A Plasminogen-Activating Protease Specifically Controls the Development of Primary Pneumonic Plague

See allHide authors and affiliations

Science  26 Jan 2007:
Vol. 315, Issue 5811, pp. 509-513
DOI: 10.1126/science.1137195


Primary pneumonic plague is transmitted easily, progresses rapidly, and causes high mortality, but the mechanisms by which Yersinia pestis overwhelms the lungs are largely unknown. We show that the plasminogen activator Pla is essential for Y. pestis to cause primary pneumonic plague but is less important for dissemination during pneumonic plague than during bubonic plague. Experiments manipulating its temporal expression showed that Pla allows Y. pestis to replicate rapidly in the airways, causing a lethal fulminant pneumonia; if unexpressed, inflammation is aborted, and lung repair is activated. Inhibition of Pla expression prolonged the survival of animals with the disease, offering a therapeutic option to extend the period during which antibiotics are effective.

Of the three species of Yersinia pathogenic to humans, Y. pestis is infamous owing to its ability to cause plague. Although usually transmitted by an arthropod vector, it is during cases of secondary pneumonic plague that Y. pestis can be spread from person to person through the inhalation of respiratory droplets carrying the bacteria (1). This method of transmission may initiate an epidemic of primary pneumonic plague, which is fatal if not treated early. Pneumonic plague is the most likely form to result in the event of a biowarfare attack with aerosolized Y. pestis (2).

Yersinia virulence in mammals requires the Ysc type III secretion system (T3SS), which is shared among all three pathogenic species (3). Unlike Y. pestis, Y. enterocolitica and Y. pseudotuberculosis are foodborne pathogens and usually result in self-limiting gastrointestinal infections (46). Thus, the presence of the Ysc T3SS alone is not sufficient to cause a rapidly progressing respiratory infection.

Y. pestis also carries pPCP1, a 9.5-kb plasmid that encodes the plasminogen activator Pla, a surface protease that is thought to promote plasmin degradation of fibrin clots (1, 7). In models of bubonic plague, Pla promotes the invasion of Y. pestis from subcutaneous sites of inoculation into the lymphatic system and deeper tissues but is dispensable for growth at the site of inoculation (8, 9). When introduced by aerosol, Y. pestis lacking Pla was reported to be equivalent or near equivalent in virulence to wild-type by median lethal dose (LD50) analysis (10, 11), but the progression of lung and systemic disease has never been evaluated in a model of primary pneumonic plague. On the basis of these studies, we predicted that respiratory infection with a strain of Y. pestis lacking Pla would proceed normally in the lungs and result in a lethal pneumonia but that fibrin deposition would restrict the ability of bacteria to escape the respiratory system.

We infected C57BL/6 mice intranasally with wild-type Y. pestis CO92, an isogenic Y. pestis strain lacking Pla (CO92 Δpla), or the Δpla strain complemented with the coding sequence for Pla. Mice given wild-type Y. pestis CO92, a strain isolated from a fatal case of pneumonic plague (12), succumbed to the infection in a highly synchronous manner. In contrast, only 50% of the mice infected with the Δpla strain developed terminal plague after 7 days, and the rate at which the mice died was less synchronous than the rate of those infected with the wild-type strain (Fig. 1A). Complementation of the mutant with the coding sequence for Pla fully restored virulence. Thus, the lack of Pla substantially delayed the time to death resulting from respiratory infection.

Fig. 1.

