The lack of good antibodies makes it difficult to assess the impact of circadian time, phase shifting stimuli, or clock gene mutations on clock gene protein products. While there are commercially available antibodies, many have not been subjected to validation to assess their usefulness in immunocytochemistry. Here we generated several antibodies against the clock proteins PER1, PER2, BMAL1 and CLOCK, and tested them for staining in the SCN of both mice and hamsters (, , and ), identifying several that gave excellent results (, , and ).
A previously described PER1 antibody, generated in rabbit to the amino terminus of mouse PER1 (residues 6–21), called 1177, has been used successfully in many publications 
. This antibody gives good labeling of SCN cells in mouse, and no staining in Per1−/−
. In the hamster, 1177 stained SCN cells at about the same intensity as R43 and GP91 when used at the same concentrations. The 1177 antibody labels cells and fibers of the magnocellular neurons in the paraventricular and supraoptic nuclei, as previously described 
, while R43 and GP91 do not ().
Photomicrographs of hamster SCN sections immunostained with PER1 antibodies.
For PER2, R38 and R39 antibodies produced good results in mouse SCN. Neither of these antibodies worked in hamster SCN. R38 has also worked in Nile grass rats (Arvicanthis niloticus
and lab rats (Rattus norvegicus
The PER2 antibody GP87 produced non-specific staining in mouse, as it labeled SCN cells with the same intensity at ZT0 and 12, and staining intensity was not reduced in the SCN of Per2−/− mice (, ). While GP87 appeared to label hamster SCN cells strongly at ZT12 and weakly at ZT0, in view of the lack of specificity seen in mouse, this staining may not be specific to PER2 (see below for further consideration).
Although PER1 and PER2 expression are higher at ZT12 than at ZT0, some cells in the central SCN show high PER levels at ZT0 
. It may be that these cells express PER constitutively, or they may express PER in antiphase with the larger population 
, or the decline in PER expression in these cells may be delayed relative to the larger population 
CLOCK protein is generally thought to be constitutively expressed in the SCN 
, although there is a report indicating that CLOCK is constitutively expressed in young but rhythmically expressed in old C57BL/6J mice 
. Here, R41 worked in both mouse and hamster where it was highly expressed at both ZT0 and 12, suggesting constitutive expression. Immunostaining of the mouse SCN, and the absence of staining in Clock−/−
mice, has recently been reported by another group using antibody R41 
The previously published evidence for rhythmicity in BMAL1 protein levels is not consistent. Some studies describe rhythmic BMAL1 protein levels in the SCN, either with a nocturnal peak in rats (at ZT or CT14–22 in Wistar rats in LD or after 2 or 3 days in dim red light 
), or with higher levels in daytime than nighttime in CD-1 mice in a 12
12 LD cycle 
. Other studies show no rhythm of BMAL1 staining in the SCN of C3H/HeJ mice kept for 1 day in dim red light after being in a 12
12 LD cycle 
or in young and old C57BL/6J mice 
. Here, three antibodies to BMAL1 (R37, GP85, GP86) stained the mouse SCN well, while two of these antibodies (R37 and GP85) also stained the hamster SCN. In both species, with each of these antibodies, BMAL1 expression was high at ZT0 and low at ZT12, indicating a rhythm in BMAL1 in the SCN. Rhythmic expression of Bmal1
RNA levels, with low levels during mid-day, has been reported in several studies; the phase of the BMAL1 protein rhythm reported here seems consistent with the 4–6 hr delay between transcript peak and protein peak typical of circadian clock proteins 
. Further analysis of BMAL1 protein rhythms is warranted.
While some antibodies labeled cells in extra-SCN regions, here we focused on SCN staining, as not all brain regions were tested for all antibodies and the time of peak expression in extra-SCN regions differs from the SCN 
. Nevertheless, antibodies producing robust signal in the SCN are likely to be useful for assessing protein rhythms in other neural sites.
Our systematic comparison of a set of polyclonal antisera to four key circadian proteins reveals some unexpected differences between the performance of these antisera in mouse and hamster. Most notably, three PER2 antisera useful in mouse SCN appear not to be useful in hamsters. As noted above, one of these PER2 antisera, R38, has been used successfully in the SCN and extra-SCN regions of Nile grass rats and lab rats 
. The single antibody that produced what appeared to be an SCN-specific signal with rhythmic labeling in the hamster was GP87. This antibody produced non-specific labeling in the mouse SCN, however, raising doubts as to its specificity in hamster as well. This species difference in PER2 staining with these antisera may be due to differences in the N-terminal 200 amino acid residues of the PER2 sequence used as the antigen. A 171-residue amino-terminal fragment of hamster PER2 
has 11 mismatches and requires 3 gaps to align with the mouse sequence it overlaps within the amino-terminal 200 residues. In contrast, the 200 amino-terminal residues of Arvicanthis niloticus
PER2 align with the mouse sequence without gaps, and the sequences differ at only 8 residues (96% identity). Similarly, the lab rat PER2 sequence aligns to the mouse sequence without gaps and differs at only 7 residues (96.5% identity). Thus, despite the use of large (200 to 400 residue) fragments of the circadian proteins as immunogens (intended to produce a polyclonal response and maximizing the probability of reactivity in multiple species), the usefulness of these antisera in other species will need to be examined. Nevertheless, our study identifies a series of antibodies to these key circadian proteins that are useful in mice, and which may be useful in other species, possibly depending on the extent of species conservation of the amino acid sequences involved. An advantage of these antisera over some that are commercially available is that the antigens used here are clearly disclosed, so sequence alignment can be performed to predict potential reactivity in other species. (Four antibodies described here, PER1-R1177, PER2-R38, CLOCK-R41 and BMAL1-GP85 are commercially available through Millipore Corporation as product numbers AB2201 through AB2204). A further advantage is our use of genetically modified mice and two time-points with expected differences in protein expression levels, allowing functional assessment of the antibodies' performance. While each lab will need to verify the utility of these antibodies for their own species of choice and staining protocol, the results reported here indicate a starting point that should facilitate other work on immunocytochemical localization of circadian clock proteins in neural tissues.