A Clinical Scoring System for the Evaluation of Corynebacterium bovis-associated Disease in NSG Mice

Abstract

Clinical signs of Corynebacterium bovis infections are well-known in athymic nude mice. However, C. bovis can also infect and cause clinical signs in many hirsute, immunocompromised mouse strains such as NSG (NOD.Cg-Prkdc^scidIl2rg^tm1Wgl/SzJ). Typically, the clinical assessment of C. bovis-infected mice begins when overt clinical signs are initially observed and thus the early course of infection has not been thoroughly described. The goal of this study was to characterize the clinical progression of C. bovis infection in NSG mice under experimental conditions and develop a quantifiable clinical scoring system. For the development and application of this clinical scoring system, 54 naïve NSG mice were exposed to soiled bedding from clinically ill C. bovis-infected NSG mice and the emergence of clinical signs was monitored and scored weekly for 8 wk. Overall, we identified 6 benchmark changes associated with C. bovis clinical infection. Four changes were the appearance of the eyes, ears, hair coat, and posture. Two behavioral changes were increased grooming activity and rapid head shaking. All clinical signs appeared consistently and progressed temporally with increasing clinical severity. Characterization of clinical signs and scoring of clinical disease will aid veterinarians in the assessment of C. bovis-infected NSG mice and may help in the evaluation of current and future clinical interventions used to prevent or treat C. bovis-infected immunodeficient mice.

Introduction

The gram-positive bacteria Corynebacterium bovis is a pathogen of immunodeficient rodents used in biomedical research. C. bovis is one of the few pathogens of research rodents that produces clinical signs and remains enzootic in a large percentage of modern research vivaria. For over 3 decades, detection of C. bovis contamination in immunodeficient mouse colonies have relied primarily on characteristic clinical signs. Despite this, the clinical progression of C. bovis infection has not been completely characterized in commonly used immunodeficient mouse strains such as NSG (NOD.Cg-Prkdc^scid Il2rg^tm1Wgl/SzJ). Similarly, the incubation period prior to the development of clinical signs, such as skin hyperkeratosis commonly associated with C. bovis-infected athymic nude mice, has only been estimated.^2,4,10,16 Clinical signs for hirsute severe combined immunodeficient (SCID) mice have been described.¹⁷ However, these data do not characterize the clinical progression or address the potential for spontaneous clinical resolution that has been documented for athymic nude mice.^2,16

The goal of this study was to experimentally induce clinical C. bovis infections in NSG mice and characterize the clinical signs and progression of clinical disease to develop a quantifiable scoring system. The intent of a valid and practical clinical scoring system is not to subvert a definitive diagnosis by molecular diagnostics. Rather, a clinical scoring system can aid veterinarians in predicting clinical progression of infected NSG mice used in preclinical research studies. Similarly, providing temporal benchmarking of clinical signs may help assess the benefit of current and future clinical interventions used to prevent or treat infections and ultimately help eliminate C. bovis from research facilities.

Materials and Methods

Fifty-four female, 7-wk-old, NOD.Cg-Prkdc^scid Il2rg^tm1Wgl/SzJ mice (NSG, strain number 005557) were obtained from Jackson Laboratories (Bar Harbor, ME). This strain was selected because it is popular for preclinical cancer research and to our knowledge, was the first commercially available SCID mouse with a mutation in the ɣ chain of the IL-2 receptor.¹⁸ The vendor documented the mice to be free of endoparasites, ectoparasites, excluded microbial pathogens including follicular mites, and C. bovis. A representative skin swab was taken from newly arriving mice to confirm the absence of C. bovis by qPCR. Husbandry practices that included a specific disposable glove change technique for working with C. bovis-infected and uninfected mice were implemented as described previously.^9,14 All animal manipulations were approved by the IACUC of CU Anschutz Medical Campus, an AAALACi-accredited institution.

