the utilization of a drug repurposing approach with Food and Drug Administration (FDA)-approved therapeutics, we identified three drugs, doxapram (DXP; a breathing stimulant), amoxapine (AXPN; an antidepressant), and trifluoperazine (TFP; an antipsychotic), that mitigate fatal Yersinia pestis infection in a model of pneumonic plague (9). Y. pestis is classified as a reemerging pathogen by the World Health Organization (WHO) and a tier 1 select agent by the Centers for Disease Control and Prevention (CDC) (10, 11). Plague is responsible for over 200 million deaths worldwide and remains a prevalent public health threat, as it is endemic to many regions of the world, with recent cases being reported in Colorado (2015) and New Mexico (2017) (12, 13). Alarmingly, in a current ongoing epidemic in Madagascar, >1,200 cases of plague, with ∼65% of cases being the pneumonic form, and over 100 deaths have been reported (14). The pneumonic form of plague rapidly spreads from person to person, resulting in a high case fatality rate (12–14). Although plague is treatable with antibiotics and levofloxacin (Levo; Levaquin) and moxifloxacin (Avelox) were approved for use for the treatment of plague by the FDA in 2012 and 2015, respectively, these antimicrobials must be administered within 20 to 24 h after the onset of symptoms to be effective (11, 15–17). With the existence of MAR Y. pestis strains in nature as well as strains being genetically engineered for possible use as a bioweapon, the value of antibiotic treatment is further diminished (16, 18–20).

In addition to Y. pestis, we also earlier reported on the efficacy of TFP and AXPN in a murine model of Clostridium difficile infection when they were administered prophylactically (9). The incidence of C. difficile infections and the rate of isolation of MAR strains have increased substantially in the last decade, leading it to currently being classified as an urgent threat by the CDC (1). This bacterium causes life-threatening diarrhea and is associated with a significant number of hospitalizations and high rates of morbidity and mortality (21). Recurrence is a substantial problem, with 25% of treated patients experiencing recurrence and about 40 to 60% of patients with a first recurrence developing subsequent recurrences (22, 23). Additionally, other MAR pathogens of high public health importance include Acinetobacter baumannii and Klebsiella pneumoniae, which are currently classified as serious and urgent threats by the CDC, respectively (1). Both of these pathogens are associated with several hospital-acquired complications, including urinary tract infections (UTIs), bloodstream infections, wound infections, meningitis, the formation of liver abscesses and endophthalmitis, and ventilator-associated pneumonia (24–30).

Therefore, to further evaluate the effectiveness and broad applicability of TFP, AXPN, and DXP as therapeutics against bacterial pathogens, in the studies described here we utilized a delayed treatment approach in murine models of both pneumonic and bubonic plague and in mice orally infected with C. difficile spores. Additionally, we determined the efficacy of these drugs when they were administered in combination with a standard-of-care antibiotic for each model of infection. We also evaluated the efficacy of AXPN in a mouse model of K. pneumoniae respiratory infection. Finally, we optimized our previously reported macrophage cytotoxicity assay (9) to identify additional nonantibiotic drugs effective against K. pneumoniae and A. baumannii infections. The results presented in this study demonstrate the broad applicability of an in vitro screening technique, based on host cell cytotoxicity, to identify potential novel therapeutics with activity against a wide array of bacterial pathogens. Thus, the potential use of FDA-approved nonantibiotic drugs to combat infections caused by MAR pathogens could become a reality after performing further safety testing in a broader human population.

RESULTS

TFP, AXPN, and DXP exhibited therapeutic efficacy in a murine model of pneumonic plague.

Based on their previously reported efficacies against Y. pestis when administered at the time of infection (9), we evaluated the therapeutic potential of TFP, AXPN, and DXP following delayed treatment (6, 12, or 24 h postinfection [p.i.]) in a mouse model of pneumonic plague. For TFP, treatment at all time points p.i. significantly protected the animals from pneumonic plague, with 40 to 60% of mice surviving. However, it was noted that the efficacy of TFP diminished somewhat as the time to dosing was delayed (Fig. 1A). Likewise, delayed treatment with AXPN and DXP provided significant protection to animals across all time points (Fig. 1B and C). However, unlike in TFP-treated animals, for both AXPN and DXP, delayed treatment resulted in an increase in the level of drug efficacy, with the rates of survival of animals treated with either of these drugs at 24 h p.i. ranging from 70 to 100% (Fig. 1B and C).

FIG 1.

Survival analysis of Y. pestis CO92-infected mice treated with TFP, AXPN, or DXP at delayed time points after infection in a model of pneumonic plague. Mice were challenged by the i.n. route with 8 LD₅₀ (1 LD₅₀ = 500 CFU) of WT strain CO92 and administered TFP (1.5 mg/kg; n = 7 mice) (A), AXPN (3 mg/kg; n = 7 mice) (B), or DXP (20 mg/kg; n = 7 mice) (C) by the i.p. route at the indicated time points p.i. A group of untreated, infected animals (which were treated with PBS; n = 5 mice) served as a control. The animals were dosed with DXP and AXPN for 3 days (at 24-h intervals) or received only 1 dose of TFP at the indicated times. Mice were monitored for signs of morbidity and mortality for 21 days. The data were analyzed for significance by the use of Kaplan-Meier survival estimates. The P values were determined on the basis of a comparison of the results for each of the drug treatment groups to the results for the untreated and infected control group of mice (which were treated with PBS).

Treatment with TFP, AXPN, and DXP during the initial infection protected the animals upon rechallenge with Y. pestis CO92.

In order to evaluate whether a protective immune response was generated in treated animals that survived the initial infection, at 21 p.i. we rechallenged the surviving mice in the groups indicated in Fig. 1 (see also Fig. S1 in the supplemental material) by the intranasal (i.n.) route with wild-type (WT) strain CO92 (8 50% lethal doses [LD₅₀]) to monitor morbidity and mortality. For all groups of treated animals, some protection from subsequent challenge was observed, with survival rates ranging from 25 to 100%, while all naive control animals succumbed to the infection (Fig. S1).

Treatment with TFP, AXPN, and DXP in combination with levofloxacin extended the therapeutic window for intervention in a murine model of pneumonic plague.

The efficacy of antibiotic therapy wanes in a murine model of pneumonic plague when it is administered past 42 h p.i., even at high concentrations (15). As all three drugs significantly increased the rate of animal survival when they were administered at 24 h p.i. (Fig. 1), we evaluated the potential of these drugs to extend the therapeutic window through treatment in combination with levofloxacin. Both a subinhibitory concentration of levofloxacin (0.25 mg/kg of body weight) and a therapeutic dose (5 mg/kg) that was previously reported to be 100% effective when administered at up to 42 h p.i., but provided only limited efficacy when administered at 48 h p.i. (9, 15), were evaluated.

