Fluctuations of brain states, such as arousal and attention, strongly impact sensory processing, decision-making, and animal behavior (Harris and Thiele, 2011; Lee and Dan, 2012; Thiele and Bellgrove, 2018; Petersen, 2019; McCormick et al., 2020). It is thus critical to understand the neural substrates of brain states and how state changes can account for the variability embedded in neuronal and behavioral data (Cano et al., 2006; Poulet and Petersen, 2008; Polack et al., 2013; Fu et al., 2014; Zagha and McCormick, 2014; Lee et al., 2020). Changes in pupil diameter under constant luminance are tightly linked to states of arousal and attention (McGinley et al., 2015a; Joshi and Gold, 2020). Dynamic pupil responses are associated with membrane potential fluctuations, sensory evoked responses, salience detection, error estimation, decision-making, and task performance (Kahneman and Beatty, 1966; Bijleveld et al., 2009; Nassar et al., 2012; de Gee et al., 2014; de Gee et al., 2020; Reimer et al., 2014; Vinck et al., 2015; McGinley et al., 2015b; Lee and Margolis, 2016; Schriver et al., 2018; Schriver et al., 2020; Kucewicz et al., 2018; Ebitz and Moore, 2019). As a result, pupil diameter has been widely used to monitor brain states and their neural substrates.

A multitude of neural circuits have been implicated in mediating brain state and pupil size changes, most notably the neuromodulatory systems (Yu and Dayan, 2005; Lee and Dan, 2012; Thiele and Bellgrove, 2018; Joshi and Gold, 2020). The locus coeruleus (LC)–noradrenergic system has long been thought to play a critical role in controlling arousal and attention (Aston-Jones et al., 1999; Berridge and Waterhouse, 2003; Aston-Jones and Cohen, 2005; Sara, 2009; Sara and Bouret, 2012), and LC activity closely tracks brain states and cognitive processes (Foote et al., 1980; Aston-Jones and Bloom, 1981; Berridge and Foote, 1991; Aston-Jones et al., 1994; Rajkowski et al., 1994; Usher et al., 1999; Takahashi et al., 2010; Carter et al., 2010; Eschenko et al., 2012; Vazey and Aston-Jones, 2014; Kalwani et al., 2014; Varazzani et al., 2015; Bouret and Richmond, 2015; Fazlali et al., 2016; Swift et al., 2018). Importantly, work mainly in the past decade has provided correlative and causal evidence linking pupil size changes to LC activity (Rajkowski et al., 1994; Murphy et al., 2014; Varazzani et al., 2015; Joshi et al., 2016; Reimer et al., 2016; de Gee et al., 2017; Liu et al., 2017; Breton-Provencher and Sur, 2019; Hayat et al., 2020; Privitera et al., 2020), leading to the increased utilization of pupil diameter as a noninvasive readout of LC (Aston-Jones and Cohen, 2005; Gilzenrat et al., 2010; Preuschoff et al., 2011; Konishi et al., 2017; Gelbard-Sagiv et al., 2018; Zhao et al., 2019; Aminihajibashi et al., 2020; Clewett et al., 2020). However, a few recent studies demonstrated that the correlation between pupil and LC could be neuron- and task epoch-specific (Joshi et al., 2016; Breton-Provencher and Sur, 2019; Yang et al., 2021), raising the possibility that pupil diameter can be dissociated from LC activity. To the best of our knowledge, we do not know to what extent pupil diameter is linked to LC activity. More importantly, we do not know whether and how pupil diameter can be used to make accurate inferences of LC activity on a moment-by-moment basis.

To address these questions, we recorded spiking activity from optogenetically tagged LC neurons simultaneously with pupil diameter in head-fixed mice trained to perform a tactile detection task (McBurney-Lin et al., 2020; Yang et al., 2021). We established a graded relationship between pupil and LC, where pupil dilation increased monotonically with LC spiking activity. However, this relationship exists with substantial variability such that pupil size changes can only accurately predict a small fraction of LC spiking on a moment-by-moment basis. Using optogenetics to activate LC neurons, we showed that pupil responses exhibited large session-to-session fluctuations to identical optical stimulation, despite stable LC responses. Notably, decision bias-related behavioral variables explained the variations in the pupil–LC relationship. Together, our data show that substantial variability exists in an overall positive relationship between pupil diameter and LC activity, and that only under limited conditions can pupil be used as an accurate real-time readout of LC. Our work further suggests that brain states dynamically modulate the coupling between pupil and LC.

We recorded spiking activity from optogenetically tagged single units in the LC together with pupil diameter in head-fixed mice during behavior (Figure 1a). To quantify a graded relationship between pupil size changes and LC spiking, we first grouped adjacent spikes into individual clusters (Hahn et al., 2010; Yu et al., 2017) based on each unit’s median interspike interval (Figure 1b, Figure 1—figure supplement 1, Methods). The magnitude of pupil responses following a spike cluster (quantified in a 6-s window from cluster onset) progressively increased with cluster size (the number of spikes in a cluster, Figure 1c, d). The latency of peak pupil diameter did not systematically vary with cluster size and ranged between 2.5 and 4 s (Figure 1—figure supplement 2). This latency is consistent with our previous report (Yang et al., 2021). Overall, we found a positive, monotonic relationship between peak pupil diameter and LC cluster size in the majority of paired recordings (linear regression R² > 0.6 in 13 out of 19 paired recordings, Figure 1e), in line with previous findings in nonhuman primates (Varazzani et al., 2015; Joshi et al., 2016). Similar relationships held when pupil responses were quantified as % changes from baseline (Cazettes et al., 2021) or the time derivatives of pupil (Reimer et al., 2016; Yang et al., 2021; Figure 1—figure supplement 3). However, substantial variations existed in the relationship (linear slopes ranging from 0.12 to 0.51. 0.24 ± 0.11, mean ± standard deviation [SD], n = 13), indicating variable couplings between pupil and LC neurons.

Figure 1 with 5 supplements see all

Download asset Open asset

Figure 1—source data 1

Source data for Figure 1.

https://cdn.elifesciences.org/articles/70510/elife-70510-fig1-data1-v1.zip

Download elife-70510-fig1-data1-v1.zip

Although pupil diameter exhibited an overall monotonic relationship with LC spiking, it did not necessarily warrant pupil diameter being an accurate readout of LC activity. We tested the extent to which pupil size changes can be used as a proxy for LC activity, that is, can we use pupil diameter to make accurate inferences of LC spiking on a moment-by-moment basis? We asked how well an ideal observer (Green and Swets, 1966) can predict LC cluster size given the associated peak pupil responses (Methods). Receiver-operating-characteristic (ROC) analysis showed that as cluster size increased, peak pupil diameter can better predict LC activity (Figure 1f, g, Figure 1—figure supplement 4). However, only peak pupil diameter associated with large clusters (≥5–6 spikes) can achieve a performance threshold of d′ = 1 (translates to ~0.75 area under the curve Simpson and Fitter, 1973, Figure 1g). Since larger clusters occurred less frequently (Figure 1d, h, Figure 1—figure supplement 5), our data suggest that pupil dilation cannot accurately represent the majority of LC spiking activity but can serve as a good proxy for the infrequent (<10%) and strong LC activity in real time (Figure 1h).

Perhaps what is more interesting (and useful) is to assess whether we can directly use pupil diameter to infer the ‘ground truth’ – LC activity, without recording from LC. To do so, we first detected pupil dilation events based on zero-crossings of pupil derivatives (Joshi et al., 2016; Figure 2a) and quantified LC spike counts immediately preceding each dilation event (Methods). Compared with the analyses in Figure 1, this method did not require prior knowledge of LC activity for identifying pupil responses and yielded a similar pupil–LC relationship (Figure 2—figure supplement 1). Overall, LC spike counts were monotonically associated with pupil dilation amplitudes (Figure 2b–d). However, a wide range of spike counts preceded pupil events of similar sizes (Figure 2b, c). We asked how well an ideal observer can predict pupil dilation events given the associated LC activity and found that as pupil dilations became larger, LC spike counts could make better predictions on a moment-by-moment basis. However, only LC activity preceding the infrequent (<10%), large dilation events (>1.5–2 SD, Figure 2—figure supplement 2) performed beyond 75% threshold (Figure 2e). Finally, we tested how well we can use the detected pupil dilation events to predict LC activity. Similar to the previous results (Figure 1g), we found that only large pupil events can achieve good predictions (Figure 2f). Taken together, our data show that pupil diameter and LC spiking are well correlated in a graded manner and that the infrequent (<10%) but strong (>1.5–2 SD) pupil dilation events can be used to accurately and reliably predict LC activity in real time.

Figure 2 with 2 supplements see all

Download asset Open asset

Figure 2—source data 1

Source data for Figure 2.

https://cdn.elifesciences.org/articles/70510/elife-70510-fig2-data1-v1.zip

Download elife-70510-fig2-data1-v1.zip

Our data presented so far were based on paired pupil–LC recordings, each consisting of a single opto-tagged LC unit. Next, we sought to test whether pupil size changes better reflect population-level LC activity instead of single neurons. To this end, we optogenetically activated groups of LC neurons and quantified the evoked pupil responses. Based on the stimulation parameters, we estimated an excitable volume on the order of 0.05–0.1 mm³, containing hundreds of LC neurons (Figure 3a, b, Figure 3—figure supplement 1, Methods). In a subset of experiments, the putatively same LC units were tracked (typically 1–5 days), based on opto-tagging, spike clustering, and waveform comparison (Figure 3c, d). Waveforms from the putatively same units were more similar than the waveforms from the putatively different units (Figure 3e–g). These putatively same units responded similarly to optical stimulation in different sessions (Figure 3h), suggesting a consistent transduction of optical stimulation to LC spiking activity. In contrast to stable LC responses, the same pupil exhibited variable dilations to optical stimulation under awake, nontask performing conditions (Figure 4a, b). Importantly, baseline pupil diameters were similar (0.71 vs. 0.75 mm, Figure 4—figure supplement 1) and thus cannot explain the differences in evoked pupil responses. Group data from multiple mice further demonstrated that significant session-to-session fluctuations of pupil responses were prevalent but not directional (solid lines in Figure 4c, d), that is, pupil responses in an earlier session (session 1) were not consistently higher or lower than in a later session (session 2). Therefore, such session-to-session fluctuations were not observable from group comparisons (Figure 4—figure supplement 2, Privitera et al., 2020). To further test whether the variable pupil responses were due to (1) weak LC stimulation with 10 ms pulses, or (2) strong spontaneous pupil fluctuations during wakefulness, we performed the following experiments. First, we evoked pupil responses with stronger stimulation (50 ms pulses instead of 10 ms) in the awake condition. While baseline pupil diameters were similar between sessions, evoked pupil responses still fluctuated significantly (Figure 4e, f). In a subset of experiments, pupil exhibited substantial fluctuations in two sessions just several hours apart (4–6 hr, magenta arrows in Figure 4c–f). Further analysis showed that across-session variability of pupil responses was comparable to within-session variability (Figure 4—figure supplement 3, Methods). In addition, for the paired sessions that exhibited significantly different responses to optical stimulation (solid lines in Figure 4c–f), only a small subset exhibited larger across-session variability than within-session variability (2 pairs out of 12 under 10 ms condition, and 3 pairs out of 11 under 50 ms condition, Methods). Next, we stimulated LC with 10 ms pulses under anesthesia (2% isoflurane) to minimize spontaneous pupil fluctuations (Figure 4—figure supplement 4). Evoked pupil responses were noticeably larger compared with the awake condition in the example recordings, possibly due to a more constricted baseline pupil size under anesthesia (Figure 4g, h, left vs. Figure 4a, b, 0.3 vs. 0.7 mm). Nevertheless, pupil responses to optical stimulation exhibited substantial session-to-session fluctuations (Figure 4g, h). Additional examples of a simultaneously recorded LC unit and pupil diameter in responses to optical stimulation are in Figure 4—figure supplement 5. In summary, pupil responses showed large session-to-session fluctuations to identical LC stimulation.

Figure 3 with 1 supplement see all

Download asset Open asset

Figure 3—source data 1

Source data for Figure 3.

https://cdn.elifesciences.org/articles/70510/elife-70510-fig3-data1-v1.zip

Download elife-70510-fig3-data1-v1.zip

Figure 4 with 6 supplements see all

Download asset Open asset

Figure 4—source data 1

Source data for Figure 4.

https://cdn.elifesciences.org/articles/70510/elife-70510-fig4-data1-v1.zip

Download elife-70510-fig4-data1-v1.zip

What may underlie the variable pupil responses? We found that the variations in the relationship between peak pupil diameter and LC cluster size (as in Figure 1e) were strongly correlated with hit rate and decision bias during task performance (Figure 5a). This effect was not likely due to linear fitting of nonlinear relationships (all linear fits are of R² > 0.85. 0.92 ± 0.05, mean ± SD, n = 9), and the results held when the analysis of pupil–LC relationship was restricted to nonlicking periods only (Figure 5b, Methods). Therefore, although mice licked more during sessions of higher hit rate and lower decision bias, the results cannot be fully explained by a potentially stronger pupil–LC coupling during licking periods. Based on these findings, we conclude that decision bias-related behavioral variables could explain, at least in part, the variations in the pupil–LC relationship. Since fluctuations of these behavioral variables reflect state changes such as impulsivity, motivation, and task engagement (Dickinson and Balleine, 1994; Mayrhofer et al., 2013; Berditchevskaia et al., 2016; Allen et al., 2019; McBurney-Lin et al., 2020), our results suggest that the coupling between pupil and LC is state dependent.