Pla is required for Y. pestis to cause a fulminant infection of the lungs. (A) Survival of C57BL/6 mice infected intranasally with Y. pestis CO92 (black squares), CO92 Δpla (white squares), or CO92 Δpla complemented with pla (white diamonds). (B and C) Kinetics of infection with Y. pestis CO92 (black) or CO92 Δpla (white). Bacteria were introduced intranasally, and at various times CFU per organ in the lungs (B) and spleen (C) were determined. Each point represents the numbers recovered from a single mouse. The limit of detection is indicated by a dashed line. Symbols below the limit of detection represent mice that survived but did not have detectable numbers of bacteria; an “X” indicates a mouse that succumbed to the infection. A solid line indicates the median of CFU recovered. *P = 0.037, **P = 0.002, and ***P < 0.001 (unpaired t test). (D) Gross weight of lungs from mice infected with Y. pestis CO92 (black) or CO92 Δpla (white). C57BL/6 mice were infected as above, and at various times lungs were excised and weighed. *P = 0.01 and **P < 0.001. Each experiment was repeated twice. Error bars represent standard deviation.

Although the kinetics of bacterial growth during infection with CO92 proceeded as expected (13), bacterial outgrowth in the Δpla-infected mice was significantly altered (Fig. 1, B and C). After 24 hours, 100- to 1000-fold fewer bacteria were recovered from the lungs of Δpla-infected mice compared with the lungs with wild-type infection. Over the next 2 days, the numbers of Δpla bacteria in the lungs did not substantially change, whereas wild-type bacteria increased by almost 6 logs. In contrast, we detected bacteria in the spleens of all mice by 72 hours, indicating that escape of the Δpla strain from the lungs to distal organs still occurred (Fig. 1C). Indeed, at later times, one of five mice had a bacterial burden approaching 108 to 109 colony-forming units (CFU) in the spleen. This corresponded with increased numbers of bacteria in the lungs; however, this is likely due to the recirculation of systemic organisms back into the lungs (13) rather than outgrowth of bacteria in this organ.

These data show that Pla controls the proliferation of Y. pestis in the lungs but is not essential for bacteria to disseminate. This is a distinct phenotype from that attributed to Pla in models of bubonic plague: When Pla-negative Y. pestis is introduced subcutaneously, dissemination is dramatically reduced, but bacterial outgrowth at the local site of infection is unaffected (8, 9). Indeed, we show that rates of dissemination from the initial site of colonization to the spleen were substantially increased when Pla-negative bacteria were introduced intranasally compared with the subcutaneous route (table S2). This may be due to the highly vascularized nature of the lung, allowing escape of a few Pla-negative bacteria through an alveolar capillary and thus initiating systemic infection.

A hallmark of fatal bacterial pneumonia is the development of edema in the lungs, which can be measured by a change in gross lung weight as fluid and cells contribute to increased mass of the organ. Although mouse lungs infected with wild-type Y. pestis weighed significantly more than uninfected lungs, the lungs of mice infected with the Δpla strain of Y. pestis showed no change in weight, even after 7 days (Fig. 1D), suggesting that the death of mice infected with this strain is not due to pneumonia but rather is caused by systemic infection. Our results, therefore, may explain the similar LD50 values for wild-type and Pla-negative strains when inhaled, even though Y. pestis requires Pla to cause a severe pneumonia.

We also examined lung histology of wild-type and mutant Y. pestis–infected mice to resolve an existing controversy regarding pathology and inflammation at the site of infection (8, 9, 13). An influx of inflammatory cells was detected in the lungs in both wild-type and Δpla infections 36 hours after inoculation; in both cases, the predominant infiltrating cell type was polymorphonuclear (Fig. 2). Although the size of the pulmonary lesions in the wild-type infection increased over time, resulting in tissue destruction and hemorrhage, the foci of inflammation in mice infected with CO92 Δpla remained relatively constant and restricted. In addition, we examined infected lung sections by using immunofluorescence with an antibody against Y. pestis. Numerous extracellular bacteria were associated with inflammation in the lungs of wild-type–infected mice, but relatively few bacteria were detected in the Δpla infection and were restricted to the much smaller inflammatory lesions (Fig. 2). Thus, both bacterial outgrowth and subsequent inflammation in the lungs were dependent on the Pla surface protease.

Fig. 2.

