Science in medicine
2868? The?Journal?of?Clinical?Investigation http://www.jci.org Volume 119 Number 10 October 2009
Sepsis and sepsis-induced acute kidney injury:
a life-threatening condition
Sepsis is a characteristic set of systemic reactions to overwhelming
infection. Sepsis, severe sepsis, and septic shock are defined accord-
ing to established criteria (Table 1) (1). Discovery of antibiotics has
dramatically improved the morbidity and mortality of the infectious
diseases for the last decades; indeed, antibiotics and volume resus-
citation are the first line of sepsis treatment strategy (2). However,
overwhelming inflammatory response accompanied by depression
in immunological function causes multiple organ injury and deter-
mines clinical outcomes. In addition to inflammation and immu-
nological dysregulation, a number of different mechanisms contrib-
ute to sepsis at different phases (Figure 1). For instance, systemic
hemodynamics evolves from an early hyperdynamic (“warm shock”)
state to a late hypodynamic (“cold shock”) state.
Sepsis is the leading cause of death in critically ill patients, and the
incidence of sepsis is increasing (3, 4). The mortality rate of severe
sepsis is very high (up to 70%), and the calculated costs exceed $15
billion per year in the United States (3). The rate of severe sepsis dur-
ing hospitalization almost doubled during the last decade and is
considerably greater than previously predicted (5). Sepsis causes mul-
tiorgan failure, including acute kidney injury (AKI) (6), and patients
with both sepsis and AKI have an especially high mortality rate (7).
AKI is diagnosed by a sudden decrease in glomerular filtration rate
(GFR), the primary measure of kidney function, which is currently
detected clinically as a rise in serum creatinine. A multinational pro-
spective observational study including 29,269 critically ill patients
revealed that the occurrence of AKI in the intensive care unit (ICU)
was approximately 6%, the most frequent contributing factor to AKI
being sepsis (50%) (8). Other reports showed that between 45% and
70% of all AKI is associated with sepsis (9–11). Several different patho-
physiological mechanisms have been proposed for sepsis-induced
AKI: vasodilation-induced glomerular hypoperfusion, dysregulated
circulation within the peritubular capillary network, inflammatory
reactions by systemic cytokine storm or local cytokine production
(12), and tubular dysfunction induced by oxidative stress (13).
Continuing concern over the efficacy and safety of the only FDA-
approved therapy for severe sepsis (activated protein C) highlights
the critical need to improve our understanding of the pathophysi-
ology of sepsis and sepsis-induced AKI and to develop novel treat-
ment strategies for critically ill patients (14). A multitude of poten-
tial drug targets have been identified in animal models of sepsis;
however, translation from animals to humans has been exceed-
ingly difficult. Several reviews have pointed out that the failure
to translate results from animals to humans has been attributed
to disease characteristics of sepsis (complexity and heterogeneity),
inappropriate clinical trials (study of ineffective drugs, inadequate
clinical trial designs), and animal models that do not fully mimic
human sepsis (14–17).
Requirements for animal models of sepsis
Human sepsis is currently hypothesized to involve at least two stages:
an initial proinflammatory burst responsible for hypotension and
organ dysfunction, followed by a compensatory antiinflamma-
tory immune response that leads to an immunosuppressed state
often called immune depression or immune dysfunction; however,
these stages can overlap temporally (Figure 1). The latter consists
of altered monocyte antigen presentation, decreased lymphocyte
proliferation and responsiveness, and lymphocyte apoptosis and
anergy (18, 19), which accounts for nosocomial infections and late
deaths in sepsis. Animal models of sepsis need to reproduce the
complexity of human sepsis and its treatment in the ICU. Ideally,
animal models should mimic the pace and severity of human sep-
sis; reproduce key hemodynamic (warm shock followed by cold
shock) and immunologic (proinflammatory stimulation, antiin-
flammatory counterregulation, i.e., immune depression) stages;
mimic histology findings in key organs (lung, liver, spleen, kidney,
etc.) that are frequently modest; and —perhaps counterintuitively
for animal modelers — exhibit variability among animals.
Standard animal models of sepsis
Sepsis animal models can be divided into three categories: (a)
injection of an exogenous toxin (e.g., LPS); (b) alteration of the
animal’s endogenous protective barrier, such as intestinal leak-
age (e.g., cecal ligation and puncture [CLP] or colon ascendens
Animal models of sepsis and sepsis-induced
Kent Doi,1 Asada Leelahavanichkul,2,3 Peter S.T. Yuen,2 and Robert A. Star2
1Department of Nephrology and Endocrinology, University of Tokyo, Tokyo, Japan. 2NIDDK, NIH, Bethesda, Maryland, USA. 3Division of Nephrology,
Department of Medicine, Chulalongkorn University, Bangkok, Thailand.
Conflict?of?interest: The authors have declared that no conflict of interest exists.
Citation?for?this?article: J. Clin. Invest. 119:2868–2878 (2009). doi:10.1172/JCI39421.
science in medicine
? The?Journal?of?Clinical?Investigation http://www.jci.org Volume 119 Number 10 October 2009
stent peritonitis [CASP]); and (c) infusion or instillation of exog-
enous bacteria (Table 2).
LPS-induced inflammation models. Endotoxin, a component of
the outer membrane of Gram-negative bacteria, is involved in the
pathogenesis of sepsis, and an LPS infusion/injection model has
been widely used for sepsis research. LPS administration induces
systemic inflammation that mimics many of the initial clinical fea-
tures of sepsis, including increases in proinflammatory cytokines
such as TNF-α and IL-1, but without bacteremia (20–22). Treat-
ment of LPS-injected animals with neutralizing antibody against
TNF-α or IL-1 resulted in improved outcomes for this model (23,
24). A case report describes a patient who self-administered a large
dose of LPS, and the full clinical manifestations of septic shock
developed (25). LPS infusion also causes renal injury, includ-
ing decreased GFR, increased blood urea nitrogen (BUN), and
increased renal neutrophil infiltration (26–28).
Several clinical trials of anti–TNF-α and anti–IL-1 therapy were
performed based on the promising results in LPS animal studies;
however, these trials failed to improve survival of septic patients
(29, 30). LPS causes much earlier and higher peak levels of cytokine
expression compared with levels observed in human sepsis, with
the notable exception of meningococcal sepsis, a rare, patho-
gen- and site-specific form of sepsis wherein cytokine levels are
comparable to those observed in LPS animal models (21, 31, 32).
Also, some features of LPS infusion such as renal hypoperfusion
and increased BUN are alleviated by volume replacement, which
is routinely performed in clinical management of sepsis (33, 34).
Nevertheless, LPS infusion remains a useful tool for interrogating
a simpler subset of the complex trajectory of sepsis. The LPS dose
can be titrated to mimic early sepsis without hemodynamic com-
promise, which has been useful for studying systemic and renal
responses during the initial phases of sepsis; doses of LPS typi-
cally used induce systemic hypotension and decrease glomerular
perfusion, whereas lower doses of LPS do not cause any systemic
hypotension but still decrease glomerular perfusion (35, 36).
CLP of polymicrobial sepsis. CLP is currently the most widely used
animal model of sepsis (15, 37, 38). CLP surgery is straightforward:
ligation distal to the ileocecal valve and needle puncture of ligated
cecum cause leakage of fecal contents into the peritoneum, with
subsequent polymicrobial bacteremia and sepsis (39). This surgical
manipulation, while not well standardized, allows the severity to be
adjusted by the length of ligated cecum and the size and/or number
of the puncture. Supportive treatment with fluids and antibiotics
is quite variable across laboratories and almost always inadequate,
since typically only a single fluid and/or antibiotic dose is given
(40). Multiple species of bacteria are found in the bloodstream (41,
42), and progressive systemic inflammatory response syndrome fol-
lowed by septic shock and multiorgan injury ensues (34, 43, 44).
Mice subjected to CLP generally became severely hypotensive with-
out an apparent hyperdynamic phase (45), although more vigor-
ous fluid resuscitation can result in an early hyperdynamic phase
detected by echocardiography (46). CLP-induced sepsis models
show a cytokine profile similar to that in human sepsis (21, 31, 34),
and anti–TNF-α treatment fails to alleviate sepsis in CLP models as
in human sepsis (31, 47, 48). Notably, treatment with recombinant
human TNF-α reduced mortality in CLP-induced sepsis (41). As
described above, human sepsis is considered to have two immuno-
logically different stages: a proinflammatory phase and a compen-
satory antiinflammatory phase. CLP-induced sepsis increased lym-
phocyte apoptosis, which mimics immunosuppression at the later
phase of human sepsis (18, 49, 50). In this respect, CLP-induced
sepsis is completely different from LPS-induced sepsis and more
closely mimics human sepsis.
However, the standard CLP model does not develop reproducible
acute kidney or lung injury. AKI has been detected by changes in
BUN or creatinine in some (51–55) but not other studies (56, 57).
