Standards of evidence in chronobiology: critical review of a report that restoration of Bmal1 expression in the dorsomedial hypothalamus is sufficient to restore circadian food anticipatory rhythms in Bmal1-/- mice

Abstract

Daily feeding schedules generate food anticipatory rhythms of behavior and physiology that exhibit canonical properties of circadian clock control. The molecular mechanisms and location of food-entrainable circadian oscillators hypothesized to control food anticipatory rhythms are unknown. In 2008, Fuller et al reported that food-entrainable circadian rhythms are absent in mice bearing a null mutation of the circadian clock gene Bmal1 and that these rhythms can be rescued by virally-mediated restoration of Bmal1 expression in the dorsomedial nucleus of the hypothalamus (DMH) but not in the suprachiasmatic nucleus (site of the master light-entrainable circadian pacemaker). These results, taken together with controversial DMH lesion results published by the same laboratory, appear to establish the DMH as the site of a Bmal1-dependent circadian mechanism necessary and sufficient for food anticipatory rhythms. However, careful examination of the manuscript reveals numerous weaknesses in the evidence as presented. These problems are grouped as follows and elaborated in detail: 1. data management issues (apparent misalignments of plotted data), 2. failure of evidence to support the major conclusions, and 3. missing data and methodological details. The Fuller et al results are therefore considered inconclusive, and fail to clarify the role of either the DMH or Bmal1 in the expression of food-entrainable circadian rhythms in rodents.

Figure 1

Duplicate 'actogram style' charts (modified, with permission, from Fuller et al [12]^© (2008) AAAS , Figures 3B and S3B, original supplementary online materials). The blue arrow indicates the ~3-h section that differs between the two versions of these data.

Figure 2

Average waveforms of body temperature in food restricted mice (modified, with permission, from Fuller et al [12]^© (2008) AAAS , Figure S3C). In both waveforms, temperature peaks prior to mealtime and begins dropping before mealtime, with no evidence of feeding induced thermogenesis. See also Fig. 3C (adapted from Fuller et al Figure S3C).

Figure 3

Actogram-style plots and corresponding average waveforms of body temperature in food restricted mice (modified, with permission, from Fuller et al [12]^© (2008) AAAS , Figure S3). For clarity, we have placed the waveform figures under the corresponding actogram-style figures. According to the text, these 2 waveforms were derived from days 10–14 of restricted feeding. We have aligned the waveforms and corresponding actogram-style plots, and drawn a blue line through the trough of body temperature that occurred (without explanation) in the middle of mealtime in waveform C, and through the peak in temperature that occurred in the middle of mealtime in waveform D. Clearly, the peak in temperature in D is not reflected by the density of the same data in B. In addition, the temperature curve in both waveforms is a mirror image on either side of the blue line. Therefore, in the actogram-style plots of the same data, the dark sections (indicating higher temperature) should also be symmetrical on either side of the blue line. They are not. In actogram-style plot B, high temperature is indicated during the first 1–2 h of mealtime, yet the corresponding waveform shows a lower temperature (likely below the daily mean) during that time.

Figure 4

The thermogenic effect of midday feeding in rats (R. Mistlberger, B. Kent, G. Landry, unpublished). Group mean average waveforms of core body temperature measured via implanted transponders in rats (N = 11) under adlib food access (thin black line), 4 h/day restricted feeding (heavy line = day1, heavy green line = day 21), and total food deprivation (heavy red line = day1). Temperature rises dramatically within 10 min of meal onset on days 1 and 21 of restricted feeding, and remains elevated throughout mealtime on the meal omission day after day 21.

Figure 5

The thermogenic effect of midday feeding in mice, adapted from Moriya et al [16]. Group mean average waveforms of core body temperature measured via implanted transponders in mice (sham lesion = open circles, DMH lesion = closed circles) under 4 h/day restricted feeding in LD (days 2 and 13) and total food deprivation in DD. Temperature rises dramatically within 15 min of meal onset on days 2 and 13 of restricted feeding, and remains elevated throughout mealtime on the meal omission day.

Figure 6

Bmal1 and Per1 expression in a Bmal1+/- control mouse and a Bmal1-/- mouse that received intra-DMH AAV-BMAL1 injections bilaterally (modified, with permission, from Fuller et al [12]^© (2008) AAAS , Figure S4). Bmal1 expression was restored bilaterally and symmetrically by AAV-BMAL1 injections into the SCN (Panel D) or DMH (panel H) in Bmal1-/- mice. The panels do not include other structures that normally express Bmal1 in control mice to confirm that Bmal1 expression was restricted to the SCN or DMH in null mutants. Panels E and G illustrate Per1 expression in full coronal sections from a Bmal1+/- control mouse and a Bmal1-/- mouse that received an intra-DMH AAV-BMAL1 injection.

Figure 7

Actogram-style plots of body temperature during restricted feeding from Bmal1-/- mice with or without AAV-BMAL1 injections into the DMH (modified, with permission, from Fuller et al [12]^© (2008) AAAS , Figure 3). Panel B. Bmal1-/- mouse that received AAV-BMAL1 injection to DMH. Panel C. Bmal1-/- mouse that received no injection. Red line denotes mealtime. Red arrow denotes 24 h food deprivation test.

Figure 8

Body temperature in Bmal1-/- mice during restricted daily feeding (modified, with permission, from Fuller et al [12]^© (2008) AAAS , Figure 2B, left, and S3D, right). Neither mouse received AAV-BMAL1 injections. The black regression lines were added here.

Figure 9

Percent change of body weight in wildtype and Bmal1-/- mice during restricted daily feeding (J. Pendergast and S. Yamazaki, unpublished). Body weights of wildtype mice (grey lines, N = 6) and Bmal1-/- mice (red and black lines, N = 7) during ad-lib food access (days -5 to 0) and 4-h/day restricted food access (days 1–10), expressed as percent change from day 0. A 25% body weight loss was established as an endpoint criterion, at which time mice were returned to ad-lib food access, to prevent mortality. Only one Bmal1-/- mouse (black line) remained above the endpoint criterion over the 10 days of restricted feeding.