Infection of neonatal mice with bioluminescent E. coli O1:K1:H7 results in a septic infection with significant pulmonary inflammation and lung pathology. Here, we describe procedures to model and further study neonatal sepsis using longitudinal intravital imaging in parallel with enumeration of systemic bacterial burdens, inflammatory profiling, and lung histopathology.
Neonates are at an increased risk of bacterial sepsis due to the unique immune profile they display in the first months of life. We have established a protocol for studying the pathogenesis of E. coli O1:K1:H7, a serotype responsible for high mortality rates in neonates. Our method utilizes intravital imaging of neonatal pups at different time points during the progression of infection. This imaging, paralleled by measurement of bacteria in the blood, inflammatory profiling, and tissue histopathology, signifies a rigorous approach to understanding infection dynamics during sepsis. In the current report, we model two infectious inoculums for comparison of bacterial burdens and severity of disease. We find that subscapular infection leads to disseminated infection by 10 h post-infection. By 24 h, infection of luminescent E. coli was abundant in the blood, lungs, and other peripheral tissues. Expression of inflammatory cytokines in the lungs is significant at 24 h, and this is followed by cellular infiltration and evidence of tissue damage that increases with infectious dose. Intravital imaging does have some limitations. This includes a luminescent signal threshold and some complications that can arise with neonates during anesthesia. Despite some limitations, we find that our infection model offers an insight for understanding longitudinal infection dynamics during neonatal murine sepsis, that has not been thoroughly examined to date. We expect this model can also be adapted to study other critical bacterial infections during early life.
Bacterial sepsis is a significant concern for neonates that exhibit a unique immune profile in the first days of life that does not provide adequate protection from infection1. Neonatal sepsis continues to be a significant U.S. healthcare problem accounting for greater than 75,000 cases annually in the U.S alone2. To study these infections in depth, novel animal models that recapitulate aspects of human disease are required. We have established a neonatal mouse infection model using Escherichia coli, O1:K1:H73. E. coli is the second leading cause of neonatal sepsis in the U.S., but responsible for the majority of sepsis-associated mortality4,5. However, it is the leading cause when pre-term and very low birth-weight (VLBW) babies are considered independently5. The K1 serotype is most frequently associated with invasive bloodstream infections and meningitis in neonates6,7. Currently, there are no other treatment options beyond antibiotics and supportive care. Meanwhile, rates of antibiotic resistance continue to rise for many pathogenic bacteria, with some strains of E. coli resistant to a multitude of antibiotics commonly used in treatment8. Thus, it is imperative that we continue to generate methods to study the mechanisms of sepsis and the host response in neonates. These results can help to improve upon current treatments and infection outcomes.
The immune state of neonates is characterized by both phenotypic and functional differences compared to adults. For instance, elevated levels of anti-inflammatory and regulatory cytokines, such as interleukin (IL)-10 and IL-27, have been shown to be produced by cord blood-derived macrophages and are present at greater levels in the serum of murine neonates9,10,11. This is consistent with lower levels of IFN-α, IFN-ɣ, IL-12, and TNF-α that are frequently reported from neonatal cells compared with adult counterparts10. Additionally, the neonatal immune system is skewed toward a Th2 and regulatory T cell response as compared to adults12. Elevated numbers of neutrophils, T cells, B cells, NK cells, and monocytes are also present in neonates, but with significant functional impairments. This includes defects in expression of cell surface markers and antigen presentation that suggest immaturity13,14,15. Additionally, neonatal neutrophils are significantly deficient in their ability to migrate to chemotactic factors16. Myeloid-derived suppressor cells (MDSCs) are also found at elevated levels in neonates and recently shown to be a source of IL-2711. MDSCs are highly suppressive toward T cells17. Collectively, these data demonstrate limitations in neonatal immunity that lend to increased susceptibility to infection.
