Defective Innate Cell Response and Lymph Node
Infiltration Specify Yersinia pestis Infection
Franc ¸oise Guinet1*, Patrick Ave ´2, Louis Jones3, Michel Huerre2, Elisabeth Carniel1
1Unite ´ des Yersinia, Institut Pasteur, Paris, France, 2Unite ´ de Recherche et d’Expertise d’Histotechnologie et Pathologie, Institut Pasteur, Paris, France, 3Groupe Logiciels
et Banques de Donne ´es, Institut Pasteur, Paris, France
Since its recent emergence from the enteropathogen Yersinia pseudotuberculosis, Y. pestis, the plague agent, has acquired
an intradermal (id) route of entry and an extreme virulence. To identify pathophysiological events associated with the Y.
pestis high degree of pathogenicity, we compared disease progression and evolution in mice after id inoculation of the two
Yersinia species. Mortality studies showed that the id portal was not in itself sufficient to provide Y. pseudotuberculosis with
the high virulence power of its descendant. Surprisingly, Y. pseudotuberculosis multiplied even more efficiently than Y. pestis
in the dermis, and generated comparable histological lesions. Likewise, Y. pseudotuberculosis translocated to the draining
lymph node (DLN) and similar numbers of the two bacterial species were found at 24 h post infection (pi) in this organ.
However, on day 2 pi, bacterial loads were higher in Y. pestis-infected than in Y. pseudotuberculosis-infected DLNs. Clustering
and multiple correspondence analyses showed that the DLN pathologies induced by the two species were statistically
significantly different and identified the most discriminating elementary lesions. Y. pseudotuberculosis infection was
accompanied by abscess-type polymorphonuclear cell infiltrates containing the infection, while Y. pestis-infected DLNs
exhibited an altered tissue density and a vascular congestion, and were typified by an invasion of the tissue by free floating
bacteria. Therefore, Y. pestis exceptional virulence is not due to its recently acquired portal of entry into the host, but is
associated with a distinct ability to massively infiltrate the DLN, without inducing in this organ an organized
polymorphonuclear cell reaction. These results shed light on pathophysiological processes that draw the line between a
virulent and a hypervirulent pathogen.
Citation: Guinet F, Ave ´ P, Jones L, Huerre M, Carniel E (2008) Defective Innate Cell Response and Lymph Node Infiltration Specify Yersinia pestis Infection. PLoS
ONE 3(2): e1688. doi:10.1371/journal.pone.0001688
Editor: Kirsten Nielsen, University of Minnesota, United States of America
Received October 26, 2007; Accepted January 22, 2008; Published February 27, 2008
Copyright: ? 2008 Guinet et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: Funding was from the Pasteur Institute, which had no role in study design, data collection and analysis, decision to publish or preparation of the
Competing Interests: The authors have declared that no competing interests exist.
Bubonic plague is an acute bacterial disease which, if untreated,
leads to death in 50–90% of the cases  in generally less than
5 days . In spite of major improvements in the management
and control of the disease since the ravaging pandemics of the past,
the disease is far from eradicated and trends show that it is
expanding or re-emerging in many countries [3,4].
Plague is an anthropozoonosis affecting primarily rodents [2,5]
and the disease has a host-vector-host transmission cycle. The
vectors are fleas which transmit the disease by intradermal (id)
biting. Humans are generally contaminated through the bite of an
infected rodent flea. Experimental id or subcutaneous (sc) injection
of the plague agent into laboratory rodents, e.g. guinea pigs, mice
and rats, causes a disease similar to naturally acquired bubonic
plague [6–12]. From the inoculation site the infection proceeds to
the draining lymph node (DLN) via lymphatic channels. The
infected node increases in size and, within a few days, gives rise to
the so-called bubo, hallmark of the disease. Clinically, the bubo is
a voluminous and exceedingly painful lymphadenitis, whose
sudden appearance coincides with the onset of fever. Histologi-
cally, buboes are typically heavily infected and display over-
whelming inflammatory and necrotic alterations [6,8,13,14].
Septicemia and hematogenous spread to distal lymph nodes and
deep organs appear to follow an initial phase of containment of the
infection in the proximal lymph node [11,15]. The septicemia-
induced shock is believed to be the proximal cause of death.