When TFP alone was administered to mice at 48 h p.i., there was no significant difference in the rate of survival compared to that of infected mice in the untreated (phosphate-buffered saline [PBS]-treated) control group as well as the group of infected mice treated with 0.25 mg/kg levofloxacin (Fig. 2A). However, when TFP was administered in combination with a low dose (0.25 mg/kg) or a high dose (5 mg/kg) of Levo, an increase in the survival rate (57 to 100%) in comparison to that for the untreated and infected control group was observed (Fig. 2A). Additionally, the group of infected mice treated with TFP-Levo (5 mg/kg) exhibited 100% survival, which was significantly higher than that in the groups of infected animals treated with TFP or levofloxacin alone (Fig. 2A).

FIG 2.

Survival analysis of Y. pestis CO92-infected mice treated with TFP, AXPN, or DXP alone or in combination with levofloxacin (Levo) at 48 h p.i. in a model of pneumonic plague. Mice were challenged by the i.n. route with 9 LD₅₀ of WT strain CO92 (n = 5 to 7 mice per group) and administered TFP (A), AXPN (B), or DXP (C) alone or in combination with levofloxacin (5 mg/kg or 0.25 mg/kg) by the i.p. route at the indicated time points p.i. The animals were dosed for up to 3 days (at 24-h intervals) beginning at 48 h p.i. (as described in the Fig. 1 legend). Untreated and infected mice (which were treated with PBS) served as the control group. Mice were monitored for signs of morbidity and mortality for 14 days. The data were analyzed for significance by the use of Kaplan-Meier survival estimates. The P values were determined on the basis of a comparison of the results for each of the drug treatment groups to the results for the untreated and infected control group (which was treated with PBS) or the indicated group.

AXPN treatment of the mice at 48 h p.i. resulted in a survival rate (72%) significantly higher than that for animals in the untreated control group, with all animals in the latter group succumbing to infection (Fig. 2B). However, an interaction between AXPN and levofloxacin was observed when they were administered together by the intraperitoneal (i.p.) route, resulting in a high mortality rate (data not shown).

All mice in the groups treated with DXP alone as well as DXP in combination with levofloxacin exhibited significantly higher survival rates (50 to 100%) than the controls (Fig. 2C). It was also noted that the group of mice treated with DXP-Levo (5 mg/kg) had a statistically significantly higher survival rate (100%) than the group of mice treated with DXP alone, further indicating an advantage of dual-drug (DXP and levofloxacin) treatment (Fig. 2C).

Following 14 days p.i., the surviving animals in the groups treated with single drugs were humanely euthanized and their lungs, livers, and spleens were collected for bacterial enumeration. For all surviving animals evaluated, no detectable bacteria were observed, indicating the clearance of Y. pestis from these animals (data not shown).

Treatment with TFP, AXPN, and DXP in combination with levofloxacin provided protection against fatal bubonic plague and extended the mean time to death.

We next evaluated the efficacies of these drugs in a murine model of bubonic plague. For these studies, the drugs alone, as well as in combination with levofloxacin (5 mg/kg), were administered to animals by the i.p. route at 48 h p.i. As an interaction between AXPN and levofloxacin was observed in the model of pneumonic plague, for these studies the dosing of each drug was staggered, with levofloxacin being administered 3 h after AXPN treatment. For the groups treated with one drug alone, only the group treated with levofloxacin exhibited an increased mean time to death compared to that for untreated and infected control animals as well as a survival rate statistically significantly higher than that for the untreated and infected control animals, with all animals treated with TFP, DXP, and AXPN alone succumbing to infection (Fig. 3). Combination therapy with TFP, DXP, or AXPN and levofloxacin, however, did result in both an increase in the mean time to death and significantly higher survival rates (42 to 85%) in comparison to those for the control group of animals (Fig. 3). Furthermore, the infected mice in the group treated with the DXP-Levo combination also exhibited a significantly higher survival rate than the infected mice in the group treated with Levo alone (Fig. 3). Following 24 days p.i., no detectable bacteria were enumerated from the lungs, livers, or spleens of the surviving animals, indicating the clearance of Y. pestis from these animals (data not shown).

FIG 3.

Survival analysis of Y. pestis CO92-infected mice treated with TFP, AXPN, or DXP alone or in combination with levofloxacin (Levo) at 48 h p.i. in a model of bubonic plague. Mice were challenged by the s.c. route with 12 LD₅₀ (1 LD₅₀ = 50 CFU) of WT strain CO92 (n = 5 to 7 per group) and administered TFP, AXPN, or DXP alone or in combination with levofloxacin (5 mg/kg) by the i.p. route at the indicated staggered time points p.i. (levofloxacin was administered 3 h after treatment with the other drugs). The animals were dosed for up to 3 days (at 24-h intervals) beginning at 48 h p.i. (as described in the Fig. 1 legend). Untreated and infected mice (which were treated with PBS) served as the control group. Mice were monitored for signs of morbidity and mortality for 24 days. The data were analyzed for significance by the use of Kaplan-Meier survival estimates. The P values were determined on the basis of a comparison of the results for each of the drug treatment groups to the results for the untreated and infected control group (which was treated with PBS) or the indicated group.

Treatment with AXPN or DXP did not alter the serum resistance of Y. pestis CO92.

We previously reported no direct bactericidal or bacteriostatic effects of these drugs (TFP, AXPN, and DXP) on Y. pestis CO92 in vitro. However, in vivo, these drugs can be metabolized to various products (31, 32), which may be directly bactericidal in nature. Additionally, these drugs or their metabolites may activate complement, which would lead to increased bacterial killing in vivo. Therefore, to determine whether drug metabolites or activation of complement was linked to the efficacy of these drugs in in vivo models of infection, we administered AXPN and DXP to uninfected mice and collected serum at 6 h after drug administration, with a total of three doses of each drug being administered to the animals. We limited these studies to AXPN and DXP, as an increase in efficacy was observed for both of these drugs as the treatment time was further delayed up to 24 h p.i. with WT strain CO92 (Fig. 1). As observed in Fig. S2, our data showed no differences in bacterial killing between the serum from drug-treated mice and that from naive mice, with the bacteria in all samples exhibiting ≥100% survival, indicating bacterial growth.

Treatment with AXPN or DXP exhibited a limited effect on cytokine/chemokine production in murine lungs.

To determine whether AXPN or DXP treatment alone alters cytokine/chemokine levels in the lungs, each drug was administered to mice at 24-h intervals a total of 3 times. At 6 h after each drug administration, the lungs of the animals were collected for cytokine/chemokine analysis. Again, we limited these studies to AXPN and DXP. Out of 23 cytokines/chemokines analyzed, only 4 exhibited levels in treated animals significantly altered from those in the naive control animals (Fig. S3). In AXPN-treated animals, the levels of interleukin-1α (IL-1α; after the 3rd dose), macrophage inflammatory protein 1α (MIP-1α; after the 2nd dose), and RANTES (regulated on activation, normal T cell expressed and secreted; after the 1st dose) were increased in comparison to those in the naive animals (Fig. S3A, C, and D). In DXP-treated animals, the levels of IL-1α (after the 3rd dose) and macrophage chemoattractant protein 1 (MCP-1; after the 1st dose) were significantly increased in comparison to those in naive mice (Fig. S3A and B).

