Euthanasia of Mouse Fetuses and Neonates

BRENDA A. KLAUNBERG, MS, VMD,1 JAMES O’MALLEY, DVM, MPH,2 TERRI CLARK, DVM,3 AND JUDITH A. DAVIS, DVM, MS2*

We sought to determine whether any of the common methods of euthanasia for adult rodents would lead to an acceptable death for fetuses or neonates. We wanted to identify a method that was rapid, free of signs of pain or distress, reliable, and minimally distressful to the person performing the procedure and that minimized the amount of handling required to perform the procedure. We evaluated six methods of euthanasia, with and without anesthesia, in three age groups of mice: gravid mice (E14-20) and neonatal pups (P1-P7 and P8- P14). Euthanasia methods included: halothane inhalation, carbon dioxide inhalation, intraperitoneal sodium pentobarbital, intravenous potassium chloride, and cervical dislocation with and without anesthesia. Noninvasive echocardiography was used to assess heartbeat during euthanasia. With cardiac arrest as the definition of death, no method of euthanasia killed fetal mice. Halothane inhalation (5% by vaporizer) was not an acceptable method of euthanasia for mice of the age groups tested. Intraperitoneal administration of sodium pentobarbital for euthanasia required a higher dose than the previously established dose, and there is a risk of reduced efficacy in pregnant animals due to potential intrauterine injection. Carbon dioxide asphyxiation was the most efficient method of euthanasia for neonatal mouse pups P1-14. For pregnant adult mice, intravenous potassium chloride under anesthesia, carbon dioxide asphyxiation, and cervical dislocation alone or under anesthesia were excellent methods of euthanasia.

Timed-pregnant animals constitute a predominant percentage of methods of euthanasia for adult rodents would also result in an ac- animal use for basic research in our institute. Vendors frequently ceptable death for the fetus or neonate. We wanted to find a method include extra dams to ensure that the scientist receives the requested that was rapid, free of signs of pain or distress, reliable, and mini- number of animals at the appropriate day of gestation (e.g., embry- mally distressful to the person performing the procedure and that onic day 12 [E12], E14, etc.). Consequently, facility staff must minimized the amount of handling required to perform the proce- euthanize pregnant animals if they are unable to locate other poten- dure. We investigated six methods of euthanasia, but not every method tial users of the extra animals. Discussions during training classes on was applied to each group. The study was divided into two phases, proper euthanasia techniques revealed employee concern about fetal which were based upon the ages of the pups. Phase one examined death. Employees asked how they would know whether the fetuses fetal mice at 14 days of gestation or more (≥ E14). Phase two exam- were dead. Removing the uterus and physically examining each fetus ined neonatal mice, which were divided into two groups: pups 1 to 7 is an option to ensure that fetuses are dead, but we consider that days old (P1-P7) or 8 to 14 days old (P8-P14). impractical. We were unable to find documented evidence that commonly used euthanasia methods rapidly and painlessly kill fetuses. Materials and Methods The only published references to fetuses or neonatal animals in rela- Mice. Virus antibody-free mice (n = 166) identified as “culls” of tion to euthanasia were admonitions against the use of carbon dioxide various genotypes were used. The experimental group included in- because newborn animals are more resistant to hypoxia (1, 2). These bred, outbred, hybrid, and mutant mice. The mice were housed in a publications further state that inhalation anesthetics will not kill fe- National Institutes of Health animal facility that is accredited by the ≤ tuses and that newborn pups 10 days of age seem to react more like Association for Assessment and Accreditation of Laboratory Animal fetuses than adult animals (3). Care, International. The mice were maintained in a specific patho- Complicating the problem of fetal and neonatal euthanasia is the gen-free animal care holding room; sentinel Swiss-Webster mice phenomenon of neonatal hypoxia tolerance, in which adaptive mecha- (Taconic, Germantown, N.Y.) were on racks in the same animal nisms to acute oxygen loss include decreased metabolism, decreased holding room and were evaluated monthly to monitor the colony for body temperature, and reduced blood pH (4). Any mention of proper pathogen exposure. In addition, colony animals identified for eutha- euthanasia of fetuses and neonates cautions that the animals may nasia were monitored for pathogen exposure. Sera from the sentinel appear insensate but are still alive. Recommendations for fetal eutha- and cull mice consistently tested negative for the following microor- nasia include removal from the uterus followed by decapitation or ganisms: CAR bacillus, ectromelia virus, mouse rotavirus, mouse < concussion (3) or, if the fetus weighs 4 g, it may be placed directly encephalomyelitis virus, lymphocytic choriomeningitis virus, murine into liquid nitrogen for euthanasia (3). None of these methods were cytomegalovirus, mouse hepatitis virus, mouse adenovirus, minute considered practical by our staff. virus of mice, Mycoplasma pulmonis, mouse parvovirus, polyoma vi- We describe the use of ultrasound imaging to determine cardiac rus, pneumonia virus of mice, reovirus 3, and Sendai virus. Mice arrest after administration of a euthanasia method to mice. The pur- were free of pinworms and fur mites. The animal holding room was pose of this study was to determine whether any of the common maintained with restricted access via electronic cipher-coded door locks. Mice were housed in polycarbonate (19.56 cm × 30.91 cm × 10 Center Drive, Bldg 10, Room B1D69, NMRF, DIR, NINDS, NIH, DHHS, Bethesda, 14.92 cm) ventilated cages (Thoren Caging Systems, Hazelton, Pa.) Maryland 208921; 36 Convent Drive, Bldg 36, Room B306, AHCS, DIR, NINDS, NIH, DHHS, Bethesda, Maryland 20892-44142; 31 Center Drive, Bldg 31, Room on CareFRESH Ultra bedding (Quality Lab Products, Inc., Elkridge, B1C37, OACU, OIR, OD, NIH, DHHS, Bethesda, Maryland 20892-22523 Md.). Cages were changed twice weekly. The animal holding room *Corresponding author was maintained under environmental conditions of 22 ± 1°C, rela- Work was conducted at the Mouse Imaging Facility, National Institutes of Health, tive humidity of 50% ± 10%, 20 air changes/h, and a 12:12-h Bethesda, Maryland The views expressed are those of the authors and do not reflect the official policy or light:dark cycle. Mice received pelleted rodent diet (NIH Open For- position of the National Institutes of Health or the United States Government. mula Rodent Diet NIH-07, Zeigler Brothers, Inc., Gardners, Pa.)

