Weight Loss and Reduced Body Temperature Determine Humane Endpoints in a Mouse Model of Ocular Herpesvirus Infection

Abstract

Herpes simplex virus (HSV) has been studied in well-established mouse models to generate latently infected animals for investigations into viral pathogenesis, latency mechanisms, and reactivation. Mice exhibit clinical signs of debilitating infection, during which time they may become severely ill before recovery or die spontaneously. Because the cohort of mice that does survive provides valuable data on latency, there is keen interest in developing methodologies for earlier detection and treatment of severe disease to ultimately increase survival rates. Here, BALB/c mice were inoculated ocularly with either a wildtype (LAT⁺) or mutant (LAT^–) strain of HSV1. Mice were monitored daily through day 30 after infection; trigeminal ganglia were harvested at day 60 to assess viral DNA load. Cages were provided with nesting material, and fluid supplementation was administered to mice with body temperatures of 35 °C or lower, as measured by subcutaneous microchip thermometry. The results showed that infected mice with temperatures less than 34.5 °C did not recover to normothermia and were euthanized or spontaneously died, regardless of infective viral strain. By using a combination of criteria including body temperature (less than 34.5 °C) and weight loss (more than 0.05 g daily) for removal of animals from the study, approximately 98% of mice that died spontaneously could have been euthanized prior to death, without concern of potential recovery to the experimental endpoint (100% specificity). Frequent monitoring of alterations to general wellbeing, body temperature, and weight was crucial for establishing humane endpoints in this ocular HSV model.

Herpes simplex virus (HSV) is an important human pathogen, with the potential to cause mucocutaneous lesions, encephalitis, and complications with keratitis and blindness.¹⁸ HSV1 forms latent infections in neurons of the peripheral nervous system, particularly in the trigeminal and sacral ganglia, and less than 40% of treated human patients return to functional (preinfection) health status after infection.³⁰ Although human HSV has been studied in the well-established mouse model of peripheral nervous system infection, a major obstacle to HSV research is the limited quantity of viral material that can be recovered from the immune-privileged neural tissue in latently infected animals. Specific research goals of the current study were to generate latently infected mice for continued assessment of HSV pathogenesis and for determination of the mechanism of HSV latency and reactivation at a molecular level. Latency is defined as the absence of any detectable replicating virus, proteins of the lytic viral replication cycle, or viral transcripts. During HSV infection of mice by the ocular (corneal) route, infectious virus can be recovered from the trigeminal ganglia (TG) of mice until days 7 to 10 after infection. To discern that latency has been accomplished, mice must be maintained beyond day 30 after infection. Estimates of the percentage of latently infected neurons are in the range of 0.1% to 1%, based on reactivation from TG cell preparations.^16,44 For this HSV mouse model, prolonged periods of survival after infection are necessary to ensure viral latency.^10,11

Clinical veterinary challenges for this model included the inability to predict which infected mice would develop severe acute disease with death as an outcome. IACUC protocols for a variety of infectious disease studies often indicate that the work precludes the use of supportive treatments, opioids, or other analgesics known to modify immune responses and affect data outcomes; therefore, clinical investigations of modest supportive care measures and developing methods to avoid spontaneous death are invaluable for improved laboratory animal welfare. The IACUC-approved protocol for this HSV model estimated that as many as 50% of HSV-infected mice would die, without the ability to predict which mice would become severely ill and which would survive. Furthermore, because of potential effects on neurochemical, respiratory, and vascular systems, suitable interventions for pain relief were not feasible for administration to infected mice. Subsequently, the current study was undertaken to determine whether, in fact, certain mice could predictably be identified for removal from study prior to eventual death and, furthermore, whether modest supportive care measures could be instated. To our knowledge, none of these points have been studied previously in this model.