Figure 5

Download asset Open asset

Figure 5—source data 1

Source data for Figure 5.

https://cdn.elifesciences.org/articles/70510/elife-70510-fig5-data1-v1.zip

Download elife-70510-fig5-data1-v1.zip

In the current study, we have shown that pupil diameter has an overall positive and monotonic relationship with LC spiking activity. However, substantial variability exists in this relationship that only the infrequent and large pupil dilation events (>1.5–2 SD amplitude, <10% occurrence) can accurately predict LC spiking on a moment-by-moment basis. In addition, pupil responses exhibit large session-to-session fluctuations to identical optical stimulation in the LC. Decision bias-related behavioral variables could explain the variations in the pupil–LC relationship. Together, our results strongly caution treating pupil dilation as a real-time readout of LC activity. Averaging multiple repeats/trials of similar pupil responses would yield a much more accurate prediction of LC activity.

We used two methods to establish the pupil–LC relationship: detecting LC activity then linking to the following pupil responses (Figure 1); and detecting pupil dilation then linking to the preceding LC activity (Figure 2). Both methods yielded similar pupil–LC relationships with the conclusion that only the infrequent, large pupil responses can accurately predict LC spiking on a moment-by-moment basis. Large pupil or LC responses have been reported to correlate with a variety of task-related processes, including sensory cue, decision formation, positive feedback, choice bias, and action (Rajkowski et al., 1994; Usher et al., 1999; Kalwani et al., 2014; Bouret and Richmond, 2015; de Gee et al., 2017; de Gee et al., 2020; Schriver et al., 2020; Yang et al., 2021). In light of this, our work suggests that the infrequent but strong pupil dilation events can be used as an accurate inference of LC activation in response to sensory stimuli and decision-making processes. However, as discussed below, in general pupil and LC likely respond to task-related processes differently, leading to variations in their relationship.

Recent evidence has uncovered considerable heterogeneity within the LC nucleus, including molecular compositions, physiological properties, released transmitters, and projection targets (Robertson et al., 2013; Chandler et al., 2014; Chandler et al., 2019; Schwarz and Luo, 2015; Schwarz et al., 2015; Kempadoo et al., 2016; Hirschberg et al., 2017; Uematsu et al., 2017; Totah et al., 2018; Borodovitsyna et al., 2020). Our data support these findings (Figure 4—figure supplement 6). Therefore, it is possible that pupil diameter is dynamically coupled with different LC subgroups that are differentially engaged during cognitive processes. However, this is insufficient to explain the session-to-session fluctuations of pupil responses to LC stimulation, since we likely activated a heterogenous group of LC neurons that exhibited similar session-to-session responses to optical stimulation.

The fact that the putatively same neurons tracked across days exhibited similar responses to optical stimulation cannot fully establish the long-term stability of population LC response because slow changes in the tissue due to tetrode/optical fiber implant (gliosis, inflammation, etc.) could alter light transmission to the neurons that were not recorded. However, several lines of evidence in our study did not favor this possibility: (1) Pupil responses in a later session did not systematically or progressively differ from an earlier session (e.g., consistently larger or smaller, Figure 4—figure supplement 2); (2) Significant pupil response variability can be observed in sessions that were a few hours apart (Figure 4); (3) Across-session variability of pupil responses was largely comparable to within-session variability (Figure 4—figure supplement 3). However, optogenetic stimulation tends to synchronize neuronal activity, which may not reflect the physiological condition (Totah et al., 2018). Future experiments with the ability to record from multiple opto-tagged LC neurons simultaneously will further investigate the relationship between pupil diameter and population-level LC activity.

During wakefulness, the state of the brain is constantly fluctuating, both in the presence and absence of external stimuli (Kenet et al., 2003; Fox et al., 2006; Fox and Raichle, 2007; Sakata and Harris, 2009; Luczak et al., 2009; Berkes et al., 2011; Harris and Thiele, 2011; Mohajerani et al., 2013; Romano et al., 2015; Petersen, 2019; McCormick et al., 2020). Pupil response profiles can reflect different behavioral processes (Schriver et al., 2020), and pupil responses also can be dissociated from cognitive processes (Podvalny et al., 2019). Our data extend these observations, supporting that LC and pupil respond to behavioral and cognitive variables differently (Yang et al., 2021).

Fluctuations of hit rate and decision bias reflect state changes such as impulsivity, motivation, and task engagement (Dickinson and Balleine, 1994; Mayrhofer et al., 2013; Berditchevskaia et al., 2016). Although mice licked more during high motivation or high engagement trials (Berditchevskaia et al., 2016; Allen et al., 2019; McBurney-Lin et al., 2020), our data show that licking alone cannot account for the tight correlation between the variations of the behavioral variables and the variations in the pupil–LC relationship (Figure 5), suggesting that pupil–LC coupling is brain state dependent.

How may brain states modulate pupil–LC coupling? Pupil size changes have been linked to activity in other brain areas and neuromodulatory systems, including the medial prefrontal cortex, the inferior colliculus, and cholinergic signaling (Joshi et al., 2016; Reimer et al., 2016; Okun et al., 2019; Kucyi and Parvizi, 2020; Pais-Roldán et al., 2020; Sobczak et al., 2021). A recent study found that pupil responses to dorsal raphe stimulation exhibited task uncertainty-dependent variations (Cazettes et al., 2021). Therefore, it is possible that in high motivation/engagement states, multiple circuits including the LC synergistically influence pupil size changes, resulting in the apparently stronger pupil–LC coupling. Future experiments are needed to elucidate how pupil and LC interact with these brain circuits during different behavioral contexts and cognitive processes.

Another possibility is that higher engagement states may be intimately associated with more ‘uninstructed’ movements as revealed by recent work (Musall et al., 2019), which can drive robust neuronal activity throughout the brain (Musall et al., 2019; Steinmetz et al., 2019; Stringer et al., 2019; Salkoff et al., 2020). Future studies with comprehensive movement monitoring will determine whether more frequent movements, both task-related and task-unrelated, during periods of high motivation/engagement underlie the stronger pupil–LC coupling.

All procedures were performed in accordance with protocols approved by UC Riverside Animal Care and Use Committee. Mice were DBH-Cre (B6.FVB(Cg)-Tg(Dbh-cre) KH212Gsat/Mmucd, 036778-UCD, MMRRC); Ai32 (RCL-ChR2(H134R)/EYFP, 012569, JAX), or DBH-Cre injected with AAV5-EF1α-DIO-hChR2(H134R)-EYFP (UNC Vector Core), singly housed in a vivarium with reverse light–dark cycle (12 hr each phase). Male and female mice of 8–12 weeks were implanted with titanium head posts as described previously (Yang et al., 2016). Procedures for microdrive construction and LC recording have been described previously (Yang et al., 2021). Briefly, custom microdrives with eight tetrodes and an optic fiber (0.39 NA, 200 µm core) were built to make extracellular recordings from LC neurons. Microdrive was implanted in the left LC. LC neurons were identified by optogenetic tagging of DBH+ neurons expressing ChR2, tail pinch response, and post hoc electrolytic lesions (Yang et al., 2021). For Figures 1 and 2, 19 single unit recordings (cluster quality measure L_ratio: 0.01 ± 0.005; firing rate: 1.65 ± 0.25 spikes/s; percent ISI <10 ms: 0.11% ± 0.1%) from 7 mice performing the single-whisker detection task (see below) were extracted using MClust (Redish, 2014), among which six recordings were from our previous dataset (Yang et al., 2021). For Figure 3, 5 units from 5 mice were tracked over time (between 1 and 5 days). For Figure 4, 68 pupil sessions (34 baseline pupil-matched session pairs) to LC stimulation were acquired from 8 mice, 4 of which were implanted with an optical fiber only (0.39 NA, 200 µm core), and the time between sessions 1 and 2 was 4.4 ± 0.9 days. At the conclusion of the experiments, brains were perfused with PBS followed by 4% paraformaldehyde, postfixed overnight, then cut into 100-μm coronal sections and stained with anti-tyrosine hydroxylase antibody (Thermo Fisher OPA1-04050).

Behavior task was controlled by BControl (C. Brody, Princeton University) or custom-based Arduino hardware and software as described previously (Yang et al., 2016; Yang et al., 2021; McBurney-Lin et al., 2020). In brief, mice were trained to perform a head-fixed, Go/NoGo single-whisker detection task, in which mice reported whether they perceived a brief deflection (0.5 s, 40 Hz or 0.2 s, 25 Hz sinusoidal deflection) to the right C2 whisker by licking toward a water port. A 0.1-s auditory cue (8 kHz tone, ~80 dB SPL) was introduced starting 1–1.5 s before stimulus onset. During all sessions, ambient white noise (cutoff at 40 kHz, ~80 dB SPL) was played through a separate speaker to mask any other potential auditory cues associated with movement of the piezo stimulator. Video of the left pupil (ipsilateral to LC recording and stimulation) was acquired at 50 Hz using a PhotonFocus camera and StreamPix 5 software, or at 20 Hz using a Basler acA1300-200 µm camera and Pylon software. 450 nM blue diode lasers (UltraLasers, MDL-III-450–200 mW) controlled by WaveSurfer (https://www.janelia.org/open-science/wavesurfer) were used for optogenetic stimulation. Electrophysiology, pupil tracking, and optogenetic stimulation were synchronized via a common TTL pulse train. The mating sleeve connecting two ferrules was covered with black tape to prevent light leak. An ambient blue LED was used to constrict the pupil and to mask any potential light leak. <15 mW (RMS) of blue light was measured at the tip of optical fiber. We estimated an excitable volume on the order of 0.05–0.1 mm³ for a 30° cylindrical cone based on 10-mW light power, 2.5 mW/mm² excitation threshold and 1.4 refractive index of brain tissue (Boyden et al., 2005; Sun et al., 2012) (brain tissue light transmission calculator: https://web.stanford.edu/group/dlab/cgi-bin/graph/chart.php), containing hundreds of neurons in the LC. Stimulation patterns were delivered every 10–30 s and randomized.

For Figure 1, in each recording if the interval between two adjacent spikes was shorter than median inter-spike interval of that unit, the spikes were grouped into a single cluster. Using other time windows (0.1–0.5 s) to group spikes did not affect this analysis for the majority of recordings (data not shown). Peak pupil dilation was defined as the absolute maximum value in a 6-s window following the onset of each cluster (time of the first spike). ROC analysis in Figure 1f–h was performed between peak pupil diameter associated with clusters of a given size and number-matched, randomly selected pupil diameter. For Figure 2, pupil traces were first smoothed with a 500-ms window to avoid false-positive slope detections. Pupil slopes were then estimated every 200 ms, and a pupil dilation event was defined as the maximum pupil size between sequential positive zero-crossings of the slopes (Joshi et al., 2016). For each dilation event, LC spikes were quantified in a −2 to −4 s window from the event. Using a −1 to −3 s window did not affect this analysis (data not shown). Pupil dilation events falling in a bin of 0.3 SD were considered of similar sizes. ROC analysis in Figure 2e was performed between LC spike counts associated with pupil dilation events of a similar size and LC spike counts associated with number-matched, randomly selected pupil sizes. ROC analysis in Figure 2f was performed the same way as in Figure 1. For Figure 4, pupil responses in each session were first bootstrapped 100 times with replacement to estimate the mean and confidence interval.Pupil responses to the same optical stimulation were pooled from the two different sessions, and then randomly assigned to session 1 or 2 with replacement. The reported p value represented the proportion of iterations where mean peak pupil responses from the two permutated sessions exceeded the observed difference from 1000 iterations. For Figure 5, 9 recordings (out of 13 shown in Figure 1e) from 4 mice during behavior were included with >100 trials and R² > 0.6. For Figure 5b, LC clusters occurring within ±0.5 s from each licking event were excluded from analyzing the pupil–LC relationship as in Figure 1e. This window was chosen based on previous results that LC spiking peaked within a few hundred milliseconds of licking onset (Yang et al., 2021). For Figure 4—figure supplement 3, across-session variability (standard deviation of peak pupil responses) was estimated by resampling trials pooled from all sessions in each condition. The iteration of resampling matched the total number of sessions in that condition. To test whether within-session variability was similar to across-session variability for individual session pairs which exhibited significantly different pupil responses, we first estimated the distribution of across-session variability by resampling trials pooled from both sessions for 1000 iterations and examined whether the variability of individual sessions fell outside 5% of the distribution.