To study the progression of the bacterial burden and dissect protective host immune responses during neonatal sepsis, we have developed a novel infection model. Neonatal mice at days 3-4 of life are difficult to inject in the intraperitoneal space or tail vein. In our model, day 3 or 4 pups are administered the bacterial inoculum or PBS subcutaneously into the scapular region. A systemic infection develops and using luminescent E. coli O1:K1:H7, we can longitudinally image individual neonatal mice to follow the disseminated bacterial burden in peripheral tissues. This is the first reported model to utilize intravital imaging to understand the kinetics of dissemination of bacteria during sepsis in murine neonates3.
Here, we describe a protocol to induce septic E. coli infections in neonatal mice3. We describe how to prepare the bacterial inoculum for injection, and how to harvest tissue for assessment of pathology, measurement of inflammatory markers by gene expression analysis, and enumeration of the bacterial burden. In addition, the use of luminescent E. coli for intravital imaging of infected neonates and quantification of bacterial killing by neonatal immune cells is also described. These protocols may also be adapted to study other important bacterial infections in neonates. The data presented here represents an overall novel approach to understanding infection dynamics in a translatable neonatal sepsis model.
All procedures were approved by the West Virginia Institutional Animal Care and Use Committees and conducted in accordance with the recommendations from the Guide for the Care and Use of Laboratory Animals by the National Research Council18.
1. Preparation of Bacterial Inoculum
2. Animal identification
3. Subscapular inoculation
NOTE: For this study, 2 experiments were performed with a low-dose and high-dose group designated for each experiment. In the first experiment, 7 pups were given the low dose inoculum (4 pups were used as controls), and 5 pups from a separate litter were given the high dose (3 pups were used as controls). The pups from experiment 1 provided data for only the 24 h timepoint. In the second experiment, 8 pups were given the low dose inoculum (2 pups were used as controls), and 6 pups were given the high dose inoculum (2 pups were used as controls). Pups from experiment 2 provided data for the 0, 10, and 24 h timepoints.
4. Evaluation of disease and endpoint criteria
5. In vivo imaging of bacterial burden
6. Euthanasia
7. Tissue harvest
8. RNA isolation from lung tissue for gene expression
9. cDNA synthesis
10. Real-time quantitative PCR (qPCR) cycle
11. Lung histopathology
12. In vitro bacterial killing assay
This protocol induced bacterial sepsis in neonatal mice, and we used longitudinal intravital imaging, enumeration of bacteria in the blood, histological assessments of pathology, and inflammatory cytokine expression profiles to study the course of disease. Signs of morbidity were observed in neonatal pups infected with both low (~2 x 106 CFUs) and high (~7 x 106 CFUs) inoculums of E.coli over time. Pups that received the greater inoculum displayed more prominent signs of distress that included reduced mobility, the inability to correct their posture, and impaired ability to maintain an upright position by 24 h post-infection (hpi). There was, however, a range of morbidity as some pups appeared worse than others. Immediately following infection, one low-dose animal died due to isoflurane exposure during an imaging session to establish baseline. By 24 hpi, two of six high-dose animals succumbed to the systemic infection (33.3% mortality). Infected pups that received either a high or low dose inoculum weighed significantly less than their control littermates at 24 hpi (Figure 1A,B). All the pups that received the higher inoculum met endpoint criteria at 24 hpi. As such, all the infected pups in this group were euthanized following imaging. Bacteria in the blood were enumerated for a subset of mice that received the lower inoculum, and for all animals that received the higher inoculum since they were all euthanized. The results from two experiments performed similarly indicate that while most animals had high levels of bacteria in the blood (CFUs/mL) at 24 hpi, some animals did not have detectable bacteria in the blood (Figure 1C). The latter suggest that they cleared the infection by this time point. As expected, pups that received the higher inoculum had nearly three orders of magnitude more CFUs/mL at 24 hpi relative to pups that received the low dose inoculum (Figure 1C).