Thus, the standard CLP model encompasses more clinical features
and drug responses of human sepsis than the LPS model but is
still missing some key features, especially kidney and lung injury.
Bacterial infusion or instillation models. Whereas models such as
CLP and CASP are helpful in understanding polymicrobial sepsis,
human sepsis may also be caused by a single pathogen. Bacterial
infusion models can approximate introduction of a single patho-
gen in a controlled manner, allowing reproducible infection. These
Refractory septic shock
Two or more of the following conditions: temperature >38.5°C or <35.0°C; heart rate of >90 beats/min; respiratory rate
of >20 breaths/min or PaCO2 of <32 mmHg; and wbc count of >12,000 cells/ml, <4,000 cells/ml, or >10% immature
SIRS in response to documented infection (culture or Gram stain of blood, sputum, urine, or normally sterile body fluid
positive for pathogenic microorganism; or focus of infection identified by visual inspection)
Sepsis and at least one of the following signs of organ hypoperfusion or organ dysfunction: areas of mottled skin;
capillary refilling of >3 s; urinary output of <0.5 ml/kg for at least 1 h or renal replacement therapy; lactate of >2 mmol/l;
abrupt change in mental status or abnormal EEG; platelet count of <100,000 cells/ml or disseminated intravascular
coagulation; acute lung injury/ARDS; and cardiac dysfunction (echocardiography)
Severe sepsis and one of the following conditions: systemic mean BP of <60 mmHg (<80 mmHg if previous
hypertension) after 20–30 ml/kg starch or 40–60 ml/kg serum saline solution or PCWP between 12 and 20 mmHg; and
need for dopamine of >5 μg/kg/min, or norepinephrine or epinephrine of <0.25 μg/kg/min to maintain mean BP at
>60 mmHg (80 mmHg if previous hypertension)
Need for dopamine at >15 μg/kg/min or for norepinephrine or epinephrine at >0.25 μg/kg/min to maintain mean BP
of >60 mmHg (80 mmHg if previous hypertension)
SIRS, systemic inflammatory response syndrome; ARDS, acute respiratory distress syndrome; PCWP, pulmonary capillary wedge pressure.
science in medicine
2870? The?Journal?of?Clinical?Investigation http://www.jci.org Volume 119 Number 10 October 2009
models have been translated to larger animals for the study of sys-
temic and organ-specific hemodynamics (see below). Similarly,
instillation can be useful for simulating pneumonia, especially as
a nosocomial infection (58). These models provide complementary
information that is likely to be pathogen specific.
Creating new clinically relevant sepsis models
LPS and traditional CLP models mimic the clinical conditions of
meningococcal sepsis or postsurgical peritonitis of young patients.
However, animal models of sepsis differ from human sepsis because
of age, comorbidity, and use of supportive therapy (16). Starting
from a clinical perspective, we reasoned that within practical limi-
tations, animals should receive treatment comparable to the sup-
portive therapy that is standard for ICU patients. This is essential
to enable us to test whether a therapy has additional benefit beyond
that provided by conventional fluid and antibiotic therapy. We also
hypothesized that the CLP model could be improved by simulation
of one or more underlying baseline conditions typically present in
septic patients, such as advanced age or chronic disease.
Age, resuscitation, and antibiotics. Since the incidence of sepsis dra-
matically increases with age, and elderly patients are especially
prone to sepsis and sepsis-induced AKI (59), our first modifica-
tion of standard CLP models was to employ older mice, i.e., retired
breeders. We and others found that aged mice (16 to 50 weeks old)
were more susceptible to CLP and LPS than young mice around 8
to 16 weeks old (34, 60, 61). In standard mouse CLP models, ani-
mals are typically given a small amount of fluid resuscitation and
perhaps a dose of antibiotics immediately after surgery, but fluid
and antibiotic treatment are not continued. Because volume resus-
citation can reverse LPS-induced renal injury in aged mice (34), we
added volume resuscitation and antibiotic treatment to the stan-
dard CLP model. Animals became clinically ill at 5–6 hours after
surgery, at which time they had evidence of liver damage; kidney
damage was evident by MRI early (62), but serum creatinine and
BUN levels did not significantly increase until 12 hours after sur-
gery (63). With these additions, consistent histological renal dam-
age with significant increase in serum creatinine were observed,
not because fluids and antibiotics were harmful to the kidney, but
because the animals could survive long enough to develop AKI (34,
64). Acute tubular necrosis is found in ischemic and toxic AKI but
is not present in any of our mouse CLP models; in contrast, we
observed areas of proximal tubular cells that contain prominent
intracellular vacuoles of unknown composition (Figure 2). A recent
systematic review of sepsis revealed that there were only mild his-
tological changes in human and animal sepsis, and acute tubular
necrosis was relatively uncommon (65). Among these studies, Sato
et al. detected increased vacuolization and flattening of the brush
border membrane in immediate autopsy of septic patients (66).
Animal models of sepsis
CLP or CASP
Clinically relevant CLP
Infusion or instillation
of exogenous bacteria
Simple, sterile; some similarities with
human sepsis pathophysiology
Early silent period; moderate and delayed peak
of mediators; multiple bacterial flora
Replication of clinical risk factors
Early hyperdynamic state
Early and transient increases in inflammatory mediators
more intense than in human sepsis
Age and strain variability; early hemodynamic
period in some models
Difficulty in analyzing pathophysiological pathways
No change in intrarenal microcirculation;
need large animals; labor-intensive
Simplified clinical course of sepsis. Progres-
sion of disease is complex, nonlinear, and
varies from one patient to another. Shown is
an outline of selected landmark events and
processes that appear to be common among
patients and some animal models. DIC, dis-
seminated intravascular coagulation.
science in medicine
? The?Journal?of?Clinical?Investigation http://www.jci.org Volume 119 Number 10 October 2009
Although a similar tubular vacuolization is found in cyclosporine
nephropathy (67), its significance and relevance in sepsis still need
to be clarified (68).
Genetic heterogeneity. We next developed a CLP model in aged out-
bred rats (Sprague-Dawley strain). We originally switched to rats
because serum and urine are difficult to collect from mice (44). In
contrast to the mouse CLP model, which consistently develops AKI
with multiorgan injury, the rat CLP model showed heterogeneous
responses, with a wide spectrum of degrees of AKI, including clear
histological evidence for AKI. This model demonstrated early but
not late increases in IL-6 levels that corresponded to development
of AKI, similar to those seen in septic patients (69).
Could the variability and sensitivity to AKI be attributed to the
outbred Sprague-Dawley strain? We shifted our CLP mouse model
to the outbred CD-1 strain and found that CD-1 mice developed
sepsis AKI at a young age (64), whereas the inbred C57BL/6 strain
developed AKI only at an advanced age (34). The reason for this
difference in susceptibility is unknown, but the genetically hetero-
geneous human population should be more accurately represent-
ed by outbred mice, reducing the bias found in inbred strains that
might contain or lack recessive disease susceptibility loci, depend-
ing on selective pressures (70, 71).
Predisposing comorbid conditions. Animal studies typically exam-
ine sepsis and related organ failure in otherwise healthy animals,
despite numerous epidemiological studies of human sepsis that
show the importance of preexisting comorbid conditions (3, 72).
Severe sepsis occurs frequently in patients with underlying chronic
diseases (comorbidities) including chronic kidney disease (CKD),
liver disease, and diabetes and has an extremely high mortality
rate (3, 72). CKD is found in approximately 30% of AKI patients
in the ICU (8, 73). Patients with CKD also have an increased risk
of morbidity and mortality from sepsis (74–77). These findings
suggest that clinical sepsis and sepsis-induced AKI are dramati-
cally influenced by underlying diseases, which may help explain
why simple animal models of sepsis do not mimic human sepsis
and do not predict human response to therapeutics. We recently
established a two-stage mouse model of preexisting renal disease
with subsequent sepsis (CKD-sepsis) to mimic the complexity of
human sepsis; mice were given folic acid to induce renal fibrotic
injury, then subjected two weeks later to CLP surgery (78). This
CKD-sepsis model showed increased vascular permeability and
decreased bacterial clearance compared with sepsis animals with-
out the comorbidity of CKD. Combination therapy with soluble
FMS-like tyrosine kinase 1 (FLT1; also known as VEGF receptor 1)
and chloroquine, which block vascular and immunological dys-
function, respectively, resulted in the best survival rate. Both drugs
were effective individually in simple CLP models, but neither drug
alone was effective in the more complex CKD/CLP model, suggest-
ing that multiple therapeutic interventions (“combination chemo-
therapy”) may be required for the treatment of sepsis complicated
with comorbidity. Complex animal models of human sepsis may
be more pharmacologically relevant than simple animal models for
the testing of therapeutics, as they may ultimately predict human
drug responsiveness more accurately. The introduction in these
models of other disease-modifying variables (e.g. advanced age,
supportive treatment with fluids or antibiotics, or the presence of
other chronic comorbidities) appears to alter and complicate the
underlying pathophysiological mechanisms of disease. Therefore,
it is important to compare the complex models with simpler mod-
els systematically to distinguish between core versus amplifying
factors in sepsis, in hopes of discovering new methods to better
classify septic patients into informative subgroups, each with a
more uniform set of pathophysiological mechanisms.