C. bovis inoculum and inoculation.

Recently, we reported that 15% (4 of 26) NSG mice inoculated with 2 × 10⁷ CFU of C. bovis isolate CUAMC1 developed characteristic clinical signs of C. bovis infection including conjunctivitis, ear hyperemia, and a rough hair coat.¹⁴ Development of clinical signs after inoculation of immunodeficient mice with cultured C. bovis is uncommon.^4,5,10 To perpetuate the clinical expression of disease with a known bacterial isolate, we transferred 50 mL of soiled bedding from clinically-affected NSG mice to a cage containing 3, 7-wk-old, naïve NSG mice. This method resulted in the transmission of clinical C. bovis infection, with characteristic clinical signs, to all exposed mice. For the current study, NSG mice were exposed at 8 to 32 h after arrival to 50 mL of soiled bedding collected from clinically affected C. bovis-infected NSG mice at 7 to 10 wk after their exposure. Uninfected, NSG mice, used as controls, were age and sex matched. We did not provide an acclimation period of more than 32 h before exposure to mimic the arrival of naïve mice into an enzootically infected colony. After exposure of NSG to soiled bedding, a complete cage change was performed after 7 d, without transfer of bedding, nesting material, or cage furniture.

Scoring of clinical signs.

To develop the clinical scoring system, 33 NSG mice were housed as sequential cohorts of 3 to 4 mice per cage. Naive mice were exposed to C. bovis with soiled bedding as described above. Mice were initially monitored daily for characteristic clinical signs of C. bovis infection until trends were observed in clinical progression. Monitoring interval was adjusted to weekly due to the slow rate of progression. To help maintain agent containment, only clinical signs that could be observed through the transparent, polysulfone cage wall were recorded with a single exception as described below. Clinical signs were then characterized based on present or absent, progressive, order of appearance, and behaviors that changed in frequency. To capture clinical progression or regression, most clinical signs were assigned to a scale in which a value of 1 equates to normal appearance, and values greater than 1 represented mild, moderate, and severe clinical signs. Signs that were either present or absent (binary) were assigned scores of 1 or 0, respectively (Table 1).

Table 1.

Clinical scoring system for NSG mice^*

Benchmark	Score	Description
Eyes	1	Normal
	2	Conjunctivitis, blepharitis, blepharospasm
	3	Periorbital alopecia
Ears	1	Normal
	2	Hyperemia at ear base only
	3	Diffuse hyperemia, semitransparent
	4	Hyperkeratosis on pinnae
Hair Coat	1	Normal
	2	Mild rough/unkempt
	3	Alopecia on face
	4	Alopecia on forelimbs and ventral body
	5	Muzzle swelling or physically thick skin†
Grooming	1	Normal 0–2 grooming events over 60 s
	2	3 or more brief events
	3	Prolonged episode(s), but < 30 s
	4	Continuous for ≥ 30 s
Adjunctǂ	0	Normal
	1	Hunched posture present
	1	Head shaking present

^*Minimum score of 4 and maximum score of 18.

^{† Assessment that may require direct animal manipulation to physically feel skin thickness during restraint.}

^ǂBinary clinical signs that if present a score of 1 is given for each.

Ocular changes (conjunctivitis, blepharitis, and blepharospasm, either unilateral or bilateral) were the first observed clinical signs (Figure 1). If unilateral conjunctivitis presented initially, both eyes were affected by the following week in all cases. Later, increased grooming activity contributed to peri-orbital alopecia, providing a progressive step in scoring of the eyes.

Figure 1.

Photographs of the eyes of NSG mice with clinical C. bovis infections. Images represent the step-wise increase in clinical score with a concurrent increase in clinical severity. Images and scores show (1) clinically normal, (2) conjunctivitis, blepharitis, and blepharospasm, and (3) all factors in score 2 plus periorbital alopecia.

With regard to the ears, bilateral dermal hyperemia began at the base of the ears at the level of the antihelix and spread distally to the apex. Inflammation of the pinnae ultimately changed the ears’ translucence as they became more hyperemic and opaque, with eventual development of hyperkeratosis of the pinnae (Figure 2).

Figure 2.