Data were reported as mean ± standard error of the mean unless otherwise noted. We did not use statistical methods to predetermine sample sizes. Sample sizes are similar to those reported in the field. We assigned mice to experimental groups arbitrarily, without randomization or blinding.

Source data for main figures (Figures 1-5, MATLAB R2016b files) are uploaded as 'Source data' files.

1. 1. Allen WE
   2. Chen MZ
   3. Pichamoorthy N
   4. Tien RH
   5. Pachitariu M
   6. Luo L
   7. Deisseroth K

   (2019) Thirst regulates motivated behavior through modulation of brainwide neural population dynamics

   Science 364:1–10.

   https://doi.org/10.1126/science.aav3932

   • Google Scholar
2. 1. Aminihajibashi S
   2. Hagen T
   3. Laeng B
   4. Espeseth T

   (2020) Pupillary and behavioral markers of alerting and orienting: An individual difference approach

   Brain and Cognition 143:105597.

   https://doi.org/10.1016/j.bandc.2020.105597

   • PubMed
   • Google Scholar
3. 1. Aston-Jones G
   2. Bloom FE

   (1981)

   Activity of norepinephrine-containing locus coeruleus neurons in behaving rats anticipates fluctuations in the sleep-waking cycle

   The Journal of Neuroscience 1:876–886.

   • PubMed
   • Google Scholar
4. 1. Aston-Jones G
   2. Rajkowski J
   3. Kubiak P
   4. Alexinsky T

   (1994) Locus coeruleus neurons in monkey are selectively activated by attended cues in a vigilance task

   The Journal of Neuroscience 14:4467–4480.

   https://doi.org/10.1523/JNEUROSCI.14-07-04467.1994

   • PubMed
   • Google Scholar
5. 1. Aston-Jones G
   2. Rajkowski J
   3. Cohen J

   (1999) Role of locus coeruleus in attention and behavioral flexibility

   Biological Psychiatry 46:1309–1320.

   https://doi.org/10.1016/s0006-3223(99)00140-7

   • PubMed
   • Google Scholar
6. 1. Aston-Jones G
   2. Cohen JD

   (2005) An integrative theory of locus coeruleus-norepinephrine function: adaptive gain and optimal performance

   Annual Review of Neuroscience 28:403–450.

   https://doi.org/10.1146/annurev.neuro.28.061604.135709

   • PubMed
   • Google Scholar
7. 1. Barry JM

   (2015) Axonal activity in vivo: technical considerations and implications for the exploration of neural circuits in freely moving animals

   Frontiers in Neuroscience 9:1–12.

   https://doi.org/10.3389/fnins.2015.00153

   • Google Scholar
8. 1. Berditchevskaia A
   2. Cazé RD
   3. Schultz S

   (2016) Performance in a GO/NOGO perceptual task reflects a balance between impulsive and instrumental components of behaviour

   Scientific Reports 6:1–15.

   https://doi.org/10.1038/srep27389

   • Google Scholar
9. 1. Berkes P
   2. Orbán G
   3. Lengyel M
   4. Fiser J

   (2011) Spontaneous cortical activity reveals hallmarks of an optimal internal model of the environment

   Science 331:83–87.

   https://doi.org/10.1126/science.1195870

   • PubMed
   • Google Scholar
10. 1. Berridge CW
    2. Foote SL

    (1991) Effects of locus coeruleus activation on electroencephalographic activity in neocortex and hippocampus

    The Journal of Neuroscience 11:3135–3145.

    https://doi.org/10.1523/JNEUROSCI.11-10-03135.1991

    • Google Scholar
11. 1. Berridge CW
    2. Waterhouse BD

    (2003) The locus coeruleus–noradrenergic system: modulation of behavioral state and state-dependent cognitive processes

    Brain Research Reviews 42:33–84.

    https://doi.org/10.1016/S0165-0173(03)00143-7

    • Google Scholar
12. 1. Bijleveld E
    2. Custers R
    3. Aarts H

    (2009) The Unconscious Eye Opener

    Psychological Science 20:1313–1315.

    https://doi.org/10.1111/j.1467-9280.2009.02443.x

    • PubMed
    • Google Scholar
13. 1. Borodovitsyna O
    2. Duffy BC
    3. Pickering AE
    4. Chandler DJ

    (2020) Anatomically and functionally distinct locus coeruleus efferents mediate opposing effects on anxiety-like behavior

    Neurobiology of Stress 13:100284.

    https://doi.org/10.1016/j.ynstr.2020.100284

    • Google Scholar
14. 1. Bouret S
    2. Richmond BJ

    (2015) Sensitivity of Locus Ceruleus Neurons to Reward Value for Goal-Directed Actions

    The Journal of Neuroscience 35:4005–4014.

    https://doi.org/10.1523/JNEUROSCI.4553-14.2015

    • Google Scholar
15. 1. Boyden ES
    2. Zhang F
    3. Bamberg E
    4. Nagel G
    5. Deisseroth K

    (2005) Millisecond-timescale, genetically targeted optical control of neural activity

    Nature Neuroscience 8:1263–1268.

    https://doi.org/10.1038/nn1525

    • PubMed
    • Google Scholar
16. 1. Breton-Provencher V
    2. Sur M

    (2019) Active control of arousal by a locus coeruleus GABAergic circuit

    Nature Neuroscience 22:218–228.

    https://doi.org/10.1038/s41593-018-0305-z

    • PubMed
    • Google Scholar
17. 1. Cano M
    2. Bezdudnaya T
    3. Swadlow HA
    4. Alonso JM

    (2006) Brain state and contrast sensitivity in the awake visual thalamus

    Nature Neuroscience 9:1240–1242.

    https://doi.org/10.1038/nn1760

    • PubMed
    • Google Scholar
18. 1. Carter ME
    2. Yizhar O
    3. Chikahisa S
    4. Nguyen H
    5. Adamantidis A
    6. Nishino S
    7. Deisseroth K
    8. de Lecea L

    (2010) Tuning arousal with optogenetic modulation of locus coeruleus neurons

    Nature Neuroscience 13:1526–1533.

    https://doi.org/10.1038/nn.2682

    • PubMed
    • Google Scholar
19. 1. Cazettes F
    2. Reato D
    3. Morais JP
    4. Renart A
    5. Mainen ZF

    (2021) Phasic Activation of Dorsal Raphe Serotonergic Neurons Increases Pupil Size

    Current Biology 31:192–197.

    https://doi.org/10.1016/j.cub.2020.09.090

    • PubMed
    • Google Scholar
20. 1. Chandler DJ
    2. Gao WJ
    3. Waterhouse BD

    (2014) Heterogeneous organization of the locus coeruleus projections to prefrontal and motor cortices

    PNAS 111:6816–6821.

    https://doi.org/10.1073/pnas.1320827111

    • PubMed
    • Google Scholar
21. 1. Chandler DJ
    2. Jensen P
    3. McCall JG
    4. Pickering AE
    5. Schwarz LA
    6. Totah NK

    (2019) Redefining Noradrenergic Neuromodulation of Behavior: Impacts of a Modular Locus Coeruleus Architecture

    The Journal of Neuroscience 39:8239–8249.

    https://doi.org/10.1523/JNEUROSCI.1164-19.2019

    • PubMed
    • Google Scholar
22. 1. Clewett D
    2. Gasser C
    3. Davachi L

    (2020) Pupil-linked arousal signals track the temporal organization of events in memory

    Nature Communications 11:1–14.

    https://doi.org/10.1038/s41467-020-17851-9

    • Google Scholar
23. 1. Cohen JY
    2. Haesler S
    3. Vong L
    4. Lowell BB
    5. Uchida N

    (2012) Neuron-type-specific signals for reward and punishment in the ventral tegmental area

    Nature 482:85–88.

    https://doi.org/10.1038/nature10754

    • PubMed
    • Google Scholar
24. 1. de Gee JW
    2. Knapen T
    3. Donner TH

    (2014) Decision-related pupil dilation reflects upcoming choice and individual bias

    PNAS 111:E618–E625.

    https://doi.org/10.1073/pnas.1317557111

    • Google Scholar
25. 1. de Gee JW
    2. Colizoli O
    3. Kloosterman NA
    4. Knapen T
    5. Nieuwenhuis S
    6. Donner TH

    (2017) Dynamic modulation of decision biases by brainstem arousal systems

    eLife 6:e23232.

    https://doi.org/10.7554/eLife.23232

    • PubMed
    • Google Scholar
26. 1. de Gee JW
    2. Tsetsos K
    3. Schwabe L
    4. Urai AE
    5. McCormick D
    6. McGinley MJ
    7. Donner TH

    (2020) Pupil-linked phasic arousal predicts a reduction of choice bias across species and decision domains

    eLife 9:e54014.

    https://doi.org/10.7554/eLife.54014

    • PubMed
    • Google Scholar
27. 1. Dickinson A
    2. Balleine B

    (1994) Motivational control of goal-directed action

    Animal Learning & Behavior 22:1–18.

    https://doi.org/10.3758/BF03199951

    • Google Scholar
28. 1. Ebitz RB
    2. Moore T

    (2019) Both a Gauge and a Filter: Cognitive Modulations of Pupil Size

    Frontiers in Neurology 9:1–14.

    https://doi.org/10.3389/fneur.2018.01190

    • Google Scholar
29. 1. Eschenko O
    2. Magri C
    3. Panzeri S
    4. Sara SJ

    (2012) Noradrenergic neurons of the locus coeruleus are phase locked to cortical up-down states during sleep

    Cerebral Cortex 22:426–435.

    https://doi.org/10.1093/cercor/bhr121

    • PubMed
    • Google Scholar
30. 1. Fazlali Z
    2. Ranjbar-Slamloo Y
    3. Adibi M
    4. Arabzadeh E

    (2016) Correlation between Cortical State and Locus Coeruleus Activity: Implications for Sensory Coding in Rat Barrel Cortex

    Frontiers in Neural Circuits 10:1–16.

    https://doi.org/10.3389/fncir.2016.00014

    • PubMed
    • Google Scholar
31. 1. Foote SL
    2. Aston-Jones G
    3. Bloom FE

    (1980) Impulse activity of locus coeruleus neurons in awake rats and monkeys is a function of sensory stimulation and arousal

    PNAS 77:3033–3037.

    https://doi.org/10.1073/pnas.77.5.3033

    • PubMed
    • Google Scholar
32. 1. Fox MD
    2. Snyder AZ
    3. Zacks JM
    4. Raichle ME

    (2006) Coherent spontaneous activity accounts for trial-to-trial variability in human evoked brain responses

    Nature Neuroscience 9:23–25.

    https://doi.org/10.1038/nn1616

    • PubMed
    • Google Scholar
33. 1. Fox MD
    2. Raichle ME

    (2007) Spontaneous fluctuations in brain activity observed with functional magnetic resonance imaging

    Nature Reviews. Neuroscience 8:700–711.

    https://doi.org/10.1038/nrn2201

    • PubMed
    • Google Scholar
34. 1. Fu Y
    2. Tucciarone JM
    3. Espinosa JS
    4. Sheng N
    5. Darcy DP
    6. Nicoll RA
    7. Huang ZJ
    8. Stryker MP

    (2014) A cortical circuit for gain control by behavioral state

    Cell 156:1139–1152.

    https://doi.org/10.1016/j.cell.2014.01.050

    • PubMed
    • Google Scholar
35. 1. Gelbard-Sagiv H
    2. Magidov E
    3. Sharon H
    4. Hendler T
    5. Nir Y

    (2018) Noradrenaline Modulates Visual Perception and Late Visually Evoked Activity

    Current Biology 28:2239–2249.

    https://doi.org/10.1016/j.cub.2018.05.051

    • PubMed
    • Google Scholar
36. 1. Gilzenrat MS
    2. Nieuwenhuis S
    3. Jepma M
    4. Cohen JD

    (2010) Pupil diameter tracks changes in control state predicted by the adaptive gain theory of locus coeruleus function

    Cognitive, Affective & Behavioral Neuroscience 10:252–269.

    https://doi.org/10.3758/CABN.10.2.252

    • PubMed
    • Google Scholar
37. 1. Gold C
    2. Henze DA
    3. Koch C
    4. Buzsáki G

    (2006) On the origin of the extracellular action potential waveform: A modeling study

    Journal of Neurophysiology 95:3113–3128.

    https://doi.org/10.1152/jn.00979.2005

    • PubMed
    • Google Scholar
38. Book

    1. Green DM
    2. Swets JA

    (1966)

    Signal Detection Theory and Psychophysics

    John Wiley and Sons.