Volume 43, No. 5 / September 2004 CONTEMPORARY TOPICS © 2004 by the American Association for Laboratory Animal Science 29 Table 1. Experimental age groups and methods of euthanasia applied (+) turned on. While in the CO2 chamber, mice could not be physically Fetuses (E14-20) Neonates (P1-7) Neonates (P8-14) examined, so a visual assessment was made. Once the mouse lost its righting reflex, stopped moving, or appeared to be unconscious, that Carbon dioxide ++ +time was recorded, and the animal was immediately transferred to a asphyxiation custom-made facemask for continued delivery of CO2. The heart- Barbiturate + not done + beat of the neonate, or a randomly chosen fetus, was identified as overdose quickly as possible with ultrasound imaging. The times to cardiac Cervical dislocation + not done not done arrest and, if possible, respiratory arrest were recorded. Halothane overdose ++ +For barbiturate overdose, two age groups were studied: fetuses (E14- (5% by vaporizer) 20) and P8-14 neonates. Sodium pentobarbital (800 mg/kg) was Potassium chloride + not done not done administered via intraperitoneal injection to manually restrained neo- under anesthesia nates or gravid mice. The injection time was recorded. When the injected animal stopped ambulating, we identified the heartbeat of Cervical dislocation + not done not done under anesthesia the fetus or neonate with ultrasonography as quickly as possible. The times to respiratory arrest and cardiac arrest were recorded. and water ad libitum. This study was reviewed and approved by our Cervical dislocation of conscious animals was studied in gravid Institutional Animal Care and Use Committee. females only. The time the dam was euthanized by cervical disloca- Experimental design. Transthoracic echocardiography was used tion was recorded. The heartbeat of a randomly chosen fetus was to study selected methods of euthanasia for the three age groups of identified by ultrasonography, and the time to fetal cardiac arrest mice. Death was defined by cardiac arrest; cardiac arrest was defined was recorded. by cessation of cardiac contractions. Echocardiography was termi- In order to compare our experimental methods of euthanasia with nated at 20 min if cardiac arrest did not occur, and the animals a currently recommended method of fetal euthanasia, we performed subsequently were euthanized using a physical method of euthana- cesarean section and decapitation on fetuses of dead dams (n = 19) sia. For cervical dislocation, loss of consciousness and cerebral death from other study groups. Timing began at the initiation of the cesar- were considered immediate. The first study examined fetal death in ean section and ended at the time of the last pup’s decapitation. The relation to euthanasia of the dam; gravid females at 14 to 20 days of number of fetuses for each dam was counted. The number of fetuses gestation were used. The second study examined neonatal euthana- per dam varied, so an average time of procedure per gravid dam was sia. Neonatal pups were divided into two experimental age groups: calculated. P1 to P7 days old and P8 to P14 days old. Anesthesia methods. For general anesthesia, mice were placed in The pregnant animals used in this study were in the late stages of the previously described chamber for induction, and 5% halothane pregnancy (E14-20). Gravid animals < 14 days gestation were not or isoflurane (Fluotec or Isotec, SurgiVet/Anesco, Waukesha, Wis.) used for a variety of reasons, but the most important one is that fetal was delivered to the chamber with 100% oxygen at a rate of 400 ml/ development of the heart and vascular system do not achieve defini- min. When the animal lost its righting reflex, stopped moving, or tive prenatal configuration until E15 (5). appeared to be unconscious, the animal was transferred to a custom- Six methods of euthanasia were studied, but not every method was made facemask for continued delivery of anesthetic gasses. Anesthesia applied to every group (Table 1). Three methods were used with was maintained with 1% to 2% halothane or isoflurane and an oxy- animals under general anesthesia, and three methods were used in gen flow rate of 250 ml/min. Halothane was used only for inhalant unanesthetized animals. Methods of euthanasia performed under anesthetic overdose, and isoflurane was used for all other methods general anesthesia included an overdose of inhalant anesthesia deliv- involving general anesthesia. ered by a vaporizer, intravenous potassium chloride injection Every anesthetized animal was placed in dorsal recumbency on a (Pharmaceutical Partners, Inc., Los Angeles, Calif.), and cervical dis- specialized mouse board with heating elements and electrodes for location. Methods of euthanasia without general anesthesia included measuring heart rate or the electrocardiogram (THM 100 Electronic carbon dioxide asphyxiation, barbiturate overdose (Beuthanasia-D Control Module, Indus Instruments, Houston, Tex.). This board Special, Shering-Plough Animal Health Corporation, Union, N.J.), was placed on top of a 37°C heated recirculating water pad (Tem- and cervical dislocation. perature Therapy Pad TP-21B, Gaymar Industries, Orchard Park, Carbon dioxide euthanasia chamber. For carbon dioxide asphyxia- N.Y.). When possible, rectal body temperature and heart rate were tion or inhalant anesthesia induction, a clear acrylic chamber with a monitored continuously with the Indus Instruments apparatus. positive-sealing gasket lid was used (no. 941444, VetEquip, Inc., Euthanasia methods with anesthesia. Halothane overdose was stud- Pleasanton, Calif.). The total chamber volume was 2000 cc (22.86 cm ied in all age groups. Gravid dams were anesthetized as previously [width] × 9.52 cm [depth] × 9.52 cm [height]). A 0- to 500-cc/min described. Once the heartbeat of a randomly chosen fetus was identi- flow meter (Multitube Frame 03215-56 Cole-Porter Instruments Co., fied by ultrasonography, the anesthetic vaporizer was increased to 5%