Photographs of the ears of NSG mice with clinical C. bovis infections. Images represent the step-wise increase in clinical score with a concurrent increase in clinical severity. Images and scores show (1) clinically normal, (2) hyperemia at the ear base with retained translucence at the apex of the pinnae, (3) diffuse hyperemia and opaque pinnae, and (4) all characteristics in score 3 plus hyperkeratosis.

The character of the overall hair coat of infected mice is the most common observation by research staff, often referring to mice as ‘scruffy.’ Alopecia was progressive, allowing stepwise scoring based on the overall coat quality and regions of alopecia. In later stages of infection, an increase in the skin thickness of NSG mice was physically palpable when the nape of the neck was grasped for routine restraint. This was the only assessment that may require animal manipulation if muzzle swelling is not observed or recognized (Figure 3).

Figure 3.

Photographs of the hair coat of NSG mice with clinical C. bovis infections. Images represent the step-wise increase in clinical score with a concurrent increase in clinical severity. Images and scores show (1) a normal hair coat, (2) a slightly unkempt appearance, (3) an unkempt appearance with facial alopecia, (4) equivalent findings to score 3 with the addition of ventral body alopecia that includes the limbs and ventral neck [photo not shown], and (5) equivalent findings to a score of 4 with the addition of muzzle swelling and/or physically thickened skin of the scruff [palpable but not visible].

The definitions for grooming activity were loosely based on a pruritus severity scoring system developed for client-owned pruritic dogs.^6,15 To score this change in behavior, cages were observed for 1 min to score grooming activity prevalence and duration. Quantitative benchmarks were set in prevalence to define normal (n ≤ 2 events) all the way to ≥ 30 s of continuous grooming as a maximum score. Scoring grooming activity used a mix of quantitative and qualitative assessment to define intermediate values (Table 1). After the initial cage manipulation, grooming activity was scored first, and then mice were evaluated on all other parameters.

Hunched body posture is one of the binary parameters that was scored based on its presence or absence. It is often associated with illness in mice and can also be observed with C. bovis clinical disease. Rapid and vigorous head shaking was also scored based on its presence or absence. Headshaking was repeated during normal activities and was identified as a consistent behavior associated with disease progression. Total duration of cage observation for 3 mice was approximately 3 to 4 min.

For the prospective study, 21 NSG mice were obtained and housed 3 mice per cage. Only the finalized scoring system was used (Table 1). As with the development of the scoring system, all cage observations were performed by the same male individual between 1600 and 1800 hr during a daily 14-hr light cycle ending at 2000 hr. Cages were removed from the IVC rack, rotated by hand horizontally 180 degrees to wake the mice up, and then placed back onto the rack so that 4/5^th of the cage extended from the rack for easy cage-side observation. Due to the relatively long clinical progression of disease in NSG mice, body weight measurements were also obtained weekly. Similarly, during routine husbandry management of cages, water consumption was monitored based on the need for water bottle replacement. Neither body weight nor water consumption were incorporated into the scoring system in order to retain a primarily cage-side scoring system.

Photography.

Representative images of mice at various stages of clinical progression were taken using a Powershot G16 camera (Canon USA, Melville, NY), 12.1 megapixel, using the automatic setting and TTL auto focus with auto white balance. Mice were awake for image acquisition because anesthesia resulted in significant blanching of skin hyperemia. JPG images were imported into Photoshop CS5 (Adobe, San Jose, CA) and autocorrected for contrast and manually adjusted for brightness and color level to accurately represent the clinical condition observed. Finally, an auto-sharpen filter was applied while maintaining the native resolution of 180 dpi.

C. bovis analysis.

To correlate clinical signs with the absolute copy number of C. bovis DNA present on a subset of mice, a combined oral and skin swab (PurFlock Ultra, Puritan, Guilford, ME) was collected. Samples were collected at arrival (week 0), and at weeks 3, 5, and 8 after exposure (n = 4 mice/time point). These time points were selected to provide a baseline, the week clinical signs were first observed, a point with a marked increase in clinical progression, and at the time of peak clinical signs, respectively. Samples were submitted to the University of Colorado Anschutz Medical Campus Quantitative PCR Core, processed, and analyzed as described previously.¹¹

Statistics.