Treatment with DXP inhibited Y. pestis CO92 proliferation and altered the cytokine/chemokine response in the lungs of mice.

As DXP was the most efficacious drug administered alone at 24 h p.i. in a model of pneumonic plague, we evaluated the proliferation of WT strain CO92 in the lungs of infected and treated animals. At 30, 54, and 78 h p.i. (6 h after each dose), we noted that the number of CFU in DXP-treated animals was reduced in comparison to that in untreated animals, although statistical significance was observed only at 30 and 78 h p.i. (Fig. 4). It was also observed that by day 10, no bacteria could be detected in the lungs of DXP-treated and infected mice (Fig. 4).

FIG 4.

Y. pestis CO92 proliferation in lungs following DXP treatment in a model of pneumonic plague. At each indicated time point, three animals per group that had been infected with 8 LD₅₀ of WT strain CO92 by the i.n. route and treated with DXP at 24 h p.i. (DXP-24) or left untreated (but treated with PBS) were euthanized, and the lungs were harvested 6 h after the administration of each dose. The numbers of CFU were determined by homogenization of the lungs, followed by serial dilution and plating on SBA plates. Statistical analysis was performed by Student's t test, with a P value of <0.05 being considered statistically significant. d, day.

Next, we evaluated the changes in cytokine/chemokine levels in DXP-treated and untreated (PBS-treated) mice in a model of pneumonic plague. In general, the levels of cytokines/chemokines observed at 30 h p.i. were equivalent between DXP-treated and untreated (PBS-treated) Y. pestis-infected animals (Table 1). Similarly, at 54 h p.i., the levels of the cytokines/chemokines tested were generally equivalent between the two groups, with the exception of the levels of tumor necrosis factor alpha (TNF-α), IL-2, IL-6, and MCP-1 (Table 1). At this time point, the levels of TNF-α and IL-6 observed in DXP-treated and infected animals were significantly decreased compared to those seen in untreated animals, while the levels of both IL-2 and MCP-1 noted in DXP-treated and infected mice were significantly increased compared to those seen in untreated animals. However, by 78 h p.i., there was a general and significant decrease in the cytokine/chemokine levels in the lungs of DXP-treated and infected animals compared to those in the lungs of untreated mice, with the exception of the levels of IL-2, which maintained a significantly higher level of expression in DXP-treated and infected animals than in untreated and infected animals (Table 1). At 10 days p.i., while the levels of most cytokines/chemokines remained low, the levels of IL-2, IL-9, and IL-12(p70) remained high in the lungs of DXP-treated animals (Table 1).

TABLE 1.

Changes in cytokine/chemokine levels in lungs of mice infected with WT strain Y. pestis CO92 and either treated with DXP or left untreated (treated with PBS)

Cytokine or chemokinea	Concn (pg/ml) atb:
	30 h		54 h		78 h		10 days, DXP
	PBS	DXP	PBS	DXP	PBS	DXP
TNF-α	116 ± 11	104 ± 23	298 ± 28	155 ± 88*	351 ± 62	111 ± 3*	137 ± 5
IFN-γ	113 ± 4	69 ± 14	194 ± 15	320 ± 32	739 ± 25	70 ± 14*	107 ± 6
G-CSF	122 ± 25	154 ± 37	2,063 ± 277	2,502 ± 169	20,603 ± 220	142 ± 45*	214 ± 42
GM-CSF	63 ± 9	45 ± 10	214 ± 33	257 ± 26	2,086 ± 71	38 ± 4*	132 ± 27
IL-1α	261 ± 31	293 ± 30	2,221 ± 275	1,381 ± 157	5,265 ± 202	305 ± 40*	255 ± 4
IL-1β	67 ± 11	45 ± 4	745 ± 98	499 ± 56	1,518 ± 582	51 ± 10*	38 ± 2
IL-2	143 ± 3	110 ± 11	108 ± 3	143 ± 16**	108 ± 5	134 ± 10**	140 ± 6
IL-3	99 ± 20	76 ± 21	75 ± 11	96 ± 37	141 ± 37	48 ± 15*	50 ± 4
IL-4	57 ± 2	51 ± 8	53 ± 7	57 ± 5	78 ± 9	36 ± 36*	40 ± 8
IL-5	33 ± 2	28 ± 1	35 ± 6	39 ± 12	52 ± 4	36 ± 5	31 ± 1
IL-6	236 ± 148	211 ± 72	7,079 ± 985	1,678 ± 516*	21,456 ± 2,839	137 ± 7*	147 ± 8
IL-9	101 ± 7	82 ± 7	92 ± 15	104 ± 9	131 ± 4	70 ± 7*	140 ± 12
IL-10	25 ± 4	38 ± 2	47 ± 8	45 ± 4	86 ± 9	35 ± 3*	58 ± 2
IL-12(p40)	285 ± 67	558 ± 194	1,196 ± 1,007	970 ± 750	3,620 ± 388	422 ± 104*	297 ± 37
IL-12(p70)	118 ± 19	93 ± 23	118 ± 39	136 ± 19	212 ± 37	65 ± 13*	165 ± 17
IL-13	56 ± 6	29 ± 7	45 ± 8	52 ± 14	60 ± 5	48 ± 11	56 ± 3
IL-17	83 ± 9	64 ± 8	110 ± 52	275 ± 236	1,166 ± 308	73 ± 9*	81 ± 2
Eotaxin	652 ± 91	800 ± 116	1,646 ± 395	1,803 ± 554	1,481 ± 188	1,524 ± 462	1,600 ± 7
KC	171 ± 49	122 ± 15	2,892 ± 381	4,393 ± 297	14,329 ± 971	318 ± 24*	209 ± 4
MCP-1	48 ± 4	50 ± 5	179 ± 19	1,765 ± 215**	12,395 ± 309	55 ± 8*	43 ± 1
MIP-1α	141 ± 15	191 ± 63	1,266 ± 158	783 ± 60	8,423 ± 845	102 ± 10*	212 ± 30
MIP-1β	152 ± 8	127 ± 8	131 ± 19	134 ± 20	238 ± 22	115 ± 10*	115 ± 3
RANTES	342 ± 7	431 ± 80	732 ± 47	656 ± 34	2,625 ± 337	479 ± 75*	467 ± 23

^aIFN-γ, gamma interferon; G-CSF, granulocyte colony-stimulating factor; GM-CSF, granulocyte-macrophage colony-stimulating factor; KC, keratinocyte-derived chemokine. The other abbreviations are defined in the text.

^bThe data were analyzed using Student's t test, with comparisons being made between the PBS-treated control and DXP-treated groups for each time point. *, P ≤ 0.05 (significantly lower than the value for the PBS-treated group); **, P ≤ 0.05 (significantly greater than the value for the PBS-treated group).