Vernon Hills, Ill.) was connected between the CO2 tank and cham- halothane, and timing began from that point. Neonates were anesthe- ber. The gas flow rate was set to 400 cc/min to achieve 20% volume tized in the previously described chamber, and 5% halothane was displacement per minute. The CO2 gas concentration of displaced delivered to the chamber with 100% oxygen at a rate of 400 ml/min. gasses exiting the chamber was measured with a capnograph Timing began when the halothane was turned on. When the neo-

(CO2SMO Monitor, Novametrix Medical Systems, Inc., nate stopped moving or appeared to be unconscious, the time was Wallingford, Conn.). The time when the chamber measured 100% recorded, and the pup was immediately transferred to a custom-

CO2 was noted. Between animals, the chamber was turned upside made mask for continued delivery of 5% halothane. For both groups, down and emptied of residual CO2 gas, then closed and flushed with the heartbeat was identified as quickly as possible with ultrasonog-

100% oxygen until the exhaust CO2 concentration was 0%. raphy, and times to cardiac arrest and, if possible, respiratory arrest Euthanasia methods without anesthesia. Carbon dioxide asphyxi- were recorded. ation was studied in all age groups. The animal was placed in the Potassium chloride euthanasia was studied in gravid females only. chamber described above and timing began when the CO2 gas was The dam was anesthetized as previously described. A butterfly cath-

30 CONTEMPORARY TOPICS © 2004 by the American Association for Laboratory Animal Science Volume 43, No. 5 / September 2004 Table 2. Time (min; mean ± 1 standard deviation) to cardiac arrest (CA) and loss of righting reflex (RR) for each method of euthanasia Dams Fetuses (E14-20) Neonates (P1-7) Neonates (P8-14) Carbon dioxide CA 2:50 ± 0:14 ≥ 20:00 ± 0 4:41 ± 0:37 3:52 ± 1:02 asphyxiation n = 5a n = 7n = 11 n = 15 RR 1:54 ± 0:17 not determined 3:44 ± 0:43 1:56 ± 0:32 n = 7n = 11 n = 15 Barbiturate overdose CA 11:02 ± 7:35 ≥ 20:00 ± 0 not determined 6:03 ± 1:19 n = 6b n = 7n = 12 RR 1:51 ± 0:04 not determined not determined 0:30 ± 0:05 n = 7n = 12 Cervical dislocation CA 2:07 ± 0:25 ≥ 20:00 ± 0 not determined not determined n = 7n = 7 RR not determined not determined not determined not determined Halothane overdose CA 18:56 ± 2:37 ≥ 20:00 ± 0 ≥ 20:00 ± 0 19:36 ± 1:06 (5% by vaporizer) n = 6c n = 7n = 7n = 7 RR not determined not determined 3:38 ± 1:16 2:01 ± 0:26 n = 6c n = 7

Potassium chloride CA 0:11 ± 0:03 ≥ 20:00 ± 0 not determined not determined under anesthesia n = 6d n = 7 RR not determined not determined not determined not determined

