Circadian regulation of diel vertical migration (DVM) and metabolism in Antarctic krill Euphausia superba

Vertical migration

We performed two separate vertical migration experiments (Fig. 1A,B). In the first one (Fig. 1A) a group of 45 krill was exposed to LD for 48 h. In the second one (Fig. 1B) a group of 41 krill was exposed to LD for 24 h followed by constant darkness (DD) for 48 h. During LD, lights were turned on at 6:00 (corresponding to Zeitgeber Time 0 or ZT0). Light intensity gradually increased until 12:00 (ZT6) and then decreased gradually until 18:00 (ZT12), when the lights were turned off. During DD, lights were turned off at all times. In both experiments, krill could move freely within a vertical column tank (200 cm height × 50 cm diameter) filled with chilled and filtered seawater (Supplementary Fig. S1). No food was offered at any time during the experiments and an acclimation period of three days (LD, no food) was provided before each experiment. We used an infrared (IR) camera system to monitor DVM during light and dark phases and estimated the variation in mean krill depth over time. We tested the presence of rhythmic patterns using the RAIN algorithm, which allows the detection of rhythms of any period and waveform²³.

Clear rhythmic oscillations were detected in both experiments. When krill were exposed to LD, they displayed rhythmic vertical migration with a period of about 24 h. Peak upward migration occurred towards the second half of the light phase. In the first experiment (Fig. 1A), RAIN detected a period of oscillation of 24 h (p-value = 3.5e⁻⁰⁷) in day 1 and a period of 23 h (p-value = 5.4e⁻⁰⁸) in day 2, with peak upward migration at ZT9 in both days. In the second experiment (Fig. 1B), RAIN detected a period of 24 h (p-value = 4.2e⁻⁰⁴) in day 1, when krill were exposed to LD, with peak upward migration at ZT8.

Krill DVM in the field displays mostly a nocturnal migratory pattern, characterized by upward migration during the night and downward migration during the day^6,11. Therefore, we were at first surprised to observe the opposite DVM pattern in the lab, with upward migration during the day (i.e. light phase) and downward migration during the night (i.e. dark phase). However, previous studies on Drosophila already revealed that circadian behaviors might differ significantly between wild and laboratory populations of the same species²⁴. Physical constriction, exposure to artificial light regimes and feeding schedules and deprivation of ecological interactions may have a profound impact on the behavior of captured animals²⁵. In the aquarium, krill are fed mostly during the day. This, in association with the absence of predators, might have led to the reversal of the DVM rhythm. In addition, in order to avoid interaction with rhythmic cues related to food, krill were briefly starved before and during the experiment. Previous studies suggested that starved zooplankton might become more attracted towards light sources^26,27. This may have contributed to stimulate the upward movement observed during the light phases. Additional observations of DVM with (A) krill accustomed to different feeding schedules (e.g. krill fed during the night) and (B) starved krill exposed to illuminated dark phases (DL or LL) might help to clarify these points.

After we switched to DD conditions (Fig. 1B), during the first day we did not observe any significant rhythm. This might not necessarily imply that no rhythmic migration occurred. In fact, individual krill might have been still rhythmic, but they might have not been synchronized with each other due to the absence of the LD entraining cue or Zeitgeber. A similar phenomenon has already been reported for wild zooplankton in the Arctic and Antarctic during periods of continuous illumination in summer (i.e. during midnight sun)^28,29. At that time of the year, in the Arctic, the local populations of Calanus finmarchicus and C. glacialis did not display any net DVM as revealed by backscatter data, but unsynchronized individual migrations were registered instead²⁹. Implementation of individual tracking methods to apply in the lab during DVM monitoring under extreme photoperiodic conditions including constant darkness (DD) and constant light (LL) would allow further insight into this aspect.

During the second day of DD exposure, RAIN detected a period of 12 h (p-value = 2.3e⁻⁰²), with a first peak of migration in the early subjective light phase (CT29) and a second peak in the early dark phase (CT41). This suggested that in free-running conditions an endogenous rhythm of vertical migration might emerge with a period of about 12 h. This is in agreement with previous studies of swimming activity, oxygen consumption, enzyme activity and transcription in krill exposed to DD^16,18,22 and strongly suggests that the endogenous clock in krill is characterized by a free-running period significantly shorter than 24 h.