Combinatorial treatment with TFP, AXPN, or DXP and vancomycin provided protection against fatal C. difficile infection in mice.

While we have previously demonstrated that TFP and AXPN have efficacy in a mouse model of C. difficile infection when they are administered prophylactically (9), we wanted to evaluate the efficacies of all three drugs in combination with a current therapeutic, the antibiotic vancomycin. For these studies, one of the three drugs was administered in combination with vancomycin (20 mg/kg, which is a subinhibitory concentration) by the i.p. route at 24 h p.i. As in our model of bubonic plague, combinatorial dosing with AXPN and vancomycin was staggered, with vancomycin administration occurring 3 h after AXPN treatment. For animals treated with vancomycin alone, only a 20% survival rate was observed. However, combination therapy with TFP, DXP, or AXPN and vancomycin resulted in significantly higher survival rates in comparison to those for the control untreated (PBS-treated) C. difficile-infected animals as well as the group of infected animals treated with vancomycin alone (Fig. 5A). It was also noted that animals in the combination treatment groups exhibited limited symptoms of infection (i.e., weight loss, diarrhea, lethargy), whereas infected animals that were left untreated or treated with vancomycin alone did experience these symptoms.

FIG 5.

Survival analysis of C. difficile- or K. pneumoniae-infected and treated mice. (A) For the model of C. difficile infection, C57BL/6 mice were infected by oral gavage with 10⁵ C. difficile (VPI 10463) spores (n = 5 to 7 per group) and administered TFP, AXPN, or DXP alone or in combination with vancomycin (Vanco; 20 mg/kg) by the i.p. route at 24 h p.i. (vancomycin was administered 3 h after treatment with the other drugs). The drugs were administered at 24-h intervals, with TFP being administered once, AXPN and DXP being administered 3 times, and vancomycin being administered 5 times. The mice were monitored for signs of morbidity and mortality for 8 days. The data were analyzed for significance by the use of Kaplan-Meier survival estimates. P values were determined on the basis of a comparison of the results for each of the drug treatment groups to the results for the group treated with vancomycin alone. P was 0.009 for all groups treated with drug combinations in comparison to the untreated and infected control group (which was treated with PBS). (B) For K. pneumoniae infection, C57BL/6 mice were infected with 5 × 10⁵ CFU by the i.n. route and administered AXPN (3 mg/kg) up to 3 times by the i.p. route at the indicated time points at 24-h intervals. Untreated and infected animals (which were treated with PBS) served as a control. The data were analyzed by the chi-square test, with P values being based on a comparison of the results for each drug treatment group to the results for the untreated and infected group (which was treated with PBS).

Treatment with AXPN provided protection against fatal K. pneumoniae infection.

As AXPN provided the highest level of protection when it was administered alone at 48 h p.i. in a model of pneumonic plague (Fig. 2), we evaluated this drug's efficacy as protection against K. pneumoniae in a mouse model of K. pneumoniae respiratory infection. It was observed that following administration of the drug both at the time of infection and at 24 h p.i., the survival of animals was increased (40 to 60%) compared to for the PBS-treated control group of mice (Fig. 5B).

Identification of additional FDA-approved drugs that were effective at inhibiting host cell cytotoxicity during infection with K. pneumoniae or A. baumannii strains.

We utilized our previously reported drug repurposing screening technique (9) to identify additional novel therapeutics with activity against K. pneumoniae or A. baumannii. For these studies, carbapenem- and gentamicin-sensitive bacterial strains were chosen to both optimize the screen and narrow the number of drugs to be tested against MAR strains. For the K. pneumoniae 244 strain, 76 out of 780 nonantibiotic drugs were identified to decrease the infection-induced cytotoxicity of the bacterium in murine macrophages, with 29 being classified as tier 1 (Fig. S4A; Table S1) and 47 being classified as tier 2 (Fig. S4B and C; Table S2). Interestingly, 8 drugs were commonly identified to be active against K. pneumoniae and Y. pestis infections, although the tier groupings did differ (Table 2). For A. baumannii strain 8, 121 out of 780 nonantibiotic drugs were identified to decrease infection-induced host cell cytotoxicity, with 51 being classified as tier 1 (Fig. S5; Table S3) and the remaining 70 being classified as tier 2 (Fig. S6; Table S4). Overall, the drugs identified from screens with either pathogen belonged to various classes, including antiprotozoals, antineoplastics, anti-inflammatory agents, antihypertensives, antivirals, anticonvulsants, antipsychotics, antidepressants, and hormone receptor modulators (Tables S1 to S4).

TABLE 2.

Common drugs identified in screens with K. pneumoniae and Y. pestis

Drug name	Therapeutic class	Tier for:
		K. pneumoniae	Y. pestisa
Mometasone furoate	Anti-inflammatory, antiallergic	1	2
Apomorphine · HCl hemihydrate	Nonselective dopamine agonist and antiparkinsonian agent	2	1
Promethazine · HCl	Antiallergic, sedative	2	1
Betamethasone	Anti-inflammatory, immunosuppressive	2	1b
Etonogestrel	Hormonal contraceptive	2	2
Terconazole	Antifungal	2	2b
Ethinyl estradiol	Estrogen, hormonal replacement, antineoplastic	2	1
Cisatracurium besylate	Neuromuscular blocking agent	2	1b

^aDetermined from reference 9.

^bIdentified in a pretreatment screen with Y. pestis.

Subsequently, the 121 nonantibiotic drugs were tested against MAR A. baumannii strains BAA-1797 and BAA-1799. For the A. baumannii BAA-1797 and BAA-1799 strains, we identified 30 and 20 drugs, respectively, which reduced the bacterium-induced cytotoxicity in murine macrophages (Fig. 6 and 7). As these strains are antibiotic resistant, a higher concentration (500 μg/ml) of gentamicin (which has no effect on intracellular bacterial survival) was used for the assay. In total, 13 drugs significantly reduced the bacterium-induced cytotoxicity in murine macrophages caused by all three A. baumannii strains, although the tier assignment of each drug did differ among the strains (Table 3). It was also noted that three of these drugs, mifepristone, bromocriptine mesylate, and zafirlukast, were also identified in our screen of drugs with activity against Y. pestis (Table 3).

FIG 6.

Drugs identified posttreatment to be tier 1 or tier 2 drugs using a host cell-based screen to evaluate RAW 264.7 murine macrophage viability by the MTT assay following infection with A. baumannii strain BAA-1797 for 18 h. For tier 1 drugs, the drug-treated (33 μM) and infected macrophages exhibited viability equivalent to that of uninfected macrophages (no bacteria, dashed line) and viability significantly greater than that of drug-untreated and infected macrophages. *, a tier 1 drug. For tier 2 drugs, the drug-treated and infected macrophages exhibited viability values that were significantly greater than those for A. baumannii-infected and drug-untreated macrophages but that were not equivalent to those for uninfected macrophages (dashed line). Two independent experiments in which each drug was tested in duplicate were performed. The data were plotted as percentages of the values for uninfected macrophages and were analyzed by 2-way ANOVA with Tukey's post hoc test. All data presented were significant.