The plague agent is an Enterobacteriaceae named Yersinia pestis. All
other Yersiniae are transmitted through the fecal-oral route and
they are either non-pathogenic or responsible for generally self-
subsiding digestive symptoms [16,17]. Among the enteropatho-
genic Yersinia species is Y. pseudotuberculosis, from which Y. pestis
originated less than 20 000 years ago . Hence, in a short time
frame, Y. pestis has developed both an extreme virulence and a
mode of transmission unique among the Enterobacteriaceae family.
The two species are genetically nearly identical . Based on
their close relatedness it has been proposed that the two species
should be reclassified as a single one, a proposition that was
subsequently rejected in consideration of their extreme divergence
in pathogenicity, life cycle and public health impact [20,21]. The
Y. pestis/Y. pseudotuberculosis pair thus provides a unique opportunity
to explore, by comparative analysis, the pathophysiological
processes associated with a high degree of pathogenicity.
Indeed, the specific mode of action whereby Y. pestis rapidly kills
its host is still elusive. Little is known of the sequence of events
mediating the exceptional severity of plague, including the time
and place where these events appear during the course of disease
and the impact of the newly acquired dermal portal in plague
PLoS ONE | www.plosone.org1February 2008 | Volume 3 | Issue 2 | e1688
pathogenesis. To address these questions, we undertook a study of
the disease progression of bubonic plague in a mouse model. To
highlight pathophysiological events likely to be critical to plague-
specific pathogenesis and to identify histological lesions that
differentiate plague from other bacterial infections, the findings
were contrasted with the disease induced by id inoculation of Y.
pseudotuberculosis, the closely related and less virulent ancestor of Y.
1. Y. pseudotuberculosis is less virulent than Y. pestis upon
Comparison of the reported lethal doses 50 (LD50) of various
strains of Y. pestis and Y. pseudotuberculosis injected sc shows a 105–
106fold higher LD50 for Y. pseudotuberculosis [5,22–25]. To test
whether a similar difference would appear after id inoculation,
which most closely mimics the natural transmission mode of
bubonic plague, serial dilutions of Y. pestis CO92 and Y.
pseudotuberculosis IP32953 were injected id into the ear of outbred
mice and LD50s were calculated from the mortality rates after a
3 week follow-up. Y. pseudotuberculosis LD50 was found to be
,56105colony forming units (cfu), and Y. pestis LD50 was
,20 cfu. Thus, the mortality studies showed that in the id
injection model Y. pestis CO92 was more virulent than Y.
Further experiments were carried out to determine the
minimum Y. pestis infective dose that would cause death of
100% mice and the maximum Y. pseudotuberculosis dose to which
100% of the animals would survive. Infective loads between
,30 cfu to ,36107were tested on a total of 225 mice followed up
for a minimum of 7 days. The lowest Y. pestis dose to which all
mice succumbed within 4 days was between ,1,000 and
,2,000 cfu. All mice having received ,1,300 to ,17,000 Y.
pseudotuberculosis cfu were still alive on day 4 post-inoculation (pi)
and 2% of them died between day 4 and day 7 pi.
2. Bacterial loads at the injection site and in the draining
To determine whether the different degrees of severity of the Y.
pestis and Y. pseudotuberculosis infections were associated with distinct
bacterial dynamics within the host, bacterial loads at the injection
site (IS) and in the DLN of Y. pestis- and Y. pseudotuberculosis-infected
mice were quantified. Mice received intradermally (mean6SEM)
42006427 cfu of Y. pseudotuberculosis or 40006299 cfu of Y. pestis.
Enumeration of cfu recovered from the IS at time 0 was performed
in a subset of experiments to confirm, as previously published ,
that it matched the injected load. Because all of the mice used in
the experiments were naive, there were no bacteria in the DLN at
time 0. Cfu enumerations from the IS and the DLN were
performed 24 h (two groups of 16 mice infected with either Y.
pseudotuberculosis or Y. pestis, 48 h (13 Y. pseudotuberculosis- and 18 Y.
pestis–infected mice), and 72 h and 96 h (2 groups of 7 Y.
pseudotuberculosis-infected animals each) pi. On day 3 pi, the
dramatic decrease in the number of available Y. pestis-infected
mice, due to around ,80% mortality, precluded valid statistical
analysis of the cfu counts at this time point.
Y. pseudotuberculosis was able to survive and to proliferate in the
dermis during the first two days pi. Its growth rate over the first
24 h pi was even higher than that of Y. pestis (Figure 1).