Late-stage immunosuppression. Despite a focus on the proinflam-
matory aspects of sepsis, most deaths in sepsis occur from nosoco-
mial infections during a late prolonged immunosuppressed state,
even in the face of successful early, supportive therapies. Septic
patients have defects in innate and adaptive immunity, includ-
ing altered monocyte antigen presentation, decreased lympho-
cyte proliferation and responsiveness, and lymphocyte apoptosis
and anergy (18, 19). Prevention and/or treatment of this immune
deficiency should be a focal point for novel treatments; immuno-
stimulatory therapies such as IFN-γ and GM-CSF are being tested
in early-phase clinical trials (79), However, this dimension of sepsis
has been underrepresented in animal models; simple CLP mod-
els have splenic apoptosis but generally die too early before later
immunosuppression fully develops. More complex models have
been developed recently. A “two-hit” model of CLP followed by
instillation of bacteria (Pseudomonas aeruginosa or Streptococcus pneu-
moniae) mimics nosocomial infections that result from immune
depression (58, 80). In these animals, proinflammatory cytokines
(e.g., IL-6) were decreased, antiinflammatory (e.g., IL-10) cytokines
were increased, bacteria clearance was reduced, and profound
Histology of AKI in a clinically relevant model of CLP-induced sepsis.
Periodic acid–Schiff (PAS) staining of mouse kidney cortex 24 hours
after CLP surgery (A) or sham surgery (B). Pink staining of brush border
is visible in sham cortical tubules, and loss of brush border is evident,
as is mild dilation, after CLP. Vacuolization is seen after CLP in almost
all tubules, most prominently in two tubules in the upper-right corner
(circled). Original magnification, ×400. Scale bar: 200 μm. Reproduced
from the American Journal of Physiology: Renal Physiology (63).
science in medicine
2872? The?Journal?of?Clinical?Investigation http://www.jci.org Volume 119 Number 10 October 2009
lymphocyte apoptosis was observed. An alternative model for late
hypoimmune events involves CLP in mice and subsequent removal
of the necrotic cecum, which can lead to a complete recovery. Peri-
toneal macrophages isolated from CLP mice that had cecectomy
four days after sepsis induction had lower IL-6 production, indi-
cating a hypoimmune state (81). IFN-γ production by splenocytes
was suppressed in CLP mice but was reversed by cecectomy and
IL-10 injection (82). Recently, serum cytokine analysis in a com-
munity-acquired pneumonia cohort showed that mortality was
not predicted by either proinflammatory IL-6 or antiinflamma-
tory IL-10, but high levels of both cytokines were more predictive.
Thus, by the time of hospital admission, high levels of IL-6 and
IL-10 portend the worst mortality; either the early proinflamma-
tory phase had passed or the early and late phase are coincident
(83), which requires further investigation.
Additional clinically relevant factors. Other issues in model devel-
opment to be considered include, first, that host susceptibility to
pathogenic factors is species dependent. For instance, rodents are
much less sensitive to LPS than humans (67). Second, bacterial
virulence factors such as microbial toxins can accelerate sepsis.
Alverdy and colleagues (84) performed a 30% hepatectomy, which
allowed exotoxin A to be disseminated systematically after direct
injection of P. aeruginosa into the cecum. The mortality rate of
this sepsis model could be increased to 100% by modifying the
virulence of bacteria.
Large animal models and bacterial infusion/inoculation
Large animal sepsis models have been developed using LPS infu-
sion, CLP surgery, and bacterial infusion and inoculation. Since
these larger animals can be fully instrumented for measurement
of circulatory parameters and continuous infusion of fluids and
therapeutic agents, ICU fluid management protocols can be sim-
ulated, and hemodynamic status can be documented. Fink et al.
implanted a fibrin clot containing live E. coli in the peritoneal cav-
ity of dogs, which increased cardiac output and decreased blood
pressure and systemic vascular resistance (SVR) (85). This canine
sepsis model has been used for drug evaluation (86–89). Recently
a new canine sepsis model induced by intrabronchial Staphylococcus
aureus administration has been developed (90). In this model, treat-
ment with mechanical ventilation, antibiotics, fluids, vasopressors,
sedatives, and analgesics was adjusted based on algorithms similar
to the care provided for human sepsis. Renal dysfunction evaluat-
ed by serum creatinine and BUN was found in acute non-survivors
(<24 hours) and decreased urine output in subacute non-survivors
(24–96 hours). In a baboon model, a hyperdynamic state is preacti-
vated by injection of killed bacteria followed by injection with live
E. coli; renal dysfunction was also demonstrated (91–93). It is of
note that renal histological changes, including tubular epithelial
cell injury, fibrin deposits, and inflammatory cell infiltration, were
found in this model.
Hemodynamics. Failure of the renal circulation is thought to be
a crucial factor for developing sepsis-induced AKI; despite lack of
demonstrated efficacy, low-dose dopamine is often administered
to preserve renal blood flow (RBF). However, in recent reports with
a large animal sepsis model, AKI developed even with increased
RBF. In sheep infused continuously with live E. coli, serum creati-
nine increased despite a more than doubled RBF (94, 95); notably,
the expected redistribution of intrarenal circulation between cortex
and medulla (measured by implanted laser Doppler flow probes)
was also absent (96). On the other hand, rodent sepsis models of
A partial list of emerging therapeutic targets tested in animal models
? ? α-MSH
S189 HMGB1 is released from damaged cells in sepsis
HMGB1 activates NF-κB via RAGE, TLR2, TLR4
Apoptosis induces depletion of immune cells
Apoptosis impairs immunity by inducing anergy
CD95 fusion protein/siRNA
E. coli sepsis induces cytokine production via TLR4
Bacterial DNA induces cytokine production via TLR9
ODN TLR9 inhibitor
Systemic VEGF level increases in sepsis
VEGF-induced vascular leakage
C5a induces lymphocyte apoptosis and coagulation system failure
C5a induces HMGB1 release via C5L2
Neutralizing Ab to C5a
Neutralizing Ab to C5aR, C5L2
78, S138, S139
Vagus nerve stimulation attenuates an inflammatory response via α7nAChR
Acetylcholine inhibits HMGB1 release via α7nAchR
α-MSH decreases inflammatory cytokines and NO production
Ghrelin decreases HMGB1 release and has antibacterial activity
NETs trap and kill bacteria in blood and tissue
TLR4 activation induces NET formation
MSC reprograms macrophage to produce IL-10
Eritoran (TLR4 antagonist) S161
RAGE, receptor of advanced glycation end product.
science in medicine
? The?Journal?of?Clinical?Investigation http://www.jci.org Volume 119 Number 10 October 2009
LPS infusion or CLP developed renal microcirculatory failure, as
evaluated by more precise methods, including intravital two-pho-
ton video microscopy (13, 48, 97, 98). It should be noted that these
rodent sepsis models might have insufficient fluid resuscitation.
According to an alternative, inadequately tested hypothesis, AKI is
a successful adaptation that protects the body from losing sodium
by using tubuloglomerular feedback (TGF) to reduce GFR (99). In
septic humans, microcirculatory derangements can be detected by
orthogonal polarization spectral imaging (100). Indeed, mortal-
ity was better predicted by microcirculatory failure than systemic
hemodynamics (101, 102). Further investigation is necessary to
clarify the relative contribution of global RBF and microcircula-
tion to sepsis and sepsis-induced AKI and whether defects in one of
these vessel beds can be treated without compromising the other.
Potential targets for intervention
We summarize here a few of the many recently emerging potential
therapeutic targets for intervention in sepsis and sepsis-induced
AKI (Table 3). Developers of potential therapeutic targets still face a
daunting set of conceptual and logistical hurdles to improve upon
the current translational success rate. Ideally, they should be evalu-
ated in several different clinically relevant sepsis animal models to
minimize failures of translation to clinical trials. Timing should
also be considered: most therapeutic candidates are administered
to animal models before or at the time of sepsis initiation, but they
should also be tested for their limits of efficacy after delayed treat-
ment, i.e., their window of therapeutic opportunity. For instance,
treatment by ethyl pyruvate or antibodies against high-mobility
group box 1 (HMGB1) improved sepsis when administered as late
as 24 hours after onset (34, 103–105). However, these time frames
would be difficult to directly translate to clinical trials without
mechanism-based surrogate biomarkers, considering that animal
models diverge more at later times.