Humane interventions and endpoints continue to be evaluated in a variety of rodent biomedical models by using objective scoring systems to quantify rankings of criteria for exclusion of animals from further study.^{1,9,21,22,25,28,33, 34,40,41,43} Nontransient decreases in whole-body temperature, as an early predictor of impending death, have been demonstrated in mice with specific experimental LD50, bacterial, viral, and fungal infections.^{2,26,34,36, 40,41,43,45,46} Microchip transponder systems that remotely read subcutaneous or intraperitoneal body temperatures have been tested and shown to be sufficiently similar in readings to traditional rectal thermometry^6,19 and intraabdominal telemetry.⁴³ Infrared surface thermometers are another method of temperature acquisition considered to be entirely noninvasive and to provide rapid and repeatable temperature readouts (within seconds). These thermometers are considerably less expensive than are other methods and are readily portable between rooms and facilities; we therefore were quite interested to determine their applicability to the HSV mouse model. When aimed at a consistent location on the animal's ventral aspect, infrared thermometry has been shown to be a reliable form of temperature measurement for laboratory mice.^2,45

On the basis of thermometry studies in other infectious disease models, we hypothesized that a nontransient decrease in body temperature (as measured both by microchip and infrared thermometry), in conjunction with diminished overall wellbeing and loss of body weight, would be predictive of alternative nondeath humane endpoints for this ocular mouse model of HSV latency.

Materials and Methods

Animals.

Clinically healthy BALB/cJ females (8 wk old; Jackson Laboratory, Bar Harbor, ME) were procured and maintained on an approved IACUC protocol in housing conditions compliant with AAALAC accreditation and meeting the room humidity and temperature expectations in the Guide for the Care and Use of Laboratory Animals.¹³ All animals were housed at a 12:12-h light:dark cycle at a density of 5 mice per static polycarbonate microisolator cage (Max 75, Alternative Design, Siloam Springs, AR) on disposable bedding (0.12-in. diameter Bed-O-Cobs, Animal Specialties and Provisions, Quakertown, PA). Wire-lid food hoppers within cages were filled to capacity with rodent chow (Lab Diet 5010, Animal Specialties and Provisions) and mice were maintained on water supplied by bottle. The housing room was classified for Animal Biosafety Level 2 experiments, requiring a doubling of disposable personal protective equipment for entry and removal of the outer layers of protective gear for exit from the room.

Sentinel mice at our institution were tested inhouse for 3 quarters annually and were found to be free from fur mites and pinworms (Syphacia spp. and Aspiculuris spp., by cecal exam). Sentinel mice also tested negative for antibodies to pathogens including mouse hepatitis virus, mouse parvoviruses, rotavirus, ectromelia virus, pneumonia virus of mice, Theiler murine encephalomyelitis virus, and Sendai virus. One quarter each year, sentinels from the housing facility were tested by an outside contract laboratory and were found to be free from all pathogens contained on a comprehensive assessment panel (HM Assessment Plus panel, Charles River Laboratories, Wilmington, MA).

Microchip placement and animal inoculations.

All described inoculation experiments were performed twice (experiments A and B) to confirm reproducibility of nondeath endpoints; the final experimental design and selected interventions were based on 2 independent pilot experiments for this disease model. At 3 d prior to inoculation, conscious mice were implanted subcutaneously on the dorsum between scapulae with individual transponders (model IPTT-300, BMDS, Seaford, DE) that transmit body temperature to a scanning device system (model DAS-6007, BMDS). All microchip transponders were inserted as directed by the manufacturer and successfully provided data throughout the course of the experiments; manufacturer instructions did not indicate that calibration, beyond initial factory settings, was necessary.

For each iteration, mice were randomly allocated as uninfected controls (n = 4 or 5), a group (n = 30) inoculated with the LAT⁺ viral strain (which produces the viral latency-associated transcript), and a group (n = 30) inoculated with the LAT^– viral strain. To ensure accuracy of inoculum administration, mice were anesthetized (ketamine [70 mg/kg IP]–xylazine [7 mg/kg IP]) and received droplets (2 μL) of HSV bilaterally onto the eyes to induce encephalitis; corneal scarification was not required for entry of the virus into the corneal epithelium. Control mice were handled first at all monitoring time points.

Viral stock preparation.