Cervical dislocation CA 3:17 ± 2:09 ≥ 20:00 ± 0 not determined not determined under anesthesia n = 7n = 7 RR not determined not determined not determined not determined adid not measure time to CA in the first two dams. bmissed CA time point in one animal. cdiscarded one value greater than 2 standard deviations from the mean. dmissed one dam’s time to CA. eter (27 gauge × 1/2", Surflo winged infusion set, Terumo Corpora- Results tion, Tokyo, Japan) was inserted into the tail vein of the anesthetized Study 1: fetal death relative to the dam’s. According to our defi- gravid dam. Once the heartbeat of a randomly chosen fetus was iden- nition of death by cardiac arrest, no euthanasia method used in this tified by ultrasonography, a lethal dose of KCl (1 to 2 mmol/kg) was study successfully caused fetal death (Table 2). With dams breathing administered. The time of injection was recorded. The times of res- 5% halothane anesthesia delivered by a vaporizer (n = 7), fetal heart- piratory arrest of the dam and cardiac arrest for both the dam and beats continued beyond the 20-min limit. Most dams (n = 6; 1 data fetus were recorded. point more than 2 standard deviations [SD] from mean) also failed Euthanasia by cervical dislocation under anesthesia was studied in to undergo cardiac arrest within 20 min of breathing 5% halothane. gravid dams only. The dam was anesthetized and maintained under Intraperitoneal injections of sodium pentobarbital (800 mg/kg) caused anesthesia as previously described. Once the heartbeat of a randomly cardiac arrest in the dams at approximately 6 min (n = 6, one data chosen fetus was identified by ultrasonography, an assistant performed point not recorded). Fetal heart rates continued beyond the cut-off cervical dislocation of the dam. The times of respiratory arrest for of 20 min, but there was considerable variation in the rates (24 to the dam and cardiac arrest for both the dam and fetus were recorded. 148 beats per min [bpm]) under pentobarbital. In addition, fetal Ultrasound image acquisition. Transthoracic echocardiography cardiac arrest did not occur within 20 min after the dam received was performed with an Acuson Sequoia Echocardiography system intravenous KCl, carbon dioxide asphyxiation, or cervical disloca- (model C256, Siemens Medical Solutions, Mountain View, Calif.). tion performed alone or under anesthesia. Conversely, these were The variable range transducer offered ultrasound frequencies of 8, excellent methods of euthanasia for the dam. After KCl injection, 12, and 14 MHz. These frequencies generated images with micro- the mean time until cardiac arrest of the dams was 11 sec (n = 6, one µ scopic resolution of approximately 200 m. Two-dimensional value not recorded). Dams developed cardiac arrest under carbon B-mode images were obtained from parasternal long and short axis dioxide inhalation in an average of 2:50 min (n = 5, 2 data points not views by using the 14-MHz linear transducer with optimized gain recorded). Pregnant animals that were first anesthetized with isoflurane and depth settings. Some data were stored in DICOM file format or then euthanized by cervical dislocation experienced cardiac arrest in video images. B-mode images were obtained for all pups, and when an average of 3:17 min (n = 7); however, the fetal hearts continued to possible, Doppler blood flow images were obtained also. Images were beat after 20 min, with an average of 31 bpm. Performing cervical viewed and analyzed with the Acuson software (DIMAQ worksta- dislocation on conscious pregnant females (n = 7) yielded similar tion). If necessary, the hair was removed from the mouse’s ventral results. The average time of death of the dam was 2:07 min, but 20 thorax prior to imaging. min later, fetal hearts were still beating at an average rate of 22 bpm. Time points were measured with a clock timer and recorded in The time required for cesarean section and decapitation of fetuses minutes. The data points were converted to seconds for calculations will vary with the number of fetuses present in the uterus. In our and then converted back to minutes for reporting. For statistical analy- study, dams selected for the mock cesarean section and decapitation sis, the value of 1200 sec (20 min) was used for data points that of fetuses experiment had an average of 10 fetuses per dam. The exceeded the 20-min limit. Calculations and Student t tests were entire procedure averaged 2:13 min per dam. performed using Excel (Microsoft, Redmond, Wash.). Study 2: neonatal mice. For neonatal mice, we found that the

Volume 43, No. 5 / September 2004 CONTEMPORARY TOPICS © 2004 by the American Association for Laboratory Animal Science 31 most rapid method of euthanasia was carbon dioxide asphyxiation respiratory pattern continued for 1.5 to 2.5 min prior to cessation of (Table 2). We examined two age groups of neonates, pups 1 to 7 all respiratory efforts. It is important to note that we observed asys- days old and those 8 to 14 days old. The average time for cardiac arrest tolic neonatal pups take deep agonal breaths that could revive them. after carbon dioxide inhalation was 4:41 min in P1-7 pups (n = 11) and As the pup took a deep agonal breath, its body would flex convul- 3:52 min in P8-14 pups (n = 15). The higher dose (800 mg/kg) of sively like a person doing a sit-up exercise. On two occasions, a intraperitoneal sodium pentobarbital produced acceptable euthana- cyanotic and asystolic pup exposed to room air turned pink and a sia results for P8-14 neonates, in which cardiac arrest occurred at an regular cardiac rhythm returned after this convulsive gasping proce- average time of 6:03 min (n = 12). Sodium pentobarbital was not evalu- dure. Indeed, the pups began crawling! It was as if the agonal gasps ated in the younger age group. The neonates confirmed our earlier and body contractions of the pup caused auto-cardiopulmonary re- findings that 5% halothane inhalation delivered by a vaporizer is not suscitation (CPR). This CPR phenomenon only occurred in “pinky” an effective means of euthanasia for mice. After 20 min of breathing pups (P1-7) and was not observed in older pups (P8-14). Duffy et al. 5% halothane, neonatal heartbeats remained strong and regular, with (6) found that the length of terminal gasping correlated well with the an average rate of 404 bpm for P1-7 (P8-14, data not recorded). length of anoxic survival, whereas Fazekas et al. (7) observed that respi- Both methods of euthanasia studied in the P1-7 group involved ratory gasps in many neonatal animals could be evoked by peripheral inhalation techniques. Although the carbon dioxide asphyxiation stimulation. Viemari et al. (8) also described this phenomenon; their produced acceptable times to cardiac arrest in P1-7 pups, there was study documented the relationship of hypoxia to gasping and its role in no significant difference in the time to loss of righting reflex when establishing a respiratory rhythm in E18 and P0-2 pups. We recom-