Toll-like receptors: hitting the snooze button
Inflammation in sepsis is largely initiated by TLRs, which detect
not only a wide range of microbial diversity, but also cellular con-
stituents released due to tissue injury, triggering innate immune
responses (106–110) (Table 4). HMGB1 can potentiate TLR ligand
signals (111) and perhaps even act as a TLR
ligand (112). Myeloid differentiation factor 88
(MyD88) is an adapter protein for all TLRs,
except TLR3, linking receptors with down-
stream kinases (106). These MyD88-depen-
dent pathways induce activation of NF-κB
and cytokine production such as TNF-α.
TLR4 and MyD88. TLRs are expressed pri-
marily in immune cells but also in solid
organs including brain, heart, lung, liver, and
kidney (113–117). TLRs expressed in the kid-
ney play a crucial role in ischemia/reperfusion
injury (118, 119) but not LPS-induced injury
(26). In a study with renal cross-transplanta-
tion between wild-type mice and mice from
the C3H/HeJ strain, which lacks TLR4 func-
tion due to a missense mutation, the authors
concluded that nonrenal TLR4 plays a major
role, but renal TLR4 plays only a minor role in
LPS-induced, TNF-α–mediated renal injury
consisting of tubular and vascular apoptosis.
However, because only BUN was used to evaluate renal function,
prerenal azotemia (from systemic TLR4-mediated inflammation)
cannot be ruled out, which would support the opposite conclu-
sion (26). TLR4 was induced in proximal and distal tubules after
CLP, perhaps as a protective response to endogenous TLR4 ligands
and/or anticipated endotoxin, and post-CLP injection of LPS was
taken up by proximal tubular cells (120). Direct LPS-induced renal
tubular cell damage might partly explain why many antiinflamma-
tory treatments failed to prevent renal injury. It is reported that
polymicrobial sepsis in rodents or humans is not dependent on
TLR4 (50, 121–123). Sepsis-induced AKI and other organ injury
were compared in wild-type mice, TLR4-deficient C3H/HeJ mice,
and MyD88-knockout mice (50). While sepsis-induced AKI was
dependent on MyD88 but not on TLR4, liver injury was indepen-
dent of either MyD88 or TLR4. On the other hand, blockade of
TLR4 pathway in a single bacteria (E. coli) injection model showed
remarkable protection (124), demonstrating that TLR4 inhibition
can be highly effective when given early, despite the presence of
other TLR ligands, and suggests that the uniformity of a bolus
injection of a single pathogen may be more amenable to treatment
than a sustained polymicrobial infection. Moreover, genetic dis-
ruption of MyD88 in mice improved survival rate in a CASP model
but not in a standard CLP model (122, 125). Survival analysis was
not performed in our CLP model in aged, fluid- and antibiotic-
treated mice (50). These disparate results highlight the importance
of systematically analyzing differences among sepsis animal mod-
els. It is of note that humans lacking MyD88 activity experience
only a limited set of severe infections; they are susceptible to strep-
tococcal and staphylococcal bacteria infection but typically resis-
tant to other microbes (S126). As described above, TLR4 pathway
blockade was effective in a mouse sepsis model of E. coli injection.
These data indicate that TLRs may play different roles in different
types of sepsis.
TLR9. The TLR9 pathway is involved in antiviral immune
responses but also recognizes bacterial DNA and specific unmeth-
ylated CpG–containing synthetic immunostimulatory oligonu-
cleotides (ODNs) (S127). TLR9 deficiency or an ODN inhibitor
of TLR9 improved CLP sepsis-induced mortality and AKI, with
improvements in downstream systemic pro- and antiinflamma-
Characteristics of TLRs
LPS, fibrinogen, heparan
sulfate, hyaluronic acid,
science in medicine
2874? The?Journal?of?Clinical?Investigation http://www.jci.org Volume 119 Number 10 October 2009
tory cytokines, bacteria clearance, and splenic apoptosis in both
standard (young, untreated) and aged fluid/antibiotic-treated
mouse CLP models (ref. 42, S128). TLR9 is primarily expressed on
dendritic cells, which can play a protective role in CLP-induced
sepsis (S129). The antimalarial drug chloroquine, which inhibits
the TLR9 pathway in vitro by preventing lysosomal acidification
(S130), also improves sepsis with a 6-hour therapeutic window
(ref. 42, S131). A TLR9-selective inhibitor, iCpG, was still effective
12 hours after CLP (S128), opening up the possibility that TLR9
is not limited to simple detection of bacterial CpG and could be
involved in amplification or resolution of inflammation. However,
in some other infection models, such as meningococcal sepsis and
Gram-negative bacterial pneumonia, blocking the TLR9 pathway
was harmful (S132, S133), whereas it was beneficial in the CLP
model (42). These discrepancies may be pathogen specific or sever-
ity specific (i.e., the intensity of infection alters the expression and
timing of the target).
Preventing end-organ damage: softening the blow
To prevent end-organ damage by sepsis, it will be important not
only to control dysregulated immunological reactions against
infection but also to reduce subsequent activation of nonim-
munological responses that damage organs. Many experimental
therapies are highly effective when given early — either prior to the
septic insult or shortly thereafter; however, they lose effectiveness
if given after the animal becomes clinically sick, corresponding
to the typical delay in diagnosis of sepsis. This is especially true
with antiinflammatory therapies, where a therapeutic target may
only be active transiently. While we do not often have the luxury
of such early treatment in patients as we do with animal models,
end-organ failure occurs well after diagnosis, which provides an
opportunity to intercept these late signals and mediators to mini-
mize irreversible organ damage. If organ failure can be decreased,
it should have a significant impact on long-term outcomes for
patients who survive sepsis.
VEGF. VEGF promotes microvascular integrity (S134), but high
levels of VEGF can cause vascular leakage by destruction of vascu-
lar barrier function (S135). In human sepsis, plasma VEGF levels
were increased, and VEGF concentration at admission correlated
with the severity of multiple organ dysfunction (S136, S137). By
trapping circulating VEGF, soluble FLT1 peptide improved sur-
vival of a mouse CLP model, even when administered 24 hours
after sepsis induction (S138, S139). In the mouse two-stage model
of CKD-sepsis described above, plasma VEGF levels in CKD ani-
mals were higher than normal and significantly increased by sep-
sis. VEGF is cleared by the kidneys, which could explain why CKD
enhances sepsis (78).
C5a. Extensive studies point to the complement system, particu-
larly C5a, as a therapeutic target (reviewed in ref. S140). Recently,
Rittirsch et al. reported that a C5a receptor, C5L2, which was previ-
ously considered a decoy receptor for C5a, contributes to the devel-
opment of sepsis by inducing HMGB1 release from macrophages
(S141). Anti-C5a antibody treatment improved CLP-induced renal
pathological injury (S142).
The neuroimmune axis: keeping a clear head
Stress, from both a neuroendocrine and cellular point of view, is an
added dimension that complicates the ability of the host to fight
infections. Much has been written about glucocorticoids as anti-
inflammatory agents (S143, S144) and catecholamines as pressors,
but usually the central nervous system is an afterthought in terms
of pathophysiology or considered too complex to study.
Increasing parasympathetic outflow. Vagus nerve stimulation attenu-
ates an inflammatory response induced by LPS, including TNF-α
release from activated macrophages (S145). This cholinergic anti-
inflammatory pathway is thought to act through the nicotinic ace-
tylcholine receptor α7 subunit (α7nAChR) (S146). A simple, direct
mechanism may be involved, as acetylcholine inhibits HMGB1
release from isolated macrophages through α7nAchR (S147). In
vivo, treatment with nicotine decreases serum HMGB1 levels and
improves survival of LPS- and CLP-induced sepsis animal models
(S147). Splenectomy and selective abdominal vagotomy dimin-
ish the antiinflammatory effects (TNF-α and HMGB1) of vagus
nerve stimulation or nicotine treatment in LPS and CLP sepsis
mice (S148). In splenectomized mice, nicotine treatment not only
fails to improve sepsis, but it worsens survival and increases serum
HMGB1 levels. These divergent results illustrate that interaction
between central nervous system and innate immune defense sys-
tem is not straightforward. In addition to the parasympathetic
nervous system, the sympathetic nervous system is also thought
to play a role in sepsis (reviewed in ref. S149).
Pituitary hormones. Several neuroendocrine mediators have recent-
ly been revealed to have immunomodulatory actions (reviewed in
ref. S150). We describe two recently reported examples. α-Melano-
cyte–stimulating hormone (α-MSH) is an antiinflammatory proo-
piomelanocortin-derived (POMC-derived) peptide from pituitary
gland, with additional effects including regulation of food intake
and energy expenditure (S151, S152). POMC levels are decreased
in septic rats (S153) and patients (S154). α-MSH and its analog
attenuate sepsis-induced AKI and mortality in the mouse CLP
model by improving systemic hemodynamics, pro- and antiinflam-
matory actions, renal and splenic NF-κB activation, and splenocyte
apoptosis (ref. 64, S155). Ghrelin is a gastrointestinal tract–derived
orexigenic peptide with potent antiinflammatory properties (S156,
S157). Ghrelin improved sepsis induced by LPS injection or CLP
up to 24 hours later by decreasing HMGB1 secretion from mac-
rophages (S158). Further, some of the beneficial effects of ghrelin
were sensitive to vagotomy (S159). Therefore, a thematic framework
appears to be coalescing around HMGB1 as an important nexus
between the immune response and the sympathetic, parasympa-
thetic, and hypothalamic-pituitary-adrenal axes.
