Abstract
Noninvasive real-time in vivo bioluminescent imaging was used to assess the spread of Streptococcus pneumoniae throughout the spinal cord and brain during the acute stages of bacterial meningitis. A mouse model was established by lumbar (LP) or intracisternal (IC) injection of bioluminescent S. pneumoniae into the subarachnoid space. Bacteria replicated initially at the site of inoculation and spread progressively from the spinal cord to the brain or from the brain down to the cervical part of the spinal column and to the lower vertebral levels. After 24 hr, animals showed strong bioluminescent signals throughout the spinal canal, indicating acute meningitis of the intracranial and intraspinal meninges. A decline in bacterial cell viability, as judged by a reduction in the bioluminescent signal, was observed over time in animals treated with ceftriaxone, but not in untreated groups. Mice treated with the antibiotic survived infection, whereas all mice in untreated groups became moribund, first in the IC group then in the LP group. No untreated animal survived beyond 48 hr after induction of infection. Colony counts of infected cerebrospinal fluid (CSF) correlated positively with bioluminescent signals. This methodology is especially appealing because it allows detecting infected mice as early as 3 hr after inoculation, provide temporal, sequential, and spatial distribution of bacteria within the brain and spinal cord throughout the entire disease process and the rapid monitoring of treatment efficacy in a nondestructive manner. Moreover, it avoids the need to sacrifice the animals for CSF sampling and the potential manipulative damage that can occur with other conventional methods.
Introduction
Experimental animal models have been instrumental in understanding the pathophysiology and treatment of bacterial meningitis [2,5,10,15,20]. Unfortunately, there are no convenient methods to confirm whether an animal has been successfully inoculated. This is especially true at early stages of the infection prior to the manifestation of disease symptoms or before bacterial counts and biochemical analysis of the cerebrospinal fluid (CSF) have been performed [7]. Additionally, repeated sampling of an individual animal's CSF to determine bacterial counts for assessment of the host's response to infection or antibiotic treatment is restricted, as the manipulative damages can cause neuronal destruction or even death to the animal [12]. In rodent models, access to their small central nervous system (CNS) compartments for CSF sampling represents a formidable technical task and, in most instances, is the end point of the experiment [11], thus not permitting the disease course to be followed in the same animal and comparative values to be assessed. These procedures thus result in large animal-to-animal variations and use of high numbers of experimental animals. Although the rabbit model of meningitis does allow for the investigator to obtain a limited number of sequential sampling of CSF over time from one animal [5,12,20] as with other such models, it does not provide data in real time or supply information such as spatial distribution or temporal patterns of the infectious agent within the compartments of CNS as the infection progresses.
With the recent advances in imaging technology and bioluminescent reporter systems, researchers can now monitor infections noninvasively and nondestructively from various sites within the live infected animal [3,8,13,19]. To facilitate early detection of bacterial meningitis and follow the disease course in real time, noninvasively, we used a strain of bioluminescent Streptococcus pneumoniae and low light imaging system for direct visualization of infection from outside intact live animals. In addition, we demonstrated the convenience of this approach in assessing the in vivo efficacy of antibiotic therapy for meningitis in real time in live animals.
Materials and Methods
Bacterial Strain and Growth Conditions
The bacterial strain used in this study was S. pneumoniae A 66.1 (type 3) that was made bioluminescent by the integration of a modified lux operon onto its chromosome [8] and designated as S. pneumoniae Xen 10. The organism was grown in brain heart infusion supplemented with 10% fetal calf serum, and the inoculum was prepared as previously described [12].