HSV1 was prepared as previously described; the highly virulent McKrae (LAT⁺) and McKrae dl2903 (LAT^–) strains, were used for the experiments.²⁹ In brief, McKrae LAT^– virus carries a deletion in the promoter and exon 1 region of the LAT gene. LAT⁺ serves as wild type; it expresses the LAT gene during latency. LAT^– serves as the mutant strain. The function of LAT may be toward antiapoptotic preservation of the latent neuron³¹ and in miRNA silencing of viral transactivating genes, which are important for the lytic replication cycle.⁴² Overall, because LAT has not been shown to be a virulence gene, it is unclear whether 1 of these 2 strains is more virulent; nor did our experiments seek to determine relative virulence. However, the LAT⁺ strain reactivates more efficiently than the LAT^– strain,^29,32 and this characteristic may be, in part, associated with the ability of the LAT to protect neurons from granzyme B-induced apoptosis and CD8 T-cell killing.¹⁴ To achieve maximal levels of latent viral material, mice were infected with high doses of virus (1.0 × 10⁵ pfu in experiment A; 3.0 × 10⁵ pfu in experiment B).

Monitoring assessments, infrared thermometry, and supportive care.

For each monitoring point, the chronology of assessments was consistent: relative scored activity (Figure 1), followed by a protocol (Figure 2) of subcutaneous temperature reading, surface temperature reading (for experiment B animals only), body weight recording, documentation of ocular lesions, and then administration of supportive care measures.

Figure 1.

Activity scoring criteria for evaluation of mice inoculated with HSV.

Figure 2.

Health monitoring criteria for evaluation of mice inoculated with HSV.

Mice were monitored once daily (0800) beginning on day 0 (day of inoculation) until day 4 (end of the incubation period). From day 5 through day 15 after infection, mice were monitored twice daily, at a 12-h interval (0800; 2000), due to the predictable development of severe disease. From day 15 through day 30, daily monitoring (0800) resumed, and surviving mice were assumed to be free of lytic infection and to have developed latency. In experiment A, at each time point, body temperature was measured by using only the subcutaneous transponders. In experiment B, the temperatures from transponders were compared with temperature readings collected from a noncontact infrared surface thermometer (Fluke 62 Mini; www.amazon.com) aimed at the xiphoid process from a distance of approximately 3 inches, as described previously.⁴⁵ This instrument works in a range of −30° to 500 °C (−20° to 932 °F); manufacturer instructions did not indicate that calibration, beyond initial factory settings, was necessary. Determination of a consensus on overall health and activity scoring was made through hands-on training of colleagues by the principal investigator (FCH).

Between days 5 and 15 after infection, every cage received hydrated gel (ClearH₂O, Portland, ME), a solid form of fluid replacer that was maintained off the bedding in a disposable dish. A fresh pad of nesting material (Nestlets, Ancare, Bellmore, NY) was provided for every cage on every day of the study. Topical antibiotic ophthalmic ointment (Antibiotic Ointment, CVS Pharmacy brand) was applied to any mouse that developed periocular lesions. For those mice that had a subcutaneous microchip temperature reading less than 35 °C, supportive fluid care (1 mL IP; room temperature 0.9% NaCl) was administered. Surviving mice were not expected to exhibit additional health concerns and therefore were checked daily by animal care staff from day 31 through day 60 after infection, until the final experimental endpoint for evaluation of viral latency in TG tissues (Figure 3).

Figure 3.

Timeline of experimental design. Microchips were implanted subcutaneously on day –3, and mice were monitored once daily (SID) until days 5 through 15 after infection, when they were monitored twice daily (BID). Once-daily checks resumed at day 16 after infection. For mice that spontaneously died or ultimately reached the experimental endpoint, trigeminal ganglia were harvested to assess viral DNA load.

According to our pilot studies, initial exclusion criteria were proposed to euthanize mice with severely compromised health status presumed to be irreversible. Any mouse that experienced prolonged inactivity or moribundity, as categorized by a score of 3, over a 24-h period (or over 3 consecutive observation points during the critical infection phase), and any mouse with periorbital self-injury was euthanized by CO₂ narcosis. Any mice that spontaneously died due to infection were removed immediately from the cages. Surviving mice (until day 60 after infection) were euthanized by cervical dislocation at their experimental endpoint to avoid alterations induced by CO₂ narcosis on brain tissue that would bias interpretation of gene expression data.

Quantification of viral DNA load.

Viral DNA amounts were determined by qPCR using primers to the TK gene region as previously described.²⁷ In brief, TG were harvested from mice that spontaneously died and from those that reached the end of the 60-d experimental protocol to assess the viral DNA load. Brain tissues were removed from the calvaria, and TG were dissected and homogenized by sonication (model GE100, Ultrasonic Processor, General Electric, Piscataway, NJ) to extract DNA. DNA was purified by using commercial extraction kits (AllPrep DNA/RNA mini kit, Qiagen, Valencia, CA).

