We found that rendering mice immunocompromised makes them hypersusceptible to P. aeruginosa pneumonia. This is characterized by a decrease in the LD50 after i.n. challenge, with as few as 20 CFU of a noncytotoxic P. aeruginosa strain causing lethality; it was also noted that there were high numbers of bacteria found in the lungs, livers, and spleens of these mice. This susceptibility was decreased, based on better survival and lower bacterial loads, if mice were vaccinated prior to immunocompromise and infection or if, after immunocompromise, antibodies were given to mice at the time of infection.
Dissemination of P. aeruginosa
from the gastrointestinal tract has been previously observed in immunocompromised mice. Koh et al. (12
) showed that P. aeruginosa
could colonize the gastrointestinal tract of immunocompetent mice but did not readily cause bacteremia or disseminate. However, administration of Cy to these infected mice resulted in bacteremia, dissemination, and death in the mice given either cytotoxic or noncytotoxic strains of P. aeruginosa
PAO1. Administration of the RB6-8C5 antibody also resulted in increased mortality in mice that had been previously colonized with P. aeruginosa
, again independent of the cytotoxic phenotype of the colonizing strain. Mortality was also shown to be dependent on the dose of Cy given, as higher Cy doses resulted in increased mortality by strains lacking ExoU. Complete mortality was seen with all strains at doses of Cy of ≥125 mg/kg (12
), the dose of Cy that was used for our experiments. We also saw that after the Cy treatment, a low infectious dose of a noncytotoxic P. aeruginosa
strain resulted in mortality and increased dissemination and bacteremia, compared to our previous findings with immunocompetent mice (unpublished data).
Dissemination of P. aeruginosa
in a pneumonia challenge model in neutropenic mice was previously investigated by Vance et al. (24
). They used this model to evaluate the virulence of TTSS mutants, performing competitive infections using both wild-type and mutant strains. Dissemination to the spleen and blood was seen in Cy-treated and doubly Cy- and RB6-8C5-treated mice infected with the noncytotoxic P. aeruginosa
strain PAO1, but all data were expressed as competitive indices, so the actual numbers of bacteria that disseminated are not known (24
). Consequently, we do not know how well the level of dissemination we observed correlates with what was previously seen by Vance and colleagues.
Cytotoxic strains of P. aeruginosa
that express the phospholipase ExoU (15
) have been shown to disseminate more readily than noncytotoxic strains that express ExoS and are also associated with a higher patient mortality rate (21
). However, Ader et al. showed that strains lacking ExoU that still have an intact TTSS are still more virulent, causing higher mortality and bacterial burdens as well as a more robust neutrophil influx and cytokine response in the lung, compared to P. aeruginosa
strains that have a completely nonfunctional TTSS (1
). Here, we were able to replicate the pattern of increased dissemination as well as increased lung permeability by the cytotoxic ExoU-containing strain PA103 compared to the noncytotoxic strain, 9882-80. When mice given Cy were infected with our challenge strain, 9882-80, a high degree of dissemination was seen, despite the fact that 9882-80 lacks ExoU (2
Cy-treated mice infected with strain 9882-80 had bacteria present in the BAL and spleen at 24 h at similar levels both in the presence and absence of the Evans blue dye. Concurrent with the dissemination, the lungs of Cy-treated mice infected with strain 9882-80 were more permeable than uninfected controls. This effect appears to be dependent on the dose of bacteria given, as the number of CFU detected in the BAL correlated with the amount of blue dye found in the BAL. The higher levels of vascular leakage in the Cy-treated mice infected with strain 9882-80 compared to nontreated mice also infected with 9882-80 were not related to the Cy treatment itself, as Cy-treated mice given PBS instead of bacteria had no differences in lung permeability compared to control, saline-treated mice instilled with PBS instead of bacteria. This leads to the conclusion that it is the accumulation of bacteria in the lungs that results in the dissemination to the spleen and liver.