All graphs and statistical analyses were performed by using SigmaPlot 11.2 (Systat Software, Point Richmond, CA). To detect a clinically relevant difference in clinical scores, we used pilot data to perform a t test sample size calculation. This analysis showed that a total of 12 mice in the experimental group and 9 mice in the control group would allow detection of a difference in means of 1.8 with a SD of 1.2 with an ⟨ equal to 0.05 and a power of 0.9 when comparing clinically affected and naïve controls at week 3. Because body weight data failed a parametric test, a Mann–Whitney rank sum test was performed to compare the clinical scores and body weight between 2 groups at any given time point. Descriptive data are expressed as mean ± SEM.

Results

All mice exposed to 50 mL of soiled bedding from NSG mice with clinical C. bovis infection developed clinical signs. Conjunctivitis was the first clinical sign observed, presenting as early as week 2 (P ≤ 0.05, mean = 2.8 ± 0.8 wk, Figure 4A) postexposure. At 3 wk after exposure, changes to the ears were observed (P ≤ 0.05, 3.5 ± 0.5 wk, Figure 4B). A decrease in hair coat quality was first observed at wk 4 (P ≤ 0.05, 4.4 ± 0.5 wk, Figure 4C) after exposure, as was the increase in grooming behavior. Time to increased grooming activity was variable, presenting as early as week 3, but most commonly at week 5 (P ≤ 0.05, 5.3 ± 1.2 wk, Figure 4D) after exposure. The mean onset of hunched posture (5.2 ± 0.4 wk, Figure 4E) and head shaking (6.3 ± 0.9 wk, Figure 4E) were the final scored components to present clinically; these were never observed in control mice.

Figure 4.

Panel of figures showing individual scores for each of the 6 benchmark clinical signs and cumulative score. Scores based on a progressive scale for clinical C. bovis-infected mice (n = 12) include (A) eyes, (B) ears, (C) hair coat quality, and (D) grooming behavior, all as compared with control mice (n = 9). (E) Binary clinical signs (that is, absent or present) are hunched posture and head shaking, presented in a reverse Kaplan-Meier curve. (F) Stacked bar chart showing the cumulative clinical score as the sum of each benchmark component, allowing the clinical signs to be assessed relative to each other and temporally. Mean ± SD, P ≤ 0.05

The presentation of the clinical signs occurred in a stepwise fashion, with benchmark changes in the eyes, ears, hair coat, grooming activity, hunched posture, and head shaking occurring weekly from wk 2 through 6 (Figure 4F). The overall clinical score of C. bovis-infected NSG mice with clinical disease is statistically different from uninfected control mice at week 3 (P ≤ 0.05; Figure 5A). At no point did the total clinical score decrease for mice with clinical C. bovis infections over the 8-wk study.

Figure 5.

Cumulative clinical score for C. bovis-infected mice (n = 12) depicted as a line plot with mean ± SD (A) as compared with the stacked bar chart in Figure 4E. Temporal correlation of C. bovis specific qPCR data for the detection of absolute C. bovis copies on the skin of C. bovis-infected NSG mice and control mice (n = 4/group) (B). Mean ± SD, P ≤ 0.05

At 3 wk after exposure, C. bovis-infected mice clearly had C. bovis DNA on their skin, with more than 10⁶ C. bovis DNA copies per swab (Figure 5B). With increasing severity of clinical signs, mice with clinical C. bovis began drinking more water at week 6 and 7 as determined by the need for water bottle replacement (data not shown), and demonstrated a significant 8% decrease in body weight at both weeks 7 and 8 as compared with uninfected control mice (Figure 6).

Figure 6.

