The role of neutrophils and monocytic cells in controlling the initiation of Clostridium perfringens gas gangrene


  • Present addresses: David K. O'Brien, Division of Bacteriology, United States Army Medical Research Institute for Infectious Diseases, Fort Detrick, 1425 Porter Street, Frederick, MD 21702, USA.
    Michael E. Woodman, Department of Microbiology, Immunology, and Molecular Genetics, University of Kentucky College of Medicine, Lexington, KY 40536-0298, USA.

  • Editor: Patrik Bavoil

Correspondence: Stephen B. Melville, Department of Biological Sciences, 2119 Derring Hall, Virginia Tech, Blacksburg, VA 24061, USA. Tel.: +1 540 231 1441; fax: +1 540 2319307; e-mail:


Clostridium perfringens is a common cause of the fatal disease gas gangrene (myonecrosis). Established gas gangrene is notable for a profound absence of neutrophils and monocytic cells (phagocytes), and it has been suggested that the bactericidal activities of these cells play an insignificant role in controlling the progression of the infection. However, large inocula of bacteria are needed to establish an infection in experimental animals, suggesting phagocytes may play a role in inhibiting the initiation of gangrene. Examination of tissue sections of mice infected with a lethal (1 × 109) or sublethal (1 × 106) inoculum of C. perfringens revealed that phagocyte infiltration in the first 3 h postinfection was inhibited with a lethal dose but not with a sublethal dose, indicating that exclusion of phagocytes begins very early in the infection cycle. Experiments in which mice were depleted of either circulating monocytes or neutrophils before infection with C. perfringens showed that monocytes play a role in inhibiting the onset of gas gangrene at intermediate inocula but, although neutrophils can slow the onset of the infection, they are not protective. These results suggest that treatments designed to increase monocyte infiltration and activate macrophages may lead to increased resistance to the initiation of gas gangrene.


Clostridium perfringens is a gram-positive anaerobic bacterium that is the most common cause of clostridial gas gangrene (myonecrosis) in humans. As a species, C. perfringens produces at least 13 different toxins, which are the major virulence factors in causing disease (Rood, 1998). Of these toxins, PLC (α-toxin) and PFO (θ toxin) are thought to be most important in the pathogenesis of gas gangrene (Stevens, 2000; Stevens & Bryant, 2002). If left untreated, gas gangrene is always fatal due to the disruption of cardiac functions and severe shock that follows the release of toxins into the bloodstream (Stevens et al., 1988; Asmuth et al., 1995).

A distinctive characteristic of established gas gangrene is the paucity of phagocytic cells in the area of the infection and this is considered to be one of the primary factors leading to uncontrolled spread of the infection (Bryant & Stevens, 1997; Stevens, 1997). PLC- and PFO-mediated effects that contribute to the lack of phagocytic cells include: the cytotoxic effects of PFO on leukocytes (Stevens et al., 1987, 1997; Stevens & Bryant, 1993, 2002; Stevens, 2000); the ability of PLC to activate synthesis of cell adhesion molecules, causing neutrophils to adhere and transvasate at locations remote from the site of infection (Bryant & Stevens, 1996; Bunting et al., 1997); and PLC-mediated activation of platelet gpIIbIIIa, inducing the formation of vascular occlusions which block the flow of blood to the site of infection (Bryant et al., 2000a, b; Stevens & Bryant, 2002). The latter process, along with platelet P-selectin (CD62-P), was recently shown to impede neutrophil diapedesis (Bryant et al., 2006).