    • Google Scholar
39. 1. Hahn G
    2. Petermann T
    3. Havenith MN
    4. Yu S
    5. Singer W
    6. Plenz D
    7. Nikolic D

    (2010) Neuronal avalanches in spontaneous activity in vivo

    Journal of Neurophysiology 104:3312–3322.

    https://doi.org/10.1152/jn.00953.2009

    • PubMed
    • Google Scholar
40. 1. Harris KD
    2. Thiele A

    (2011) Cortical state and attention

    Nature Reviews. Neuroscience 12:509–523.

    https://doi.org/10.1038/nrn3084

    • PubMed
    • Google Scholar
41. 1. Hayat H
    2. Regev N
    3. Matosevich N
    4. Sales A
    5. Paredes-Rodriguez E
    6. Krom AJ
    7. Bergman L
    8. Li Y
    9. Lavigne M
    10. Kremer EJ
    11. Yizhar O
    12. Pickering AE
    13. Nir Y

    (2020) Locus coeruleus norepinephrine activity mediates sensory-evoked awakenings from sleep

    Science Advances 6:eaaz4232.

    https://doi.org/10.1126/sciadv.aaz4232

    • PubMed
    • Google Scholar
42. 1. Hirschberg S
    2. Li Y
    3. Randall A
    4. Kremer EJ
    5. Pickering AE

    (2017) Functional dichotomy in spinal-vs prefrontal-projecting locus coeruleus modules splits descending noradrenergic analgesia from ascending aversion and anxiety in rats

    eLife 6:e29808.

    https://doi.org/10.7554/eLife.29808

    • Google Scholar
43. 1. Joshi S
    2. Li Y
    3. Kalwani RM
    4. Gold JI

    (2016) Relationships between Pupil Diameter and Neuronal Activity in the Locus Coeruleus, Colliculi, and Cingulate Cortex

    Neuron 89:221–234.

    https://doi.org/10.1016/j.neuron.2015.11.028

    • PubMed
    • Google Scholar
44. 1. Joshi S
    2. Gold JI

    (2020) Pupil Size as a Window on Neural Substrates of Cognition

    Trends in Cognitive Sciences 24:466–480.

    https://doi.org/10.1016/j.tics.2020.03.005

    • PubMed
    • Google Scholar
45. 1. Kahneman D
    2. Beatty J

    (1966) Pupil diameter and load on memory

    Science 154:1583–1585.

    https://doi.org/10.1126/science.154.3756.1583

    • PubMed
    • Google Scholar
46. 1. Kalwani RM
    2. Joshi S
    3. Gold JI

    (2014) Phasic Activation of Individual Neurons in the Locus Ceruleus/Subceruleus Complex of Monkeys Reflects Rewarded Decisions to Go But Not Stop

    The Journal of Neuroscience 34:13656–13669.

    https://doi.org/10.1523/JNEUROSCI.2566-14.2014

    • Google Scholar
47. 1. Kempadoo KA
    2. Mosharov EV
    3. Choi SJ
    4. Sulzer D
    5. Kandel ER

    (2016) Dopamine release from the locus coeruleus to the dorsal hippocampus promotes spatial learning and memory

    PNAS 113:14835–14840.

    https://doi.org/10.1073/pnas.1616515114

    • PubMed
    • Google Scholar
48. 1. Kenet T
    2. Bibitchkov D
    3. Tsodyks M
    4. Grinvald A
    5. Arieli A

    (2003) Spontaneously emerging cortical representations of visual attributes

    Nature 425:954–956.

    https://doi.org/10.1038/nature02078

    • PubMed
    • Google Scholar
49. 1. Konishi M
    2. Brown K
    3. Battaglini L
    4. Smallwood J

    (2017) When attention wanders: Pupillometric signatures of fluctuations in external attention

    Cognition 168:16–26.

    https://doi.org/10.1016/j.cognition.2017.06.006

    • PubMed
    • Google Scholar
50. 1. Kucewicz MT
    2. Dolezal J
    3. Kremen V
    4. Berry BM
    5. Miller LR
    6. Magee AL
    7. Fabian V
    8. Worrell GA

    (2018) Pupil size reflects successful encoding and recall of memory in humans

    Scientific Reports 8:1–7.

    https://doi.org/10.1038/s41598-018-23197-6

    • Google Scholar
51. 1. Kucyi A
    2. Parvizi J

    (2020) Pupillary dynamics link spontaneous and task-evoked activations recorded directly from human insula

    The Journal of Neuroscience 40:2020.

    https://doi.org/10.1523/JNEUROSCI.0435-20.2020

    • Google Scholar
52. 1. Lee SH
    2. Dan Y

    (2012) Neuromodulation of Brain States

    Neuron 76:209–222.

    https://doi.org/10.1016/j.neuron.2012.09.012

    • PubMed
    • Google Scholar
53. 1. Lee CR
    2. Margolis DJ

    (2016) Pupil Dynamics Reflect Behavioral Choice and Learning in a Go/NoGo Tactile Decision-Making Task in Mice

    Frontiers in Behavioral Neuroscience 10:1–14.

    https://doi.org/10.3389/fnbeh.2016.00200/full

    • Google Scholar
54. 1. Lee CCY
    2. Kheradpezhouh E
    3. Diamond ME
    4. Arabzadeh E

    (2020) State-Dependent Changes in Perception and Coding in the Mouse Somatosensory Cortex

    Cell Reports 32:108197.

    https://doi.org/10.1016/j.celrep.2020.108197

    • Google Scholar
55. 1. Liu Y
    2. Rodenkirch C
    3. Moskowitz N
    4. Schriver B
    5. Wang Q

    (2017) Dynamic Lateralization of Pupil Dilation Evoked by Locus Coeruleus Activation Results from Sympathetic, Not Parasympathetic, Contributions

    Cell Reports 20:3099–3112.

    https://doi.org/10.1016/j.celrep.2017.08.094

    • PubMed
    • Google Scholar
56. 1. Luczak A
    2. Barthó P
    3. Harris KD

    (2009) Spontaneous events outline the realm of possible sensory responses in neocortical populations

    Neuron 62:413–425.

    https://doi.org/10.1016/j.neuron.2009.03.014

    • PubMed
    • Google Scholar
57. 1. Mayrhofer JM
    2. Skreb V
    3. von der Behrens W
    4. Musall S
    5. Weber B
    6. Haiss F

    (2013) Novel two-alternative forced choice paradigm for bilateral vibrotactile whisker frequency discrimination in head-fixed mice and rats

    Journal of Neurophysiology 109:273–284.

    https://doi.org/10.1152/jn.00488.2012

    • PubMed
    • Google Scholar
58. 1. McBurney-Lin J
    2. Sun Y
    3. Tortorelli LS
    4. Nguyen QAT
    5. Haga-Yamanaka S
    6. Yang H

    (2020) Bidirectional pharmacological perturbations of the noradrenergic system differentially affect tactile detection

    Neuropharmacology 174:108151.

    https://doi.org/10.1016/j.neuropharm.2020.108151

    • PubMed
    • Google Scholar
59. 1. McCormick DA
    2. Nestvogel DB
    3. He BJ

    (2020) Neuromodulation of Brain State and Behavior

    Annual Review of Neuroscience 43:391–415.

    https://doi.org/10.1146/annurev-neuro-100219-105424

    • PubMed
    • Google Scholar
60. 1. McGinley MJ
    2. Vinck M
    3. Reimer J
    4. Batista-Brito R
    5. Zagha E
    6. Cadwell CR
    7. Tolias AS
    8. Cardin JA
    9. McCormick DA

    (2015a) Waking State: Rapid Variations Modulate Neural and Behavioral Responses

    Neuron 87:1143–1161.

    https://doi.org/10.1016/j.neuron.2015.09.012

    • Google Scholar
61. 1. McGinley MJ
    2. David SV
    3. McCormick DA

    (2015b) Cortical Membrane Potential Signature of Optimal States for Sensory Signal Detection

    Neuron 87:179–192.

    https://doi.org/10.1016/j.neuron.2015.05.038

    • Google Scholar
62. 1. Mohajerani MH
    2. Chan AW
    3. Mohsenvand M
    4. LeDue J
    5. Liu R
    6. McVea DA
    7. Boyd JD
    8. Wang YT
    9. Reimers M
    10. Murphy TH

    (2013) Spontaneous cortical activity alternates between motifs defined by regional axonal projections

    Nature Neuroscience 16:1426–1435.

    https://doi.org/10.1038/nn.3499

    • PubMed
    • Google Scholar
63. 1. Murphy PR
    2. O’Connell RG
    3. Sullivan M IH
    4. Robertson IH
    5. Balsters JH

    (2014) Pupil diameter covaries with BOLD activity in human locus coeruleus

    Human Brain Mapping 35:4140–4154.

    https://doi.org/10.1002/hbm.22466

    • PubMed
    • Google Scholar
64. 1. Musall S
    2. Kaufman MT
    3. Juavinett AL
    4. Gluf S
    5. Churchland AK

    (2019) Single-trial neural dynamics are dominated by richly varied movements

    Nature Neuroscience 22:1677–1686.

    https://doi.org/10.1038/s41593-019-0502-4

    • PubMed
    • Google Scholar
65. 1. Nassar MR
    2. Rumsey KM
    3. Wilson RC
    4. Parikh K
    5. Heasly B
    6. Gold JI

    (2012) Rational regulation of learning dynamics by pupil-linked arousal systems

    Nature Neuroscience 15:1040–1046.

    https://doi.org/10.1038/nn.3130

    • PubMed
    • Google Scholar
66. 1. Okun M
    2. Steinmetz NA
    3. Lak A
    4. Dervinis M
    5. Harris KD

    (2019) Distinct structure of cortical population activity on fast and infraslow timescales

    Cerebral Cortex 29:2196–2210.

    https://doi.org/10.1093/cercor/bhz023

    • Google Scholar
67. 1. Pais-Roldán P
    2. Takahashi K
    3. Sobczak F
    4. Chen Y
    5. Zhao X
    6. Zeng H
    7. Jiang Y
    8. Yu X

    (2020) Indexing brain state-dependent pupil dynamics with simultaneous fMRI and optical fiber calcium recording

    PNAS 117:6875–6882.

    https://doi.org/10.1073/pnas.1909937117

    • Google Scholar
68. 1. Petersen CCH

    (2019) Sensorimotor processing in the rodent barrel cortex

    Nature Reviews Neuroscience 20:533–546.

    https://doi.org/10.1038/s41583-019-0200-y

    • Google Scholar
69. 1. Podvalny E
    2. Flounders MW
    3. King LE
    4. Holroyd T
    5. He BJ

    (2019) A dual role of prestimulus spontaneous neural activity in visual object recognition

    Nature Communications 10:1–13.

    https://doi.org/10.1038/s41467-019-11877-4

    • Google Scholar
70. 1. Polack PO
    2. Friedman J
    3. Golshani P

    (2013) Cellular mechanisms of brain state-dependent gain modulation in visual cortex

    Nature Neuroscience 16:1331–1339.

    https://doi.org/10.1038/nn.3464

    • PubMed
    • Google Scholar
71. 1. Poulet JFA
    2. Petersen CCH

    (2008) Internal brain state regulates membrane potential synchrony in barrel cortex of behaving mice

    Nature 454:881–885.

    https://doi.org/10.1038/nature07150

    • PubMed
    • Google Scholar
72. 1. Preuschoff K
    2. ’t Hart BM
    3. Einhäuser W

    (2011) Pupil dilation signals surprise: evidence for noradrenaline’s role in decision making

    Frontiers in Neuroscience 5:1–12.

    https://doi.org/10.3389/fnins.2011.00115

    • Google Scholar
73. 1. Privitera M
    2. Ferrari KD
    3. von Ziegler LM
    4. Sturman O
    5. Duss SN
    6. Floriou-Servou A
    7. Germain P-L
    8. Vermeiren Y
    9. Wyss MT
    10. De Deyn PP
    11. Weber B
    12. Bohacek J

    (2020) A complete pupillometry toolbox for real-time monitoring of locus coeruleus activity in rodents

    Nature Protocols 15:2301–2320.

    https://doi.org/10.1038/s41596-020-0324-6

    • Google Scholar
74. 1. Rajkowski J
    2. Kubiak P
    3. Aston-Jones G

    (1994) Locus coeruleus activity in monkey: phasic and tonic changes are associated with altered vigilance

    Brain Research Bulletin 35:607–616.

    https://doi.org/10.1016/0361-9230(94)90175-9

    • PubMed
    • Google Scholar
75. 1. Rall W
    2. Shepherd GM

    (1968) Theoretical reconstruction of field potentials and dendrodendritic synaptic interactions in olfactory bulb

    Journal of Neurophysiology 31:884–915.

    https://doi.org/10.1152/jn.1968.31.6.884

    • PubMed
    • Google Scholar
76. Software

    1. Redish AD

    (2014) MClust Spike sorting toolbox Documentation for version 4.4, version 4.4

    Redishlab.

    http://redishlab.neuroscience.umn.edu/MClust/MClust-4.4.pdf
77. 1. Reimer J
    2. Froudarakis E
    3. Cadwell CR
    4. Yatsenko D
    5. Denfield GH
    6. Tolias AS

    (2014) Pupil Fluctuations Track Fast Switching of Cortical States during Quiet Wakefulness