CO2 was compared to 5% halothane. P1-7 pups lost their righting mend that when mouse pups (P1-7) are euthanized in a CO2 chamber, reflexes after breathing carbon dioxide for an average of 3:44 min (n they should be kept in the chamber for several minutes after they are = 11) and 3:38 min (n = 6, 1 data point > 2 SD from mean) after cyanotic and all signs of life have ceased. To ensure death with no breathing 5% halothane. Similarly, there was no significant differ- chance of revival, pups should be carefully removed from the cham- ence in the time to loss of righting reflex of the P8-14 pups when ber with minimal stimulation and placed directly into the freezer. CO (1:56 min, n = 15) was compared with 5% halothane (2:01 Perhaps a secondary physical method of euthanasia should be per- 2 = min, n 7). It is noteworthy, however, that the time to loss of right- formed when euthanizing neonates with CO2. ing reflex for the P1-7 pups is nearly twice that of the older pups for Our results support those in the current literature and demon- = -5 = both CO2 (P 0.18E ) and halothane (P 0.0237). It is well known strate the fetal and neonatal mouse’s tolerance to hypoxia. In the that the neonatal mouse is resistant to the effects of hypoxia (4), and CO2 chamber, the P1-7 pups took nearly twice the amount of time this phenomenon may account for the delayed loss of righting re- to lose their righting reflex (3:44 min, n = 11) when compared to flexes in the younger pups. pups aged 8 to 14 days (1:56 min, n = 15). A similar phenomenon occurred with pups breathing 5% halothane. The P1-7 pups aver- Discussion aged 3:38 min (n = 6, one value 2 SD from mean) to lose their Selection of an appropriate euthanasia method for pregnant ro- righting reflex whereas the P8-14 pups lost their righting reflex in = dents is a difficult task, and one must consider several factors. If 2:01 min (n 7). A number of physiological mechanisms enable euthanasia is to be performed in the laboratory, the use of fetal or young animals to tolerate hypoxia. In contrast to that in adults, any neonatal tissue will influence the method of euthanasia chosen. If impairment to the oxygen supply of a fetal or neonatal animal trig- the animal facility staff performs euthanasia, one must consider the gers bradycardia, which is accompanied by shunting of blood from human emotions involved and choose a method that is not offensive. the periphery to central organs (4). Singer (4) describes the left shift In this situation, efficiency and human sensitivities must be balanced of the fetal oxyhemoglobin dissociation curve, in which the aerobic with the knowledge that the selected method will humanely and rap- environment at the cellular level is prolonged. Furthermore, fetal and idly kill both the dam and the fetuses or neonates. neonatal animals have lower cerebral metabolic needs compared with In our study, the time-to-death (cardiac arrest) of the dams was those of adults (7). much longer than anticipated for most euthanasia methods, although Body temperature is another important factor in metabolism and the expected variability between the animals’ time-to-death was not tolerance to hypoxia. As the fetus or neonate begins to lose body as large as anticipated (Table 2). As there was almost no variation in temperature, its metabolic rate slows to exert a protective effect against fetal experimental time points, we reduced the experimental group hypoxia. In fact, a reduction in ambient temperature is most effec- number from 10 to 7 animals. We terminated echocardiography af- tive in prolonging the survival time of asphyxiated mammalian ter 20 min if the fetal or neonatal hearts continued to contract. The neonates (7). During the study, all pups were placed on a recirculat- dams and fetuses were then euthanized using a physical method of ing warm water blanket when placed in a facemask for delivery of CO , but despite our efforts, pups lost body heat and became palpa- euthanasia. In a pilot study, we found that none of the adult animals 2 experienced cardiac arrest within 20 min of breathing 5% isoflurane bly cool. We suspect that hypothermia may have prolonged the time delivered through a vaporizer. Further investigation of this method to cardiac arrest because of the slowing of metabolic rate; therefore, of euthanasia was abandoned, and halothane was used for inhalant we suggest that maintaining a warm environment may reduce the tolerance to hypoxia when using CO as the method of euthanasia anesthesia overdose instead of isoflurane. We were surprised that no 2 method of euthanasia studied rapidly killed fetuses aged E14 or older. for neonatal rodents. We found that fetal cardiac contraction occurred long after the dam The use of 5% halothane for euthanasia did not produce accept- was clinically dead. able results in our study. The time to cardiac arrest for all groups was Although carbon dioxide asphyxiation appears to be an effective either or close to 20 min or not achieved (Table 2). When compar- method of euthanasia for adult, gravid, and neonatal mice, it did not ing the time to cardiac arrest with 5% halothane with that of the other inhalational method (CO ), statistical significance was found cause fetal death by our criteria. Within 1.5 to 3 min after initial 2 = -5 exposure to 100% CO , adult mice lost their righting reflex, exhib- between all groups (Student’s t test; dams, P 0.205E ; pups P1-7, 2 = -15 = -12 ited muscle flaccidity, and (in some cases) were unresponsive to deep P 1.820E ; pups P8–14, P 2.199E ) This statistical significance was not surprising because CO2 inhalation produced acceptable pressure applied to the hind paw. Under 100% CO2, animals dis- played short-term apnea and brief muscle twitching. This agonal times to cardiac arrest, whereas 5% halothane delivered by vaporizer