However, the role that neutrophils and macrophages play in preventing the initiation of gas gangrene has not been clearly defined. Clinically, this is an important phase of the disease because once the infection progresses to the point where it is visually apparent, extensive tissue debridement, amputation or even mortality will be the outcome, as even with treatment, death occurs in ∼25% of all cases (Present et al., 1990). Neutrophils and macrophages possess strong antimicrobial activities even under anaerobic conditions (Spitznagel, 1984), so they should, in theory, be able to eliminate C. perfringens from contaminated wounds even if the tissue is hypoxic. Yet, in the mouse model most often used to study C. perfringens gas gangrene, in which C. perfringens bacteria are injected into the mouse femoral muscle and the development of gas gangrene is monitored, the infectious doses are relatively high. The strain used most often, the genetically-tractable strain 13 (Rood & Cole, 1991), requires an infectious dose of 1 × 108–1 × 109 bacteria to initiate an infection (Lyristis et al., 1994; Awad et al., 1995; Stevens et al., 1997). The requirement of high infectious doses may indicate that the bacteria need to overcome an initial phagocytic cell response before becoming established in host tissues. In the present study, we used two approaches to define the respective roles that neutrophils and monocytic cells play in the initiation of gas gangrene: measurements of the rate of infiltration of phagocytes into the wound site immediately after infection and the effects of depleting phagocytes on the survival of mice infected with C. perfringens at increasing inocula.

Materials and methods

Bacterial strains and growth conditions

Clostridium perfringens strain 13 (obtained from C. Duncan) was used for all studies. The bacteria were grown in PGY media (Melville et al., 1994) in a Coy anaerobic chamber (Coy Laboratory Products).

The kinetics of leukocyte recruitment in the early stages of an infection

Clostridium perfringens was grown to mid-log phase in PGY medium and washed twice with phosphate buffered saline (PBS). Samples were resuspended in PBS to give a final concentration of ∼1 × 1010 CFU mL−1 or diluted to a final concentration of ∼1 × 107 CFU mL−1. The number of CFU in each inoculum was confirmed by serial dilution and plating on PGY medium. Twelve BALB/c mice were inoculated in the right hind leg with 0.1 mL of the 1010 or 107 CFU mL−1 bacterial suspensions (i.e. 109 or 106 CFU). Three mice each were euthanized at 10, 60, 120, or 180 min postinfection. The hind leg muscle was rapidly excised, fixed in neutral-buffered formalin, and stored in 70% ethanol. The muscles were then embedded in paraffin and 3-μm-thick sections were made at 2-mm intervals and the sections were then fixed to glass slides. Sections were stained with hematoxylin and eosin (H/E) and examined for the colocalization of leukocytes and bacteria. A Leitz Dialux 20 microscope with an attached Moticam 1300 camera was used to capture high magnification color images. For each mouse, 66–238 images within the tissues were examined. Each image was nonoverlapping and contained, by definition, at least one C. perfringens bacterium. Each image captured a field that measured 100 μm in diameter at × 1000 magnification. The presence of one or more leukocyte within 45 μm of one (or more) bacterial cell was scored as positive. The percent of fields that contained neutrophils, monocytic cells, or lymphocytes colocalizing with bacteria was then calculated.

Depletion of neutrophils

Six BALB/cJ mice received an intraperitoneal injection of 100 μg of the monoclonal antibody RB6-8C5 (Cedarlane), which is specific for the Gr-1 surface antigen present on granulocytes of mice but not on monocytes or lymphocytes (Rogers & Unanue, 1993). To verify that the neutrophils were depleted, blood smears were taken from mice at 1, 3, and 5 days postantibody injection, stained using Hema 3 Stain Set (Fisher), and the total number of neutrophils mm−3 was determined. Six control mice received intraperitoneal injections of PBS instead of the monoclonal antibody.

Depletion of peripheral monocytes

Liposomes were prepared as described (Qian et al., 1994). Monocytes were depleted by intravenous tail-vein injection of 100 μL of clodronate-filled liposomes (or 100 μL of PBS-filled liposomes for control mice) into BALB/c mice. To determine the level of monocyte depletion, 200-μL blood samples were taken from the tail-vein of each mouse before and 24 h after liposome injection. Each blood sample was incubated with a F4/80-specific monoclonal antibody labeled with Alexa Flour 488 (Molecular Probes) and then treated with Pharm Lyse (Becton Dickenson) to remove red blood cells from the sample. The total leukocyte cell number in each blood sample was determined on a hemocytometer and the number of monocytes labeled with F4/80 in each sample was quantified using fluorescence at 488 nm. Four mice were tested for depletion with each type of liposome.