    Neuron 84:355–362.

    https://doi.org/10.1016/j.neuron.2014.09.033

    • PubMed
    • Google Scholar
78. 1. Reimer J
    2. McGinley MJ
    3. Liu Y
    4. Rodenkirch C
    5. Wang Q
    6. McCormick DA
    7. Tolias AS

    (2016) Pupil fluctuations track rapid changes in adrenergic and cholinergic activity in cortex

    Nature Communications 7:13289.

    https://doi.org/10.1038/ncomms13289

    • Google Scholar
79. 1. Robertson SD
    2. Plummer NW
    3. de Marchena J
    4. Jensen P

    (2013) Developmental origins of central norepinephrine neuron diversity

    Nature Neuroscience 16:1016–1023.

    https://doi.org/10.1038/nn.3458

    • Google Scholar
80. 1. Romano SAA
    2. Pietri T
    3. Pérez-Schuster V
    4. Jouary A
    5. Haudrechy M
    6. Sumbre G

    (2015) Spontaneous Neuronal Network Dynamics Reveal Circuit’s Functional Adaptations for Behavior

    Neuron 85:1070–1085.

    https://doi.org/10.1016/j.neuron.2015.01.027

    • PubMed
    • Google Scholar
81. 1. Sakata S
    2. Harris KD

    (2009) Laminar structure of spontaneous and sensory-evoked population activity in auditory cortex

    Neuron 64:404–418.

    https://doi.org/10.1016/j.neuron.2009.09.020

    • PubMed
    • Google Scholar
82. 1. Salkoff DB
    2. Zagha E
    3. McCarthy E
    4. McCormick DA

    (2020) Movement and Performance Explain Widespread Cortical Activity in a Visual Detection Task

    Cerebral Cortex 30:421–437.

    https://doi.org/10.1093/cercor/bhz206

    • PubMed
    • Google Scholar
83. 1. Sara SJ

    (2009) The locus coeruleus and noradrenergic modulation of cognition

    Nature Reviews. Neuroscience 10:211–223.

    https://doi.org/10.1038/nrn2573

    • PubMed
    • Google Scholar
84. 1. Sara SJ
    2. Bouret S

    (2012) Orienting and reorienting: the locus coeruleus mediates cognition through arousal

    Neuron 76:130–141.

    https://doi.org/10.1016/j.neuron.2012.09.011

    • PubMed
    • Google Scholar
85. 1. Schriver BJ
    2. Bagdasarov S
    3. Wang Q

    (2018) Pupil-linked arousal modulates behavior in rats performing a whisker deflection direction discrimination task

    Journal of Neurophysiology 120:1655–1670.

    https://doi.org/10.1152/jn.00290.2018

    • PubMed
    • Google Scholar
86. 1. Schriver BJ
    2. Perkins SM
    3. Sajda P
    4. Wang Q

    (2020) Interplay between components of pupil‐linked phasic arousal and its role in driving behavioral choice in Go/No‐Go perceptual decision‐making

    Psychophysiology 57:1–20.

    https://doi.org/10.1111/psyp.13565

    • Google Scholar
87. 1. Schwarz LA
    2. Luo L

    (2015) Organization of the Locus Coeruleus-Norepinephrine System

    Current Biology 25:R1051–R1056.

    https://doi.org/10.1016/j.cub.2015.09.039

    • Google Scholar
88. 1. Schwarz LA
    2. Miyamichi K
    3. Gao XJ
    4. Beier KT
    5. Weissbourd B
    6. DeLoach KE
    7. Ren J
    8. Ibanes S
    9. Malenka RC
    10. Kremer EJ
    11. Luo L

    (2015) Viral-genetic tracing of the input–output organization of a central noradrenaline circuit

    Nature 524:88–92.

    https://doi.org/10.1038/nature14600

    • Google Scholar
89. 1. Simpson AJ
    2. Fitter MJ

    (1973) What is the best index of detectability?

    Psychological Bulletin 80:481–488.

    https://doi.org/10.1037/h0035203

    • Google Scholar
90. 1. Sobczak F
    2. Pais-Roldán P
    3. Takahashi K
    4. Yu X

    (2021) Decoding the brain state-dependent relationship between pupil dynamics and resting state fMRI signal fluctuation

    eLife 10:e68980.

    https://doi.org/10.7554/eLife.68980

    • PubMed
    • Google Scholar
91. 1. Steinmetz NA
    2. Zatka-Haas P
    3. Carandini M
    4. Harris KD

    (2019) Distributed coding of choice, action and engagement across the mouse brain

    Nature 576:266–273.

    https://doi.org/10.1038/s41586-019-1787-x

    • Google Scholar
92. 1. Stringer C
    2. Pachitariu M
    3. Steinmetz N
    4. Reddy CB
    5. Carandini M
    6. Harris KD

    (2019) Spontaneous behaviors drive multidimensional, brainwide activity

    Science 364:255.

    https://doi.org/10.1126/science.aav7893

    • PubMed
    • Google Scholar
93. 1. Sun J
    2. Lee SJ
    3. Wu L
    4. Sarntinoranont M
    5. Xie H

    (2012) Refractive index measurement of acute rat brain tissue slices using optical coherence tomography

    Optics Express 20:1084–1095.

    https://doi.org/10.1364/OE.20.001084

    • PubMed
    • Google Scholar
94. 1. Sun SH
    2. Almasi A
    3. Yunzab M
    4. Zehra S
    5. Hicks DG
    6. Kameneva T
    7. Ibbotson MR
    8. Meffin H

    (2021) Analysis of extracellular spike waveforms and associated receptive fields of neurons in cat primary visual cortex

    The Journal of Physiology 599:2211–2238.

    https://doi.org/10.1113/JP280844

    • PubMed
    • Google Scholar
95. 1. Swift KM
    2. Gross BA
    3. Frazer MA
    4. Bauer DS
    5. Clark KJD
    6. Vazey EM
    7. Aston-Jones G
    8. Li Y
    9. Pickering AE
    10. Sara SJ
    11. Poe GR

    (2018) Abnormal Locus Coeruleus Sleep Activity Alters Sleep Signatures of Memory Consolidation and Impairs Place Cell Stability and Spatial Memory

    Current Biology 28:3599–3609.

    https://doi.org/10.1016/j.cub.2018.09.054

    • PubMed
    • Google Scholar
96. 1. Takahashi K
    2. Kayama Y
    3. Lin JS
    4. Sakai K

    (2010) Locus coeruleus neuronal activity during the sleep-waking cycle in mice

    Neuroscience 169:1115–1126.

    https://doi.org/10.1016/j.neuroscience.2010.06.009

    • PubMed
    • Google Scholar
97. 1. Thiele A
    2. Bellgrove MA

    (2018) Neuromodulation of Attention

    Neuron 97:769–785.

    https://doi.org/10.1016/j.neuron.2018.01.008

    • PubMed
    • Google Scholar
98. 1. Totah NK
    2. Neves RM
    3. Panzeri S
    4. Logothetis NK
    5. Eschenko O

    (2018) The Locus Coeruleus Is a Complex and Differentiated Neuromodulatory System

    Neuron 99:1055–1068.

    https://doi.org/10.1016/j.neuron.2018.07.037

    • PubMed
    • Google Scholar
99. 1. Uematsu A
    2. Tan BZ
    3. Ycu EA
    4. Cuevas JS
    5. Koivumaa J
    6. Junyent F
    7. Kremer EJ
    8. Witten IB
    9. Deisseroth K
    10. Johansen JP

    (2017) Modular organization of the brainstem noradrenaline system coordinates opposing learning states

    Nature Neuroscience 20:1602–1611.

    https://doi.org/10.1038/nn.4642

    • PubMed
    • Google Scholar
100. 1. Usher M
     2. Cohen JD
     3. Servan-Schreiber D
     4. Rajkowski J
     5. Aston-Jones G

     (1999) The Role of Locus Coeruleus in the Regulation of Cognitive Performance

     Science 283:549–554.

     https://doi.org/10.1126/science.283.5401.549

     • Google Scholar
101. 1. Varazzani C
     2. San-Galli A
     3. Gilardeau S
     4. Bouret S

     (2015) Noradrenaline and dopamine neurons in the reward/effort trade-off: a direct electrophysiological comparison in behaving monkeys

     The Journal of Neuroscience 35:7866–7877.

     https://doi.org/10.1523/JNEUROSCI.0454-15.2015

     • PubMed
     • Google Scholar
102. 1. Vazey EM
     2. Aston-Jones G

     (2014) Designer receptor manipulations reveal a role of the locus coeruleus noradrenergic system in isoflurane general anesthesia

     PNAS 111:3859–3864.

     https://doi.org/10.1073/pnas.1310025111

     • PubMed
     • Google Scholar
103. 1. Vinck M
     2. Batista-Brito R
     3. Knoblich U
     4. Cardin JAA

     (2015) Arousal and locomotion make distinct contributions to cortical activity patterns and visual encoding

     Neuron 86:740–754.

     https://doi.org/10.1016/j.neuron.2015.03.028

     • PubMed
     • Google Scholar
104. 1. Yang H
     2. Kwon SE
     3. Severson KS
     4. O’Connor DH

     (2016) Origins of choice-related activity in mouse somatosensory cortex

     Nature Neuroscience 19:127–134.

     https://doi.org/10.1038/nn.4183

     • PubMed
     • Google Scholar
105. 1. Yang H
     2. Bari BA
     3. Cohen JY
     4. O’Connor DH

     (2021) Locus coeruleus spiking differently correlates with S1 cortex activity and pupil diameter in a tactile detection task

     eLife 10:e64327.

     https://doi.org/10.7554/eLife.64327

     • PubMed
     • Google Scholar
106. 1. Yu AJ
     2. Dayan P

     (2005) Uncertainty, Neuromodulation, and Attention

     Neuron 46:681–692.

     https://doi.org/10.1016/j.neuron.2005.04.026

     • Google Scholar
107. 1. Yu S
     2. Ribeiro TL
     3. Meisel C
     4. Chou S
     5. Mitz A
     6. Saunders R
     7. Plenz D

     (2017) Maintained avalanche dynamics during task-induced changes of neuronal activity in nonhuman primates

     eLife 6:e27119.

     https://doi.org/10.7554/eLife.27119

     • PubMed
     • Google Scholar
108. 1. Zagha E
     2. McCormick DA

     (2014) Neural control of brain state

     Current Opinion in Neurobiology 29:178–186.

     https://doi.org/10.1016/j.conb.2014.09.010

     • PubMed
     • Google Scholar
109. 1. Zhao S
     2. Chait M
     3. Dick F
     4. Dayan P
     5. Furukawa S
     6. Liao HI

     (2019) Pupil-linked phasic arousal evoked by violation but not emergence of regularity within rapid sound sequences

     Nature Communications 10:1–16.

     https://doi.org/10.1038/s41467-019-12048-1

     • Google Scholar

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting the paper "A graded yet variable relationship between pupil diameter and locus coeruleus activity" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The reviewers have opted to remain anonymous.

We are sorry to say that, after extensive consultation with the reviewers, we have decided that this work will not be considered further for publication by eLife.

Although all reviewers agreed that your in vivo spiking data from the LC is impressive and quite valuable, and the attempt to clarify the relationship with pupil measurements is impactful, there were major concerns about several aspects of the paper that preclude publication in eLife. Reviewers thought that several conclusions were not strongly supported by the data, including the suggestion that there were two distinct classes of LC neurons and more importantly that variability across cells reflected true variability in the LC-pupil relationship rather than uncontrolled experimental factors. For the optogenetic experiments, the reviewers were not convinced by the limited evidence that the same units were recorded chronically across days, opening up the possibility that variability in the population of neurons recruited by the optogenetic stimulation might be the source of variability in pupil changes. In addition, several key features of the study were not well described, especially the spike clustering method, the anesthesia, and the parameters of the behavioral task.

Reviewer #1:

Megemont, McBurney-Lin, and Yang simultaneously recorded pupil diameter and spiking from brainstem noradrenergic locus coeruleus (LC) neurons in awake and anesthetized mice in order to further probe the relationship between LC activity and pupil diameter. Numerous variables are known to correlate with pupil diameter, including behavioral variables (e.g., locomotion, behavioral task-related cognitive processes) and neuronal activity patterns (e.g., cholingergic neurons, inferior colliculus, anterior cingulate cortex). Here, the authors aimed to determine how accurately LC activity can be predicted from pupil diameter, which is a critical question given the relationship between pupil diameter and numerous other variables. To those ends, the authors convincingly show that there is a strong correlation (R^2>0.6) between number of spikes and pupil diameter in 68% of the units (13 out of 19). The authors propose, furthermore, that the activity of only a small proportion of LC units are related to pupil diameter; that specific sub-types of LC neurons have diverse coupling with pupil diameter; that pupil diameter cannot be used to predict the majority of LC spiking activity; and that pupil diameter is not related to LC population firing patterns. However, these latter claims are not supported by the data.

