All mice were housed in a standard animal maintenance facility under a 12 h light: 12 h dark (LD) cycle. The Pml^−/− (129sv) animals were obtained from the Mouse Models of Human Cancers Consortium repository, National Cancer Institute [7]. The mPer2^−/− (129sv/C57/b6) and Per1^−/− (129sv/C57/b6) mice were generated as previously described [4]. Mice deficient in either PML/PER1 or PML/PER2 were generated by initially breeding Pml^−/− with Per1^−/− or Per2^−/− animals to produce the heterozygote F1. The double mutant animals for Per1^−/−/Pml^−/− or Per2^−/−/Pml^−/− were then obtained from heterozygote F1 breeding and were genotyped by PCR with primers specific to the respective genes. The Per1^−/−, Pml^−/− and Per2^−/− mice used for these behavioral studies were derived from this breeding scheme. Both male and female mice deficient in either Per1^−/−/Pml^−/− or Per2^−/−/Pml^−/− were viable, fertile and were morphologically indistinguishable from wild type. Male siblings of the respective genotypes aged about 16 weeks from the backcross were used in the wheel-running studies. Wheel-running activity was monitored as described [6]. Briefly, mice were initially entrained to a LD cycle for at least 2 weeks, followed by their release into a constant darkness free-running condition. To assess the stability of the clock, the same mice were re-entrained to an LD cycle for 2 weeks and then released again into a free-running environment. For phase shift studies, the protocol was as previously described [8,9]. Briefly, a light pulse of 15 min with an intensity of 480lux was given using white fluorescent tube light. A linear regression using data from at least 10 days following the light pulse was used to determine the level of phase shift. Zeitgeber time zero (ZT0) is light on and ZT12 is light off.

All animal experiments in this study were carried out under an institutional approved animal protocol: HSC-AWC-06-077.

Male wild type, Pml^−/−, Per1^−/− and Per1^−/−/Pml^−/− mice were analyzed for their circadian behavior by monitoring their wheel-running activity in a LD cycle followed by a free-running period of 12 h dark/12 h dark (DD) cycles. For all four genotypes, the mice entrained to an LD cycle and were rhythmic in the free-running environment throughout our analysis. Interestingly, when released into a free-running state, the Per1^−/−/Pml^−/− mice displayed an “after effect of entrainment” where their daily period gradually shortened over the initial 20 days (Figure 1A, B). The period lengths of Per1^−/−/Pml^−/− mice were eventually shortened to an average 22.2 h in the free-running environment (Figure 1B). The Per1^−/−/Pml^−/− mice period lengths from day 20–40 and 20–60 were not significantly different. Under a free-running environment, the wild type, Per1^−/− and Pml^−/− mice displayed periods that averaged 23.4, 22.8 and 23.2 h, respectively (Table 1). The Per1^−/−/Pml^−/− mice displayed a significant level of activity prior to light off at ZT12 during the LD cycle (Figure 1A, also see Figure 2). On average, the Per1^−/−/Pml^−/− mouse phase angle of entrainment activity onset in the LD cycle started about 1.5 h prior to ZT12 (Figure 3). To a much lesser extent, the Per1^−/− mice displayed some activity prior to ZT12 consistent with previous observations [9]. Wild type and Pml^−/− mice showed no significant wheel-running activity before light off at ZT12. The current observations indicate that Per1^−/−/Pml^−/− mice have a significantly advanced phase angle of entrainment compared to Pml^−/−, Per1^−/− or wildtype mice during LD cycles.

Next we examined whether the phase shift behavior of Per1^−/−/Pml^−/− mice was different from that of wild type, Pml^−/− or Per1^−/− mice. A protocol in which the light pulse was given at either ZT14 or ZT22 on the last day of entrainment and prior to release into a free-running environment was used [8,9]. With this protocol, wildtype mice displayed a −1.3 ± 0.1 h and +0.8 h ± 0.2 h phase delay and phase advance response to ZT14 and ZT22 light pulses, respectively (Table 1). The Per1^−/− mice displayed a −1.5 ± 0.2 h and +0.5 h ± 0.3 h phase delay and phase advance response to ZT14 and ZT22 light pulses, respectively. The Pml^−/− mice displayed a −1.0 ± 0.1 h and +0.9 h ± 0.3 h phase delay and phase advance response to ZT14 and ZT22 light pulses, respectively. The Per1^−/−/Pml^−/− mice displayed a −2.0 h ± 0.2 h and +1.5 h ± 0.3 h phase delay and phase advance response, respectively (Figure 2A-D, Table 1). These observations demonstrate that the Per1^−/−/Pml^−/− mice have a significantly altered phase shift response, shorter average period length and increased advance of phase angle of entrainment when compared to wild type, Pml^−/− or Per1^−/− mice. Together, these behavioral observations indicate that the loss of PML and PER1 functions further altered the mammalian clock in vivo from that observed for the single gene mutant mice.

Next we examined the circadian behavior of Per2^−/−/Pml^−/− mice using a similar wheel-running analysis. When released into a free-running environment, some of the Per2^−/−/Pml^−/− mice displayed a transient rhythm that deteriorated into arrhythmic activities (Figure 4A-C). These behaviors were also observed in Per2^−/− mice. Previous studies of Per2^m/m mice described a similar behavioral phenotype [6]. This similarity in the Per2^−/− and Per2^−/−/Pml^−/− mice behavior suggests that the primary cause of the loss of rhythm behavioral phenotype in the free-running state is a loss of PER2 function. On the other hand, all arrhythmic Per2^−/− mice (n = 4) treated with a light pulse transiently returned to rhythmic behavior, as previously observed (Figure 4B) [9]. However, the transient post-light pulse rhythm seen in the arrhythmic Per2^−/− mice was not apparent in 50% of arrhythmic Per2^−/−/Pml^−/− mice (n = 4) that were similarly treated (Figure 4C). These observations suggest that the endogenous clock of Per2^−/−/Pml^−/− mice is further compromised from that of Per2^−/− mice.

The phase shift response of Per2^−/−/Pml^−/− mice was also examined using a similar protocol in which a light pulse was given at either ZT14 or ZT22 on the last day of entrainment prior to release into a free-running environment. The phase shift response could not be determined accurately due to immediate loss of rhythm in many of the Per2^−/−/Pml^−/− mice when released into a free-running condition. Comparing Per2^−/−/Pml^−/− mice to Per2^−/− mice, it was apparent that the Per2^−/−/Pml^−/− mice were much more active in the LD cycle (Figure 4C). The daily activity onset of Per2^−/−/Pml^−/− mice in an LD cycle was highly variable and changed from day to day, unlike that seen in Per2^−/− mice (Figure 4B, C). On average, the phase angle of entrainment was about 2.5 h prior to ZT12 (Figure 3). This is significantly greater than the value of 0.4 h prior to ZT12 previously reported for Per2^m/m mice [9]. These observations suggest that the robustness of LD entrainment is affected and the phase angle of entrainment is also advanced when mice are deficient in both PER2 and PML when compared to PER2 or PML deficient mice.