Clostridium. perfringens infections of phagocyte-depleted mice

Six to eight weeks old female BALB/c mice were used for the monocyte depletion experiments and 6–8 weeks old female BALB/cJ mice for neutrophil depletion experiments. Preceding infection with C. perfringens, mice were depleted of either their neutrophils or peripheral monocytes as described above. Clostridium perfringens cultures were grown to mid-log phase, washed three times in PBS, and 1 × 109, 108, 107, or 106 CFU were injected into the right hind calf muscle. Once the onset of gangrene symptoms (i.e. edema and discoloration in the leg muscles, blackening of the infected foot, lethargy, and a scruffy appearance) was apparent, the mice were euthanized to eliminate unnecessary suffering in accordance with humane animal care practices. Any mouse that did not show signs of gangrene was euthanized 36 h after the start of the experiment. Six mice were used for each inoculum.

All animal experiments were performed with the approval of the Institutional Animal Care and Use Committee at Virginia Tech.

Statistical analysis

The LIFETEST procedure in the sas system (version 9.1.3; sas Institute Inc., Cary, NC) was used to perform comparisons of mouse survival times after infection. All other statistical analyses were done using instat 3 software (Graphpad Inc.). For all statistical analyses, P values of <0.05 were considered significant.

Results and discussion

To determine whether the exclusion of leukocytes from a gangrene site was dose-dependent and, if so, what the kinetics of the exclusion process were, mice were injected with a lethal dose (1 × 109 CFU) or sublethal dose (1 × 106 CFU) of C. perfringens. As an indicator of infiltration rates of leukocytes into the site of injection, H/E stained tissue sections were examined for the colocalization of phagocytes and bacteria at 10, 60, 120, or 180 min postinfection. Representative images showing the state of the tissue samples at the site of injection at 10 or 120 min postinfection are shown in Fig. 1. The relatively large bacteria stained dark purple and were clearly visible in the images. The H/E staining and high magnification images allowed us to differentiate leukocyte types as neutrophils, monocytes, macrophages, or lymphocytes based on characteristic nuclear morphology, cytoplasmic staining and overall cell morphology (Fig. 1).

Figure 1.

 Representative images showing colocalization of leukocytes and bacteria at the sites of infection. Tissue sections were obtained from mice injected in the hind leg muscle with 1 × 106 (a and b) or 1 × 109 (c and d) CFU of Clostridium perfringens 10 min (a and c) and 120 min (b and d) postinfection. Sections were made from paraffin-embedded tissue and stained with hematoxylin and eosin. Arrows point to macrophages, arrowheads to bacteria, open arrowheads to neutrophils, asterisks indicate red blood cells. M, muscle fibers; NM, necrotic muscle fibers. Bars=10 μm.

The images in Fig. 1 and similar images were used to determine the timing of phagocyte infiltration and the results are shown in Fig. 2. When the bacterial inoculum was 1 × 106 CFU, the influx of colocalizing neutrophils present in the observed fields increased from 17% to 71% from 10 min to 120 min postinfection (Fig. 2a). Over the same time span, colocalizing neutrophil numbers remained static (15–21%) in the mice infected with 1 × 109 CFU and were at statistically significant lower numbers than in the mice infected with the smaller inoculum at the 60 and 120 min times postinfection (Fig. 2a).

Figure 2.