Mouse Model of Meningitis
All experimental procedures for the animals were performed in accordance with guidelines of the Institutional Animal Care and Use Committee. A total of 51 female Balb/C mice (Charles River, Wilmington, MA) weighing 18–22 g were anesthetized with 2.5% isoflurane and their head and spine area were shaved to expose the inoculating sites, allowing accurate inoculation of the pathogen to the subarachnoid space. Using a 30.5-gauge needle and a Hamilton syringe, 10 μL of a suspension containing 104 CFU of S. pneumoniae Xen 10 in sterile pyrogen-free saline was slowly delivered intracisternally (IC) or via lumbar puncture (LP) at the interspace between the first and second lumbar vertebrae. Another group of mice were mock infected with sterile saline and served as uninfected controls. Mice were observed until they recovered from anesthesia, at which time they showed no evidence of behavioral abnormalities. Twelve hours after infection, a group of IC-infected mice (n = 13) were treated subcutaneously with 100 mg/kg ceftriaxone [6,9] in 0.1 mL saline. The treatment was repeated every 12 hr for 3 days (total of 4 treatments). The minimal inhibitory concentration/minimal bactericidal concentration of ceftriaxone for S. pneumoniae Xen 10 was 0.015/0.03 μg/mL, respectively. At indicated time points, two mice from each experimental group were euthanized, and a skin incision was made over the suboccipital muscles and thoracolumbar junction under a preparation microscope. The cisterna magna was punctured as described by Carp et al. [1] and CSF was drawn using a 27.5-gauge butterfly cannula connected to a 1.0-mL syringe. After CSF collection, brains were quickly removed, weighed, and homogenized in 1.0 mL PBS. For the determination of bacterial titers, serial dilutions of the CSF, blood and tissue homogenates were plated on sheep blood agar and incubated overnight at 37°C with 5% CO2 to allow bacterial colonies to develop.
Imaging Procedure
Prior to imaging, mice were anesthetized with 2–2.5% isoflurane gas. Animals were then imaged for a maximum of 5 min at various times following inoculation and treatment using an IVIS Imaging System 100 Series (Xenogen, Alameda, CA). During the imaging, mice were maintained in an anesthetized state by constant delivery of 2.5% isoflurane through an IVIS anesthesia manifold placed inside the imaging chamber. Total photon emissions from defined regions of interest (ROIs) within the images of each mouse were quantified using the Living Image software (Xenogen, Alameda, CA) as described previously [13].
Results
Visualization of Infection
Progression of S. pneumoniae infection was noninvasively monitored using bioluminescent imaging. This allowed the pathogen burden, serial quantification, localization, and spread to be clearly seen in live animals. Mice could be imaged as early as 3 hr following inoculation to show bioluminescent bacteria around the LP and IC inoculation sites in all S. pneumoniae Xen 10 infected animals (Figure 1). The signal intensities increased and expanded in a time-dependent fashion in both groups of animals, indicating the growth and spread of bacteria beyond the inoculated sites. The dissemination of the bacteria from occipital squama and throughout the cerebral hemisphere started as early as 10 hr postinfection in IC-inoculated animals. Approximately 22 hr after LP infection, bioluminescence signal was detected in the head, suggesting the transition of infection from spine to the brain. Quantitative analysis of bacterial titers in CSF confirmed that pneumococcus was indeed present and had spread to the brain. The photon intensity from regions of interest in the head and spine increased exponentially over the next few hours and reached approximately 108 photons/sec around the brain in the IC-inoculated group (Figure 2A). The brain signal in LP-inoculated animals also increased but required more time to reach the levels of IC-inoculated mice. The signal intensity in the brains of IC-challenged mice was nearly 1–2 logs more than the signal observed in the heads of LP-inoculated mice, especially during the early stages of the infection (Figure 2A and B). The total photon emission from the brains of IC- and LP-challenged mice after 24 hr was 2 × 107 and 2 × 106 photons/sec, respectively. The number of bacteria enumerated by CFU in the brains of IC-inoculated mice was also higher than in brains of LP-inoculated mice, confirming bioluminescence data. The extension of the bioluminescence signal from the brain down the spinal canal of IC-inoculated animals started approximately 22 hr after inoculation, suggesting the dissemination of infection from the brain to the spinal cord took longer than the spread from the spine to the