Two major steps in the evolution from Y. pseudotuberculosis to Y. pestis have been the acquisition of an id portal of infection and a sharp increase in virulence, raising the question of a possible causal link between the two events. However, in our system this route of entry did not endow Y. pseudotuberculosis with the killing power of its descendant, indicating that the high virulence of Y. pestis is a genuine property of this species.
Previous images of dermis sections of human and animal plague cases have indicated that Y. pestis
penetration into the dermis is followed by local bacterial expansion 
. This finding is here confirmed and quantified. However, Y. pestis
dermal expansion is not superior to that of the less virulent species, indicating it does not account for the exceptional severity of plague. In fact, unexpectedly, there were even higher numbers of Y. pseudotuberculosis
than Y. pestis
within the IS at 24 h and 48 h pi. This difference does not seem to be linked to a lower in situ inflammatory reaction to Y. pseudotuberculosis
because similar inflammatory lesions were observed in the dermis of mice injected with either Yersinia
species. At 24 h pi, similar numbers of Y. pestis
and Y. pseudotuberculosis
were present in the DLN. Altogether these observations show that the first steps of the disease, including a phase of bacterial multiplication in the DLN within the first 24 h, proceeded similarly during the two Yersinia
infections, indicating that the expression of Y. pestis
unique pathogenicity is delayed until later stages of the infectious process.
In contrast, Y. pestis
loads in the DLN were higher than those of Y. pseudotuberculosis
on day 2, indicating that, in this organ, Y. pestis
growth was not controlled as efficiently as that of Y. pseudotuberculosis
past the first 24 h of infection. Our findings show that the large accumulation of bacteria in the draining lymph node previously reported in descriptions of human and animal plague buboes 
is specific to Y. pestis
compared to its sister species. As has been suggested, accumulation of Y. pestis
in the extravascular lymph node reservoir, followed by brutal release into the blood stream, may be important for the generation of the high level septicemia necessary for transmission to a new host by the blood sucking vector 
. Therefore, the difference in the intranodal bacterial accumulation of the two species might be critical to their different pathogenic potentials.
Significantly different histopathological patterns were associated in the DLN with Y. pestis
and Y. pseudotuberculosis
infections. This conclusion was reached through decomposition of images of the histological sections into elementary lesions and subsequent analysis by statistical tools able to deal with large sets of data with no assumed theoretical distribution of the variables or the individuals 
. To our knowledge the approach used here, which combines the use of clustering analysis methods, MCA and test-value calculations, has never been reported so far for comparative analysis of histological images. While this study focused on the host response to Yersinia
infection, another potential application of this approach is the comparison of the lesions induced by Y. pestis
mutants in order to determine the role of specific genes to plague pathogenesis. These mutants would likely express phenotypes less distant from Y. pestis
than does Y. pseudotuberculosis
. However, the technique was developped to detect small differences in complex data sets so it could probably detect differences more subtle than between the two Yersinia
species, although it would probably give lower percentages of inertia for each axis.
The MCA analysis yielded a list of elementary lesions that specified the two types of infection with respect to one another. All the abscess-type lesions, whether wedge-shaped, polar or organized as a peripheral band, were characteristic of Y. pseudotuberculosis
infection and, in these structures, the infectious foci were kept separate from the normal lymph node tissue. Thus, an organized PMN response able to contain the bacteria was a hallmark of Y. pseudotuberculosis
infection, which implies, conversely, that one specificity of plague lesions was the lack of an organized innate cell reaction. Indeed, PMNs in buboes were often disposed as a thin, irregular and fragmented layer, a feature highly discriminatory towards plague. Abscesses are the expression of an acute cellular response to bacterial invasions and play a major role in the control of bacterial infections 
. Therefore the lack of an abscess-type structure is certainly a major impediment to the control of Y. pestis
proliferation in the DLN and probably accounts largely for the high representation of plague bacilli in this organ.
The molecular and cellular mechanisms underlying abscess formation remain poorly defined 
, except for the step of PMN recruitment from the blood stream about which several inflammatory cues and molecular mechanisms have been uncovered 
. The Y. pestis
-associated deficit in abscess formation is not the result of a massive defect of PMN recruitment to the DLN, because, from the morphological data, there was no gross difference in the amount of PMNs in the DLN during the two Yersinia
infections. Therefore it is possible that functional alterations of the PMNs, or of other cells involved in abscess formation, are induced by the plague agent.
Bacterial extensions from the subcapsular sinus or from colonies, seen at low magnifications as «flames» and at high magnification as bacterial strands streaking between host cells, and free floating isolated bacteria, all reflected an infiltrating character of the plague infection. The images of isolated host cells within bacterial areas might be the result of this infiltrating process around host cells. The above criteria all had discriminatory values above 5, making the infiltrative character of the infection the most specific feature of plague in the comparison with Y. pseudotuberculosis infection. It remains to be determined whether specific mechanisms, such as selective tissue destructions, are responsible for this infiltrating behavior, in addition to the failure of the inflammatory cells to contain the infection.
In conclusion, no major differences were noted between the two Yersinia infections during the progression to DLN, and in the DLN before day 2 pi. At this point, Y. pseudotuberculosis infection had induced an organized PMN reaction and was contained, unlike Y. pestis growth which was not controlled, leading to death within hours. These findings point to the population of PMNs recruited to the DLN as a likely primary or secondary target of the Y. pestis specific strategy. More work directed at characterizing the mechanisms by which Y. pestis specifically prevents the formation of an effective abscess-type defense should further our understanding of plague pathogenesis.