Oxygen consumption

An endogenous rhythm of activity occurring at the organismic level might contribute to the occurrence of DVM in zooplankton^19,30. This might apply to krill as well, since they are known to display daily rhythms in metabolism and physiology^15,16. Therefore, we monitored daily rhythms of oxygen consumption in krill exposed to LD and DD and associated them with the observed DVM patterns. Due to technical limitations, we could not monitor oxygen consumption directly within the vertical migration tank. We used individual krill incubated in 2 L Schott bottles filled with oxygen-saturated chilled and filtered seawater instead. We followed the decrease in pO₂ over a period of 48 h and tested the presence of rhythmic consumption patterns using the RAIN algorithm. Clear rhythmic oscillations were detected in both conditions.

In LD, two out of six tested animals (33%) displayed rhythmic oxygen consumption (Fig. 2A,B). Both individuals displayed clear oscillations in the circadian range (i.e., approx. 24 h period): the first one (krill#4, Fig. 2A) displayed a period of 21 h in day 1 (p-value = 1.9e⁻⁰⁴) and 24 h in day 2 (p-value = 3.7e⁻⁰²), while the second one (krill#6, Fig. 2B) displayed a period of 24 h in both days (day 1, p-value = 5.3e⁻¹¹; day 2, p-value = 3.9e⁻¹¹). The oscillation was always strongly synchronized to the LD cycle, with peak oxygen consumption occurring at ZT11 (day 1) and ZT12 (day 2) in krill#4 (Fig. 2A) and ZT12 (day 1) and ZT13 (day 2) in krill#6 (Fig. 2B), corresponding to the light/dark transitions. Considering that the animals were incubated within a small volume of water (2 L) and their ability of swim freely might have been severely reduced, changes in swimming activity might not have greatly contributed to the oxygen uptake oscillation. Therefore, this could be interpreted as the result of an internal metabolic oscillation instead. This would support the hypothesis that an internal rhythm of activity might contribute to DVM in krill, even if the mechanisms still remain unknown. In the Calanoid copepod Calanus finmarchicus similar oscillations of oxygen consumption were interpreted as a metabolic anticipation of DVM¹⁹.

In DD, four out of seven tested animals (57%) displayed rhythmic oxygen consumption (Fig. 3A–D). Three of them (krill#3, #4 and #7, Fig. 3B–D) showed a period of about 24 h. The peak of oxygen uptake was advanced compared to LD and occurred during the subjective light phase. Krill#3 (Fig. 3B) displayed a period of 23 h in day 1 (p-value = 1.0e⁻⁰⁷) and 24 h in day 2 (p-value = 4.2e⁻⁰⁶), with peak oxygen consumption at CT11 and CT34. Krill#4 (Fig. 3C) displayed a period of 21 h in day 1 (p-value = 2.0e⁻⁰⁵) and 24 h in day 2 (p-value = 1.6e⁻⁰⁶), with peak consumption at CT9 and CT28. Krill#7 (Fig. 3D) displayed a period of 24 h in day 2 (p-value = 2.2e⁻⁰⁷), with peak consumption at CT28. A fourth individual (krill#2, Fig. 3A) displayed a period of 12 h in day 1 (p-value = 1.7e⁻⁰³), with a first peak of consumption at CT3 during the early subjective light phase and a second peak 12 h later (CT15) during the early dark phase. The general advance of the oxygen uptake peak observed in krill#3, 4 and 7 together with the 12 h rhythm displayed by krill#2 suggested that in free-running DD conditions a process of period shortening similar to the one observed for DVM might occur also for oxygen consumption. Previous studies suggested that in krill similar 12 h free-running endogenous rhythms might occur also at the level of metabolic gene expression and enzymatic activity¹⁶. This, together with the recent analysis of krill circadian transcriptome revealing a significant 12 h free-running oscillation at the molecular level, strongly suggests that a short endogenous period of oscillation might characterize the circadian clock of Antarctic krill.