 Kinetics of infiltration of leukocytes into the site of infection. Hematoxylin and eosin stained slides of mouse muscle tissue were examined for the colocalization of bacteria and leukocytes. The mice were inoculated with 1 × 106 bacteria (squares) or 1 × 109 bacteria (diamonds). The mean and SD (shown as error bars) for each group of three mice at each time point is shown. The movement of neutrophils into the site of infection (a) was significantly higher at 1 and 2 h for the 1 × 106 inoculum than the 1 × 109 inoculum (P<0.0005 and P<0.0290, respectively, using the two-tailed Student's t-test). The infiltration of macrophages (b) was significantly higher with the 1 × 106 inoculum than the 1 × 109 inoculum at 1 and 2 h (P<0.0037 and P<0.0257, respectively, using the two-tailed Student's t-test). The presence of lymphocytes was variable at the site of infection and not significantly different with the two inocula (c).

At 180 min postinfection with a 1 × 106 inoculum, bacteria could not be found within the tissues despite examining several hundred images. However, neutrophils and macrophages were visible in the tissue sections in approximately the same numbers seen in tissue sections from the 1 × 106 inocula at 120 min (data not shown). To determine whether the absence of bacteria in the tissue sections was confined to mice in a single experimental group, the entire 180-min postinfection experiment was repeated with the 1 × 106 inoculum, but bacteria were still not clearly identifiable in the mouse tissue sections. Therefore, no data are shown in Fig. 2 at that time and dosage. In a previous report, we have shown that up to 6 h postinfection with 1 × 106 CFU, ∼1 × 106 viable bacteria were still present within the tissues (O'Brien & Melville, 2004). We are not sure why the bacteria could not be seen in these sections, but there are at least two possibilities: (1) the bacteria are present within the tissues as single cells only and are difficult to locate using H/E-stained sections, (2) the bacteria are within phagocytes and not easily recognized, as proposed by us based on in vitro results with macrophages (O'Brien & Melville, 2003, 2004). This question is currently under investigation in our laboratory.

Monocytic cells were observed to colocalize with bacteria by 10 min postinfection in mice infected with 1 × 106 or 1 × 109 bacteria but nearly all of them appeared, based on their cell morphology (e.g. Fig. 1a–c), to be macrophages rather than peripheral blood monocytic cells (Fig. 2b). The mean value for the percent of fields in which bacteria co-localized with macrophages at 10 min postinfection was 13% for the 1 × 106 inoculum and 5.8% for the 1 × 109 inoculum, but this difference was not statistically significant. The percent of fields showing co-localized bacteria and macrophages declined between 10 and 120 min in both sets of mice but there was a statistically higher level in the mice infected with 1 × 106 than 1 × 109 CFU bacteria (Fig. 2b). In fact, at the 60- and 120-min time points with the 1 × 109 inoculum, mean values of the percent of fields showing colocalization with macrophages declined to 1% and <0.5%, respectively, which is a more dramatic decline than that observed with neutrophils (Fig. 2a).

The number of lymphocytes that colocalized with bacteria varied over the course of the experiment (Fig. 3c) but there was no significant difference seen between the different inocula of C. perfringens. Lymphocytes are most likely present due to vascular leakage at the site of infection rather than as a response to a chemotactic signal (Bunting et al., 1997).

Figure 3.

 Time course showing depletion of neutrophils using monoclonal antibody RB6-8C5. The mean and SD (error bars) of the number of neutrophils/ cu mm−1 of peripheral blood of BALB/cJ mice was determined after injection with the monoclonal antibody (circles) or PBS alone (squares).

These results indicate that with a large inoculum, 1 × 109 CFU, there was little recruitment of phagocytes in the first 3 h of infection. This supports previous studies that implicate C. perfringens toxins PLC and PFO in blocking chemotaxis and recruitment of phagocytes to the site of infection (Bryant & Stevens, 1997; Stevens, 2000; Stevens & Bryant, 2002). However, there was a significant level of recruitment of phagocytes in response to a nonlethal bacterial inoculum, 1 × 106 CFU. These results suggest that inhibition of chemotaxis to the site of infection is a dose-dependent event and that at small inocula the bacteria are not effective at keeping phagocytes at a distance, which probably accounts for the lack of gangrene development at these concentrations. If exclusion of phagocytes is an important element in the initiation of gangrene, the question then arises as to which main phagocyte type, neutrophils or monocytic cells, plays a more important role in controlling the initiation of the infection. To answer this question, mice were depleted of circulating neutrophils and monocytes and the effects of depletion were measured using a mouse gas gangrene model.