Histology and presence of bacteria in the lungs of mice during the progression of pneumonic plague. Mice were infected intranasally; at various times, lungs were inflated and fixed with 10% neutral buffered formalin and embedded in paraffin, and 5-μm sections were stained with hematoxylin and eosin or a Y. pestis antibody. The images shown are representative of experiments repeated twice. Scale bars indicate 200 μm.

These results suggest that host immunity controls the pulmonary infection without developing an overwhelming inflammatory reaction to Pla-negative bacteria. Therefore, we assessed the amount of immune activation in the lungs by using quantitative reverse-transcription polymerase chain reaction (qRT-PCR) to determine changes in transcript quantities of multiple inflammatory mediators. Consistent with our previous observations (13), mice infected with CO92 remained unresponsive early in the infection but showed significant cytokine up-regulation by 48 hours (Fig. 3A). Similarly, the cytokine transcript numbers in the lungs of mice infected with Δpla were also relatively unchanged early in the infection. After 48 hours, however, cytokine transcript numbers in response to Δpla were unchanged or only slightly increased, and by the following day of infection transcript numbers for most cytokines decreased, suggesting down-regulation of the inflammatory response to Δpla. Thus, the data reveal that, in the absence of Pla, an anti-inflammatory state is maintained in the lungs and the infection is unable to progress to the pro-inflammatory phase that we described previously (13).

Fig. 3.

Progression of the inflammatory response to Y. pestis CO92 or CO92 Δpla. (A) Mice were uninfected or infected intranasally, and at various times lungs were removed and immersed in RNAlater (Ambion, Woodward, TX). RNA was extracted and reverse-transcribed, and relative cytokine transcript quantities compared with those of uninfected mice were determined by qRT-PCR using the ΔΔCt method (22) and normalized to glyceraldehyde-3-phosphate dehydrogenase for Y. pestis CO92 (black) or Δpla-infected mice (white). IL, interleukin; MIP, macrophage inflammatory protein; and TNF, tumor necrosis factor. (B and C) PCNA antibody stain and hematoxylin counterstain of lungs infected with Y. pestis CO92 or CO92 Δpla after 48 hours. PCNA-positive cells (arrows) are present in the lungs of Δpla-infected mice but largely absent in the wild-type infection. Extranuclear granular staining in the CO92-infected lungs correlate with the presence of bacteria. Scale bars, 50 μm.

That cytokine transcript amounts appeared to stabilize and then decrease in the Δpla-infected mice suggested the pulmonary inflammatory lesions were resolving. We immunostained infected lungs for the proliferating cell nuclear antigen (PCNA), a marker for host cellular DNA synthesis (14). Whereas the cells of wild-type–infected lungs were almost uniformly PCNA-negative (Fig. 3B), large numbers of PCNA-positive cells were present in Δpla-infected mice (Fig. 3C), indicating active cell proliferation and regeneration of tissue in the lungs. Lung repair at this stage of the infection is further evidence that fatalities among Δpla-infected mice are not a consequence of airway inflammation or damage but instead are the result of systemic spread of the bacteria.

If Pla alone controls the ability of Y. pestis to cause pneumonic plague, we hypothesized that experimental induction of pla expression midway during the aborted pulmonary disease would be sufficient to turn the nonpneumonic infection into a pneumonic one. To test this, we adapted the tetracycline-responsive promoter system (15) to exogenously control gene expression in Y. pestis during infection (Materials and Methods and figs. S1 and S2). We cultured the Δpla strain of Y. pestis carrying pla under control of the tetracycline-responsive promoter (Y. pestis CO92 Δpla Ptet-pla, strain YP138) in pla-repressing conditions (i.e., absence of anhydrotetracycline, or ATC). Thirty-six hours after intranasal inoculation, we induced pla expression by injecting ATC and then followed the progression of the infection. Bacteria in the pla-repressed state established a nonprogressive lung infection in a manner similar to that of the Δpla-infected mice. However, once ATC was administered and Pla expression was up-regulated, the condition of these mice quickly converted to a disease with all the features of pneumonic plague: rapid proliferation of bacteria with development of visible microcolonies (Fig. 4, A and B), unrestricted inflammatory infiltrate, tissue damage (Fig. 4C), and shortened time to death. Thus, the absence of Pla stalls the development of disease in the early anti-inflammatory phase but does not eliminate the potential of these organisms to cause pneumonic plague. Ultimately, the block in the progression of infection by the respiratory route is completely reversible by the expression of Pla.