brain (Figure 1). Temporal progression of infection as determined by lateral images demonstrated the distribution of bacteria within the entire spinal column, regardless of route of inoculation. Interestingly, the dorsal view of LP-inoculated mice showed two distinct regions with bioluminescent signal in the brain and lower vertebral regions, while the lateral view showed a continuous signal along the entire length of spinal column. The lack of signal between the head and the lumbar region could be due to several reasons. The relative difference in bacterial titers in this anatomical region may give rise to a low level of bioluminescence or the forward angulations of the back bone, which is a typical clinical sign of meningitis, allow the head and mid region of the spine to be closer to the camera. Because the current bioluminescence technology collects two-dimensional images, the regions closer to the camera potentially could give relative greater light emission. This limitation could be overcome by imaging animal in several different positions (Figure 1). To better identify sites of infection, a limited number of animals from IC- and LP-inoculated groups were sacrificed and excised brain and spinal cords were imaged. The source of light was indeed confirmed to be emanating from these tissues, substantiating data obtained in intact animals (data not shown). An association between disease severity and signal intensity was seen in both IC- and LP-inoculated mice. Peak levels of brain signal in IC-infected animals occurred around 36 hr postinfection and the animals turned to be moribund ~40 hr after the infection. In contrast, total photon emission in the brains of LP-inoculated mice peaked at around 48 hr, coinciding with moribund animals. Severely sick animals (IC or LP inoculated) showed bioluminescence signals outside the brain and spinal cord, demonstrating the dissemination of bacteria beyond intracranial and intraspinal meninges. Quantification of bacteria in blood from these animals confirmed bacteria in the blood stream. Control animals injected with sterile saline showed no altered health status.
In vivo bioluminescence monitoring of S. pneumoniae Xen 10 in the mouse model of meningitis. A representative animal from IC- or LP-inoculated group is shown. Imaging in multiple projections shows the spatial and temporal distribution of pathogen within the brain or spinal column during acute meningitis. The total photon emission from defined ROIs (ROI for brain and spinal column, shown on 0 hr image) were quantified using Living Image software. Arrows indicates the IC and LP injection sites.
Real-time monitoring of the spread of S. pneumoniae Xen 10 after IC or LP injection. The total photon intensity from ROI of (A) the brain and (B) the spinal cord was quantified using Living Image Software, and cumulative results are shown as bacterial growth. Each data point is the mean and standard error of three or four mice.
Real-Time Monitoring of Antimicrobial Therapy
To determine whether bioluminescence imaging can be used to effectively monitor efficacy of antibiotic therapy for the treatment of pneumococcal meningitis in real time, a group of mice with S. pneumoniae infections due to IC inoculation were treated with ceftriaxone. This antibiotic has been demonstrated to be effective in treating pneumococcal meningitis in both mouse models and human patients [9,10,18]. Treatment with ceftriaxone was started 12 hr after infection and continued every 12 hr for three consecutive days. Prior to treatment, the mean bacterial titer in the CSF was approximately 3.8 × 106 CFU/mL, corresponding to 4.2 × 106 photons/sec. In the treated groups, a marked reduction in the bioluminescent signal was observed after the first treatment, indicating a decline in bacterial cell viability (Figure 3). After the second treatment, there was almost a 100% reduction in the signal by Day 2, indicating clearance of the infection. The quantitative time-course analysis of emitted bioluminescence in untreated animals continued to increase over time, reaching ~108 photons/sec at the end of Day 2. The severity of clinical symptoms among these animals also increased with time, and became moribund ~48 hr postinfections.
Real-time monitoring of the effect of ceftriaxone on S. pneumoniae Xen 10 meningitis in the mouse model following IC infection. Viable bacteria in CSF were determined by culture and are reported as CFU per μL of CSF, and bioluminescence is represented as the total number of photons measured (photons/sec) using the IVIS Imaging System 100 Series, and the data plotted with respect to time. Arrows indicate the days of antibiotic administration. Each data point is the mean and standard error of three to four mice.