32 CONTEMPORARY TOPICS © 2004 by the American Association for Laboratory Animal Science Volume 43, No. 5 / September 2004 appears to be inadequate for euthanasia. Interestingly, there was no 0:25 min (n = 7). Interestingly, dams that had CD under isoflurane statistical significance between the effects of 5% halothane and CO2 anesthesia took longer to achieve CA (3:17 min), but the standard on the righting reflex of neonates within the age groups. We believe deviation was 2:09 min (n = 7). We have no explanation for this that the data support the theory of neonatal tolerance to hypoxia. difference; perhaps there was variability in the CD method. When In our hands, the use of intraperitoneal pentobarbital at the pub- the time to CA for CD alone versus CD under anesthesia is com- lished dosage of 200 mg/kg did not produce euthanasia within an pared using Student’s t test, there is no statistical significance. acceptable period of time. In a pilot study, adult non-gravid mice We concluded that cardiac arrest was not a good measure of death. dosed with 200 mg/kg intraperitoneally did not undergo cardiac ar- Our definition of cardiac arrest was “cessation of all cardiac contrac- rest until approximately 15:16 ± 7:07 min (n = 11); increasing the tions.” We did not anticipate that “cardiac arrest” would be so difficult dose to 400 m/kg reduced the time to cardiac arrest to 5:18 ± 2:18 to assess. Although the hearts continued to “beat,” rates were often min (n = 10). Time to loss of righting reflex also decreased with the dramatically lower than normal, and we suspect that blood was not increased dose, requiring an average of 2:20 (n = 10, one point not moving effectively in either pups or dams. When we described our recorded) or 1:36 min (n = 10) for the 200 or 400 mg/kg dose, re- observations to a pediatric cardiac specialist (9), she suggested that spectively. We were not satisfied with the time of death until we our observations were similar to those in human infants. In termi- increased the pentobarbital dose to 800 mg/kg, when cardiac arrest nally ill premature human neonates for whom life support is of adult non-gravid animals occurred in 2:40 ± 0:30 min (n = 10). discontinued, the heart activity may persist for an hour or more. In Acceptable results also were achieved when 800 mg/kg was adminis- human medicine, a similar clinical condition called pulseless electri- tered intraperitoneally to neonatal pups (P8-14). The P8-14 pups cal activity (PEA) is a well-recognized phenomenon. Physicians lost their righting reflexes in an average of 0:30 min (n = 12) and characterize PEA as the presence of coordinated electrical activity of appeared to be unconscious; however, cardiac arrest did not occur the heart muscle without peripheral pulses. In humans, PEA is asso- until an average of 6:03 min (n = 12) from the time of administra- ciated with major cardiac insult, frequently secondary to respiratory tion of pentobarbital. arrest or hypoxia (10). Progressive acidosis exacerbates PEA. In hu- In contrast to the non-gravid mice in the pilot study, mans, the heartbeat is often the last parameter to cease in death, echocardiography of pregnant females given 800 mg/kg of intraperi- usually because of overwhelming metabolic acidosis from major or- toneal pentobarbital was stopped at 20 min because none of the fetuses gan system failure (10). Furthermore, PEA is not considered to be a had achieved cardiac arrest. Pentobarbital-injected pregnant females conscious state. In our pilot study, we observed electrical activity by took longer to lose their righting reflexes (1:51 min, n = 7) than did electrocardiogram monitoring that did not always correspond to car- non-gravid adults (1:04 min, n = 10). Pregnant animals consequently diac contractions in dying adult mice. After learning about PEA, we took longer to develop cardiac arrest (11:02 min, n = 6, one value suspect that these mice may have been in PEA when their heart rate not recorded) than did non-gravid adults (2:40 min, n = 10). The fell below the normal range. Mice in this condition (PEA) may also increased length of time needed for pentobarbital to have the antici- be considered unconscious and not suffering. pated effect