Live animal imaging of luminescent bacteria further confirmed the dissemination of bacteria and increase in growth in neonatal pups over time at 10 and 24 hpi (Figure 2 and Figure 3). Additionally, using intravital imaging with the microCT, we were able to identify infection foci, including the brain (Figure 2B), lungs (Figure 2B, Figure 3A,B), and other peripheral tissues (Figure 2B). The lungs of some highly infected mice demonstrated opaque regions consistent with inflammatory consolidation that co-localized to luminescent bacterial signal (Figure 3A). These regions of presumed inflammatory exudate are not found in uninfected control lungs (Figure 3A). Further evidence of a pronounced inflammatory cytokine response within the lungs of infected pups is demonstrated by gene expression analysis of IL-1β, IL-6, and TNF-α. A significant increase in expression relative to controls was observed for all three cytokines in both the low and high inoculum groups (Figure 4A). Histopathology of the lung was also examined at 24 hpi in control and infected pups. Despite similar inflammatory cytokine profiles, a progressive increase in pathology was commonly observed from the lower to the higher inoculum. Compared with tissue from uninfected controls, the lungs of infected pups showed notable inflammatory changes, thickening of the alveolar wall, increased alveolar hemorrhaging, and inflammatory infiltration (Figure 4B). In the most severe infections, the pulmonary congestion and areas of hemorrhage contributed to a massive reduction in open air space (Figure 4B). Collectively, these results demonstrate that in our model of early onset neonatal sepsis, dissemination of luminescent bacteria can be followed over time from a subscapular inoculation site to important infection foci and cause significant inflammation and pathology in severely infected animals.
To study host factors that contribute to bacterial killing by innate immune cells such as monocytes, macrophages, and neutrophils, we developed a sensitive in vitro assay to measure bacterial clearance. Ly6B.2+ cells isolated from the spleens of neonatal mice were infected with bioluminescent E. coli at a range of MOIs for 1 h and then treated with gentamicin to kill extracellular bacteria. At 3, 6, 20, and 48 hpi, intracellular luminescence was measured with a multimode reader. As expected, with increasing MOI, more luminescent signal was recorded at 3 h (Figure 5). Gradually, this signal was lost, indicative of bacterial clearance (Figure 5). This assay is amenable to supplemented cytokines, neutralization of secreted effectors, and the addition of pharmacological inhibitors of cellular pathways to study interventions that may promote bacterial clearance and serve to improve outcomes in the neonatal sepsis model described here.
Figure 1: Changes in body weight and bacterial replication in septic neonatal mice.
(A,B) Individual mouse weights within a group (low and high) expressed as a percentage of the mean weight of littermate control pups. Data are presented as mean percentage ± SEM. Individual t-tests at each post-infection time point reveal significant differences at 24 h between control pups and pups that received the low inoculum (p<0.0001) (A), or between control pups and pups that received the high inoculum (p=0.0031) (B). (C) CFU/mL in the blood at 24 hpi were log transformed and presented as the mean ± SEM. Mann-Whitney test reveals a trend towards significance between the low and high dose inoculums (p=0.0882). Please click here to view a larger version of this figure.
Figure 2: Intravital imaging demonstrates dissemination of bacteria in neonatal mice over time.
(A) A representative neonatal mouse (#1) infected with an inoculum of ~2 x 106 CFUs is shown at time 0, 10, and 24 hpi. A colorimetric scale with the minimum and maximum radiance values per time point are displayed for each time point. Mice at 0 and 10 h are displayed on both their time point scale and the 24 h scale to demonstrate changes in bacterial growth over time. (B) Representative 3D reconstructed microCT images of the same neonatal mouse at 10 and 24 hpi are shown. Each time point has images at overhead, transaxial, and coronal perspectives. In the transaxial image at 24 hpi, the plane has moved toward the periphery of the mouse to better display infection foci in the peripheral tissues. White arrows indicate the brain and kidney at 10 hpi and the kidney and lung at 24 hpi. Please click here to view a larger version of this figure.
Figure 3: Lungs are a site of major infection during bacterial sepsis in neonates.