Statistical methods.

We summarized the body temperature and weight by mean, standard error, and minimum and maximum values for the experimental animals. Of note, day 0 measurements were removed from the statistical analysis due to potential bias induced by anesthesia used to restrict movement of animals for accuracy of ocular inoculation. For each animal, we calculated the minimal temperature, largest temperature decrease from baseline, and activity scores, by using all of the temperatures and activity scores measured for days 1 through 30. Using all weight measures recorded over days 1 through 30, we determined the lowest weight, largest weight decrease from baseline, and the weight growth rate for each mouse by using the linear regression model. For days 5 through 15, when temperature and weight were measured twice daily (at both 0800 and 2000), we used the lower temperature for the prediction of death, and weights measured at these different time points were included in the linear regression model to estimate the weight growth rate for an animal; this approach was undertaken to minimize the effect of circadian rhythms on temperature and weight. One-way ANOVA was then performed for the comparison of temperature measures (minimal temperature, largest temperature decrease from baseline), body weight alterations, and highest activity score among the 3 groups of animals (alive, euthanized, and spontaneously dead) for each experiment.

To evaluate how each of these measures discriminated between surviving and deceased mice, we first determined the cutpoint that maximized the specificity at 100% to avoid euthanizing animals that likely would survive until day 60 after infection. Next, the sensitivity and specificity corresponding to the cutpoint of each temperature measure, body weight, and activity score were calculated.

For experiment B, to evaluate the agreement between subcutaneous temperature microchips and surface temperatures taken at the level of the xiphoid process, we calculated the paired difference between subcutaneous and surface temperatures and summarized the agreement by using mean difference and 95% confidence limits of agreement, supplemented with Bland–Altman plots.³

All statistical analyses were performed in SAS version 9.2 (SAS Institute, Cary, NC) and a 2-sided P value of less than 0.05 was considered to be statistically significant.

Results

Prediction of death from measures of body temperature, activity score, and weight loss.

Control mice remained uninfected throughout the experiments. For experiment A controls (n = 4), the mean lowest subcutaneous temperature was 35.3 °C, and the weight gain rate from day 1 through day 30 was 0.05 g/d. For experiment B control mice (n = 5), the mean lowest subcutaneous temperature was 36.0 °C, and the weight gain rate from day 1 to 30 was 0.03 g/d.

After infection with HSV, mice became ill; however, the manifestation of the infection and severity of clinical signs differed widely between animals within cages and across groups (Figure 4). Mortality rates differed between experimental groups, such that fewer mice survived during experiment B (mortality rate, 88%) than in experiment A (mortality rate, 48%). Regardless of overall mortality rates, the lowest temperature that a surviving mouse was recorded to achieve in both experiments was 34.5 °C. There was not a significant difference in the mean temperatures recorded at baseline (prior to infection) between mice that survived, were euthanized for humane reasons, or that died spontaneously (Tables 1 and 2). However, after the infectious course began, body temperatures differed significantly at the lowest recorded measurements between those animals that survived and those that were euthanized or died spontaneously (experiment A, P < 0.0001; experiment B, P = 0.009).

Figure 4.

Variability in disease severity between similarly HSV-inoculated mice on day 12 after infection. during experiment A. (A) Differences in clinical signs between cagemates were demonstrated. For example, one mouse (left) had unilateral conjunctivitis with periocular alopecia and swelling (chip temperature, 35.6 °C; weight, 21.8 g), whereas another (right) had no overt clinical signs of infection (chip temperature, 36.5 °C; weight, 21.7 g). Both mice were assigned an activity score of 1 and both ultimately survived the infection. (B) A mouse in another cage was hunched and moved slowly around the cage after prompting (chip temperature, 27.2 °C; weight, 14.5 g; activity score, 3); this animal died on day 13. (C) In a third cage, a mouse was hunched, inactive, and demonstrated neurologic deficits, with an inability to grip the wire-bar lids (chip temperature, 25.8 °C; weight, 15.5 g; activity score, 3); this animal died on day 12.

Table 1.