To deplete neutrophils, mice were injected with the monoclonal antibody RB6-8C5 (see Materials and methods). To determine the effectiveness of the RB6-8C5 antibodies at depleting neutrophils, circulating neutrophils levels in five mice injected with RB6-8C5 or PBS were measured 1, 3, and 5 days postinjection. The maximal level of neutrophil depletion was observed 24 h postinjection, in comparison with the control mice injected with PBS alone (Fig. 3). Therefore, in vivo infections were done 24 h after RB6-8C5 injection. No measurable effects were seen on the levels of other circulating blood cell types, as measured by differential counts in blood smears (data not shown).

The time that elapsed from the time of injection until the onset of gangrene symptoms (at which time the mice were euthanized, see Materials and methods) in mice depleted of neutrophils was examined after infection with 1 × 109, 108, 107, or 106 bacteria. Neutropenic mice showed a statistically significant decrease in survival rate after infection with 1 × 109 bacteria, in comparison with the control mice (Fig. 4, upper left panel, Table 1). However, the depletion of neutrophils did not affect the overall ability of mice to survive during the course of the experiment as all the mice died by 24 h postinfection (Fig. 4, upper left panel). Mice infected with 108, 107, and 106 CFU did not show a statistically significant difference in the time to the onset of gangrene between the control and neutropenic mice (Fig. 4, Table 1).

Figure 4.

 Survival of neutrophil-depleted mice after infection with Clostridium perfringens. BALB/cJ mice were depleted of neutrophils using monoclonal antibody RB6-8C5 (cross hairs) and control mice were injected with PBS alone (circles). Mice were infected with the Clostridium perfringens inoculum shown in each panel.

Table 1.   Results of statistical analyses comparing mean times to the onset of gangrene (i.e. survival) in mice depleted of neutrophils or circulating monocytes vs. control groups
1 × 1091 × 108
Mean survival time (h)Chi-squarePMean survival time (h)Chi-squareP
  • The LIFETEST procedure, using the log-rank test, was used to determine the P values shown. The survival times of individual mice are shown graphically in Figs 4 and 5. None of the infections at inocula of 1 × 107 or 1 × 106 CFU of Clostridium perfringens in either the neutropenic mice or monocyte-depleted mice exhibited a statistically significant difference between the treatment and control groups (not shown). P values <0.05, considered statistically significant, are shown in bold.

  • *

    All mice survived for the entire course of the experiment.

Neutrophil depletion8.714.0310.044711.40.02430.876
Neutrophil control13.3  13.3  
Monocyte depletion6.80.07650.782115.95.60420.0179
Monocyte control6.71  36*  

For the experiments designed to measure the effects of depleting circulating monocytes, mice received intravenous injections of either clodronate- or PBS-filled liposomes. Mice injected with clodronate-filled liposomes showed a 4.5-fold decrease in their level of peripheral blood monocytes 24 h postinjection, whereas control mice showed no significant decrease in monocyte numbers (Table 2). Monocyte-depleted and control mice were then infected 24 h after liposome treatment with 1 × 109, 108, 107, or 106 CFU. Mice depleted of their peripheral monocytes and infected with 1 × 109 bacteria showed no significant difference in the time to the onset of gangrene symptoms (i.e. survival) in comparison with control mice (Fig. 5, upper left panel, and Table 1). However, monocyte-depleted mice infected with 108 bacteria began to show a decrease in the time to the onset of gangrene symptoms at 12 h postinfection relative to the control mice (Fig. 5, upper right panel). By 24 h postinfection, only 33% of the clodronate-treated mice had survived, in comparison with 100% of the PBS-treated mice (Fig. 5, upper right panel) and this difference was shown to be significant (Table 1). With inocula of 1 × 107 and 1 × 106 CFU, none of the mice from either treatment developed gas gangrene (Fig. 5, lower panels).