Cellular therapies: fighting smarter
Conventional therapeutic strategies are systemic in nature, but
many symptoms and complications of sepsis emanate from a local
site of initial infection. Rather than applying a uniform treatment
globally, cellular therapies seek to enhance responses to local con-
ditions. Cellular therapies cannot replace antibiotics, but serve as
an adjunct to fight pockets of infection or regulate inflammation
on a local scale.
Neutrophil extracellular traps. After bacteria bind to platelet
TLR4, neutrophils are activated, resulting in the production of
extracellular fibers called neutrophil extracellular traps (NETs)
composed of granule proteins and chromatin that also kill bacteria
in blood and tissue (S160–S162). NET formation may also injure
endothelium and tissue, as neutrophil or platelet depletion attenu-
ated endothelial and hepatic injury in an LPS-induced sepsis model
(S161). Depletion of platelets also reduced neutrophil recruitment
to the lung and lung edema in CLP-induced sepsis (S163). Because
of its dependence on TLR4, NET formation may account for some
science in medicine
? The?Journal?of?Clinical?Investigation http://www.jci.org Volume 119 Number 10 October 2009
differences in AKI found in LPS versus CLP models; however, the
effect may be indirect, as NETs were found in liver sinusoids and
pulmonary capillaries but not in the kidney (S161).
Mesenchymal stem cells/bone marrow stromal cells. Mesenchymal
stem cells (MSCs) have been proposed to have immunosuppres-
sive properties and reduce inflammation (S164, S165) by inhibit-
ing T cell proliferation and B cell function (S166). MSC treatment
improved severe acute human graft-versus-host disease (S167) and
LPS-induced acute lung injury (S168). MSCs reduced nonseptic
AKI through paracrine effects (S169, S170). Recently, we dem-
onstrated that MSCs reduced mortality and multiple organ fail-
ure in a mouse CLP model (S171). MSCs transiently moved near
monocytes/macrophages and reprogrammed them to produce the
antiinflammatory cytokine IL-10 via PGE2 signaling. MSCs from
adipose tissue were also effective in treating CLP-induced sepsis
(S172). Because allogeneic MSCs can be injected intravenously
into the host without any immunosuppression treatment and can
be expanded ex vivo, MSC therapy has great potential for human
sepsis. Systemic PGE2 administration has not been demonstrated
to treat sepsis, whereas MSCs can respond to changing conditions
to deliver a localized, transient, proportionate release of PGE2. This
smarter local delivery vehicle should be effective in a broader range
of patients and reduce side effects including immunosuppression.
New biomarkers in sepsis-induced AKI
Despite numerous sophisticated scoring systems (acute physiol-
ogy and chronic health evaluation [APACHE; ref. S173], sepsis-
related organ failure assessment [SOFA, ref. S174], etc.), organ
failure metrics (e.g., ventilator-free days, dialysis-free days, etc.),
and pro- and antiinflammatory biomarkers (e.g., IL-6, procalcito-
nin, triggering receptor expressed on myeloid cells–1 [TREM-1],
IL-10, etc.) mortality is still used as the gold standard for evaluat-
ing new therapeutic agents and strategies. Because mortality from
sepsis is increased dramatically when complicated by AKI, early
detection and accurate evaluation of AKI is important in septic
patients. AKI has been mainly diagnosed by serum creatinine con-
centration, which can partly reflect GFR as a function of solute
elimination by the kidney. A recent international consensus defi-
nition for AKI attempts to adjust for large increases in risk that
accompany small increases in serum creatinine (6). However, this
definition still depends on changes in serum creatinine concen-
tration and urine output. Unfortunately, serum creatinine does
not accurately reflect the GFR in critically ill patients with fluid
shifts who are not in steady state. We recently discovered that
sepsis also decreases creatinine production (S175). In bilaterally
nephrectomized (BNx) mice, sepsis reduced
serum creatinine but increased nonrenal
organ injury markers and serum cytokine
levels, as expected. Treatment of sepsis with
chloroquine decreased nonrenal organ inju-
ry markers but paradoxically increased serum
creatinine. These data suggest that evalu-
ation of kidney injury by serum creatinine
alone would severely underestimate renal
injury, a serious failure of early diagnosis of
sepsis-induced AKI, and lead to an incorrect
assessment of drug efficacy.
One approach to finding a better alterna-
tive to serum creatinine is to test new renal
biomarkers that were initially developed to
detect other forms of AKI (S176). A few studies of novel renal bio-
markers have been reported in patients with sepsis-induced AKI
(reviewed in refs. S177, S178 and summarized in Table 5).
The development of drug-biomarker pairs has been critical in
other areas of medicine, especially oncology. Many agents, such as
TNF-α, can have opposing effects on sepsis in different contexts
(ref. 41, S179). Therefore, it will be critical to match agents with
a particular clinical context or “niche”: agent(s) must be given in
the correctly defined subpopulation that may be susceptible to
the agent, as in the case of the anticancer drugs Gleevec and Her-
ceptin (S180). Biomarkers can also be used to determine when the
drug target is available and active; how to adjust the dose to satu-
rate the target; and when to stop due to the recovery from disease
or no expected response. Giving agents to all the septic patients
without any stratification by biomarkers has been unsuccessful at
best, and in many cases harmful (S144, S181, S182). Considering
the complexity of sepsis within the patient population, as well as
the dynamic nature of sepsis in an individual patient, a panel of
biomarkers may be needed, based on each available therapy. For
example, patients can be screened for low endogenous protein C
before administration of activated protein C (S183). The use of
a genetically heterogeneous animal model may be advantageous
over the use of inbred strain-based models for development of bio-
markers and biomarker-therapeutic pairs.
Lessons learned. The failure to translate benefits seen in animal
models to the clinical setting has caused much soul searching in
the sepsis community (14–16). In the last several decades, animal
modelers have focused primarily on replicating clinical stages
(hemodynamic, inflammatory, and, most recently, immunosup-
pression) and outcomes (organ failure, mortality rate). We have
incorporated results from human epidemiological studies to
design and validate our animal models. We found that replicat-
ing genetic heterogeneity and critical clinical risk factors such as
advanced age and comorbid conditions (e.g., CKD) led to improved
models of sepsis and sepsis-induced AKI. Such disease models have
given the field new ways to addresses the vexing problem of how
to mimic a complex human disease in animal models. In the LPS
or simple CLP models without the genetic diversity or risk factors
(age, underlying disease), we had great difficulty titrating the dose
of LPS or CLP to produce AKI; a mild insult did not cause AKI, but
a slightly stronger insult caused death without AKI. Addition of
genetic diversity or risk factors made the animals more susceptible
to sepsis. When the severity of insult is then decreased, these mod-
Biomarker candidates for sepsis-induced AKI
Proximal tubule damage
Proximal tubule damage
Tubule transcription factor
Proximal tubule damage
Pediatric sepsis AKI
Adult AKI including sepsis
Adult sepsis AKI
Pediatric sepsis AKI
Adult sepsis AKI
NGAL, neutrophil gelatinase–associated lipocalin; L-FABP, L-type fatty acid–binding protein;
ATF3, activating transcription factor 3.
science in medicine
2876? The?Journal?of?Clinical?Investigation http://www.jci.org Volume 119 Number 10 October 2009
els became less pathogen dependent and more host dependent,
allowing the development of multiple organ failure to unfold dur-
ing the course of sepsis.
The added risk factors enabled a failed human clinical trial to be
replicated in an animal model of disease (48), an often overlooked
criterion for validating animal models. For any animal model of
sepsis, it will be important to demonstrate that the model possess-
es a “negative predictive value,” i.e., if a therapeutic lacks efficacy in
the treatment of human disease, does the animal model similarly
predict therapeutic failure? Addition of risk factors also changed
the responsiveness to treatments of sepsis. Monotherapies effective
in a simpler CLP model were no longer effective in more complex
models; combination therapies were needed to prevent disease in
the more complex models (63, 78). Perhaps similar drug combina-
tions will be required for human sepsis, analogous to combination
chemotherapy for cancer (S184) or HIV (S185).
Future directions for animal models. The disease models to date have
successfully mimicked the initial stages of sepsis; however, the later
stages of immunosuppression and secondary infections have only
been addressed by a few studies, and many mechanistic questions
have yet to be answered. When comparing different treatment win-
dows of opportunity in different animal models, some normal-
izing principle may be necessary, such as initiation of treatment
relative to mean survival time.