The comparison of measures among 3 groups of mice (experiment A) from day 1-30 postinoculation

	Alive (n = 31)	Euthanized (n = 6)	Spontaneous death (n = 23)	Pa
Baseline temperature (°C; day 0)
Mean ± SE	37.2 ± 0.08	37.1 ± 0.20	37.1 ± 0.20	0.65
Minimum, maximum	36.3, 37.9	36.7, 37.9	36.2, 38.0
Lowest subcutaneous temperature (days 1–30)
Mean ± SE	35.6 ± 0.07	33.8 ± 1.7	30.8 ± 0.9	<0.0001
Minimum, maximum	35.0, 36.4	25.5, 36.3	23.4, 36.5
Largest decrease of subcutaneous temperature from baseline
Mean ± SE	−1.6 ± 0.1	−3.3 ± 1.7	−6.3 ± 0.9	<0.0001
Minimum, maximum	–2.5, –0.9	–12.0, –0.8	–13.0, –0.5
Baseline weight (g; day 0)
Mean ± SE	20.0 ± 0.24	20.0 ± 0.47	20.2 ± 0.41	0.92
Minimum, maximum	17.2, 22.8	18.6, 21.4	15.6, 22.8
Weight gain rate (g/d)
Mean ± SE	0.07 ± 0.01	−0.28 ± 0.13	−0.51 ± 0.07	<0.0001
Minimum, maximum	–0.05, 0.28	–0.70, 0.01	–1.00, 0.31
Highest (worst) activity score
Mean ± SE	2.1 ± 0.1	2.7 ± 0.2	2.7 ± 0.1	0.0005
Minimum, maximum	1, 3	2, 3	1, 3

^aP value reflects the comparison among all 3 groups.

Table 2.

The comparison of measures among 3 groups of mouse (experiment B)

	Alive (n = 7)	Euthanized (n = 21)	Spontaneous death (n = 32)	Pa
Baseline temperature (°C; day 0)
Mean ± SE	36.5 ± 0.26	37.0 ± 0.14	36.8 ± 0.13	0.21
Minimum, maximum	35.1, 37.3	35.6, 38.3	34.2, 38.2
Lowest subcutaneous temperature (days 1–30)
Mean ± SE	35.5 ± 0.20	30.3 ± 1.1	32.3 ± 0.6	0.009
Minimum, maximum	34.5, 35.9	22.5, 36.0	26.0, 36.7
Largest decrease of subcutaneous temperature from baseline
Mean ± SE	−1.0 ± 0.2	−6.7 ± 1.1	−4.4 ± 0.6	0.005
Minimum, maximum	–1.8, –0.4	–15, –0.6	–11, 1.3
Baseline weight (g; day 0)
Mean ± SE	20.0 ± 0.56	19.9 ± 0.25	20.1 ± 0.25	0.89
Minimum, maximum	18.5, 22.2	18.0, 22.0	17.6, 23.4
Weight gain rate (g/d)
Mean ± SE	0.03 ± 0.02	−0.58 ± 0.08	−0.62 ± 0.04	<0.0001
Minimum, maximum	–0.05, 0.07	–1.00, 0.02	–1.2, –0.17
Highest (worst) activity score
Mean ± SE	2.3 ± 0.2	2.7 ± 0.1	2.3 ± 0.1	0.02
Minimum, maximum	2, 3	2, 3	1, 3

^aP value reflects the comparison among all 3 groups.

Weight measurements were collected at the same frequency as were temperatures and recorded throughout the experiment from days 1 through day 30 after infection. Similar to body temperature, the baseline weights of mice that survived or were euthanized or died did not differ, regardless of experiment. However, over the course of the study, weight changes were significantly (P < 0.0001) different between those mice that survived and those that did not. Surviving mice gained weight (experiment A: mean, 0.07 g daily; experiment B: mean, 0.03 g daily) over the 30-d monitoring period; mice that did not survive lost approximately 0.05 g daily. Weigh growth rate was calculated as the slope estimated from the linear regression models based on all the weight data, thus any potential bias on weight due to the randomized administration of fluids to a particular mouse at a single time point was minimized by linear regression analysis.