(A) Representative 3D reconstructed microCT images of a neonatal mouse (#5) infected with an inoculum of ~7 x 106 CFUs are shown at 24 hpi compared to an uninfected control. Both mice are displayed in the transaxial perspective and lungs are indicated by white arrows. The infected mouse was placed on two radiance (photons/sec) scales. Scale #1 includes all 6 wavelengths (500, 520, 560, 580, 600, 620 nm) and scale #2 includes only 500, 520, and 560 nm wavelengths. This second scale allowed us to visualize an increased signal in bacteria in the lungs because lower wavelengths are more highly absorbed by tissue and produce stronger signal. (B) Representative 3D reconstructed microCT images of a neonatal mouse (#4) infected with an inoculum of ~7 x 106 CFUs are shown at 24 hpi. This time point has images at the overhead, sagittal, transaxial, and coronal perspectives. White arrows indicate infection foci in the lungs. Please click here to view a larger version of this figure.
Figure 4: Inflammation and associated histopathological findings in the lungs of septic neonates.
At 24 hpi the lungs were harvested from pups that received ~2 x 106 or 7 x 106 CFUs or uninfected controls. (A) RNA was isolated and the expression of IL-1β, IL-6, or TNF-α as determined relative to uninfected controls by quantitative real-time PCR using the formula 2-ΔΔCt. The data is shown as the mean log2 transformed change in expression ± SEM for each inoculum as indicated. Statistical significance was determined using unpaired t-tests of ΔCt values between individual cytokine genes and the internal control in the 95% confidence interval. Asterisks indicate p<0.01. (B-D) Histopathologic sections of H&E stained lung tissues (20x, area of interest constructed into clipping mask and enlarged for clarity) are shown. Lung tissues from a representative uninfected control (B) or infected neonate at the low (C) or high (D) inoculum are shown. Yellow arrows indicate alveolar thickening (C) or hemorrhaging (D). Scale bar = 500 μm. Please click here to view a larger version of this figure.
Figure 5: An in vitro assay for bacterial clearance.
Ly6B.2+ cells were isolated from the spleens of uninfected control neonates. Cells were seeded in 96-well plates and infected with luciferase-expressing E. coli O1:K1:H7 at a multiplicity of infection (MOI) of 10, 50, or 250 as indicated. After 1 h, the medium was replaced with fresh that contained gentamicin (100 µg/mL). Mean relative light units (RLU) ± SE for an individual experiment representative of multiple are shown. Statistical significance in the 95% confidence interval was determined using unpaired t tests with Welch’s correction; asterisks indicate p<0.05. Please click here to view a larger version of this figure.
Our subscapular infection model for inducing bacterial sepsis in neonatal mice is a novel method to study the longitudinal spread of bacterial pathogens in real time. Intravital imaging provides the opportunity to explore bacterial dissemination in real time in neonates. This is critical to understand the kinetics of bacterial dissemination and to further study the host response and damage at the appropriate phase of disease. Mouse pups are administered a subcutaneous, subscapular injection of bacterial inoculum. This injection technique is simpler than other commonly used alternatives, such as the tail vein and intraperitoneal infections, as it requires less precision within an injection site. This is important given the small size of the pups. The intravital imaging allows for a longitudinal assessment of bacterial proliferation and dissemination into peripheral tissues and the central nervous system over time without the need to sacrifice the animal. Similar imaging approaches and technologies have been used for the study of cancer biology and metastasis20,21. Furthermore, while another study has cited the use of bioluminescent imaging during an E. coli infection in neonatal rats22, here, we have applied the approach to neonatal mice, wherein our methodology allows evaluation of bacterial kinetics during murine sepsis. Visualization of the bacteria is based on emission of bioluminescent light at various wavelengths from bacteria (e.g., bacterial luciferase activity) within the animal. Bioluminescence is then visualized through a cooled charged coupled device (CCD) camera. The resulting visualized bioluminescence can then be reconstructed into a 3D image that shows both spatial- and temporal-dependent effects of bacteria within an animal. For an added, more nuanced layer of data acquisition, successful animal identification through tail tattoo allows for a repeated measures assessment of individual pups across time and the identification of possible outliers within a given experimental group.