Mean body weight ± SEM for clinically C. bovis-infected mice (n = 12) as compared with control mice (n = 9) over the duration of the study. P ≤ 0.05

Discussion

Herein, we describe the creation and use of a novel clinical scoring system for the assessment of NSG mice clinically infected with C. bovis. The goal of this effort was to define the clinical signs of C. bovis infection, the sequence in which they appear, and their rate of progression for this popular immunodeficient mouse strain for a period of 8 wk after exposure.

The incubation time prior to a statistically significant increase in clinical score for NSG mice was 3 wk under these experimental conditions. Despite the emergence of clinical signs at 3 wk, qPCR data demonstrates a tremendous amount of C. bovis present on the skin of these mice at 3 wk (Figure 5). It is at this point when marked bacterial shedding is likely and clinical signs are just emerging when significant environmental contamination occurs from the lack of infection awareness. Nevertheless, clinical signs began with the eyes, which developed changes that included conjunctivitis, blepharitis, and blepharospasm. However, in the absence of other clinical signs, these changes are less likely to be immediately linked to a systemic skin infection like C. bovis. By week 4, changes to the visual appearance of the ears became more prominent. The combination of ear-base hyperemia and ocular changes are likely to be identified as abnormal by skilled observers such as animal care and veterinary staff. By week 5, the ears are bright red, in stark contrast to the white hair coat, making this clinical sign very easy to observe even from a distance. In addition, the combination of squinty, swollen eye lids, bright red ears, decrease in hair coat quality, and hunched posture are likely to be easily identified by the novice observer as being abnormal. Thus, week 5 is when we expect that veterinary consultation is most commonly sought by research or husbandry personnel. Clinical severity rapidly increases after week 5, with progressive and diffuse alopecia and the striking behavior of intermittent, rapid bouts of head shaking. Based on our observations, we hypothesize that the excessive grooming and rapid head shaking are due to pruritus rather than irritation from whole body skin hyperkeratosis. However, further investigation is needed to confirm the physiologic mechanism causing these behaviors.

Beginning at week 6, 375-mL water bottles had to be replaced weekly rather than biweekly. This difference represented approximately 15 to 18 mL/mouse/day of water intake, which exceeds the traditionally accepted intakes of 5 to 10 mL/mouse/day.^1,12 We hypothesize this increase was due to C. bovis-associated skin pathology, which allows for transepidermal water loss (TEWL). The skin is an important barrier for water retention. With skin pathology, the disruption of normal skin architecture allows more evaporative water loss.¹⁹ Increases in TEWL have been shown previously in 2 C. bovis-infected transgenic immunocompetent mouse lines with clinical signs similar to those described here for NSG mice.^7,8

Our data showed that mice began to lose body weight by week 7. We can only speculate on the cause of body weight loss, but hypothesize that it is due to a combination of TEWL, a chronic and systemic skin inflammatory response, prolonged activity associated with excessive grooming behavior, and muzzle swelling that potentially impedes food acquisition through a wire bar feeder. Supporting this hypothesis, we observed that additional food provided on the floor of the cage was readily consumed by mice and helped to slow the rate of weight loss but did not permit a return to a normal body weight (data not shown). Additional studies would be necessary to fully elucidate all contributing factors to the observed weight loss.

Finally, by week 8, the majority of mice had reached the maximum total score of 18 and consistently resemble the image in Figure 3 which depicts the poorest hair coat quality as a score of 5. Also seen in this image is concurrent periorbital alopecia, and ear hyperemia with concurrent hyperkeratosis. What is not represented by this image is the physical thickening of the skin of the nape, excessive grooming behavior, hunched posture and head shaking which all contribute to a total maximum score of 18 (Figure 4F).