Activity scores were recorded at the same frequency as were the weight and temperature measurements (Tables 1 and 2). In experiment A, there was a significant (P = 0.0005) difference in mean highest (worst) activity scores between surviving mice and those that were euthanized or died. This difference was significant (P = 0.02) also in experiment B. Across both experiments, only one mouse with 2 consecutive scores of 3 returned to a healthier status of activity; all others were euthanized or died spontaneously.

Sensitivity and specificity of prediction of spontaneous death by using single or multiple measures.

With application of the calculated cutoff temperature point to euthanize mice at 34.5 °C, approximately 70% of the animals that spontaneously died (n = 55) could have been euthanized without concern of potential recovery for collection of TG tissues. If only the loss of weight was used as a cutoff point (using a weight loss of 0.051 g daily) over the experimental course, approximately 96% of these same mice (n = 55) could have been euthanized. If these parameters of temperature were combined with assessments of weight loss over the study time course, approximately 98% of those mice that died spontaneously could have been euthanized prior to death without concern of potential recovery for TG collection. For the mice that survived (n = 38), all had recorded temperatures and weights at higher than the cutoff points listed above, resulting in 100% specificity for success in reaching the planned experimental endpoint.

Comparison of subcutaneous temperatures with surface temperatures.

In experiment B, the average subcutaneous temperature across the course of the repeated experimental protocol was 36.7 °C, and the corresponding surface temperature was 34.1 °C. Despite the fact that the paired subcutaneous and surface temperatures were highly correlated (Pearson correlation coefficient, 0.08; P < 0.0001), these temperature readings were not interchangeable, because the measurements differed significantly (P < 0.001) by approximately 2.57 °C, with a 95% limit of agreement. Furthermore, the divergence between the methods of thermometry increased significantly (P < 0.0001) when temperatures of mice were greater than or equal to 35.5 °C, whereas the divergence did not change significantly (P = 0.91) when the subcutaneous temperature of mice fell below 35.5 °C.

Relative amounts of TG viral DNA in representative mice.

TG tissue from mice, infected with either type of HSV, were assessed by PCR to determine viral DNA amounts (Figure 5). During the replicative phase of viral infection, large amounts of viral DNA were synthesized, as demonstrated in representative mice –28 and –53, both of which were found dead on day 10. This high level of detectable viral DNA dropped precipitously by day14 after infection. No significant difference in the amount of viral DNA harvested from TG tissue was noted between the viral strains used in these experiments.

Figure 5.

Viral DNA quantification in trigeminal ganglia of infected mice. After corneal infection with a virulent strain of HSV1, mice were monitored for as long as 60 d after infection. Representative surviving mice (–33, –66, 38, 39, 47, 55, 62) then were euthanized and trigeminal ganglia removed. DNA was extracted from the ganglia, and the amount of HSV DNA present (in pg) was quantified by PCR. The viral DNA from the trigeminal ganglia of mice (–28, –53, –29, 46) that died on various days after infection was determined also. HSV strain LAT+ infection is designated by the positive animal numbers and LAT– infection is designated as ‘–’ in front of identification numbers. Mock2 is a representative control animal.

Discussion

Contemporary guidance clarifies that experimental protocols that result in “significant alteration of the animals’ ability to maintain normal physiology, or adequately respond to stressors, should include descriptions of appropriate humane endpoints.”¹³ Inherent subjectivity may bias the recognition of animals experiencing suboptimal health; therefore, relying on an animal to reach a moribund state to qualify for removal from study is of questionable benefit.^38,43 Humane endpoints should be determined prior to the start of experiments; importantly, humane endpoints differ from experimental endpoints, which are defined as the end of animal use due to achievement of scientific aims.¹³ The establishment of reliable humane interventions and endpoints is predicated according to the principals of reduction, replacement, and refinement in the use of animals for research.³⁵ Refinements are directly related to improvements in animal welfare, exemplified by establishing well-defined humane intervention and endpoints, and are essential to eliminate unnecessary suffering of animals and consequently improve the quality of data obtained from experiments.^12,25 Ultimately, balancing humane endpoints with scientific data collection and veterinary intervention is crucial for laboratory animals and typically should involve input from investigators, veterinarians, and the IACUC.^13,23

Herpesviral studies in the described mouse model were aimed at further understanding the complex interaction, at a molecular level, of the virus with the neuronal cell. Within the known framework of critical disease manifestation during days 5 through 15 after infection, our goals were to determine criteria for intervention by euthanasia prior to spontaneous death, with the hope that viral latency would be achieved in mice surviving the critical phase of the infection. Although multiple mouse strains have been assessed for susceptibility to HSV,¹⁵ our model assessed humane endpoints for female BALB/c mice exclusively.