FIG 7.

Drugs identified posttreatment to be tier 1 or tier 2 drugs using a host cell-based screen to evaluate RAW 264.7 murine macrophage viability by the MTT assay following infection with A. baumannii strain BAA-1799 for 18 h. (A) For tier 1 drugs, the drug-treated (33 μM) and infected macrophages exhibited viability equivalent to that of uninfected macrophages (no bacteria, dashed line) and viability significantly greater than that of untreated and infected macrophages. (B) For tier 2 drugs, the drug-treated and infected macrophages exhibited viability values that were significantly greater than those for A. baumannii-infected and untreated macrophages but that were not equivalent to those for uninfected macrophages (dashed line). Two independent experiments in which each drug was tested in duplicate were performed. The data were plotted as percentages of the values for uninfected macrophages and were analyzed by 2-way ANOVA with Tukey's post hoc test. All data presented were significant.

TABLE 3.

Common drugs with efficacy against all 3 A. baumannii strains screened and status in screens with Y. pestis

Drug name	Therapeutic class	Tier for:
		A. baumannii clinical isolate 8	A. baumannii BAA-1797	A. baumannii BAA-1799	Y. pestis CO92
Telmisartan	Antihypertensive	1	2	2	NAa
Mifepristone	Contraceptive	1	1	1	2
Primaquine phosphate	Antimalarial, antiprotozoal	2	2	2	NA
Loteprednol etabonate	Anti-inflammatory, antiallergy	1	2	2	NA
Repaglinide	Antihyperglycemic	1	2	2	NA
Bromocriptine mesylate	Antiparkinsonian, antidyskinetic	1	1	1	2
Calcitriol	Antihypocalcemic, antihypoparathyroid, vitamin (vitamin D), bone density conservation	1	2	2	NA
Sildenafil citrate	Erectile dysfunction, vasodilator, pulmonary, arterial hypertension treatment	2	1	1	NA
Candesartan	Antihypertensive	2	2	1	NA
Calcipotriene	Dermatologic, antipsoriatic	1	2	1	NA
Zafirlukast	Antiasthmatic	1	2	1	1
Rivastigmine tartrate	Neuroprotective, dementia treatment	2	2	1	NA
Ergotamine tartrate	Vasoconstrictor	1	2	1	NA

^aNA, not applicable.

Thirteen drugs exhibited no bactericidal or bacteriostatic effects on any of the three A. baumannii strains evaluated.

None of these 13 drugs (at the 33 μM concentration used in the in vitro screening assay) (9) showed any bactericidal or bacteriostatic effect on A. baumannii strains, namely, strains 8, BAA-1797, and BAA-1799, when they were tested for 24 h. The MICs of these 13 drugs for each A. baumannii strain were also analyzed and observed to be >200 μg/ml.

Quantification of intracellular A. baumannii survival in macrophages.

Focusing on the 13 drugs identified to be efficacious against all three A. baumannii strains, we tested the effects of the drugs at a concentration of 33 μM on intracellular bacterial survival in RAW 264.7 macrophages. Eight of the 13 drugs affected the bacterial intracellular survival of at least 1 strain, and 6 of these 8 drugs reduced the bacterial intracellular survival of all three strains: primaquine phosphate, loteprednol etabonate, bromocriptine mesylate, sildenafil citrate, calcipotriene, and zafirlukast (Fig. 8). None of these 13 drugs alone induced any cytotoxicity in macrophages (data not shown).

FIG 8.

Quantification of intracellular survival of A. baumannii strain 8, BAA-1797 (97), or BAA-1799 (99) in RAW 264.7 murine macrophages. RAW 264.7 macrophages were infected with one of the A. baumannii strains for 2 h at an MOI of 100. The cell monolayers were then treated with gentamicin at a concentration of 200 to 500 μg/ml for 2 h, washed twice with PBS, and incubated with the same concentration of gentamicin. At this time, each drug at a concentration of 33 μM or PBS as a control was added to the cell monolayers and the cells were incubated for 4 h. At this point, the macrophages were lysed and the number of CFU was determined by serial dilution and plating on SBA plates. Two independent experiments in which each drug was tested in duplicate were performed. The data were analyzed by comparison of the results for untreated and infected controls (which were treated with PBS) to those for the treatment groups by using a one-way ANOVA with Tukey's post hoc test. *, P < 0.01.

DISCUSSION

In efforts to identify alternative therapeutics to combat MAR pathogens, we evaluated the efficacies of three drugs (TFP, AXPN, and DXP) as treatment options. These three drugs were previously characterized to provide protection in mice when administered immediately following Y. pestis infection in a model of pneumonic plague (9). Our earlier data also indicated that these drugs target host function and not bacterial growth or virulence determinants (9). Since the drugs mentioned above modulate the function of the central nervous system (CNS) to achieve clinical efficacy in patients, it is possible that common drug-induced neuroimmune signals are broadly protective against infectious diseases. It is also important to note that the doses used in mice in these studies were much lower than the human-equivalent doses required to achieve clinical efficacy, and, thus, an allometrically scaled dose in humans should be effective with no side effects (33, 34).

In pneumonic plague, for which the therapeutic window in both humans and rodents is narrow (25, 26), all three drugs were effective when administered at up to 24 h p.i. (Fig. 1). Remarkably, this therapeutic intervention window could be extended by combinatorial drug treatment with TFP or DXP and levofloxacin, representing a significant advancement in the development of new modalities to treat such a deadly disease (Fig. 2). For AXPN, however, an adverse interaction was observed during coadministration with levofloxacin, resulting in the rapid deaths of the animals. Therefore, our further studies staggered the administration of AXPN in combination with an antibiotic in an attempt to alleviate the adverse effects observed.

While pneumonic plague is the most fatal and difficult-to-treat form of plague, the majority of plague cases currently reported globally are of the bubonic form, and, therefore, the evaluation of potential therapeutics to treat this form of the disease is imperative (20, 21). Unlike what we observed in the model of pneumonic plague, none of the drugs provided any protection from fatal infection in mice when they were administered alone at 48 h p.i. (Fig. 3). However, dosing of each drug in combination with levofloxacin provided significant protection to all groups according to the increased survival compared to that of WT strain CO92-infected and untreated control animals (Fig. 3). Similarly, in C. difficile-infected animals, dosing of all three drugs in combination with vancomycin provided statistically significant protection according to the increased survival compared to that of both C. difficile-infected and untreated controls and animals treated with vancomycin alone (Fig. 5A). Finally, the protective effect of AXPN in a model of K. pneumoniae respiratory infection further showed its effectiveness in combating bacterial pneumonias (Fig. 5B). Importantly, the clinical symptoms of the diseases in the animals significantly improved after drug treatment. Together, these data provide evidence of the enhanced efficacy of these drugs when they are administered postinfection in combination with antibiotics when neither the drugs nor the antibiotics alone were protective. Further, our data suggest that the protective efficacies of these drugs are dependent upon the mechanism(s) of disease pathogenesis, with involvement of the lungs during pneumonic plague and respiratory infection caused by K. pneumoniae, targeting of the lymphatic system during bubonic plague, and an attack on the gastrointestinal system during C. difficile infection (10, 35, 36).