The efficacy of other P. aeruginosa
vaccines has been investigated in immunocompromised mouse models. Active vaccination against OprF or OprI was shown to increase survival and raise the LD50
of Cy-treated mice to i.p. challenge of P. aeruginosa
). Vaccination with a Salmonella enterica
serovar Dublin strain expressing P. aeruginosa
OprI was able to protect about 10 to 20% of Cy-treated vaccinated mice from oral challenge with P. aeruginosa
). However, these mice received eight vaccinations or boosters to achieve this low level of protection, while our mice received only one vaccination. Administration of monoclonal antibodies to P. aeruginosa
lipoproteins was shown to raise the LD50
of a subcutaneous challenge in mice rendered leukopenic by Cy treatment (8
). Passive vaccination strategies using transfer of rabbit serum to SCID mice showed that mice could be completely protected from an i.p. P. aeruginosa
challenge dose of up to 500 CFU for one epitope (25
). Also, serotype-specific antilipopolysaccharide antibodies were found to raise the LD50
of a challenge directed into an incision on the backs of leukopenic mice (3
). This protection at low bacterial challenge doses is consistent with what we saw in our experiments. These experiments were encouraging in that protection, although sometimes at low levels, could be achieved in immunocompromised mice, although none of these prior studies looked at a pneumonia challenge model. Here we have shown that neutrophils are required for the complete protection from P. aeruginosa
challenge seen with the vaccine, although low levels of protection remain in vaccinated neutropenic mice.
Vaccination prior to rendering the mice immunocompromised made them less susceptible to the P. aeruginosa infection. However, this protection was still not to the same level as would be seen in wild-type mice. This is due to the loss of the immune cells, as the antibody response was unaffected by either treatment. The Cy treatment affects both neutrophils and lymphocytes, while the RB6-8C5 treatment is more specific for neutrophils but still causes a small decrease in lymphocyte numbers. Further investigations will look at the role of lymphocytes in protection, which could validate that the changes in protection in the RB6-8C5-treated mice are from the lack of neutrophils.
The presence of antibodies seems to be enough to slow the infection, although some degree of protection could be from the few remaining lymphocytes or neutrophils in these immunocompromised mice. When antibodies are added directly to the site of infection, the mice fare even better than when they have been vaccinated, suggesting that it is indeed antibodies in the lung that are slowing the progression of the disease and allowing the mice to survive longer. Work by others showed that intravenous administration of antibodies or vaccination of sheep with an O antigen-specific vaccine prior to infection with P. aeruginosa
was able to prevent dissemination but not lung injury, while administration of antibodies to the respiratory tract was able to prevent both dissemination and lung injury (19
). This could be because the circulating antibodies need lung damage to enter the lungs, while the administered antibodies are present directly at the site of infection before the lung damage occurs. These data correlate with what was seen in our immunocompromised mouse model. We have not investigated the lung permeability of the Cy-treated mice with our vaccination strategies to see if the results mirror what was seen in the sheep.
Either of our vaccination strategies could be a potential treatment in the immunocompromised patient population, particularly the passive administration of antibodies at the site of infection. Although neither active nor passive vaccination strategies were able to completely protect the mice, they did manage to slow the progression of disease. If these treatments were used in conjunction with antibiotic therapy in an immunocompromised patient, the outcome would likely be better. We have not tested this possibility in our mouse infection model. We also did not test cytotoxic strains in our immunocompromised mouse models. We have previously seen that cytotoxic strains are more virulent than noncytotoxic strains, having lower LD50s and also having a more rapid progression of disease and earlier dissemination. The infection process of cytotoxic strains in immunocompromised mice could be different, as such strains are likely to cause lung damage and disseminate to other organs at a faster rate than the noncytotoxic strain we tested. However, the presence of specific antibodies could also slow the progression of disease.
Our vaccine is to a single O antigen serogroup, and many serogroups have been implicated in disease. Although there are at least 20 serogroups of P. aeruginosa
, not all of them are implicated in disease, making an O antigen-specific vaccine within reach. Analysis of acute and chronic lung isolates of P. aeruginosa
yielded a higher prevalence of strains of serogroups O6, O1, O11, and O4 (6
). Interestingly, strains belonging to serogroup O11 had a much higher prevalence of cytotoxicity than other serogroups and were also associated with a high mortality rate (6
Here we have shown that treating mice with Cy renders them leukopenic and highly susceptible to P. aeruginosa infection, characterized by increased lung permeability and decreased LD50s. This susceptibility is also seen in mice rendered neutropenic by RB6-8C5 and can be reduced in both groups of immunocompromised mice by administration of our vaccine prior to treatment. Administration of antibodies directly to the lungs at the time of infection seems to afford better protection to the mice and could be combined with antibiotic therapies in clinical settings.