Can animal models help lead us to develop new therapeutic strat-
egies and diagnostic tools? Although there has been some success
with early goal-directed therapy (S186), this preemptive strategy has
a short/narrow window of opportunity, probably because the hemo-
dynamically amenable phase is short lived, then patients diverge
into multiple mechanistic directions. Because patients with sepsis
are exceedingly heterogeneous, with individual “disease clocks,” we
need to improve our ability to monitor and ultimately predict how
patients move along their individual disease trajectory (S187).
Furthermore, we are accustomed to having a single fixed thera-
peutic target (e.g., HMG-CoA reductase for hypercholesterolemia)
(S188), but in sepsis, the therapeutic targets vary with the under-
lying preexisting conditions, etiology of sepsis, time course, and
disease trajectory. Subsets of the septic patient population, such
as those with meningococcemia, can be replicated individually in
animal models. However, cases are typically more complex, and we
need new classification schemes and biomarkers to identify which
informative subset a patient belongs to, how far along the disease
trajectory a patient has progressed, and which biological targets
are active at any given time. Ideally, drug target assays would be
developed that indicate whether an administered drug was hit-
ting and saturating the target, allowing proper dose titration and
anticipation of side effects. Such schemes might even allow for the
selection of a rationally designed, synergistic combination therapy.
The predictive accuracy of such schemes can then be used to design
more accurate animal models in an interactive, iterative process.
Note: References S126–S198 are available online with this article;
This research was supported by the Intramural Research Program
of the NIH, NIDDK. We apologize to all the authors whose work
we could not cover due to space limitations.
Address correspondence to: Peter S.T. Yuen, Renal Diagnostics and
Therapeutics Unit, NIDDK, NIH, 10 Center Drive, Room 3N108,
Bethesda, Maryland 20892-1268, USA. Phone: (301) 402-6702;
Fax: (301) 402-0014; E-mail: email@example.com.
1. Levy, M.M., et al. 2003. 2001 SCCM/ESICM/ACCP/
ATS/SIS International Sepsis Definitions Confer-
ence. Crit. Care Med. 31:1250–1256.
2. Dellinger, R.P., et al. 2008. Surviving Sepsis Cam-
paign: international guidelines for management of
severe sepsis and septic shock: 2008. Crit. Care Med.
3. Angus, D.C., et al. 2001. Epidemiology of severe
sepsis in the United States: analysis of incidence,
outcome, and associated costs of care. Crit. Care
4. Martin, G.S., Mannino, D.M., Eaton, S., and Moss,
M. 2003. The epidemiology of sepsis in the United
States from 1979 through 2000. N. Engl. J. Med.
5. Dombrovskiy, V.Y., Martin, A.A., Sunderram, J.,
and Paz, H.L. 2007. Rapid increase in hospitaliza-
tion and mortality rates for severe sepsis in the
United States: a trend analysis from 1993 to 2003.
Crit. Care Med. 35:1244–1250.
6. Mehta, R.L., et al. 2007. Acute Kidney Injury Net-
work: report of an initiative to improve outcomes
in acute kidney injury. Crit. Care. 11:R31.
7. Russell, J.A., et al. 2000. Changing pattern of organ
dysfunction in early human sepsis is related to
mortality. Crit. Care Med. 28:3405–3411.
8. Uchino, S., et al. 2005. Acute renal failure in criti-
cally ill patients: a multinational, multicenter
study. JAMA. 294:813–818.
9. Bagshaw, S.M., et al. 2005. Prognosis for long-term
survival and renal recovery in critically ill patients
with severe acute renal failure: a population-based
study. Crit. Care. 9:R700–R709.
10. Neveu, H., Kleinknecht, D., Brivet, F., Loirat, P.,
and Landais, P. 1996. Prognostic factors in acute
renal failure due to sepsis. Results of a prospec-
tive multicentre study. The French Study Group
on Acute Renal Failure. Nephrol. Dial. Transplant.
11. Silvester, W., Bellomo, R., and Cole, L. 2001. Epi-
demiology, management, and outcome of severe
acute renal failure of critical illness in Australia.
Crit. Care Med. 29:1910–1915.
12. Schrier, R.W., and Wang, W. 2004. Acute renal fail-
ure and sepsis. N. Engl. J. Med. 351:159–169.
13. Wu, L., Gokden, N., and Mayeux, P.R. 2007. Evi-
dence for the role of reactive nitrogen species in
polymicrobial sepsis-induced renal peritubular
capillary dysfunction and tubular injury. J. Am. Soc.
14. Riedemann, N.C., Guo, R.F., and Ward, P.A. 2003.
The enigma of sepsis. J. Clin. Invest. 112:460–467.
15. Rittirsch, D., Hoesel, L.M., and Ward, P.A. 2007.
The disconnect between animal models of sepsis
and human sepsis. J. Leukoc. Biol. 81:137–143.
16. Dyson, A., and Singer, M. 2009. Animal models of
sepsis: why does preclinical efficacy fail to translate
to the clinical setting? Crit. Care Med. 37:S30–S37.
17. Esmon, C.T. 2004. Why do animal models (some-
times) fail to mimic human sepsis? Crit. Care Med.
18. Hotchkiss, R.S., and Karl, I.E. 2003. The patho-
physiology and treatment of sepsis. N. Engl. J. Med.
19. Riedemann, N.C., Guo, R.F., and Ward, P.A. 2003.
Novel strategies for the treatment of sepsis. Nat.
20. Wichterman, K.A., Baue, A.E., and Chaudry, I.H.
1980. Sepsis and septic shock--a review of laborato-
ry models and a proposal. J. Surg. Res. 29:189–201.
21. Remick, D.G., Newcomb, D.E., Bolgos, G.L., and
Call, D.R. 2000. Comparison of the mortality and
inflammatory response of two models of sepsis:
lipopolysaccharide vs. cecal ligation and puncture.
22. Michie, H.R., et al. 1988. Detection of circulating
tumor necrosis factor after endotoxin administra-
tion. N. Engl. J. Med. 318:1481–1486.
23. Tracey, K.J., et al. 1987. Anti-cachectin/TNF mono-
clonal antibodies prevent septic shock during
lethal bacteraemia. Nature. 330:662–664.
24. McNamara, M.J., Norton, J.A., Nauta, R.J., and Alex-
ander, H.R. 1993. Interleukin-1 receptor antibody
(IL-1rab) protection and treatment against lethal
endotoxemia in mice. J. Surg. Res. 54:316–321.
25. Taveira da Silva, A.M., et al. 1993. Brief report:
shock and multiple-organ dysfunction after self-
administration of Salmonella endotoxin. N. Engl.
J. Med. 328:1457–1460.
26. Cunningham, P.N., Wang, Y., Guo, R., He, G., and
Quigg, R.J. 2004. Role of Toll-like receptor 4 in
endotoxin-induced acute renal failure. J. Immunol.
27. Tiwari, M.M., Brock, R.W., Megyesi, J.K., Kaushal,
G.P., and Mayeux, P.R. 2005. Disruption of renal
peritubular blood flow in lipopolysaccharide-
induced renal failure: role of nitric oxide and cas-
pases. Am. J. Physiol. Renal Physiol. 289:F1324–1332.
28. Knotek, M., et al. 2001. Endotoxemic renal failure
in mice: role of tumor necrosis factor indepen-
dent of inducible nitric oxide synthase. Kidney Int.
29. Fisher, C.J., Jr., et al. 1996. Treatment of septic
shock with the tumor necrosis factor receptor:Fc
fusion protein. N. Engl. J. Med. 334?:1697–1702.
30. Fisher, C.J., Jr., et al. 1994. Recombinant human
interleukin 1 receptor antagonist in the treatment
of patients with sepsis syndrome. Results from a
randomized, double-blind, placebo-controlled
trial. JAMA. 271?:1836–1843.
31. Eskandari, M.K., et al. 1992. Anti-tumor necrosis
science in medicine
? The?Journal?of?Clinical?Investigation http://www.jci.org Volume 119 Number 10 October 2009
factor antibody therapy fails to prevent lethality
after cecal ligation and puncture or endotoxemia.
J. Immunol. 148:2724–2730.
32. Brandtzaeg, P., et al. 1988. Systemic meningococcal
disease: a model infection to study acute endotox-
inemia in man. Prog. Clin. Biol. Res. 272:263–271.
33. Johannes, T., Mik, E.G., and Ince, C. 2009. Nonre-
suscitated endotoxemia induces microcirculatory
hypoxic areas in the renal cortex in the rat. Shock.
34. Miyaji, T., et al. 2003. Ethyl pyruvate decreases sep-
sis-induced acute renal failure and multiple organ
damage in aged mice. Kidney Int. 64:1620–1631.
35. Wang, W., et al. 2002. Protective effect of renal
denervation on normotensive endotoxemia-
induced acute renal failure in mice. Am. J. Physiol.
Renal Physiol. 283:F583–F587.
36. Wang, W., et al. 2003. Interaction among nitric
oxide, reactive oxygen species, and antioxidants
during endotoxemia-related acute renal failure.