While we hypothesize that these drugs activate host defenses to aid with the clearance of the bacteria, the true mechanism(s) of action of these drugs is unknown. Activation of neither the complement system nor the metabolites of DXP (ketodoxapram) and/or AXPN (8-hydroxyamoxapine and 7-hydroxyamoxapoine) contributed to the enhanced killing of Y. pestis (see Fig. S2 in the supplemental material). Further assessment of the cytokine/chemokine levels in the lungs revealed increased levels of IL-1α, MIP-1α, and RANTES in AXPN-treated animals and increased IL-1α and MCP-1 levels in DXP-treated animals (Fig. S3A and B), which suggests the activation and recruitment of macrophages, neutrophils, and T cells to the lung environment. This enhanced recruitment of leukocytes would aid with the rapid clearance and resolution of infection, as seen in our infection models.

In terms of the changes in cytokine/chemokine levels in Y. pestis-infected animals, their levels were generally comparable between DXP-treated and untreated infected animals at both 30 and 54 h p.i. (Table 1). However, at 78 h p.i., DXP-treated animals exhibited a dramatic reduction in cytokine/chemokine levels, possibly related to a reduction in the bacterial load (Fig. 4), while the levels in untreated and infected animals continued to increase (Table 1). At 54 h p.i., a reduction in the levels of TNF-α and IL-6, both of which are proinflammatory cytokines upregulated during infection (35), by DXP (Table 1) suggests the development of a more controlled inflammatory response that is normally absent in later stages of pneumonic plague (35, 37, 38). Additionally, MCP-1 and IL-2 levels were significantly altered in DXP-treated and infected animals, with both exhibiting increased levels in the lungs at 54 p.i. (Table 1). While IL-2 is mainly involved in T cell differentiation and the establishment of enduring cell-mediated immunity (39, 40), MCP-1 plays several physiological and pathophysiological roles. In terms of bacterial pneumonias, the absence of MCP-1 has been linked to an increased bacterial burden and decreased neutrophil influx in models of Escherichia coli and K. pneumoniae infection (41, 42). This early increase in MCP-1 levels in DXP-treated animals may aid with the clearance of bacteria (Fig. 4). Furthermore, while most cytokine/chemokine levels remained low by day 10 p.i., the levels of IL-2, IL-9, and IL-12(p70) remained high in the lungs of DXP-treated and infected animals (Table 1). All three of these cytokines play a key role in T cell differentiation and/or T cell proliferation (40, 43–45), and the prolonged upregulation of these cytokines provides evidence of the activation of a targeted cell-mediated immune response. It is important to note that, given the proinflammatory roles of these mediators, there is a risk of overactivation of the inflammatory pathways. However, the clearance of bacteria from drug-treated animals supports the notion that a controlled immune response was mounted without the exacerbation of inflammatory pathways, which normally occurs during Y. pestis infection.

While antibiotic-resistant strains of Y. pestis, specifically, strains resistant to the antimicrobials used for the treatment of plague, such as chloramphenicol, streptomycin, and doxycycline, have been isolated from patients (19, 20), infections with MAR strains of K. pneumoniae and A. baumannii have recently expanded dramatically worldwide (1, 27). Infections caused by K. pneumoniae, such as UTIs, are becoming unmanageable, and cases of bacteremia and pneumonia caused by MAR K. pneumoniae strains are increasingly life-threatening (1, 26, 27). Similarly, the increasing rates of resistance of A. baumannii to antimicrobials, in part due to its versatile genetic machinery that allows the quick evolution of resistance, have made UTIs and wound infections caused by this pathogen unmanageable in hospital settings (24, 46). Therefore, to combat these increasingly resistant pathogens, we identified several nonantibiotic drugs limiting bacterium-induced cytotoxicity in macrophages, with 76 drugs with activity against K. pneumoniae and 121 drugs with activity against A. baumannii being identified (Fig. S4 and S5; Tables S1 to S4).

The use of host-directed therapeutics for the treatment of infections has continued to garner much attention (47–49). As such, several of the drugs identified to be effective against K. pneumoniae or A. baumannii infections in vitro have also been reported to exhibit efficacy against other pathogens. For example, three drugs, mifepristone (a contraceptive), bromocriptine mesylate (an antiparkinsonian and antidyskinetic), and zafirlukast (an antiasthmatic), reduced Y. pestis-induced macrophage cytotoxicity (Table 2) (9). Mifepristone and bromocriptine mesylate also blocked the intracellular survival of Coxiella burnetii and Legionella pneumophila in macrophages (47). Recent studies have identified bromocriptine mesylate as having antiviral activity against both Zika virus and dengue virus (50, 51), while zafirlukast has been observed to have antibacterial activity against the oral pathogens Streptococcus mutants and Porphyromonas gingivalis, as well as against Mycobacterium tuberculosis (52, 53). However, as we observed no direct antibacterial effects of zafirlukast against A. baumannii or Y. pestis (9), the drug's mechanism of action against these pathogens requires further investigation. In addition to these drugs, telmisartan (an antihypertensive) has been identified to inhibit the intracellular growth of L. pneumophila, while ergotamine tartrate (a vasoconstrictor) reduced the intracellular growth of C. burnetii in macrophages (47), further implicating the potential broad applicability of these drugs against several human pathogens.

From our screens against A. baumannii, calcitriol and calcipotriene, both of which are vitamin D supplements, were identified to control infection in macrophages. Interestingly, vitamin D deficiency has been linked to the development of A. baumannii infections and is a predictor of mortality in critically ill patients (54, 55). In terms of its mechanism of action, vitamin D has been reported to induce the antimicrobial peptide LL-37, which can effectively kill both Gram-positive and Gram-negative pathogens as well as promote bacterial phagocytosis and killing by macrophages (56). The upregulation of LL-37 by vitamin D supplementation may boost the innate immune response to better control infections and could explain the results that we observed in our screens against A. baumannii, although further studies are required to fully delineate this mechanism of action.

The results from these studies provide further evidence of the utility of in vitro screening techniques for the identification of novel host-directed therapeutics to treat infections caused by a variety of pathogens. In addition to identifying three nonantibiotic drugs with therapeutic efficacy against Y. pestis, C. difficile, and K. pneumoniae, we have identified several drugs for further testing in in vivo models of infection against MAR K. pneumoniae and A. baumannii. Although none of the three drugs, i.e., TFP, AXPN, and DXP, was identified in in vitro screens to have activity against these two pathogens, AXPN did indeed show efficacy in a model of K. pneumoniae respiratory infection (Fig. 5B). These data indicate some limitations of the in vitro screen, and, therefore, we plan to test the in vivo protective effects of all three drugs in models of K. pneumoniae and A. baumannii respiratory infections to further evaluate their potential broad applicability.