Below is a discussion of the data and authors' claims based on these data, methodological stengths and weaknesses that affect the authors' interpreation of the data, and how the findings relate to prior work:

(1) The authors claim that there is a monotonic relationship between pupil diameter and LC single unit activity. The authors show that is a strong correlation (R^2>0.6) between number of spikes and pupil diameter in 68% of the units (13 out of 19 single units). The data in Figure 1A-E support this claim. However, it is necessary to further show the inter-spike interval distributions and firing rates to demonstrate that the recorded spiking was isolated from single units. These data are qualitatively similar to the prior work cited by the authors reporting trial-by-trial correlations during a behavioral task (Joshi, et al. 2016 and Varazzani, et al. 2015). The authors extend these prior findings of LC-pupil coupling by characterizing the relationship between pupil and LC activity also in the absence of an ongoing behavioral task. Interestingly, while prior work during the behavioral task (Joshi et al., 2016) reported observing a much smaller proportion of LC single units that were significantly correlated with pupil diameter (35% in monkey Oz and 20% in monkey Ci), the task-free LC activity presented here shows the majority of LC units (68%) were coupled with pupil diameter. This is an important finding because it characterizes how the relationship between LC activity and pupil diameter changes whether assessed using the trial structure of a behavioral task versus spontaneous, task-free conditions. Given that the authors found 68% of units coupled to pupil diameter, the authors' proposal that the activity of only a small proportion of LC units are related to pupil diameter is not justified given that the majority were indeed coupled with pupil diameter.

(2) The authors claim that specific sub-types of LC neurons have diverse coupling with pupil diameter. This was substantiated by data showing that there were two subgroups of LC units with different LC firing rates and different coupling to pupil diameter (Figure S2). Here, it would be useful to know how the division was made between subgroups of units. Also, raster examples would help show how a unit can emit a brief cluster of 2 spikes or a brief cluster of 8 spikes, but in both cases spike at ~5Hz. The coupling between pupil dilation and LC unit subgroup does not appear different in Figure S2C, which is borne through by the two-tailed unpaired t-test P=0.09. Finally, the authors interpret their claim that there are 2 subgroups of LC neurons in the context of recent work showing that LC neurons projecting to primary motor cortex have a different firing rate than those projecting to the prefrontal cortex (Chander, et al. 2014) and other work showing that two subsets of LC neurons with different action potential waveform widths fire at different rates (Totah, et al. 2018). However, the comparison to this work and the authors' interpreation is not warrented. First, the difference in firing rates observed between the two subgroups of neurons reported by Chander et al. and by Totah et al. were much smaller than the differences shown in Figure S2. Moreover, the authors did not report the action potential waveform duration in their data to draw a direct comparison with the results of Totah et al.

(3) The authors next test either pupil diameter can be used to infer LC activity to an extent that would allow a direct read-out without actually recording LC neurons. It is well-known, given the number of behavioral variables and activity of various neurons, that such a 1:1 read-out is not possible unless all of these variables were somehow accounted for in the calculation. It is a laudable goal to attempt this work, but the authors do not record from other regions simultaneously with LC neuronal recordings and/or monitor other behavioral variables; therefore, their work cannot be used to assess the potential of pupil diameter as a direct read-out of LC spiking. The data presented by the authors does support clearly a relationship between spontaneous, task-free LC spiking and pupil diameter, but their statement that pupil cannot accurately represent the majority of LC spiking and can only serve as a good proxy for infrequent epochs of high spike rate must be considered in light of the baseline level of LC activity in the recorded mice. Presumably, these data were recorded in a state of "quiet wakefulness." However, if the mice were to enter a drowsy or near-sleep state (which could be quantified with EEG or LFP recordings during these experiments), then LC firing rate would be reduced below what was reported here. Pupil diameter would also be reduced. This greater dynamic range of firing rates and pupil diameter could lead to a stronger relationship between LC spike rate and pupil diameter, even at the lower end of the range of firing rates. Effectively, if the LC spike rate is low and not changing much (quiet wakefulness), then the pupil diameter will be stable at some value and only transient changes of LC spike rate and pupil diameter from that baseline would show correlations; however, if LC rate and pupil diameter are further reduced, then smaller transient changes in rate and diameter would become observable and potentially correlated.

(4) The authors assessed how pupil diameter relates to LC population activity using optogenetic stimulation of the LC. They calculated the volume stimulated and estimate that a few hundred of the ~1,500 mouse LC neurons were synchronously activated. The data show substantial session-to-session variability in the pupil dilation response to optogenetic stimulation which was dissociated from stable responses of the same LC single units recorded across two recording sessions. These data were interpreted as LC activity being largely uncoupled with pupil diameter since the response to optogenetic LC activation fluctuates widely across different recording sessions for only one of those variables (pupil diameter), but the response of the same LC single unit tracked across days does not fluctuate substantially. First, it is particularly important that the authors note that stimulating the LC is not a physiological condition and does not reflect the population activity of the LC, which consists of de-correlated LC neuronal pairs (Totah et al. 2018) and small discrete ensembles (Totah et al. 2020, bioRxiv, https://doi.org/10.1101/2020.03.30.015354). Second, a methodological question that needs clarification in order to assess these data is a discussion and supporting data for how LC neurons were tracked across days. For instance, it has been shown that using wire tetrodes in rodents (as was done in the present study) allows single units to be recorded for typically only 3 days (see Table 1 in Eschenko, O. and Sara, S. J. "Learning-Dependent, Transient Increase of Activity in Noradrenergic Neurons of Locus Coeruleus during Slow Wave Sleep in the Rat: Brain Stem-Cortex Interplay for Memory Consolidation?" Cereb Cortex. (2008) 18, 2596-2603) and, moreover, they were not the same units across days. Thirdly, beyond methodological questions about tracking LC units across days, it is interesting to note that other work that assessed pupil diameter changes in response to LC optogenetic stimulation in both awake and anesthetized mice has shown that there is little variability in pupil diameter response to LC optogenetic stimulation across recording sessions at 2, 4 and 6 week time points (Privitera, M. et al. A complete pupillometry toolbox for real-time monitoring of locus coeruleus activity in rodents. Nat Protoc (2020) 15, 2301-2320). Accordingly, given the methodological questions and also existing work, it is not appropriate to make the strong claim that activation of a large portion of the LC neuronal population is largely uncoupled to pupil diameter.

Overall, given that prior work has shown that the relationship between pupil diameter and LC activity is not 1:1 and that many factors are correlated with pupil diameter, these data provide additional context showing how spontaneous LC activity (no behavioral task) is related to pupil diameter. The impact of such work is strong given that pupil size is often used a non-invasive proxy of neuromodulatory neuronal activity.

Reviewer #2:

In this manuscript the authors characterized a graded relationship between locus coeruleus (LC) activity and pupil diameter. This is an extension of their previous work (Yang et al., eLife, 2020). Through simultaneously recording of LC activity and pupil diameter, the authors found that pupil diameter increased monotonically with the number of LC spikes, but with substantial variability. Their data also showed that pupil exhibited large session-to-session fluctuations in response to identical optogenetic stimulation of the LC. The variations in the pupil-LC relationship were correlated with all decision related behavioral parameters in mice performing a tactile detection task. The results have very important implications for the use of the pupil diameter as a non-invasive proxy of LC activity, but are perhaps not terribly surprising given previous studies showing pupil diameter is correlated with cortical cholinergic activities.

In this manuscript the authors used simultaneous recordings of LC activity/pupil, optogenetic stimulation and behavior to carefully characterize a graded relationship between LC activity and pupil diameter. The results have very important implications for the use of the pupil size as a non-invasive proxy of LC activity. With that being said, methods were not clearly described in the manuscript and some results may be misinterpreted. Additional LC units should be recorded to ensure the results presented in figure 4 are generalizable. I believe a major revision is necessary before the manuscript being considered publication in the journal.

How to group adjacent spikes into clusters was not clearly defined. It was vaguely described in line 594 but I was confused by the definition, particularly why the clustering depends on each unit's inter-spike-interval. The authors mentioned other time windows (0.1-0.5 s), but it was unclear to me which time window they used in the original analysis.

Figure 1e-h: the authors used linear regression R^2>0.6 as an inclusion criterion for analysis in figure 1e-h. Would the p value of linear regressions a better inclusion criterion since it indicates whether the fit was statistically significant? In addition, it is unclear to me why 0.6 was chosen.

Figure S2c, the p is 0.09 for slope distributions between two subgroups of LC neurons. This indicated that the two subgroups of LC units were not different in coupling with pupil diameter. Therefore, I don't quite understand why the authors argued the two subgroups of LC units coupled with pupil differently (line 99).

Line 139-142: the authors argued here that the same LC unit was recorded across sessions. It is very challenging to record the same neuron across session in chronic settings. The authors should provide more proof to demonstrate that it was the same LC neuron (e.g. waveforms, statistics of spontaneous firing, tail pinch response, etc.).

Line 151: I liked the authors' idea to use anesthetized setup to control for spontaneous pupil fluctuations. However, the authors failed to provide quantitative comparisons of spontaneous pupil fluctuations across awake and anesthetized conditions. In addition, I couldn't find which anesthetics the authors used and how to control for that the difference in evoked pupil responses across sessions was not due to difference in anesthesia levels.

Line 163-165: I don't think results from 9 LC units are able to draw solid conclusions that LC-pupil coupling was strongly correlated with Hit rate, False Alarm rate and decision bias but weakly correlated with overall performance and detection sensitivity. Significantly more LC units (probably 25-30) should be recorded to test if the conclusions are generalizable across LC neurons and animals.

Reviewer #3:

The activity of the locus coeruleus (LC) noradrenergic system is thought to play an important role in the regulation of brain states and has been linked to changes in pupil size under constant illumination. However, recent studies, including Yang et al., 2021, have shown that changes in pupil size do not always reliably correlate with LC activity. Here, building on their previous work (Yang et al., 2021), the authors propose to better characterize the relationship between LC activity and pupil dilation, and attempt to explain the source of variability observed in the correlations between these two variables.

Pupillometry, which is an easy-to-use and non-invasive technique, has been used in humans and other animals as a readout of brain states, especially those linked to LC activity. Yet, the neural mechanisms mediating changes in pupil size are complex, and a thorough characterization of the relationship between LC activity and changes in pupil size is still missing in the field. Thus, this paper addresses an important question and should provide relevant information to better understand in which context pupil size can be used to index LC activity.

The data in the paper (pupillometry during a behavioral task paired with extracellular recording of LC neurons and optogenetic stimulation of LC population) are novel and rich and should be adequate to investigate the relationship between pupil size and LC activity. However, some of the analyses performed in this paper are not well designed or properly controlled, and therefore, do not always conclusively support the claims stated in the manuscript.

Some aspects of the paper can be clarified and/or improved:

1) One of the main claims in the paper is that the coupling between LC activity and pupil dilation is variable and that this variability is dependent on specific states (i.e., the decision-bias of mice performing a go/no-go task). Yet, pupil size in awake animals, even under constant luminance and in the absence of locomotion, remains intrinsically variable. In fact, facial movements such as whisking or licking (that are also involved in this go/no-go task) have been associated with pupil dilation. Therefore, different types of movements, as well as their rate and vigor, could potentially account for some of the variability observed in the correlation between pupil and NC activity. Yet, there is no mention of such confounds in the paper.

2) Most of the analyses and conclusions of the paper are based on the graded relationship between LC activity and change in pupil size quantified in Figure 1. This relationship relies on the size of spike clusters identified in the activity of single LC neurons and on the maximum change in pupil size within a 6s window after the first spike of a given cluster. However, it is not stated in the paper whether it was ensured that no other clusters occurred within the 6s window. If this condition is not respected, the number of spikes occurring in the time window of interest may themselves be variable, and therefore, not necessarily consistent with the cluster sizes used in the analysis.

3) It is briefly mentioned in Figure S2 that subgroups of neurons have instantaneous firing rate changing differently with cluster size. Interestingly, these subgroups have different coupling with pupil dilation. Thus, it is possible that changes in instantaneous firing rate, rather than cluster size, more reliably correlate with pupil dilation, and in turn, explain away some of the variability that is observed with cluster size. This alternative has not been explored in the paper.

4) In Figure 3, the reliability of the effect of optogenetic stimulation on the population of LC activity is inferred from one single unit across two sessions. Even if one single neuron appears stable between two sessions, it is not unlikely that the population activity varies from session to session. Thus, the data reported in this figure do not convincingly support a consistent transduction of optical stimulation to LC population activity.

5) Similarly, variability of pupil responses from "session-to-session", is quantified across only two sessions (i.e., two points), in just a couple of animals. From one session to the next, some experimental parameters that cannot be controlled for may change and add variability to the stimulations and measurements. Even in anesthetized animals, the level of anesthesia could differ from one session to the next. Thus, to more convincingly describe variability across sessions, one would need to add more sessions. Alternatively, the results could be compared to a "baseline variability" quantified from matched non-stimulated trials or from stimulated trials in control mice.