The most successful application of the described model requires accuracy in preparation of the bacterial inoculum. Here, we describe an optimized method for bacterial preparation using a pre-established and validated E. coli growth curve that reduces variation between the target and actual inoculum. This allows experimental reproducibility at an intended inoculum. The inclusion of two inoculums in our model demonstrated dose-dependent outcomes in blood CFUs, mortality, and lung pathology. However, some aspects of the disease trajectory were not dose dependent. The failure to gain weight in infected animals was not dependent on the inoculum at 24 hpi. Additionally, similar levels of inflammatory cytokine expression were observed in the lung in response to infection with both inoculums. Whether or not this pattern would be replicated in all tissues where bacteria were observed, such as the kidney, liver, spleen and brain, remains to be determined. In addition to sepsis, E. coli O1:K1:H7 is associated with meningitis in the neonatal population23. This brain infection occurs when bacteria from the periphery invade and penetrate the blood brain barrier. Future studies will explore this aspect of the model through analysis of changes in tight junction protein expression, as well as test different ranges of bacterial inoculums. An additional modification during intravital imaging includes the addition of a singular cotton ball, doused in isoflurane, which is placed approximately 2-3 inches away from the mice during imaging. In response to previous experiments wherein the neonatal pups have regained consciousness during the imaging session, preventing accurate image acquisition, we now place the cotton ball close enough to the mice to keep them continuously anesthetized during imaging. However, it is important this is not done so close that they fail to recover from the anesthesia.
Although flexible and easily adaptable for the study of the kinetics of different bacteria in various animal and disease models, our protocol has some limitations to consider. The first limitation to consider is that the subscapular route of infection does not mirror a natural route of transmission. However, a primary objective in the development of our model from the outset was to establish an easily reproducible mode of delivery that could be used to establish a systemic infection that replicates aspects of human disease. Therefore, in this report, we describe a model of human early onset sepsis disease syndrome, not a model of natural transmission. There is an established model of oral delivery in neonatal rats that replicates some aspects of common human transmission, such as initial colonization of E. coli infection in the alimentary canal and subsequent dissemination to the bloodstream and peripheral tissues, including the brain22. The model established by Witcomb and colleagues also incorporates bioluminescent E. coli and intravital imaging. Moreover, it is crucial to minimize isoflurane exposure, as well as inject, tail tattoo, and handle pups as quickly as possible without compromising accuracy and precision of the techniques in an attempt to mitigate stress levels for both the neonates and the dams. In some cases, if the pups experience enhanced human-induced and/or experimental manipulations, the dams can stop nursing and caring for the pups, resulting in decreased survival unrelated to the infection. Similarly, pups that are exposed to isoflurane for prolonged periods beyond the approximate 10 minutes of an imaging session have an increased risk of death; thus it is crucial to supply just enough isoflurane to sufficiently anesthetize the mice, but not enough to euthanize them. A final point of consideration is the limit of sensitivity. Tissues in which less than 104 CFUs/mL E. coli were enumerated the luminescent signal recorded falls at the low end of the detectable range, according to the scaling method used in the imaging software3. Thus, some tissues may be colonized with low levels of bacteria but appear without visible bioluminescence.