To acquire detectable levels of latent herpesviral material from the peripheral nervous system, mice were infected with high doses of virus, leading to correspondingly high mortality rates. The mortality rate for mice was higher for experiment B than for experiment A; this difference may be related to the higher dose of virus with which mice were inoculated in experiment B. A complication of this model is establishing the appropriate viral dose for inoculation, so that a percentage of mice survive to latency (where viral DNA levels remain constant for the remaining lifetime of the mouse), while maintaining the dose so that residual viral DNA does not fall below detectable levels. High levels of viral DNA were present in the TG tissue of mice during the acute stage of infection (day 10); however, these levels were due to the presence of lytic, not latent, viral DNA. Over time, viral DNA loads decreased as the mouse immune system eliminated the infection. The drop in viral DNA was dramatic, and by day14 after infection, levels approached a steady state (0.20 to 0.40 pg per ganglion for these experiments). Importantly, HSV DNA does not reach a zero level, primarily because the immune system can eliminate only acutely infected cells that are producing viral proteins.¹⁴ It is thought that the immune system does not recognize the latently infected cells because they do not express viral proteins;¹⁴ indeed, there is documented presence of viral DNA until 60 d, a time point well past the resolution of acute viral infection.³⁷ Although viral DNA at 2 to 3 wk after infection approaches latent levels, analysis of mice at these time points shows elevated expression of immunologically associated genes.^5,7,17 Therefore mice typically are not considered reliable models for human HSV latency until at least 2 mo after infection.

We monitored mice daily by using a combination of weight assessments, activity scores, and body temperature readings throughout the initial 30 d after infection. Overall, we successfully identified weight loss parameters and a cutoff point of a nontransient decreased body temperature (relative to baseline temperatures) from which mice did not recover to survive to latency. This cutoff temperature, measured by using subcutaneous microchips, was repeatable in 2 independent experiments in which mice received different doses of virus.

Because of the manual manipulation of mice required to successfully read the subcutaneous temperature chips with the transponder wand, we were concerned about a potential bias toward prolonged handling stress and unanticipated contribution to decompensation in severely ill mice. Although the system scanner did not have to actually touch the mice for temperature readings, sometimes several repositionings of the scanner were needed to ultimately identify the animal's chip and capture the temperature, and during this time, mice were manually restrained. In addition, the use of this system requires the implantation of the microchip under the skin of the mouse at the beginning of the experiment. This process has no effect on the temperature readings at monitoring time points but does involve an invasive, albeit brief, procedure. Therefore, in the second iteration of the experiment, we compared the microchips with noninvasive infrared surface thermometers that could be aimed at the same point (theoretically) on each mouse to derive a rapid temperature reading with minimal manual restraint.

With both thermometry methods, due to cohousing of mice in defined experimental cohorts, mice had to be handled individually and lifted onto the wire-bar lids of their cages for gentle manual restraint, temperature data acquisition, and assessment of individual health; these procedures may have led to immediate and transient elevations in body temperatures.⁸ In experiment B, the readout of the subcutaneous microchip was compared with the surface temperature at the level of the xyphoid process for all animals. Despite our hope that the 2 methods would be sufficiently similar and interchangeable, data from the 2 measures were not directly correlated, with a mean significant difference of 2.57 °C per read. Furthermore, the subcutaneous and surface temperatures diverged significantly when temperatures were above that at which we provided fluid support (35.5 °C). Despite attempts to measure at a known distance from the sternum of the animals, variability likely was introduced by this method, and therefore accurate body temperatures could not be confirmed reliably.