in gravid females was unexpected. One explanation for Normal fetal development is a complex process. Production of the difference in results between gravid and non-gravid mice is acci- normal, viable offspring is dependent on the coordination of many dental in utero injection rather than injection into the peritoneal space. systems all developing at the chronologically appropriate time, de- When performing the mock C-sections, we noticed that the uteruses veloping within the correct anatomical location, and ultimately completely filled the abdomen; thus, unintended in utero injection functioning properly. Until approximately E15, the developing fetus would be an easy mistake. Perhaps gravid females (> E14) should be is completely dependent on a viable interface between fetal and ma- given pentobarbital intravenously if it is used for euthanasia. ternal tissues. This dependence is initially accomplished through According to echocardiography, fetuses of gravid females appeared placentation. The placental vascular bed of both rodents and hu- to be unaffected by intraperitoneal pentobarbital. A plausible expla- mans is classified as hemochorial. In this configuration, maternal nation for the lack of fetal effects of pentobarbital could be the blood perfuses a space lined by fetal trophoblast cells (11) where gas following: although cardiac arrest in the dam did not occur until exchange occurs. Prior to E10, fetal mouse oxygen exchange is de- 11:02 min after injection, effective blood circulation may have stopped pendent completely on the trophoblast placental system. With soon after administration. The European Union Working Party’s (1) minimal circulatory development, embryonic tissues are bathed in recommendations for euthanasia of experimental animals suggest that fetal fluids, and gas exchange is direct. fetuses often are not anesthetized by inhalation anesthetics unless the As the fetus develops, it must contribute various components to dam is maintained under anesthesia for a prolonged period of time. its own survival, despite its dependence on the dam for oxygen, nu- The report further suggests that fetuses frequently are not anesthe- trients, and waste removal. The developing embryonic mouse heart tized (any type of anesthesia) for potentially painful procedures. The achieves strong regular heartbeats at E9, yet blood components re- implication is that even if drugs cross the placenta, they rarely achieve main primitive, and the heart is not fully developed until E15 (5). the desired (or presumed) effect. We do not recommend intraperito- Bifurcation of the common carotid arteries is not detectable until neal pentobarbital as a sole method of euthanasia for fetuses > E14. late pregnancy (E17) (5). After E10, embryonic red blood cells are Potassium chloride achieved rapid death in both gravid and non- needed to carry oxygen to sustain fetal mouse life. The importance of gravid mice; however, similar to pentobarbital, KCl did not appear embryonic red blood cells to embryonic viability is demonstrated by to affect the fetuses. All echocardiography exams of gravid mice were EKLF mice, which are deficient in erythropoietic transcription fac- stopped at 20 min because fetal heartbeats were still present. We tor. The EKLF fetal mice have normal embryonic hematopoiesis but speculate that sufficient KCl did not cross the maternal–placental fail to make definitive red blood cells. The lack of normal red blood barrier due to the rapid cardiac death of the dam. cells causes these mice to die in utero between E13.5 and E14 (12, Cervical dislocation (CD), with or without anesthesia, did not re- 13). Continued fetal viability is dependent on not only an effective sult in fetal death by our criteria. Dams achieved cardiac arrest within placenta and maternal blood supply but also appropriate fetal contri- acceptable times, but it appears that isoflurane anesthesia may have butions. Failure of any component of this complex matrix will result an effect on the time to cardiac arrest after cervical dislocation is in fetal death. Thus, until E15, the fetus is dependent on the mater- applied. Dams that had CD alone achieved cardiac arrest in 2:07 ± nal-placental system for survival.