Currently, most studies utilize adult methods of bacterial dissemination, such as intraperitoneal (i.p.) and tail vein injections for neonates. Pluschke and Pelkonen analyzed the effect of E. coli K1 on neonatal mice through i.p., tail vein, and oral infections24. This study demonstrated that different genotypes of mice with immunodeficiencies are more susceptible to the K1 strain; however, many aspects of host immune response to infection as well as the mechanisms for bacterial spread are left unaddressed. Deshmukh and colleagues infected neonatal mice intraperitoneally with E. coli K1 or K. pneumoniae and measured CFUs in the spleen and liver at 72 hpi25. This study also analyzed some aspects of host response to infection based on pre-exposure of mice to antibiotics. However, thorough investigation of bacterial dissemination to peripheral tissues and blood over time in parallel with inflammatory profiling in the same tissue (other than granulocytosis) was not addressed. Other studies of neonatal sepsis in mice with Staphylococcus aureus, Staphylococcus epidermidis, Group B Streptococcus, and E. coli explore varying aspects of the host immune system in response to infection. However, none of these studies utilize intravital imaging to explore the kinetics of bacterial dissemination or localization of infection foci23,25,26,27. Our method of infection and intravital imaging, combined with bacterial burden assessment and inflammatory profiling of peripheral tissues, allows us to comprehensively examine aspects of both the host and pathogen during infection, providing a more precise understanding of host-pathogen interplay during sepsis.
We intend to utilize this infection and imaging model to further our understanding of early-onset neonatal sepsis using a variety of pathogenic bacteria commonly responsible for sepsis in neonates, including Group B streptococci, K. pneuomoniae, and Listeria monocytogenes. This infection model will allow us to longitudinally compare dissemination of different bacterial pathogens in parallel with the host response in neonates. In addition, this model is adaptable to the adoptive transfer of specific (fluorescently conjugated) immune cell types to study their migration to sites of infection and subsequent influence on the host response and control of bacteria. This grants the opportunity to better understand the host-pathogen interactions that occur during sepsis in early life in ways that have not been previously demonstrated.
The authors have nothing to disclose.
This work was supported by institutional funds to C.M.R.
1 mL Insulin Syringe | Coviden | 1188128012 | Inoculum or PBS injection |
10% Neutral Buffered Formalin | VWR | 89370-094 | Histopathology |
ACK Lysis Buffer | Gibco | LSA1049201 | Bacterial clearance assay |
Animal Tattoo Ink Paste | Ketchum | KI1482039 | Animal identification |
Animal Tattoo Ink Green Paste | Ketchum | KI1471039 | Animal identification |
Anti-Ly-6B.2 Microbeads | Miltenyi Biotec | 130-100-781 | Cell isolation |
Escherichia coli O1:K1:H7 | ATCC | 11775 | |
Escherichia coli O1:K1:H7-lux (expresses luciferase) | N/A | N/A | Constructed in-house at WVU |
E.Z.N.A. HP Total Extraction RNA Kit | Omega Bio-tek | R6812 | RNA extration |
DPBS, 1X | Corning | 21-031-CV | |
Difco Tryptic Soy Agar | Becton, Dickinson and Company | 236950 | Bacterial growth |
IL-1 beta Primer/Probe (Mm00434228) | Thermo Fisher Scientific | 4331182 | Cytokine expression qPCR |
IL-6 Primer/Probe (Mm00446190) | Thermo Fisher Scientific | 4331182 | Cytokine expression qPCR |
iQ Supermix | Bio-Rad | 1708860 | Real-time quantitative PCR |
iScript cDNA Synthesis Kit | Bio-Rad | 1708891 | cDNA synthesis |
Isolation Buffer | Miltenyi Biotec | N/A | Bacterial clearance assay |
IVIS Spectrum CT and Living Image 4.5 Software | Perkin Elmer | N/A | Intravital imaging |
LB Broth, Lennox | Fisher BioReagents | BP1427-500 | Bacterial growth |
EASYstrainer (Nylon Basket) | Greiner Bio-one | 542 040 | Cell strainer |
SpectraMax iD3 | Molecular Devices | N/A | Plate reader |
Pellet Pestle Motor | Grainger | 6HAZ6 | Tissue homogenization |
Polypropylene Pellet Pestles | Grainger | 6HAY5 | Tissue homogenization |
Prime Thermal Cycler | Techne | 3PRIMEBASE/02 | cDNA synthesis |
TNF-alpha Primer/Probe (Mm00443258) | Thermo Fisher Scientific | 4331182 | Cytokine expression qPCR |
TriReagent (GTCP) | Molecular Research Center | TR 118 | RNA extration |