Am. J. Physiol. Renal Physiol. 284:F532–F537.
37. Buras, J.A., Holzmann, B., and Sitkovsky, M. 2005.
Animal models of sepsis: setting the stage. Nat. Rev.
Drug Discov. 4:854–865.
38. Deitch, E.A. 2005. Rodent models of intra-abdomi-
nal infection. Shock. 24(Suppl. 1):19–23.
39. Rittirsch, D., Huber-Lang, M.S., Flierl, M.A., and
Ward, P.A. 2009. Immunodesign of experimental
sepsis by cecal ligation and puncture. Nat. Protoc.
40. Freise, H., Bruckner, U.B., and Spiegel, H.U. 2001.
Animal models of sepsis. J. Invest. Surg. 14:195–212.
41. Alexander, H.R., et al. 1991. Treatment with recom-
binant human tumor necrosis factor-alpha pro-
tects rats against the lethality, hypotension, and
hypothermia of gram-negative sepsis. J. Clin. Invest.
42. Yasuda, H., et al. 2008. Chloroquine and inhibition
of Toll-like receptor 9 protect from sepsis-induced
acute kidney injury. Am. J. Physiol. Renal Physiol.
43. Ertel, W., et al. 1991. The complex pattern of cytokines
in sepsis. Association between prostaglandins, cachec-
tin, and interleukins. Ann. Surg. 214:141–148.
44. Holly, M.K., et al. 2006. Biomarker and drug-tar-
get discovery using proteomics in a new rat model
of sepsis-induced acute renal failure. Kidney Int.
45. Ganopolsky, J.G., and Castellino, F.J. 2004. A pro-
tein C deficiency exacerbates inflammatory and
hypotensive responses in mice during polymicro-
bial sepsis in a cecal ligation and puncture model.
Am. J. Pathol. 165:1433–1446.
46. Hollenberg, S.M., et al. 2001. Characterization of a
hyperdynamic murine model of resuscitated sep-
sis using echocardiography. Am. J. Respir. Crit. Care
47. Echtenacher, B., Falk, W., Mannel, D.N., and Kram-
mer, P.H. 1990. Requirement of endogenous tumor
necrosis factor/cachectin for recovery from experi-
mental peritonitis. J. Immunol. 145:3762–3766.
48. Yasuda, H., Yuen, P.S., Hu, X., Zhou, H., and Star,
R.A. 2006. Simvastatin improves sepsis-induced
mortality and acute kidney injury via renal vascular
effects. Kidney Int. 69:1535–1542.
49. Ayala, A., and Chaudry, I.H. 1996. Immune dys-
function in murine polymicrobial sepsis: media-
tors, macrophages, lymphocytes and apoptosis.
Shock. 6(Suppl. 1):S27–S38.
50. Dear, J.W., et al. 2006. Sepsis-induced organ fail-
ure is mediated by different pathways in the kid-
ney and liver: acute renal failure is dependent on
MyD88 but not renal cell apoptosis. Kidney Int.
51. Hoesel, L.M., et al. 2005. Harmful and protective
roles of neutrophils in sepsis. Shock. 24:40–47.
52. Haybron, D.M., et al. 1987. Alterations in renal
perfusion and renal energy charge in murine peri-
tonitis. Arch. Surg. 122:328–331.
53. Kalechman, Y., et al. 2002. Anti-IL-10 therapeutic
strategy using the immunomodulator AS101 in
protecting mice from sepsis-induced death: depen-
dence on timing of immunomodulating interven-
tion. J. Immunol. 169:384–392.
54. Matsukawa, A., Kaplan, M.H., Hogaboam, C.M.,
Lukacs, N.W., and Kunkel, S.L. 2001. Pivotal role
of signal transducer and activator of transcription
(Stat)4 and Stat6 in the innate immune response
during sepsis. J. Exp. Med. 193:679–688.
55. Yang, S., and Hauptman, J.G. 1994. The efficacy of
heparin and antithrombin III in fluid-resuscitated
cecal ligation and puncture. Shock. 2:433–437.
56. Kuhlmann, M.K., et al. 1994. New experimental
model of acute renal failure and sepsis in rats.
JPEN J. Parenter. Enteral. Nutr. 18:477–485.
57. Pedersen, P.V., et al. 1989. Hemodynamic and
metabolic alterations during experimental sep-
sis in young and adult rats. Surg. Gynecol. Obstet.
58. Muenzer, J.T., et al. 2006. Pneumonia after cecal
ligation and puncture: a clinically relevant “two-
hit” model of sepsis. Shock. 26:565–570.
59. Neild, G.H. 2001. Multi-organ renal failure in the
elderly. Int. Urol. Nephrol. 32:559–565.
60. Saito, H., Sherwood, E.R., Varma, T.K., and Evers,
B.M. 2003. Effects of aging on mortality, hypother-
mia, and cytokine induction in mice with endotox-
emia or sepsis. Mech. Ageing Dev. 124:1047–1058.
61. Turnbull, I.R., et al. 2003. Effects of age on mor-
tality and antibiotic efficacy in cecal ligation and
puncture. Shock. 19:310–313.
62. Dear, J.W., et al. 2005. Dendrimer-enhanced MRI
as a diagnostic and prognostic biomarker of sepsis-
induced acute renal failure in aged mice. Kidney Int.
63. Leelahavanichkul, A., et al. 2008. Methyl-2-acet-
amidoacrylate, an ethyl pyruvate analog, decreases
sepsis-induced acute kidney injury in mice. Am. J.
Physiol. Renal Physiol. 295:F1825–F1835.
64. Doi, K., et al. 2008. AP214, an analogue of alpha-
melanocyte-stimulating hormone, ameliorates
sepsis-induced acute kidney injury and mortality.
Kidney Int. 73:1266–1274.
65. Langenberg, C., Bagshaw, S.M., May, C.N., and Bello-
mo, R. 2008. The histopathology of septic acute kid-
ney injury: a systematic review. Crit. Care. 12:R38.
66. Sato, T., Kamiyama, Y., Jones, R.T., Cowley, R.A.,
and Trump, B.F. 1978. Ultrastructural study on
kidney cell injury following various types of shock
in 26 immediate autopsy patients. Adv. Shock Res.
67. Liptak, P., and Ivanyi, B. 2006. Primer: histopa-
thology of calcineurin-inhibitor toxicity in renal
allografts. Nat. Clin. Pract. Nephrol. 2:398–404.
68. Wan, L., et al. 2008. Pathophysiology of septic
acute kidney injury: what do we really know? Crit.
Care Med. 36:S198–S203.
69. Chawla, L.S., et al. 2007. Elevated plasma concen-
trations of IL-6 and elevated APACHE II score
predict acute kidney injury in patients with severe
sepsis. Clin. J. Am. Soc. Nephrol. 2:22–30.
70. Doetschman, T. 1999. Interpretation of pheno-
type in genetically engineered mice. Lab. Anim. Sci.
71. Williams, S.M., Haines, J.L., and Moore, J.H. 2004.
The use of animal models in the study of complex
disease: all else is never equal or why do so many
human studies fail to replicate animal findings?
72. Guidet, B., Aegerter, P., Gauzit, R., Meshaka, P.,
and Dreyfuss, D. 2005. Incidence and impact of
organ dysfunctions associated with sepsis. Chest.
73. Mehta, R.L., et al. 2004. Spectrum of acute renal
failure in the intensive care unit: the PICARD expe-
rience. Kidney Int. 66:1613–1621.
74. Naqvi, S.B., and Collins, A.J. 2006. Infectious com-
plications in chronic kidney disease. Adv. Chronic
75. Sarnak, M.J., and Jaber, B.L. 2000. Mortality caused
by sepsis in patients with end-stage renal disease
compared with the general population. Kidney Int.
76. Thamer, M., Ray, N.F., Fehrenbach, S.N., Rich-
ard, C., and Kimmel, P.L. 1996. Relative risk and
economic consequences of inpatient care among
patients with renal failure. J. Am. Soc. Nephrol.
77. James, M.T., et al. 2008. Risk of bloodstream infection
in patients with chronic kidney disease not treated
with dialysis. Arch. Intern. Med. 168:2333–2339.
78. Doi, K., et al. 2008. Pre-existing renal disease pro-
motes sepsis-induced acute kidney injury and
worsens outcome. Kidney Int. 74:1017–1025.
79. Pugin, J. 2007. Immunostimulation is a rational
therapeutic strategy in sepsis. Novartis Found. Symp.
280:21–27; discussion 27–36, 160–164.
80. Steinhauser, M.L., et al. 1999. IL-10 is a major medi-
ator of sepsis-induced impairment in lung antibac-
terial host defense. J. Immunol. 162:392–399.
81. Xiao, H., Siddiqui, J., and Remick, D.G. 2006.
Mechanisms of mortality in early and late sepsis.
Infect. Immun. 74:5227–5235.