Furthermore, Y. pseudotuberculosis was found in the proximal lymph
node, and similar numbers of both Yersinia species were present in
the DLN at 24 h (p=0.39). On the following day, however, Y.
pestis numbers in the DLN were significantly higher than those of
Y. pseudotuberculosis. Over extended observation periods of 7 days,
DLN Y. pseudotuberculosis loads reached a plateau from day 3 on
(Fig. 1 and data not shown). Therefore, both species were able to
multiply at the inoculation site and to reach the proximal lymph
node but while, in the DLN, Y. pseudotuberculosis growth was
contained, Y. pestis multiplication was not controlled and the
animals started to die between days 2 and 3.
3. Y. pestis and Y. pseudotuberculosis induce similar
lesions at the injection site
In a separate series of experiments, the gross and microscopical
pathology at the IS was examined to compare the lesions induced
locally by the two Yersinia species at 24 h, 48 h and 72 h (for the
surviving mice) pi.
At the gross pathology level, the infected ear dorsum appeared
normal or exhibited slight inflammatory changes such as a diffuse
moderate hyperemia or a red inflammatory dot at the needle entry
point irrespective of the injected species. On microscopical
examination, edema and polymorphonuclear leukocyte (PMN)
influx were observed at the IS of most infected mice. Figure 2
shows examples of PMN reactions at the IS accompanying Y. pestis
and Y. pseudotuberculosis infections. Semi-quantitative scoring of the
intensity of the inflammatory response, performed independently
by two investigators on a set of 29 infected dermis sections,
indicated a slide to slide variation in the intensity of inflammation
but no significant impact of the injected species. Thus, Y. pestis and
Y. pseudotuberculosis induced lesions of similar type and intensity at
the injection site.
4. Y. pestis and Y. pseudotuberculosis induce different
lesions in the draining lymph node
To know whether Y. pseudotuberculosis would induce in the
proximal lymph node lesions comparable to those observed in the
Figure 1. Bacterial loads at injection site and in draining lymph
node. Data are means of log10of cfu numbers recovered from 77 mice
infected over six independent experiments. Bars=Standard Errors. Stars
indicate that the data are significantly different between the two
Yersinia species (p#0.0001). The dashed line shows the detection limit
(10 cfu). The cross symbol indicates that most Y. pestis-infected mice
died between days 2 and 3.
Specific Plague Pathogenesis
PLoS ONE | www.plosone.org2 February 2008 | Volume 3 | Issue 2 | e1688
plague bubo, DLNs collected on day 1, 2 and 3 from mice infected
with comparable numbers (3,500–6,500 cfu) of either Y. pestis or Y.
pseudotuberculosis were examined. A total of 36 DLNs were
4.1. Gross pathology.
While DLNs infected with either
Yersinia species were macroscopically subnormal on day 1 pi,
DLNs collected on days 2 and 3 pi consistently displayed
distinctive features of Y. pestis or Y. pseudotuberculosis infections. Y.
pseudotuberculosis infected DLNs were frankly purulent, whereas Y.
pestis infected DLNs were hemorrhagic, sometimes partly purulent,
and their consistency was dense and firm (Fig. 3). Therefore, the
DLN is the first organ to display species-specific alterations at the
gross pathology level during the course of infection, and this
difference appears on day 2 pi.
The lesions present in the infected
DLNs were much more complex and varied than in the dermis, so
that an objective evaluation of the impact of the species and of the
infecting dose on the DLN alterations was not possible without a
detailed statistical analysis, using tools suitable for large categorical
data sets. Cluster analysis and multiple correspondence analysis
were employed, after decomposition of the histological patterns
into elementary lesions.
a) Cluster analysis identifies histology patterns associated with either Y. pestis or Y.
Criteria aimed at assessing the PMN inflammatory reaction, the
extent and organization of bacterial involvement, and changes in
DLN morphology and density were defined (Table 1 and Fig. 4),
and DLN sections were scored as «+» or «2» for each criterion so
that to each DLN was associated a ‘‘+/2’’ sequence representing
its histopathological profile and allowing clustering analysis of the
data. The scoring was done blinded and repeated at least twice by
the same observer for each lymph node.
The cluster analysis delineated three main types of histopath-
ological patterns (Fig. 5 and Fig. 6). Type 1 patterns corresponded
to both Y. pestis and Y. pseudotuberculosis infected DLNs collected
shortly after the bacterial injection. Therefore, this type, which was
characterized by no or limited alterations of the lymph nodes,
represented early, non species-specific lesions. The type 2 group
comprised only Y. pseudotuberculosis-infected DLNs, all from day 2
or 3 pi. Hence, type 2 corresponded to advanced Y. pseudotuber-
culosis lesions. Type 3 comprised only profiles associated with Y.
pestis infection, and all of them but one were from DLNs taken on
day 2 or 3 pi. Type 3 thus represented advanced Y. pestis lesions.