Fig. 4.

Control of primary pneumonic plague progression by the tetracycline-responsive promoter system in Y. pestis. (A and B) Induction of Pla during intranasal infection. Mice were infected intranasally with Y. pestis CO92 (black) or CO92 Δpla + Ptet-pla prepared in the absence of ATC (repressed state). After 36 hours, mice infected with the Ptet-pla strain were administered phosphate-buffered solution (PBS) (blue) or ATC (red) by intraperitoneal injection bid. CFU in the lungs (A) and spleen (B) were determined at various times. Symbols and lines as in Fig. 1. (C) Lung histology of mice infected with the Ptet-pla strain and treated with PBS or ATC. Lungs were prepared as for Fig. 2. Scale bars, 200 μm. (D) Repression of pla expression midway through pneumonic plague infection. Survival of mice infected with CO92 Δpla plus Ptet-pla in the pla-induced state for the duration of the experiment (black) [mean time to death (MTD) = 3.1 days], in the pla-repressed state for the duration (blue) (MTD = 5.1 days), or in the pla-induced state for the first day followed by the pla-repressed state for the remainder of the experiment (red) (MTD = 4.6 days). See Materials and Methods for details. The bar beneath the graph approximates the period at which pla induction ends and repression begins.

One hypothesis of the mechanism by which Pla facilitates the invasive nature of Y. pestis is that the protease converts host plasminogen into plasmin while degrading the plasmin inhibitor α2-antiplasmin, releasing bacteria from the entrapment of fibrin clots (16, 17). Indeed, recent evidence has shown that fibrin deposition is an important means of immune control for a variety of pathogens (1820), and thus the subversion of the coagulation cascade may be a notable virulence mechanism. Consistent with this, we show that the plasminogen-activating activity of Pla is essential to Y. pestis virulence in the pulmonary system (Materials and Methods and fig. S3). Additionally, fibrin(ogen) deposition can be detected in the lungs of mice infected with either the wild-type or the Δpla strain, but the pattern and extent of fibrin(ogen) immunostaining is substantially altered (fig. S4). We cannot exclude, however, the possibility that other targets of Pla activity may also contribute to the development of primary pneumonic plague. Nonetheless, the role of Pla during pneumonic plague may help explain how Y. pestis acquired the ability to cause a rapid, severe respiratory infection and be transmitted from person to person by the aerosol route, whereas Y. pseudotuberculosis and Y. enterocolitica did not. Interestingly, the altered syndrome we observed with a Pla-negative strain is similar to case descriptions from the early 20th-century Manchurian epidemics in which aerosol-acquired plague resulted in fatal sepsis before a local lung disease could develop (21).

The critical role for Pla suggests that its inhibition could offer a therapeutic advantage, particularly because the rapid progression of pneumonic plague leaves little time for effective treatment. We tested this experimentally by infecting mice with the Pla-inducible Y. pestis strain YP138 prepared in the presence of ATC. We then provided ATC to the animals for only the first day of the infection, allowing the ATC to be cleared and pla expression to be repressed for the remainder of the experiment. The time until death was delayed when the Pla-induced state was switched to a Pla-repressed state during the infection (Fig. 4D); in fact, the kinetics more closely resemble those of a Δpla infection. This suggests that exogenous inhibition of Pla during primary pneumonic plague may indeed prolong the survival of the affected individual, expanding the window during which antibiotics could be successfully administered to treat the disease.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S4

Tables S1 and S2


References and Notes

View Abstract

Navigate This Article