Finally, we acknowledge that although the evaluation of FDA-approved therapeutics can lead to efficient and cost-effective means of discovery of novel approaches to combat MAR bacterial pathogens, it is critical that the use of these drugs be thoroughly assessed in a broader human population. Additionally, in considering the potential combination of these drugs with known antibiotics, the drug-drug interaction and its impact on pharmacokinetics/pharmacodynamics, absorption, distribution, metabolism, and excretion (ADME) and safety/tolerability will need to be investigated before their use is allowed in clinical practice. However, it is equally important to recognize that for major epidemics or public health threats, a certain level of safety/tolerability risk from the use of such combinations would be more acceptable, particularly when treatment options are limited, as was noted during the Ebola outbreak in 2014, than when infections present less of a danger.

MATERIALS AND METHODS

Bacterial strains and cell culture.

A highly virulent Y. pestis CO92 strain was obtained from the Biodefense and Emerging Infections (BEI) Resources Repository, Manassas, VA. Y. pestis was grown in heart infusion broth (HIB; Difco, Voigt Global Distribution, Inc., Lawrence, KS) at 28 or 37°C with constant shaking at 180 rpm or was grown for 48 h on 5% sheep blood agar (SBA) plates (Teknova, Hollister, CA). All experiments with Y. pestis were performed in a CDC-approved select agent laboratory in the Galveston National Laboratory (GNL), University of Texas Medical Branch (UTMB).

The C. difficile VPI 10463 strain was obtained from the American Type Culture Collection (ATCC; Manassas, VA). The organism was grown anaerobically in a cooked meat medium (Fluka, St. Louis, MO) as previously described (57, 58). A gentamicin- and carbapenem-sensitive K. pneumoniae clinical strain, designated strain 244, and an A. baumannii strain of human origin, designated strain 8, were obtained from UTMB. K. pneumoniae strain 43816 and MAR A. baumannii strains BAA-1797 and BAA-1799 were from patients (ATCC). These organisms were grown in Luria-Bertani (LB) broth at 37°C with constant shaking at 180 rpm or grown for 24 h on SBA plates.

Cells of the RAW 264.7 murine macrophage cell line (ATCC) were maintained in Dulbecco's modified Eagle medium (DMEM) with 10% fetal bovine serum supplemented with 1% l-glutamine (Cellgro; Mediatech, Inc., a Corning subsidiary, Manassas, VA) and 1% penicillin-streptomycin (Invitrogen, Carlsbad, CA) at 37°C with 5% CO₂.

Reagents.

The compounds in the Screen-Well FDA-approved drug library V2, consisting of 780 compounds, were provided as 10 mM stock solutions in dimethyl sulfoxide (BML-2843-0100; Enzo Life Sciences, Albany, NY). Sterile, injectable formulations of levofloxacin and doxapram (DXP) were purchased from the UTMB Pharmacy, while vancomycin was purchased from Hospira (Lake Forest, IL). Amoxapine (AXPN) and trifluoperazine (TFP) were purchased from Sigma-Aldrich (St. Louis, MO) in a dry powder form. 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) and detergent solution were purchased from ATCC.

Testing of TFP, AXPN, and DXP as therapeutics alone or in combination with levofloxacin in murine models of pneumonic and bubonic plague.

All animal studies with Y. pestis were performed in an animal biosafety level 3 (ABSL-3) facility under a protocol approved by the UTMB Institutional Animal Care and Use Committee (IACUC). Six- to 8-week-old female Swiss Webster mice (weight, 17 to 20g), purchased from Taconic Laboratories (Germantown, NY), were anesthetized by isoflurane inhalation and subsequently challenged intranasally (i.n.; model of pneumonic plague) with 8 to 9 LD₅₀ (1 LD₅₀ = 500 CFU) of wild-type (WT) strain Y. pestis CO92 or subcutaneously (s.c.; model of bubonic plague) with 12 LD₅₀ (1 LD₅₀ = 50 CFU) of WT strain CO92 (59). At 0, 6, 12, 24, or 48 postinfection (p.i.), mice were dosed through the intraperitoneal (i.p.) route with TFP (1.5 mg/kg), AXPN (3 mg/kg), or DXP (20 mg/kg) alone or with each drug in combination with levofloxacin (0.25 mg/kg or 5 mg/kg). Dosing for TFP occurred once at the time points indicated above, while AXPN, DXP, and levofloxacin were dosed a total of 3 times at 24-h intervals. Mice were assessed for morbidity and/or mortality as well as clinical symptoms for the duration of each experiment (up to 24 days p.i.). For the experiments described above, the surviving animals in each group were humanely euthanized and lungs, livers, and spleens were collected, homogenized, serially diluted in phosphate-buffered saline (PBS), and spread onto SBA plates to assess bacterial clearance from the treated animals.

For rechallenge experiments, after 21 days p.i., surviving animals were infected with WT strain CO92 by the i.n. route as described above at a dose of 8 LD₅₀. Mice were assessed for morbidity and mortality, as well as clinical symptoms, for the duration of each experiment (14 days p.i., 35 days after the original challenge).

Resistance of Y. pestis CO92 to DXP- and AXPN-treated serum.

Serum was collected from naive or DXP- or AXPN-treated mice (n = 3 per group). Mice were dosed with 20 mg/kg DXP or 3 mg/kg AXPN a total of 3 times (24 h apart), and serum was collected 6 h after each dose (i.e., 30, 54, and 78 h). WT strain CO92 was grown overnight, harvested, and then diluted in PBS to an optical density at 600 nm (OD₆₀₀) of 0.2 (10⁸ CFU/ml). A 50-μl volume of the diluted bacteria was mixed with 200 μl of serum or PBS as a control. The samples were incubated at 37°C for 2 h with shaking at 180 rpm. The number of surviving bacteria (the number of CFU) in each sample was determined by serial dilution and plating on SBA plates (59–61). Percent bacterial survival was calculated by dividing the average number of CFU in PBS by the average number of CFU in each serum sample.

Evaluation of cytokine/chemokine levels in lungs following DXP and AXPN administration to mice.

Animals were divided into three groups, consisting of naive and DXP-treated and AXPN-treated animals, the last two of which were dosed with 20 mg/kg DXP or 3 mg/kg AXPN, respectively, a total of 3 times (24 h apart). At 6 h, 30 h, and 54 h after the first dose, three mice from each group were humanely euthanized and the lungs were immediately collected and homogenized. Lung homogenates were then filtered using Costar 0.1-μm-pore-size centrifuge tube filters (Corning Inc., Corning, NY). The levels of cytokines/chemokines in sterile lung homogenate samples were analyzed using a Bio-Plex Pro mouse cytokine 23-plex assay (Bio-Rad, Hercules, CA).