Type 3 DLNs appeared to be generally more severely altered than
type 2 DLNs and four of them were totally destroyed, the normal
DLN tissue being entirely replaced by bacteria or by PMNs.
Therefore, the cluster analysis strengthened our preliminary
observation that a differential evolution of the DLN lesions
appears on day 2.
b) Exploration of an extended infecting dose range
An extended bacterial dose range (500–14,000 cfu) was
explored, to know whether a higher infecting dose of Y.
pseudotuberculosis could generate destructive lesions similar to those
induced by Y. pestis, and if, after infection with lower Y. pestis loads,
histopathological profiles similar to type 2 would appear. As seen
in Fig. 7, this did not modify the relative grouping of infected
DLNs, as all DLNs from mice infected with Y. pestis segregated in
group 1 or 3, while all DLNs but one from mice infected with Y.
pseudotuberculosis clustered in group 1 or 2. Therefore, the bacterial
load does not appear to influence the type of lesions, which is
strictly associated with the species type.
c) Identification of the criteria most discriminating between types 2 and 3
Thus, the 42 criteria together efficiently discriminated between
advanced Y. pestis and Y. pseudotuberculosis DLN histopathology
Figure 2. Inflammatory reaction at the injection site. Examples of
edema and PMN influx 48 h following injection of ,4500 Y. pestis
(panel B) or Y. pseudotuberculosis (panel C) cfu, compared to a saline-
injected control (panel A). Immunostained PMNs are brown-orange on
the preparations. Bar=20 mm.
Figure 3. Draining lymph node gross pathology. Macroscopical aspect of lymph nodes two days after inoculation of either saline, ,5500 Y.
pseudotuberculosis cfu or ,5500 Y. pestis cfu. Lymph node from Y. pseudotuberculosis-infected mice is purulent whereas Y. pestis-infected lymph node
is reddish and adherent to neighboring tissues. Bar=2 mm.
Specific Plague Pathogenesis
PLoS ONE | www.plosone.org3February 2008 | Volume 3 | Issue 2 | e1688
Table 1. Criteria used to score sections of lymph nodes infected with Y. pestis or Y. pseudotuberculosis, and their discriminating
General features explored
by criteria Description of criteriaCriterion Nu
Similar to saline-injected controls1
Extent and organisation
of lesions and of
PMNs,50% of surface area of the section2
PMNs.50% of surface area of the section3
Wedge shaped abscess4 4A
Abscess involve one pole of the DLN5
Abscess.25% of section surface6
Destructive non-abscess lesions.50% of surface7
+3.95 Lesion=100% of surface8
Peripheral band of PMNs containing bacterial foci9 4B
Layer of PMNs bordering a peripheral band of bacteria+PMNs+cell debris10 4C
Blunt border of inflammatory front 11 4B
Extent of bacterial
colonization of the LN
Bacteria in sub-capsular sinus, ,1/3 of circumference 12
Bacteria: small peripheral focus or foci 13
Patches of densely packed bacteria, bordered by PMNs14 4D
Atypical bact. patches (low density, undefined boundery, weak PMN border)15 4E
Bacteria in $ 1/3 of sub-capsular sinus 16
Bacteria involve large areas of the DLN 17 4F
Invasiveness in the LN tissueInward bacterial projections from the subcapsular sinus (like flames)18 4I
Isolated host cells within a bacterial zone194G
At high magnification: bacterial infiltration around host cells 20 4H
At high magnification: free floating bacteria21 4H
Breaching of the PMN barrier Inward bacterial infiltration beyond the PMN line 22
Bacterial clusters in direct contact with lymphocytes 23
Tissue density alterations Normal LN tissue density24
Area of reduced host cell density within bacterial areas25
Area of reduced host cell density outside bacterial areas26 4I
Area of reduced host cell density with a reticular pattern27 4J
Area of reduced host cell density without a reticular pattern 28
Area of reduced host cell density in a purulent zone 29
Area of reduced host cell density central to the inflammatory front30
Area of reduced host cell density peripheral to the inflammatory front31
Presence of numerous pycnotic cells 32 4J
«Moth eaten» appearance33 4K
Islet of apparently normal lymphocytes in an otherwise altered DLN 34
Part of the DLN is missing35
Final stages of LN destructionTotally destroyed LN massively infiltrated with bacteria 36
Totally destroyed LN massively infiltrated with PMNs 37
Enlarged blood vessels packed with RBCs40 4L
Surrounding, extranodal tissue, adhering to the LN41
More than 1 LN on the slide 42
Discriminating indexes are the test values in the right column. Only significant test values (i.e. .1.96 or ,21.96) are given. Positive and negative test-values are
distinctive of, respectively, Y. pestis and Y. pseudotuberculosis infections.