6) It is stated that pupil dilation is a good proxy for LC activity only for infrequent (<10%) and strong activity events. However, the authors do not comment on the nature of these infrequent events, especially in relation to the decision-bias.

Figure 1:

1) During the first read, I had assumed that Figure 1 showed recordings from sessions where mice were awake but passive. After reaching Figure 4, I then became unclear about whether mice were performing the go/no-go task already in Figure 1.

2) It is not specified whether the peak pupil size occurs at a consistent latency after the first cluster spike or whether the latency is cluster size dependent.

3) A different way to assess the correlation between pupil and LC activity, as well as the latency of the pupil response, is to first compute a smooth instantaneous firing rate signal (convolved with some gaussian kernels for instance), and then perform a cross-correlation between the firing rate and the pupil signal. This analysis will directly reveal the strength of the correlation and lag between the two vectors. Using the instantaneous firing rate could be informative especially with regards to the results in Figure S2.

4) Related to comment 1 of the public review: here the authors may want to condition their analysis to epochs when mice are licking, whisking, etc. (assuming that these data have been recorded from the behavioral task). If there are also some video recordings of the face (not just the pupil), the authors can even track facial movements and correlate such signals with the pupil response and LC activity.

Figure 3:

1) The authors claim that they have recorded the same unit across different days. Yet, it is not described in the text nor in the methods what criteria are used to define those units as identical.

2) I appreciate that examples of single mice are shown (Figure 3, Figure S8-10), but it would also help to see a summary plot with all the mice, all the sessions, and all the different conditions.

3) Related to the point above: assuming that there are significant session-to-session fluctuations in the pupil response to identical LC stimulation, could you describe it further? Does the stimulation of LC always induce a smaller pupil response on the second day? Is it consistent across mice?

4) The 95% confidence intervals seem extremely small compared to their respective P-values.

Figure 4:

1) This manuscript is formatted as a Research Advances and builds on Yang et al., 2021. Yet, it would help to have a sentence or two to briefly describe the task, either here or in the methods.

2) It is my understanding that the data shown here are from the same units as the data Figure 1e. However, it is unclear whether the results come from the same sessions. If they do, why are the slopes in Figure 4 not the same as the slope in Figure 1e (right)?

3) The more impulsive the animal (negative bias), the stronger the coupling between pupil and LC. Therefore, here again, we cannot rule out the implication of movements, since impulsive mice may tend to respond stronger and faster (larger false alarm, larger hit rate, which also correlates with stronger pupil-LC coupling). Without this basic control for movement, it is difficult to believe that the LC-pupil relationship directly depends on cognitive states such as motivation and engagement as stated in the paper, rather than difference in movements.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "A graded yet variable relationship between pupil diameter and locus coeruleus activity" for further consideration by eLife. Your revised article has been evaluated by Joshua Gold (Senior Editor) and a Reviewing Editor.

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

The new analyses and more tempered conclusions make the study more convincing compared to the previous version. However, there are remaining methodological and analysis concerns that should be addressed. If these concerns are addressed, there is agreement that the revised study would contribute to the field of LC physiology and how LC activity relates to pupil size, and would be suitable for publication in eLife.

Please address all the concerns outlined by the reviewers below.

In addition, it seems like it would be possible to perform an analysis of variability of optostim-to-dilation *within* each session and show that this variability is similar to the variability across sessions. This would help ameliorate concerns about variability in the LC-pupil relationship being due to variability in the long term stability of recordings and the optical fiber across sessions. The authors should provide information about the within-session vs cross-session variability.

Reviewer #1:

The authors main claim is that pupil diameter is not an accurate real-time readout of locus coeruleus activity. With the revised text, figures, and new analyses, the claim is convincing. I think the title change more accurately conveys the findings of the study.

With the addition of new analyses, the authors convincingly show that there is substantial session-to-session variability in the pupil dilation response to optogenetic stimulation (Figure 4C, D, etc.). In response to the prior review, the authors addressed the differences in relation to Privitera, et al. (2020) regarding pupil response variability across sessions. Moreover, with the addition of a new analysis of waveform correlations for LC single units across days (Figure 3G) the authors convincingly show that the same units can be tracked across days. The authors then point out that, in contrast to the pupil response to opto-stim having variable responses across days, the response of the same unit to opto-stim on different days is not variable (Figure 3H). Based on these data, the authors suggest that LC neurons are not an "accurate real-time readout of locus coeruleus activity". I think these new analyses and the data collectively do support this statement.

There is one caveat to this, which is that different size populations of LC neurons could be driven by optical stimulation on different days. This could occur because slow changes in the tissue over days (gliosis, blood flow, inflammation) could occlude light or alter light transmission. Under this scenario, the neurons tracked across days were likely not affected by changes in inflammation/gliosis (which is why they could be tracked), but this does not mean that LC neurons which could not be monitored by the electrode did not receive different light transmission on different days. If different size populations were stimulated, then the overall pupil dilation might change. This caveat is of course nothing that the authors could address given the current technology limitations and it is mentioned just for their consideration and could be discussed if they agree and find this warranted.

There is a question about the number of units in Figure 3H. The legend states 5 awake units (3H, left panel) and does not report the number of units for anesthetized (3H, right panel). However, it appears that there are only 4 awake units (dots) in the left panel.

There remains one methodological concern regarding single unit recordings.

The authors added additional quantification of spike train and spike clustering characteristics. The new Figure S1A shows that the inter-spike interval (ISI) histogram has most of its density around a mean of 20-25 msec. However, the figure legend reports a median of 200 msec. That median does not appear to match the distribution plotted, given that nearly the entire distribution plotted in Figure S1A is <0.2 sec. In prior work that recorded awake monkey LC single units, the ISI histogram has been shown to peak between ~150 msec and 350 msec with ISI's <100 msec rarely being observed (Aston-Jones, G., Rajkowski, J., Kubiak, P. and Alexinsky, T. Locus coeruleus neurons in monkey are selectively activated by attended cues in a vigilance task. J Neurosci 14, 4467-4480 (1994)). Other studies of multi-unit activity have found a median ISI of 53 msec with an inter-quartile range of 20 to 157 msec (Kalwani, R. M., Joshi, S. and Gold, J. I. Phasic activation of individual neurons in the locus ceruleus/subceruleus complex of monkeys reflects rewarded decisions to go but not stop. J Neurosci 34, 13656 13669 (2014)). This range appears to be in line with the authors' presentation of the ISI distribution in Figure S1A. On the other hand, the example waveforms look convincing. Further quantification is necessary, such as showing some ISI histograms for individual single units as well as some waveform clustering examples and average waveforms. Given the small number of units in the study, I think it would be helpful for the field to see for each unit the average waveform and the ISI histogram for each unit in a supplementary figure. This can help the field to better interpret their work in the context of others. Additionally, a presentation of the distribution of waveform durations against firing rate would also provide helpful support for interpreting the results.

While the authors' claim that pupil diameter is not an accurate real-time readout of locus coeruleus activity is supported by their data, I am still not convinced that this is not already a known and accepted idea. The field clearly knows that many different neuronal circuits affect pupil size. For example, it is known that LC output (axonal activity) tracks specific aspects of pupil dynamics but cholinergic neurons another aspect (Reimer, J. et al. Pupil fluctuations track rapid changes in adrenergic and cholinergic activity in cortex. Nat Commun 7, 13289 (2016)). With that being said, I think these data support their claim and add to the literature on this topic.

Reviewer #3:

The authors have performed new analyses (Figure 3, 4, S1, S2c, S8, S9) in an attempt to address most of the reviewers' comments. Missing information in the text and methods has been added, especially regarding the optogenetics experiments, and the discussion has been extended. This helps clarify some aspects of the paper. Some of the results are now more convincing (especially Figure 3 and 4) but other analyses remain oversimplified and therefore do not strongly support several conclusions.

1) In Figure 1 and Figure S2, depending on the method used to quantify the relationship between pupil dilation and cluster size, a different number of neurons meet the R2>0.6 criterion. This seems problematic since following analyses only focus on this subset of neurons. Are there always the same neurons that consistently fall below the 0.6 criterion? Do the results hold when all the recorded neurons are included?

2) I am still not fully convinced about the validity of the pupil dilation vs. cluster size analysis. Yet, this result (and its variability) is at the essence of the main conclusion of the paper. In particular, this analysis assumes that discrete events (spike clusters) linearly increase pupil size after a fixed latency (2-3s from Yang et al., 2021). Not accounting for the number of spikes occurring after a given cluster in the 3 s or 6 s time window disregard the fact that the latency may be variable or that the relationship between LC spiking and changes in pupil size could be nonlinear. This could reflect true variability in the LC-pupil relationship.

3) In Figure 5b, the non-liking period is defined as the activity outside a one second window around each licking events (clusters occurring within {plus minus}0.5 s from a lick were excluded). This choice is somewhat arbitrary, or at least not well motivated in the text. Besides, the pupil response is slow and the authors mentioned that the latency between peak pupil dilation and LC spikes is around 2 to 3 s. Therefore, the 1 s window may not be the most appropriate. A more convincing approach would be to condition the analysis on the number of licks performed in a trial for instance.

[Editors’ note: The authors appealed the original decision. What follows is the authors’ response to the first round of review.]

Reviewer #1:

Megemont, McBurney-Lin, and Yang simultaneously recorded pupil diameter and spiking from brainstem noradrenergic locus coeruleus (LC) neurons in awake and anesthetized mice in order to further probe the relationship between LC activity and pupil diameter. Numerous variables are known to correlate with pupil diameter, including behavioral variables (e.g., locomotion, behavioral task-related cognitive processes) and neuronal activity patterns (e.g., cholingergic neurons, inferior colliculus, anterior cingulate cortex). Here, the authors aimed to determine how accurately LC activity can be predicted from pupil diameter, which is a critical question given the relationship between pupil diameter and numerous other variables. To those ends, the authors convincingly show that there is a strong correlation (R^2>0.6) between number of spikes and pupil diameter in 68% of the units (13 out of 19). The authors propose, furthermore, that the activity of only a small proportion of LC units are related to pupil diameter; that specific sub-types of LC neurons have diverse coupling with pupil diameter; that pupil diameter cannot be used to predict the majority of LC spiking activity; and that pupil diameter is not related to LC population firing patterns. However, these latter claims are not supported by the data.

Below is a discussion of the data and authors' claims based on these data, methodological stengths and weaknesses that affect the authors' interpreation of the data, and how the findings relate to prior work:

(1) The authors claim that there is a monotonic relationship between pupil diameter and LC single unit activity. The authors show that is a strong correlation (R^2>0.6) between number of spikes and pupil diameter in 68% of the units (13 out of 19 single units). The data in Figure 1A-E support this claim. However, it is necessary to further show the inter-spike interval distributions and firing rates to demonstrate that the recorded spiking was isolated from single units. These data are qualitatively similar to the prior work cited by the authors reporting trial-by-trial correlations during a behavioral task (Joshi, et al. 2016 and Varazzani, et al. 2015). The authors extend these prior findings of LC-pupil coupling by characterizing the relationship between pupil and LC activity also in the absence of an ongoing behavioral task. Interestingly, while prior work during the behavioral task (Joshi et al., 2016) reported observing a much smaller proportion of LC single units that were significantly correlated with pupil diameter (35% in monkey Oz and 20% in monkey Ci), the task-free LC activity presented here shows the majority of LC units (68%) were coupled with pupil diameter. This is an important finding because it characterizes how the relationship between LC activity and pupil diameter changes whether assessed using the trial structure of a behavioral task versus spontaneous, task-free conditions. Given that the authors found 68% of units coupled to pupil diameter, the authors' proposal that the activity of only a small proportion of LC units are related to pupil diameter is not justified given that the majority were indeed coupled with pupil diameter.

We included new Figure S1a and b to show individual and group inter-spike interval distributions, respectively. Additional quality control measures are now included in Methods, line 667-669:

“Nineteen single unit recordings (cluster quality measure L_ratio: 0.01 ± 0.005; firing rate: 1.65 ± 0.25 spikes/s; percent ISI < 10 ms: 0.11% ± 0.1%) from seven mice performing the singlewhisker detection task (see below) were extracted using MClust (Redish, 2014)…”

68% of units showed monotonic relationship with pupil diameter. However, this relationship does not guarantee that pupil diameter can be used to predict LC activity on a moment-by-moment basis. We now make this point more clearly throughout the manuscript, e.g., line 99-100:

“Although pupil diameter exhibited an overall monotonic relationship with LC spiking, it does not necessarily warrant pupil diameter being an accurate readout of LC activity.”

We also changed the manuscript title to ‘Pupil diameter is not an accurate real-time readout of locus coeruleus activity’ to reflect this point.