Volume 43, No. 5 / September 2004 CONTEMPORARY TOPICS © 2004 by the American Association for Laboratory Animal Science 33 In addition to other developing systems, the central nervous sys- manuscript and offering many helpful suggestions. We thank Dr. Linda tem is not complete in the mouse until after birth. Pain perception is Leatherbury, a pediatric cardiologist, for stimulating conversations and in- considered unlikely in fetal mice less than E15 because neural devel- sight into the difficulty of determining pediatric death. Finally, we would like opment is incomplete. The mouse cerebral cortex is still differentiating to thank our pets, as conversations about them provided comfort to us while we performed the technical aspects of this investigation. The views, opinions, at E17 (5). Fetuses cannot survive ex utero until late E18-E19 (8, and findings contained in this report are those of the authors and do not reflect 14); however, there may be strain variation (11, 15, 16). official policy or positions of the Department of Human Health Services, the By our criteria of death, we found no method of euthanasia for National Institutes of Health, or the United States Government. fetuses that satisfied all aspects of the definition of humane euthanasia; however, for several of the physiological reasons discussed above, we believe that fetal (E15 or younger) death should be considered References equivalent to death of the dam. Although the fetal hearts appeared to 1. Close, B., K. Banister, V. Baumans, et al. 1996. Recommendations for euthanasia of experimental animals: part 1, compiled by the EU be beating, we suspect that the contractions were not effectively Working Party. Lab. Anim. 30:293-316. moving blood, and the fetuses may have been exhibiting PEA. With 2. Bertens, A. P. M. G., L. H. D. J. Booij, P. A. Flecknell, et al. 1993. the removal of a maternal blood supply (death of dam), the fetus Anesthesia, analgesia, and euthanasia, p. 267-298. In L.F.M. van presumably experiences increasing hypoxia and progressive metabolic Zutphen, V. Bauman, and A. C. Beynen (ed.), Principles of laboratory acidosis. We feel that these factors are incompatible with fetal life. animal science. Elsevier Publishing Co., Amsterdam. Because of late brain development, we also suggest that these fetal 3. Close, B., K. Banister, V. Baumans, et al. 1997. Recommendations mice are most likely not experiencing any pain or discomfort. Our for euthanasia of experimental animals: part 2, compiled by the EU assessment of fetal death was a difficult task, and the results were not Working Party. Lab. Anim. 31:1-32. 4. Singer, D. 1999. Neonatal tolerance to hypoxia: a comparative-physi- as objective as we had hoped. Perhaps a future study of effective fetal ological approach. Comp. Biochem. Physiol. Part A. 123:221-234. hemodynamics (blood flow) or blood gas analysis may yield a more 5. Kaufman, M. H. 1995. The atlas of mouse development, p. 183, 212, definitive parameter for determining the viability of a fetus. 215, 295, and 431. Academic Press, San Diego, Calif. In summary, we do not recommend inhalation anesthesia over- 6. Duffy, T. E., S. J. Kohle, and R. C. Vannucci. 1975. Carbohydrate dose (delivered by a vaporizer) for euthanasia of laboratory rodents and energy metabolism in perinatal rat brain: relation to survival in anoxia. J. Neurochem 24:271-276. of any age. If CO2 asphyxiation is used for neonatal pups (P1-7), a second method of euthanasia is strongly recommended to ensure 7. Fazekas, J. F., F. A. D. Alexander, and H. E. Himwich. 1941. Toler- ance of the newborn to anoxia. Am. J. Physiol. 134:281-287. death. We also suggest that during CO2 euthanasia, keeping pups warm may decrease the time it takes for neonatal death. In our hands, 8. Viermari, J-C., H. Burnet, M. Bevengut, et al. 2003. Perinatal matu- ration of the mouse respiratory rhythm-generator: in vivo and in vitro the commonly used euthanasia dose of intraperitoneal pentobarbital studies. Eur. J. Neurosci. 17:1233-1244. (100 to 200 mg/kg) is not effective for euthanasia; the higher dose of 9. Leatherbury, L. 2003. Personal communication. 800 mg/kg intraperitoneally produces acceptable euthanasia results 10. Martin, S. K., C. H. Shatney, J. P. Sherck, et al. 2002. Blunt trauma in non-gravid animals. If pentobarbital is used for euthanasia of gravid patients with prehospital pulseless electrical activity (PEA): poor end- mice, intravenous administration may be a better route of adminis- ing assured. J. Trauma 53:876-881. tration to achieve the desired effect. For euthanasia of adult mice, 11. Adamson, S. L., Y. Lu, K. J. Whiteley, et al. 2002. Interactions be- intravenous potassium chloride, carbon dioxide asphyxiation, and tween trophoblast cells and the maternal and fetal circulation in the cervical dislocation alone or under anesthesia are effective and ac- mouse placenta. Dev. Biol. 250:358-373. 12. Nuez, B., D. Michalovich, A. Bygrave, et al. 1995. Defective ceptable methods. These findings are in agreement with Cartner et haematopoiesis in fetal liver resulting from inactivation of the ELKF al. (17). Finally, measurement of cardiac arrest is not a good indica- gene. Nature 375:316-318. ≤ tor of fetal death. Fetuses E14 likely die at the time of maternal 13. Perkins, A. C., A. H. Sharpe, and S. H. Orkin. 1995. Lethal b- death. Because cortical structures in the embryonic mouse brain are thalassaemia in mice lacking the erythroid CACCC-transcription factor poorly differentiated during late stages of gestation (5), we feel that EKLF. Nature 375:318-322. these fetuses are not conscious or capable of perceiving pain. Because 14. Hatta, T., K. Moriyama, K. Nakashima, et al. 2002. The role of gp130 we were unable to find a method of euthanasia that caused fetal (E14- in cerebral cortical development: in vivo functional analysis in a mouse 20) death by our definition (asystole), we recommend using an exo utero system. J. Neurosci. 22(13):5516-5524. additional, physical method of euthanasia (cervical dislocation or 15. Bronson, F. H., C. P. Dagg, and G. D. Snell. 1968. Reproduction, p. 187-204. In E. L. Green (ed.), Biology of the laboratory mouse, 2nd decapitation) to ensure death for the fetuses of gravid mice in late ≥ ed. Dover Publications, Inc., New York. gestation ( E14). 16. Theiler, K. 1989. The house mouse. Atlas of embryonic development, 2nd ed. Springer-Verlag, Inc., New York. Acknowledgments 17. Cartner, S. C., S. C. Barlow, and T. J. Ness. 2003. Loss of cortical function in mice following decapitation, cervical dislocation, potas- We would like to thank Ms. Kathryn Sharer who provided pregnant fe- sium chloride injection, and CO inhalation. Poster Session, ACLAM male mice for this study and Mr. Daryl DesPres and Mr. Dan Schimel who 2 Forum, Induced animal models of human diseases, Sanibel Harbour shared their technical expertise. We thank Ms. Joan Timberlake (University Resort, Ft. Myers, Fla., 14 to 17 May 2003. of Maryland and Johns Hopkins University) for her critical reading of the