Infrared thermometry has successfully been used in mouse models of infection with Candida⁴⁵ and S. pneumoniae,² yet in our HSV model, divergence between thermometry methods was as large as 5.8 °C at some monitoring points, thus eliminating our confidence that the 2 methods were interchangeable for body temperature assessments. Industrial infrared thermometers were used in the cited publications;^2,45 however, the make and model differed from that applied in our studies. Using a veterinary infrared thermometer that would account for the potential bias of haircoat and body location could facilitate consistent, reliable, yet noninvasive body temperature readings for mice and other laboratory animal species.⁴ Ultimately, our data did not support reliance on the accuracy of infrared thermometry for establishing an endpoint cut-off temperature; if infrared thermometry alone had been used in our experiments, these significant inaccuracies may have led to the premature euthanasia of some mice, whereas others with fatal illness might have remained on study longer than necessary. Premature euthanasia of mice was particularly undesirable due to the potential to waste valuable animal lives and compromise group sizes and calculations for statistical analysis, resulting in a conflict between the collection of necessary data and minimization of animal pain and distress.³⁴

Progression of illness is accompanied most often by decreases in body condition and weight, such that monitoring of body weight with some frequency during experimental protocols is expected for determining endpoints.¹⁹ Therefore, body weight assessments also hold predictive qualities related to overall loss of body condition during experimental manipulations.³⁹ The challenge in solely using body weight changes as endpoint criteria is that substantial weight loss is not always fatal and that animals may die without any changes in weight. We assessed weight loss in the HSV model as a potential option for determining humane interventions and found a predictive value in weight loss of greater than 0.05 g daily for mice that ultimately did not recover to viral latency. Because these calculations were made based on linear regression analyses, we recommend that the average weight loss of an infected animal should be checked over a minimum of 3 d to determine the best-fit line and slope of loss.

The initial IACUC protocol assumptions that one could not predict those mice that would become ill was found to be accurate in our experience; for example, at day 12 after infection, during the acute phase of infection, certain mice were severely compromised, whereas others appeared clinically similar to uninfected control mice. The weight loss over the course of the 30-d monitoring phase was subtle, despite twice-daily assessments during the intensive infectious phase and further emphasized the importance of attention to monitoring changes in this health parameter for indications of illness. Activity measurements and ambulation have been reported to have predictive value for determining surrogate markers of impending death.²⁰ In our experiments, relative activity was assessed and found to contribute to overall health assessments yet did not provide a consistently objective criterion on which to initiate humane interventions and permit removal from study prior to spontaneous death. However, from our data, only one mouse that was moribund (that is, activity score = 3) for longer than 2 observation points returned to a more active status; therefore, we recommend euthanasia for any mouse found moribund for 24 h, to avoid their ultimate spontaneous death, as an additional humane endpoint.

Throughout the experiments, we provided cotton nesting material, diet gel supplementation, and fluid support as potential comforts to the mice. Although these additions were important for enrichment and potential relief, other studies have shown that fluids and nutritional gels may be of questionable efficacy for improving survival of mice.²⁴ An unexpected finding was the severity to which mice would rub at their eyes after viral inoculation and during the acute phase of disease. Because we could not administer analgesia due to scientific need to avoid the introduction of study bias, we applied antibiotic ointment as a modest supportive intervention to those animals with noted conjunctivitis. In a limited subset of mice, aggressive self-mutilation of the ocular area required immediate removal from study and prompted discussions regarding the need for pain relief measures in future experiments, potentially to include topical anesthetic (nonsystemic) eye drops for those mice with severe periorbital lesions.

To our knowledge, humane endpoint criteria based on objective health assessments have not previously been developed for this mouse model of ocular HSV infection. Our work indicated that body temperatures measured by using subcutaneous microchip methods were preferable to those from noncontact infrared thermometry for the establishment of endpoint criteria. According to our scoring methods and findings, mice with a subcutaneous body temperature below 34.5 °C did not recover to normothermia, regardless of infective HSV strain. If we had relied only on temperature reaching a measurement of 34.5 °C for monitoring, approximately 70% of mice that died could have been euthanized without concern of potential recovery. Furthermore, experimental mice reaching temperatures less than 34.5 °C and with body weight loss averaging 0.05 g or more daily did not survive to HSV latency. By taking advantage of these 2 described health monitoring criteria in future iterations of this infectious disease model, approximately 98% of mice that would otherwise die spontaneously can be euthanized at a humane endpoint. The assumption that one could not predict readily which mice would survive HSV infection was disproven in this investigation; this finding will contribute greatly to improvements in animal welfare during future studies.