Specific Plague Pathogenesis
PLoS ONE | www.plosone.org4February 2008 | Volume 3 | Issue 2 | e1688
characteristics. In order to assess the relative contribution of each
criterion to this discrimination, lymph node histology patterns
were further analyzed by Multiple Correspondence Analysis
(MCA), a multivariate statistical method designed to describe
objects characterized by categorical variables [27–29]. The
method reduces the complexity of the initial data set while
retaining most of the information on their relationships, and
provides graphical outputs so that the above relationships can be
visualized. This is achieved through the fitting of axes onto which
objects and variables are projected. Here, the «objects» are the
DLNs and the «variables» are the criteria. The first axis is
calculated as the best fit linear model of the dispersed data, and
each successive axis is orthogonal with the preceding axis. For
representation of the data, the minimum number of axes are
selected that maximizes this representation. All DLNs correspond-
ing to advanced lesions (types 2 and 3) were included in the
analysis, except those that were totally destroyed.
The first axis, the first two axes together and the first three axes
together represented 93.9%, 96.8% and 99.1% of the inertia,
respectively, indicating that almost the totality of the information
was contained within the first three axes and that the first axis
alone contributed most of it. Figure 8 shows a projection on the
Figure 4. Illustrations of some of the criteria used for scoring. Panels A–E, G, H and J–L display HE stained sections, panels F and I show
sections immunostained with Y. pestis specific antibodies. In the text below, the criteria are referred to according to the numbering in Table 1. A:
Wedge shaped abscess (criterion 4) indicated by arrows. Within the abscess bacterial colonies are visible as pink patches. B: Peripheral layer of PMNs
containing bacterial foci (criterion 9), with a blunt demarcation (arrows) from the lymph node tissue (criterion 11). Inset: higher magnification to show
the characteristic horseshoe shaped nuclei of the PMNs, and a bacterial focus. C: Layer of PMNs bordering a peripheral band of bacteria, cell debris
and PMNs (criterion 10). The inset shows the typical PMN morphology of the cells within the layer. Behind this layer, bacterial aggregates are seen as
pink areas (arrowheads) containing purple dots that are, as seen as higher magnification (not shown here) PMNs and cell debris. D: Patch of densely
packed bacteria, bordered by PMNs (criterion14). Packed bacteria form a pink 8-shaped area at the centre of the picture (star). E: Atypical bacterial
patch (criterion 15). At the center of the picture an aggregate of bacterial rods, not as densely packed as the preceding one, is loosely surrounded by
inflammatory cells. F: A large bacterial zone (criterion 17), stained brownish on the preparation. G: Isolated host cells within a bacterial zone (criterion
19). Isolated host cells and cell remnants are seen amid a sea of bacteria which gives a ‘‘ground glass’’ appearance to this part of the LN section. H,
left: bacterial infiltration around host cells (criterion 20). A bacterial strand (arrow), that seems to originate from a nearby colony (star), passes between
host cells. H, right: Free floating bacterium, indicated by an arrow (criterion 21). I: Zone of reduced tissular density (arrows) outside bacterial areas,
which are brownish on this preparation (criterion 26). This image also shows flame-like inward bacterial projections (criterion 18). J: Area of reduced
host cell density with a reticular pattern (criterion 27) and containing numerous pycnotic cells (criterion 32). K: Moth eaten appearance (criterion 33)
of a lymph node with areas of contrasting tissular densities. L: Vascular congestion (criterion 40), showing bright red on this preparation.
Magnifications: Panels A–C, F, I–L : bar=200 mm. Insets of panels B and C: bar=50 mm. Panels D, E, G and H: bar=10 mm.
Specific Plague Pathogenesis
PLoS ONE | www.plosone.org5 February 2008 | Volume 3 | Issue 2 | e1688