Bacterial proliferation and cytokine/chemokine levels in lungs following infection and treatment with DXP.

Mice were divided into two groups, consisting of one group that was infected with 10 LD₅₀ of WT strain CO92 by the i.n. route and that remained untreated and a second group that was treated with DXP beginning at 24 h p.i. At 30 h, 54 h, and 78 h p.i. (6 h after each dose) and at 10 days p.i., three animals from each group were humanely euthanized and their lungs were excised. As described above, the tissues were homogenized and serial dilutions were assessed for proliferation of the bacteria in the lungs. Portions of each lung homogenate were filter sterilized, and the levels of cytokines/chemokines were analyzed.

Testing of TFP, AXPN, and DXP as therapeutics in combination with vancomycin in a murine model of C. difficile infection.

All animal studies with C. difficile were performed in an ABSL-2 facility under a protocol approved by the UTMB IACUC. Female C57BL/6 mice (age, 8 weeks) were purchased from The Jackson Laboratory (Bar Harbor, ME) and housed under specific-pathogen-free conditions. Animals were administered an antibiotic cocktail (colistin, 850 U/ml; gentamicin, 0.035 mg/ml; kanamycin, 0.4 mg/ml; metronidazole, 0.215 mg/ml; vancomycin, 0.045 mg/ml) in the drinking water for 3 days and then switched to regular water (57). Two days later, the mice received a single dose of clindamycin (32 mg/kg) via i.p. injection to disrupt the normal intestinal microbiota, allowing C. difficile colonization. On the next day, the mice were infected with 10⁵ spores of C. difficile (strain VPI 10463) by oral gavage as previously described (9, 58). Following infection, the mice were divided into groups and dosed with TFP (1.5 mg/kg), AXPN (3 mg/kg), or DXP (20 mg/kg) alone or with each drug in combination with vancomycin (20 mg/kg) at 24-h intervals by the i.p. route. Animals receiving TFP were dosed once, while animals receiving AXPN or DXP were dosed for 3 days. Mice receiving vancomycin were dosed for a total of 5 days. All animals were monitored daily for 8 days p.i. for signs of infection, including weight loss, the presence of diarrhea, hunched posture, and prolonged lethargy.

Testing of AXPN as a therapeutic in a murine model of K. pneumoniae respiratory infection.

Animal studies with K. pneumoniae were performed in an ABSL-2 facility under a protocol approved by the UTMB IACUC. Female C57BL/6 mice (age, 6 to 8 weeks; The Jackson Laboratory) were anesthetized by isoflurane inhalation and subsequently challenged i.n. with 5 × 10⁵ CFU of K. pneumoniae strain 43816. At 0 or 24 h p.i., the mice were dosed with AXPN (3 mg/kg) by the i.p. route a total of 3 times at 24-h intervals. The mice were assessed for morbidity and/or mortality as well as for clinical symptoms for the duration of each experiment.

Screening for macrophage viability following infection with K. pneumoniae or A. baumannii.

To identify additional potential therapeutics from a library of 780 FDA-approved drugs, murine RAW 264.7 macrophages were seeded in 96-well microtiter plates at a concentration of 2 × 10⁴ cells/well to form confluent monolayers in a volume of 100 μl per well. Macrophages were then infected with K. pneumoniae or A. baumannii strains at a multiplicity of infection (MOI) of 100, centrifuged at 1,250 rpm for 5 min to promote bacterial contact with the host cells, and incubated at 37°C in 5% CO₂ for 120 min. The extracellular bacteria were then killed with the addition of 300 μg/ml, 200 μg/ml, or 500 μg/ml of gentamicin for the K. pneumoniae 244, A. baumannii 8, or A. baumannii BAA-1797 and BAA-1799 strains, respectively, for 120 min at 37°C in 5% CO₂. The drugs, each at a 33 μM concentration, were then added to the macrophages and the macrophages were maintained in medium containing gentamicin at the respective concentrations mentioned above for 18 h at 37°C in 5% CO₂ before performing the MTT assay (9). Following the protocol outlined by ATCC, the MTT reagent was added to the wells (10 μl/well) and the cells were incubated at 37°C in 5% CO₂ for an additional 2 h. Then, 100 μl of the detergent reagent was added to the wells, the plates were incubated in the dark at ambient temperature for 2 h, and the absorbance at 570 nm was measured in a VersaMax tunable microplate reader (Molecular Devices, Sunnyvale, CA).

Growth kinetics and sensitivity of A. baumannii strains to 13 drugs that inhibited macrophage cytotoxicity.

Overnight cultures of A. baumannii strains 8, BAA-1797, and BAA-1799, grown in LB broth at 37°C, were normalized to the same absorbance (OD₆₀₀). Subcultures were then inoculated into 100 μl of LB broth contained in 96-well plates with the 13 drugs, each at a concentration of 33 μM, which was identified to reduce host cell cytotoxicity by all three A. baumannii strains, or PBS (negative control). The plates were incubated at 37°C with agitation, and absorbance measurements were taken at the time points indicated above. For MIC determinations, the broth macrodilution method was utilized (62). Briefly, A. baumannii strains were grown to saturation at 37°C, and the growth was adjusted to a 0.5 McFarland standard before addition of the same volume of the culture to serial dilutions of each drug (the highest concentration tested was 200 μg/ml) in 100 μl of LB broth in 96-well plates. The cultures were grown for 24 h with agitation at 37°C. Bacterial growth with PBS instead of the drugs served as a negative control.

Screening for inhibition of A. baumannii intracellular survival in RAW 264.7 murine macrophages.

The three A. baumannii strains (strains 8, BAA-1797, and BAA-1799) were grown in LB broth overnight to saturation at 37°C. Macrophages were seeded in 96-well plates at a concentration of 2 × 10⁴ cells/well to achieve confluence. The plates were then infected with one of the A. baumannii strains at an MOI of 100 in DMEM, centrifuged, and incubated at 37°C in 5% CO₂ for 120 min. Infected macrophages were then treated with gentamicin and incubated with each of the drugs at a concentration of 33 μM as described above. At 4 h p.i., the macrophages were lysed with sterile cold water and the viable bacteria recovered were quantitated by culturing serial dilutions of the lysates on SBA plates.

Statistical analyses.

For the in vitro studies, two independent experiments were performed in duplicate. Whenever appropriate, one-way analysis of variance (ANOVA) or two-way analysis of variance with Tukey's post hoc test or Student's t test was employed for data analysis. For drug screening studies, the drugs were grouped into either a tier 1 or a tier 2 classification. Tier 1 drugs significantly reduced bacterial cytotoxicity for infected macrophages so that the viability of the macrophages was equivalent to that of the uninfected control macrophages, while tier 2 drugs significantly reduced the cytotoxicity of infected macrophages but viability was not reduced to the levels seen for uninfected control macrophages. Kaplan-Meier survival estimates or chi-square analyses were used for the animal studies, with P values of ≤0.05 being considered significant for all statistical tests used.