Table 2.   Mean percent F4/80-positive cells (monocytes) in murine blood samples before and after liposome treatment
TreatmentBefore treatment
After treatment
Clodronate liposomes1.8% (0.5–2.6)0.4% (0.0–1.1)
PBS liposomes0.9% (0.71–1.1)0.75% (0.6–1.0)
Figure 5.

 The survival rate of monocyte-depleted mice infected with Clostridium perfringens. BALB/c mice were injected with clodronate-filled liposomes (cross hairs) or PBS-filled liposomes (circles) and then infected with Clostridium perfringens. Mice treated with clodronate-filled liposomes showed a significant decrease in survival as compared to mice injected with PBS-filled liposomes at the 108 inoculum. Mice were infected with the Clostridium perfringens inoculum shown in each panel.

At an intermediate inoculum of 1 × 108 CFU, a difference in survival rate was seen between the control groups of BALB/cJ mice (Fig. 4, upper right panel) and BALB/c mice (Fig. 5, upper right panel), and is likely due to a slightly increased susceptibility to infection in the BALB/cJ mice. Therefore, while we cannot directly compare the mean survival rates between the neutrophil- and monocyte-depleted groups, the results with control mice showed there was a statistically significant decrease in survival with the monocyte-depleted mice infected with C. perfringens within that group of mice.

Bryant et al. (2000a) depleted neutrophils in rats using an unspecified antineutrophil serum and observed a decreased level of intravascular aggregate formation, leading them to hypothesize that platelet/neutrophil aggregation may be an important factor, along with platelet/platelet aggregation, in promoting the vascular inclusions induced by PLC activity. However, they did not examine how the depletion of neutrophils influenced survival of mice in a gas gangrene model (Bryant et al., 2000a). Our results suggest that, although neutrophils may briefly attenuate the progression of gangrene within the first few hours of infection with a high bacterial inoculum, they play neither a positive or a negative role in enhancing the overall levels of host survival, as all the mice died by 24 h postinfection (Fig. 4, upper left panel).

In contrast to the results seen with neutropenic mice, depletion of circulating monocytes did lead to decreased levels of protection at intermediate inocula (Fig. 5). These results were unexpected, as previous work in our laboratory has shown that, under in vitro conditions, C. perfringens can escape the phagosome and persist in the presence of mouse macrophage-like J774 cells, but less efficiently in thioglycollate-elicited mouse peritoneal macrophages (O'Brien & Melville, 2000, 2004). Our in vivo results suggest that tissue macrophages can kill C. perfringens effectively if they make contact with the bacteria. Preventing intimate contact between C. perfringens and macrophages may be an essential aspect of the PLC- and PFO-mediated exclusion of phagocytes from the locus of infection (Bryant & Stevens, 1997; Stevens, 2000; Stevens & Bryant, 2002).

Clodronate-filled liposomes, which were injected intravenously, deplete only peripheral monocytes and tissue macrophages in organs such as the liver and spleen, because the liposomes cannot escape the vasculature (Buiting & Van Rooijen, 1994; Van Rooijen & Sanders, 1994). However, in the experiments shown in Fig. 5, monocyte-mediated protection occurred not in the early stages of infection, as was seen with neutropenic mice, but in the period between 6 and 24 h postinfection. A mechanism that could account for the timing of this protection pattern is one in which peripheral monocytes chemotax into the infected tissues and mature into macrophages, which would then be effective in limiting the spread of the infection.

The role of neutrophil functions in limiting C. perfringens infections has dominated research on phagocyte/bacterium interactions in the past, but our results indicate that tissue macrophages may play an important role in limiting the onset of the disease. Treatments in which monocytes are stimulated to chemotax to an infected area and mature into tissue macrophages may help control the progression of gas gangrene in clinical settings.