To confirm that the decline in light emission from these animals was due to bacterial cell death, a group of mice were sacrificed for collection of CSF, brain, and spinal cord and the number of CFU was determined using the conventional plate count method. As Figure 3 shows, recovery of bacteria (CFU) from the CSF of these animals corresponded closely with the bioluminescent signal in both treated and untreated animals with a correlation coefficient of 0.99. In treated animals, the decline in metabolic activity as determined by the bioluminescent signal was immediate (Figure 3). In contrast, estimation of pathogen burden with traditional methods required overnight incubation, and took considerably longer to demonstrate a response to treatment. No bacteria could be cultured from CSF of animals that had no detectable bioluminescent signal following treatments with ceftriaxone, indicating that bioluminescence is a convenient method of monitoring infection and response to treatment.
Discussion
In the present study, we have examined the distribution of S. pneumoniae infection within the intracranial and intraspinal meninges during acute stages of meningitis. Following the delivery of this pathogen into the CNS using two different inoculation routes, we were able to visualize the disease process noninvasively and provide real-time information regarding the bacterium's spread within the CNS compartment. We also demonstrated the use of this approach to study the in vivo therapeutic response of ceftriaxone against pneumococcal meningitis.
Monitoring the progression of meningitis and the spread of bacteria within the spinal canal and brain is difficult using conventional methods due predominantly to difficulties in sampling limitations of their CSF. Demonstration of bacteria within these sites requires time- and labor-intensive methods such as microbiological, biochemical, immunological, or histological examinations of harvested tissues. In contrast, bioluminescent imaging relies upon the metabolic activity of the lux engineered bacterial cells to report light (photons) that is quantifiable, thus allowing the cell number and location to be determined noninvasively by simple measuring of the light that penetrates the tissue of the living animal. The ability to observe bioluminescent bacteria as early as 3 hr postinfection and an inoculum as low as 104 CFU shows the advantage of using in vivo biophotonic imaging for monitoring the progression of infection. Moreover, the immediate real-time detection of cell perturbation following antibiotic treatment clearly demonstrates the potential of this technology in preclinical drug efficacy studies. Similar to the results presented here, other investigators have also demonstrated that light emission in vivo closely parallels the relative CFU of bacteria expressing the lux operon; a correlation that has been established in several different disease models and during the period of action of antibiotics [3,8,14,19,21]. Compared to real-time bioluminescence monitoring, CFU determinations rely on counting viable numbers of bacteria after overnight incubation and therefore took considerably more time to demonstrate a response to treatment. In addition to quantitative time-course analysis, emitted bioluminescent signals also provided both spatial and sequential information of the progression of infection.
Biophotonic imaging of bacterial or viral meningitis has opened up the possibility of following infection in more depth, studying pathogen-host interactions, and identifying virulence factors involved in the disease process [4,6,16,17]. In this study, we have demonstrated that the external monitoring of bioluminescent signal intensity permitted rapid, noninvasive assessment of the distribution of S. pneumoniae and progression of infection within the spinal cord and brain during acute stages of infection. Because of inhalant anesthesia, mice can be repeatedly imaged with short acquisition times. Furthermore, the possibility of simultaneous measurement of four to six mice at a time is ideal for time-course studies of therapeutic agents. We believe that this approach permits a more convenient and rapid method by which to monitor an infection, allowing disease process to be seen and treated earlier noninvasively and in the same animal. The methodology is especially appealing because it is possible to visually track spatio-temporal patterns of bacteria within the spinal canal and brain and serially quantify the pathogen without sacrificing the experimental animal. This allows a more complete picture of the disease to be drawn, allowing better, more informed, therapeutic development strategies to be devised.
Footnotes
Acknowledgments
The authors thank C. Dalesio (Graphics and Communications, Xenogen) for assistance with drawings.