The relatively small percentages (33% in LD and 57% in DD) of individual krill showing significant rhythmic oxygen consumption might have been related to the small sample size in our experiment (n = 6 in LD and n = 7 in DD). However, additional measurements of krill oxygen consumption performed using the same methods in LD and DD on larger samples (n = 10), on board of RV Polarstern during Antarctic expedition PS 112 (March to May 2018), displayed similar percentages of rhythmic individuals (3 out of 10 or 30% in LD, and 4 out of 10 or 40% in DD) (Supplementary Figs. S2A–C and S3A–D). Interestingly, the on-board analyses revealed the same tendency to develop shorter periods of oscillation in DD (krill#7: T = 12 h, p-value = 8.5e⁻⁰⁴; krill#9: T = 17 h, p-value = 8.3e⁻⁰⁸) (Supplementary Fig. S3C,D). Also, on-board measurements in LD displayed almost a reversed phase of oscillation compared to those in the lab, with peak oxygen consumption occurring at the beginning of the light phase, around ZT4 (Supplementary Fig. S2A–C). It is tempting to speculate that this might be related to the reverse DVM pattern observed in the lab. Unfortunately, we do not have the corresponding DVM measurements from the field to support this hypothesis. Additional field observations of krill rhythms of oxygen consumption in association with DVM would help to clarify how these two phenomena might be related to each other.

Dissection of putative circadian oscillators in krill brain and eyestalks

It might be possible that the DVM rhythm and the metabolic oscillation observed in krill were under the influence of separated circadian oscillators. This was suggested by the different responses displayed by DVM and oxygen consumption in DD. While DVM showed a complete shift to a 12 h rhythm, oxygen consumption only showed an advance in the peaking time.

Previous studies on the circadian system of Crustaceans identified multiple oscillators located in the head and along the body³¹. In particular, a model has been proposed where the circadian oscillators in the head are situated in the brain, in the eyestalks and in the retinae of the compound eye (Supplementary Fig. S4)³¹. To investigate this hypothesis we compared, for the first time, daily patterns of clock genes expression in brain and eyestalks tissue of krill exposed to LD and DD. Krill were sampled within a 72-h time-series, 9 animals were collected every four hours. During the first 24 h (ZT0-24), krill were exposed to LD, while during the remaining 48 h (CT0-48) krill were exposed to DD. We dissected brain and eyestalks tissue (Supplementary Fig. S5) and measured relative changes in clock-related mRNAs over time during the first day in LD (ZT0-24) and during the second day in DD (CT24-48) (Supplementary Table S1). We used RAIN to check for rhythmic oscillations within tissue.

Clear rhythms of oscillation were found in both tissues. In LD, the core clock components in the brain displayed an antiphase relationship between positive (clk-cyc) and negative (per-tim) regulators, with clk-cyc peaking around ZT16-20 (clk: T = 24 h, p-value = 4.86e⁻⁰⁷; cyc: T = 24 h, p-value = 2.28e⁻¹²) and per-tim peaking around ZT4 (per: T = 24 h, p-value = 2.99e⁻⁰⁵; tim: T = 24 h, p-value = 1.31e⁻¹¹) (Fig. 4A). In the eyestalks such antiphase relationship was not present and most clock genes showed upregulation during the dark phase (ZT16-24) (Fig. 4B). The antiphase relationship is a typical feature of the circadian clock and is well-known from studies on model organisms like Drosophila and mouse². So far, previous studies on krill failed to demonstrate a clear antiphase relationship between positive and negative regulators^21,22. This was discussed as a non-canonical feature of krill circadian clock, as observed before in other Crustacean species^32,33. However, our results suggest that this might be related to the specific tissue analyzed. In fact, previous works focused mostly on the eyestalks or on the full head^21,22, possibly failing to detect the antiphase oscillation in the brain. This indicates that separate oscillators might be present in the brain and the eyestalks of krill, as already suggested for other Crustaceans³¹, and suggests that the central pacemaker might be located in the brain. The oscillator in the eyestalks, where all clock genes displayed upregulation during the dark phase, seems to be mostly influenced by the external photoperiod and might be considered as a peripheral oscillator.