82. Manley, M.O., O’Riordan, M.A., Levine, A.D., and
Latifi, S.Q. 2005. Interleukin 10 extends the effec-
tiveness of standard therapy during late sepsis with
serum interleukin 6 levels predicting outcome.
83. Kellum, J.A., et al. 2007. Understanding the inflam-
matory cytokine response in pneumonia and sep-
sis: results of the Genetic and Inflammatory Mark-
ers of Sepsis (GenIMS) Study. Arch. Intern. Med.
84. Laughlin, R.S., et al. 2000. The key role of Pseudo-
monas aeruginosa PA-I lectin on experimental gut-
derived sepsis. Ann. Surg. 232:133–142.
85. Fink, M.P., MacVittie, T.J., and Casey, L.C. 1984.
Inhibition of prostaglandin synthesis restores nor-
mal hemodynamics in canine hyperdynamic sepsis.
Ann. Surg. 200:619–626.
86. Kalil, A.C., et al. 2006. Preclinical trial of L-argi-
nine monotherapy alone or with N-acetylcysteine
in septic shock. Crit. Care Med. 34:2719–2728.
87. Natanson, C., et al. 1990. Antibiotics versus car-
diovascular support in a canine model of human
septic shock. Am. J. Physiol. 259:H1440–H1447.
88. Quezado, Z.M., et al. 1993. A controlled trial of
HA-1A in a canine model of gram-negative septic
shock. JAMA. 269:2221–2227.
89. Sevransky, J.E., et al. 1997. Tyrphostin AG 556
improves survival and reduces multiorgan failure
in canine Escherichia coli peritonitis. J. Clin. Invest.
90. Minneci, P.C., et al. 2007. A canine model of sep-
tic shock: balancing animal welfare and scien-
tific relevance. Am. J. Physiol. Heart Circ. Physiol.
91. Carraway, M.S., et al. 2003. Blockade of tissue fac-
tor: treatment for organ injury in established sep-
sis. Am. J. Respir. Crit. Care Med. 167:1200–1209.
92. Welty-Wolf, K.E., et al. 2001. Coagulation block-
ade prevents sepsis-induced respiratory and renal
failure in baboons. Am. J. Respir. Crit. Care Med.
93. Welty-Wolf, K.E., et al. 2006. Blockade of tissue
factor-factor X binding attenuates sepsis-induced
respiratory and renal failure. Am. J. Physiol. Lung
Cell. Mol. Physiol. 290:L21–L31.
94. Langenberg, C., Wan, L., Egi, M., May, C.N., and
Bellomo, R. 2006. Renal blood flow in experimental
septic acute renal failure. Kidney Int. 69:1996–2002.
95. Langenberg, C., Wan, L., Egi, M., May, C.N., and
Bellomo, R. 2007. Renal blood flow and function
during recovery from experimental septic acute
science in medicine Download full-text
2878? The?Journal?of?Clinical?Investigation http://www.jci.org Volume 119 Number 10 October 2009
kidney injury. Intensive Care Med. 33:1614–1618.
96. Di Giantomasso, D., Morimatsu, H., May, C.N.,
and Bellomo, R. 2003. Intrarenal blood flow dis-
tribution in hyperdynamic septic shock: effect of
norepinephrine. Crit. Care Med. 31:2509–2513.
97. Wu, L., et al. 2007. Peritubular capillary dysfunc-
tion and renal tubular epithelial cell stress follow-
ing lipopolysaccharide administration in mice. Am.
J. Physiol. Renal Physiol. 292:F261–F268.
98. Gupta, A., et al. 2007. Activated protein C amelio-
rates LPS-induced acute kidney injury and down-
regulates renal iNOS and angiotensin 2. Am. J.
Physiol. Renal Physiol. 293:F245–F254.
99. Thurau, K., and Boylan, J.W. 1976. Acute renal suc-
cess. The unexpected logic of oliguria in acute renal
failure. Am. J. Med. 61:308–315.
100. De Backer, D., Creteur, J., Preiser, J.C., Dubois, M.J.,
and Vincent, J.L. 2002. Microvascular blood flow
is altered in patients with sepsis. Am. J. Respir. Crit.
Care Med. 166:98–104.
101. De Backer, D., et al. 2006. The effects of dobuta-
mine on microcirculatory alterations in patients
with septic shock are independent of its systemic
effects. Crit. Care Med. 34:403–408.
102. Sakr, Y., Dubois, M.J., De Backer, D., Creteur, J.,
and Vincent, J.L. 2004. Persistent microcirculatory
alterations are associated with organ failure and
death in patients with septic shock. Crit. Care Med.
103. Yang, H., et al. 2004. Reversing established sepsis
with antagonists of endogenous high-mobility group
box 1. Proc. Natl. Acad. Sci. U. S. A. 101:296–301.
104. Wang, H., et al. 1999. HMG-1 as a late mediator of
endotoxin lethality in mice. Science. 285:248–251.
105. Ulloa, L., et al. 2002. Ethyl pyruvate prevents lethal-
ity in mice with established lethal sepsis and sys-
temic inflammation. Proc. Natl. Acad. Sci. U. S. A.
106. Akira, S., and Takeda, K. 2004. Toll-like receptor
signalling. Nat. Rev. Immunol. 4:499–511.
107. Coussens, L.M., and Werb, Z. 2002. Inflammation
and cancer. Nature. 420:860–867.
108. Medzhitov, R. 2007. Recognition of microorgan-
isms and activation of the immune response.
109. Seong, S.Y., and Matzinger, P. 2004. Hydrophobic-
ity: an ancient damage-associated molecular pat-
tern that initiates innate immune responses. Nat.
Rev. Immunol. 4:469–478.
110. Medzhitov, R. 2001. Toll-like receptors and innate
immunity. Nat. Rev. Immunol. 1:135–145.
111. Tian, J., et al. 2007. Toll-like receptor 9-dependent
activation by DNA-containing immune complexes
is mediated by HMGB1 and RAGE. Nat. Immunol.
112. van Zoelen, M.A., et al. 2009. Role of toll-like recep-
tors 2 and 4, and the receptor for advanced glyca-
tion end products in high-mobility group box 1-
induced inflammation in vivo. Shock. 31:280–284.
113. Basu, S., and Fenton, M.J. 2004. Toll-like receptors:
function and roles in lung disease. Am. J. Physiol.
Lung Cell. Mol. Physiol. 286:L887–L892.
114. Baumgarten, G., et al. 2001. In vivo expression of
proinflammatory mediators in the adult heart
after endotoxin administration: the role of toll-like
receptor-4. J. Infect. Dis. 183:1617–1624.
115. Chakravarty, S., and Herkenham, M. 2005. Toll-like
receptor 4 on nonhematopoietic cells sustains CNS
inflammation during endotoxemia, independent
of systemic cytokines. J. Neurosci. 25:1788–1796.
116. Matsumura, T., Ito, A., Takii, T., Hayashi, H., and
Onozaki, K. 2000. Endotoxin and cytokine regula-
tion of toll-like receptor (TLR) 2 and TLR4 gene
expression in murine liver and hepatocytes. J. Inter-
feron Cytokine Res. 20:915–921.
117. Tsuboi, N., et al. 2002. Roles of toll-like receptors
in C-C chemokine production by renal tubular epi-
thelial cells. J. Immunol. 169:2026–2033.
118. Leemans, J.C., et al. 2005. Renal-associated TLR2
mediates ischemia/reperfusion injury in the kid-
ney. J. Clin. Invest. 115:2894–2903.
119. Wolfs, T.G., et al. 2002. In vivo expression of Toll-
like receptor 2 and 4 by renal epithelial cells: IFN-
gamma and TNF-alpha mediated up-regulation
during inflammation. J. Immunol. 168:1286–1293.
120. El-Achkar, T.M., et al. 2006. Sepsis induces chang-
es in the expression and distribution of Toll-like
receptor 4 in the rat kidney. Am. J. Physiol. Renal
121. Feterowski, C., et al. 2003. Effects of functional
Toll-like receptor-4 mutations on the immune
response to human and experimental sepsis. Immu-
122. Weighardt, H., et al. 2002. Cutting edge: myeloid
differentiation factor 88 deficiency improves resis-
tance against sepsis caused by polymicrobial infec-
tion. J. Immunol. 169:2823–2827.
123. McMasters, K.M., Peyton, J.C., Hadjiminas, D.J.,
and Cheadle, W.G. 1994. Endotoxin and tumour
necrosis factor do not cause mortality from caecal
ligation and puncture. Cytokine. 6:530–536.
124. Roger, T., et al. 2009. Protection from lethal gram-
negative bacterial sepsis by targeting Toll-like recep-
tor 4. Proc. Natl. Acad. Sci. U. S. A. 106:2348–2352.
125. Peck-Palmer, O.M., et al. 2008. Deletion of MyD88
markedly attenuates sepsis-induced T and B lym-
phocyte apoptosis but worsens survival. J. Leukoc.