While this study focused on the characterization of clinical signs of disease, C. bovis infections can occur in several forms including an asymptomatic incubation period, clinical disease, and an asymptomatic carrier state. Our data demonstrate a 3-wk asymptomatic incubation period for NSG mice prior to the appearance of clinical signs (Figure 5A). This approximates the incubation period for athymic nude mice, which typically develop clinical signs 1 to 2 wk after introduction into an enzootically infected colony.^2,4,10,16 Clinical signs in athymic nude mice are known to spontaneously resolve within 1 wk after initial observation.^2,4,16 In our study, clinical signs did not regress at any time point in NSG mice. To investigate this finding further, we followed a subset of NSG mice out to 16 wk and did not observe a reduction of clinical signs as compared with week 8 clinical scores (data not shown). These data suggest that, in contrast to athymic nude mice, clinical signs do not appear to resolve in NSG mice. In addition, for athymic nude mice, a C. bovis asymptomatic carrier state has been documented in 67% of cages in an enzootically infected colony.¹¹ Similar data are not available for infected hirsute immunodeficient mice like the NSG. However, our institution’s extensive C. bovis monitoring program has identified 2 different strains of mice, both on a Rag1 knockout background, that were asymptomatic carriers of C. bovis. The existence of an asymptomatic incubation period and an asymptomatic carrier state hampers the ability to use clinical signs as a sensitive diagnostic tool for C. bovis detection. Thus, the sole use of clinical signs for surveillance would have a high risk of missing infected mice and allowing bacterial shedding into the environment. In short, the presence of infected mice and the resulting C. bovis environmental contamination are significant concerns for infection perpetuation within research vivariums.^3,11,13

During the development of the scoring system, significant effort was made to ensure its validity for the intended purpose of scoring clinical signs of C. bovis-infected NSG mice. To ensure validity, 33 NSG mice were used initially to observe C. bovis infection progression of clinical signs and to rank those signs numerically and temporally to limit subjectivity. When evaluating a scoring system, practicality is essential to ensure that the scoring system can actually be used under biocontainment conditions. To be practical, the scoring system must be quick to perform and based primarily on visual assessment through the cage wall; this minimizes the need for physical manipulation and thus supports biocontainment. Limitations to this scoring system are the assessment of reliability and generalizability. Reliability of scoring systems is typically assessed by comparing the results of multiple scorers to provide the interobserver error. Reliability can also be affected by the ease of learning the scoring system and the consistency of scorers over time. Because we used only one scorer, we could not assess reliability. Furthermore, using only a single scorer may have introduced bias because the observer could not be blind to the treatment groups. We recognize these limitations and contend that because the scoring system is based on well-defined incremental changes to the eyes, ears, hair coat, grooming activity, posture, and head shaking, the robust validity may prove to reduce observer subjectivity and bias when used by others. In addition, we do not know how generalizable this scoring system will be in real-world conditions. Ideally, the scoring system could aid veterinarians in determining the approximate time of NSG exposure to C. bovis. In addition, this scoring system may also be useful for other SCID mice with mutations in the ɣ chain of the IL-2 receptor. However, the infectious dose and the genetic background of the infected mouse will affect the rate of clinical progression and severity of clinical signs. Thus, additional testing of reliability and generalizability of the scoring system will be necessary for its broader application.

The development of this clinical scoring system was possible because C. bovis can be transmitted through soiled bedding and phenotypically causes clinical signs. We have previously used this method to transmit infection and clinical disease between enzootically infected and uninfected nude mice.¹⁰ However, the current study is the first time that clinical C. bovis infections have been produced using a known C. bovis isolate. Further research is needed into the genetic or microbial environment that gives C. bovis, and potentially other commensal bacteria a pathologic phenotype that induces clinical disease.

Our scoring system has a variety of applications, but it is not intended for use as a sole diagnostic tool. C. bovis-specific, PCR-based diagnostic assays are inexpensive, sensitive, and specific, and the skin swab as a diagnostic sample is too easy to obtain to consider relying on clinical signs alone. One application of this clinical scoring system is to help veterinarians anticipate the clinical progression of disease for valuable, infected immunodeficient mice in preclinical research studies. Based on our observations, we question the utility and scientific data generated from NSG mice maintained beyond the point when clinical disease results in weight loss (Figure 6). Finally, we hope this clinical scoring system can be leveraged to assess the potential benefit of current and future clinical interventions to prevent or treat infections that may ultimately help eliminate C. bovis from research facilities.