As in LD, also in DD different patterns of regulations were observed in the brain and in the eyestalks (Supplementary Fig. S6A,B). In the brain, clk, cyc and per, three of the core clock components, displayed a marked tendency towards a shortening of the oscillation period (Fig. 5A–C). Clk shifted from 24 to 12 h (p-value = 1.31e⁻¹⁶), cyc from 24 to 16 h (p-value = 2.48e⁻¹⁰) and per from 24 to 12 h (p-value = 4.05e⁻⁰⁷). In the eyestalks, only per showed a similar tendency, shifting from 24 h (p-value = 3.69e⁻⁰⁶) to 16 h (p-value = 2.36e⁻¹³) (Fig. 5D). This again suggested that separated oscillators might be present in krill brain and eyestalks. The shortening of clock gene oscillation period observed in DD in the brain and, to a lesser extent, in the eyestalks, together with the short (12–16 h) oscillation period registered for cry2 across all tissues and conditions (brain LD: T = 12 h, p-value = 2.4e⁻⁰⁸; brain DD: T = 12 h, p-value = 1.4e⁻⁰⁴; eyestalk LD: T = 12 h, p-value = 3.2e⁻⁰⁵; eyestalk DD: T = 16 h, p-value = 6.1e⁻¹³) (Supplementary Fig. S7A,B), is in agreement with previous observations of krill clock gene expression in DD²² and strongly indicates that krill endogenous circadian period might be significantly shorter than 24 h also at the molecular level.

The presence of separated oscillators in krill might help to coordinate different aspects of daily rhythmicity on different levels including physiology (e.g. oxygen consumption) and behavior (e.g. DVM). This might allow a flexible regulation of daily rhythms in response to local changes in environmental conditions driven by temporary and/or seasonal factors including photoperiod, food availability and presence/absence of predators among others.

Adaptive significance of the short endogenous free-running period in krill

The tendency in krill to display a 12 h endogenous free-running period might represent a circadian adaptation for living at high latitudes, where the photoperiodic signal displays strong seasonal variability^16,22. Surveys from plants and insects along broad latitudinal gradients indicate that high-latitude species tend to have shorter endogenous periods compared to low-latitude ones³⁴. The reasons why this happens are not fully understood yet, but it has been proposed that the short endogenous period might help the clock to entrain to a wider range of photoperiods³⁴. Indeed, krill displayed rhythmic clock gene activity during the midnight sun in summer, suggesting that their clock can successfully entrain to extremely long photoperiods^17,35. However, exposition to similar extreme photoperiods in the laboratory apparently caused the disruption of the clock, suggesting that in the field krill might switch to alternative Zeitgebers (i.e. entraining cues) to entrain the clock when photoperiod becomes extremely long/short (e.g. midsummer/midwinter)³⁶.

From an ecological perspective, having a short endogenous period might help krill to adapt to different environmental scenarios. In the presence of overt day/night cycles, the clock would entrain to the photoperiod and promote daily rhythms with a period of approx. 24 h, like for example the nocturnal DVM pattern usually observed during spring and autumn. This would help krill to anticipate the day/night cycle and maximize the costs/benefits balance between time spent feeding at the surface and time spent hiding in deeper water layers. At the same time, the clock would promote 24 h oscillations in krill metabolism and physiology and coordinate phases of high and low activity with phases of upward and downward migration during DVM.

On the other hand, in the absence of overt day/night cues, for example during summer and winter, the clock would tend to shift towards the free-running period and promote daily rhythms with a period shorter than 24 h. Observations of krill DVM in the field during winter are scarce, due to the harsh weather conditions which characterize the Southern Ocean at that time of the year. Much of the rhythmic biology of krill during winter still remains unknown. At the same time, field observations of krill DVM and swimming activity during summer suggest that a de-synchronized, “around-the -clock”, individual migratory movement might be present instead^18,28. According to some studies, frequent shallow individual migrations could occur during summer as a result of a hunger-satiation mechanism, triggered by the abundant primary production occurring in the surface layers³⁷. To get further insight into these aspects, first we would need to perform additional laboratory studies and field observations of krill rhythmic functions (oxygen consumption, swimming activity, DVM) under extreme photoperiods (LL-DD in the lab, summer–winter in the field). Second, laboratory and on-board trials with krill exposed to different food concentrations (low-medium–high) and different photoperiods (LD-DD-LL) could be used to study the interaction between food and light cues in the emergence of “opportunistic” rhythmic responses.

Source: Ecology - nature.com