We apologize for the misunderstanding. All LC-pupil recordings shown in Figure 1, 2 and 5 (old Figure 4) were acquired during behavior. We now made that clear in the manuscript, e.g., line 68-79:

“To address these questions, we recorded spiking activity from optogenetically-tagged LC neurons simultaneously with pupil diameter in head-fixed mice trained to perform a tactile detection task…”

(2) The authors claim that specific sub-types of LC neurons have diverse coupling with pupil diameter. This was substantiated by data showing that there were two subgroups of LC units with different LC firing rates and different coupling to pupil diameter (Figure S2). Here, it would be useful to know how the division was made between subgroups of units. Also, raster examples would help show how a unit can emit a brief cluster of 2 spikes or a brief cluster of 8 spikes, but in both cases spike at ~5Hz. The coupling between pupil dilation and LC unit subgroup does not appear different in Figure S2C, which is borne through by the two-tailed unpaired t-test P=0.09. Finally, the authors interpret their claim that there are 2 subgroups of LC neurons in the context of recent work showing that LC neurons projecting to primary motor cortex have a different firing rate than those projecting to the prefrontal cortex (Chander, et al. 2014) and other work showing that two subsets of LC neurons with different action potential waveform widths fire at different rates (Totah, et al. 2018). However, the comparison to this work and the authors' interpreation is not warrented. First, the difference in firing rates observed between the two subgroups of neurons reported by Chander et al. and by Totah et al. were much smaller than the differences shown in Figure S2. Moreover, the authors did not report the action potential waveform duration in their data to draw a direct comparison with the results of Totah et al.

All reviewers raised a similar point regarding the results in the old Figure S2. Although in the original manuscript we stated that the two groups of LC units ‘appeared to be coupled with pupil differently’, we completely agree that the results are not conclusive due to weak statistics. Therefore, this part of the results has been removed. In addition, we performed the Bayesian analysis suggested by this reviewer (recommendation #5) and the Bayes Factor (BF10) was 1.24, indicating inconclusive evidence.

(3) The authors next test either pupil diameter can be used to infer LC activity to an extent that would allow a direct read-out without actually recording LC neurons. It is well-known, given the number of behavioral variables and activity of various neurons, that such a 1:1 read-out is not possible unless all of these variables were somehow accounted for in the calculation. It is a laudable goal to attempt this work, but the authors do not record from other regions simultaneously with LC neuronal recordings and/or monitor other behavioral variables; therefore, their work cannot be used to assess the potential of pupil diameter as a direct read-out of LC spiking. The data presented by the authors does support clearly a relationship between spontaneous, task-free LC spiking and pupil diameter, but their statement that pupil cannot accurately represent the majority of LC spiking and can only serve as a good proxy for infrequent epochs of high spike rate must be considered in light of the baseline level of LC activity in the recorded mice. Presumably, these data were recorded in a state of "quiet wakefulness." However, if the mice were to enter a drowsy or near-sleep state (which could be quantified with EEG or LFP recordings during these experiments), then LC firing rate would be reduced below what was reported here. Pupil diameter would also be reduced. This greater dynamic range of firing rates and pupil diameter could lead to a stronger relationship between LC spike rate and pupil diameter, even at the lower end of the range of firing rates. Effectively, if the LC spike rate is low and not changing much (quiet wakefulness), then the pupil diameter will be stable at some value and only transient changes of LC spike rate and pupil diameter from that baseline would show correlations; however, if LC rate and pupil diameter are further reduced, then smaller transient changes in rate and diameter would become observable and potentially correlated.

We completely agree with the comment that ‘given the number of behavioral variables and activity of various neurons, that such a 1:1 read-out is not possible unless all of these variables were somehow accounted for in the calculation’. However, it is our impression that this is not everyone’s understanding. Actually, the main message we aimed to convey in this manuscript is, in a quantitative manner, that pupil is not a 1:1 readout of LC in most circumstances. We now include more extensive discussion on this topic, e.g., line 238-247:

“How may brain states modulate pupil-LC coupling? Pupil size changes have been linked to activity in other brain areas and neuromodulatory systems, including the medial prefrontal cortex, the inferior colliculus and cholinergic signaling (Joshi et al., 2016; Reimer et al., 2016; Okun et al., 2019; Kucyi and Parvizi, 2020; Pais-Roldán et al., 2020; Sobczak et al., 2021). A recent study found that pupil responses to dorsal raphe stimulation exhibited task uncertainty dependent variations (Cazettes et al., 2020). Therefore, it is possible that in high motivation/engagement states, multiple circuits including the LC synergistically influence pupil size changes, resulting in the apparently stronger pupil-LC coupling. Future experiments are needed to elucidate how pupil and LC interact with these brain circuits during different behavioral contexts and cognitive processes.”

Again, we apologize for the misunderstanding (response to #1). All LC-pupil recordings (Figure 1, 2, 5) were acquired during behavior. Our new Figure 5b supports that pupil-LC coupling is correlated with the state of the mice, which cannot be accounted for by the differences in licking. We have included new discussion on this topic, line 232-236:

“Although mice licked more during high motivation or high engagement trials (Berditchevskaia et al., 2016; Allen et al., 2019; McBurney-Lin et al., 2020), our data show that licking alone cannot account for the tight correlation between the variations of the behavioral variables and the variations in the pupil-LC relationship (Figure 5), suggesting that pupil-LC coupling is brain state dependent.”

Line 249-254:

“Another possibility is that higher engagement states may be intimately associated with more ‘uninstructed’ movements as revealed by recent work (Musall et al., 2019), which can drive robust neuronal activity throughout the brain (Musall et al., 2019; Steinmetz et al., 2019; Stringer et al., 2019; Salk_off et al., 2020). Future studies with comprehensive movement monitoring will determine whether more frequent movements, both task-related and task-unrelated, during periods of high motivation/engagement underlie the stronger pupil-LC coupling.”

(4) The authors assessed how pupil diameter relates to LC population activity using optogenetic stimulation of the LC. They calculated the volume stimulated and estimate that a few hundred of the ~1,500 mouse LC neurons were synchronously activated. The data show substantial session-to-session variability in the pupil dilation response to optogenetic stimulation which was dissociated from stable responses of the same LC single units recorded across two recording sessions. These data were interpreted as LC activity being largely uncoupled with pupil diameter since the response to optogenetic LC activation fluctuates widely across different recording sessions for only one of those variables (pupil diameter), but the response of the same LC single unit tracked across days does not fluctuate substantially. First, it is particularly important that the authors note that stimulating the LC is not a physiological condition and does not reflect the population activity of the LC, which consists of de-correlated LC neuronal pairs (Totah et al. 2018) and small discrete ensembles (Totah et al. 2020, bioRxiv, https://doi.org/10.1101/2020.03.30.015354). Second, a methodological question that needs clarification in order to assess these data is a discussion and supporting data for how LC neurons were tracked across days. For instance, it has been shown that using wire tetrodes in rodents (as was done in the present study) allows single units to be recorded for typically only 3 days (see Table 1 in Eschenko, O. and Sara, S. J. "Learning-Dependent, Transient Increase of Activity in Noradrenergic Neurons of Locus Coeruleus during Slow Wave Sleep in the Rat: Brain Stem-Cortex Interplay for Memory Consolidation?" Cereb Cortex. (2008) 18, 2596-2603) and, moreover, they were not the same units across days. Thirdly, beyond methodological questions about tracking LC units across days, it is interesting to note that other work that assessed pupil diameter changes in response to LC optogenetic stimulation in both awake and anesthetized mice has shown that there is little variability in pupil diameter response to LC optogenetic stimulation across recording sessions at 2, 4 and 6 week time points (Privitera, M. et al. A complete pupillometry toolbox for real-time monitoring of locus coeruleus activity in rodents. Nat Protoc (2020) 15, 2301-2320). Accordingly, given the methodological questions and also existing work, it is not appropriate to make the strong claim that activation of a large portion of the LC neuronal population is largely uncoupled to pupil diameter.

We thank the reviewer for these comments. We discuss the potential weakness of LC optical stimulation, line 216-219:

“However, optogenetic stimulation tends to synchronize neuronal activity, which may not reflect the physiological condition (Totah et al., 2018). Future experiments with the ability to simultaneously record from multiple opto-tagged LC neurons will further investigate the relationship between pupil diameter and population-level LC activity.”

We have provided new data to support that the putatively same units were tracked, based on opto-tagging, spike clustering and waveform comparison (Figure 3c-g).

We have included new data to show group pupil responses (Figure 4c-h). Pupil responses in an earlier session were not consistently higher or lower than a later session. Therefore, such session-to-session fluctuations were not observable from group data (Figure S8), which is consistent with Privitera et al., 2020. Line 144-149:

“Group data from multiple mice further demonstrated that significant session-to-session fluctuations of pupil responses were prevalent but not directional (solid lines in Figure 4c, d), i.e., pupil responses in an earlier session (session 1) were not consistently higher or lower than in a later session (session 2). Therefore, such session-to-session fluctuations were not observable from group comparisons (Figure S8, Privitera et al., 2020).”

In addition, some of the new pupil data were acquired a few hours apart (magenta arrows in Figure 4c-f) and still had substantial differences. In our opinion, it further alleviates the concern that LC responses may change over time, presumably due to slow drift.

Reviewer #2:

In this manuscript the authors characterized a graded relationship between locus coeruleus (LC) activity and pupil diameter. This is an extension of their previous work (Yang et al., eLife, 2020). Through simultaneously recording of LC activity and pupil diameter, the authors found that pupil diameter increased monotonically with the number of LC spikes, but with substantial variability. Their data also showed that pupil exhibited large session-to-session fluctuations in response to identical optogenetic stimulation of the LC. The variations in the pupil-LC relationship were correlated with all decision related behavioral parameters in mice performing a tactile detection task. The results have very important implications for the use of the pupil diameter as a non-invasive proxy of LC activity, but are perhaps not terribly surprising given previous studies showing pupil diameter is correlated with cortical cholinergic activities.

In this manuscript the authors used simultaneous recordings of LC activity/pupil, optogenetic stimulation and behavior to carefully characterize a graded relationship between LC activity and pupil diameter. The results have very important implications for the use of the pupil size as a non-invasive proxy of LC activity. With that being said, methods were not clearly described in the manuscript and some results may be misinterpreted. Additional LC units should be recorded to ensure the results presented in figure 4 are generalizable. I believe a major revision is necessary before the manuscript being considered publication in the journal.

How to group adjacent spikes into clusters was not clearly defined. It was vaguely described in line 594 but I was confused by the definition, particularly why the clustering depends on each unit's inter-spike-interval. The authors mentioned other time windows (0.1-0.5 s), but it was unclear to me which time window they used in the original analysis.

We now described in more details in Methods, line 700-701:

“For Figure 1, in each recording if the interval between two adjacent spikes was shorter than median inter-spike interval of that unit, the spikes were grouped into a single cluster.

Each unit was clustered by its own inter-spike interval (ITI) because units had different ITIs. We include a new Figure S1a, b to show the distribution of inter-spike interval.

Figure 1e-h: the authors used linear regression R^2>0.6 as an inclusion criterion for analysis in figure 1e-h. Would the p value of linear regressions a better inclusion criterion since it indicates whether the fit was statistically significant? In addition, it is unclear to me why 0.6 was chosen.

We performed this analysis and found that all included recordings (R²>0.6) had significant p values. None of the excluded recordings except one had significant p value. We now include a new Figure S1c to show the distribution of R² for all recordings, illustrating that the group with high R² values was well separated, and that setting the threshold between 0.5 and 0.8 as the inclusion criterion won’t affect any following results.

Figure S2c, the p is 0.09 for slope distributions between two subgroups of LC neurons. This indicated that the two subgroups of LC units were not different in coupling with pupil diameter. Therefore, I don't quite understand why the authors argued the two subgroups of LC units coupled with pupil differently (line 99).

We agree with this point. Please see our response to Reviewer1 comment #2 and recommendation #5. We have removed this part of the results due to weak statistics and inconclusive evidence.

Line 139-142: the authors argued here that the same LC unit was recorded across sessions. It is very challenging to record the same neuron across session in chronic settings. The authors should provide more proof to demonstrate that it was the same LC neuron (e.g. waveforms, statistics of spontaneous firing, tail pinch response, etc.).

Please see our response to Reviewer1 #4. We now provide new data to support that the putatively same units were tracked over days, based on opto-tagging, spike clustering and waveform comparison (Figure 3c-g).

Line 151: I liked the authors' idea to use anesthetized setup to control for spontaneous pupil fluctuations. However, the authors failed to provide quantitative comparisons of spontaneous pupil fluctuations across awake and anesthetized conditions. In addition, I couldn't find which anesthetics the authors used and how to control for that the difference in evoked pupil responses across sessions was not due to difference in anesthesia levels.

We thank the reviewer for this comment. We have performed new analysis to show that the occurrence and amplitude of spontaneous pupil dilation events were reduced in the anesthetized condition (Figure S9). We always used 2% isoflurane for anesthetized experiments. This is now included in the manuscript.

Line 163-165: I don't think results from 9 LC units are able to draw solid conclusions that LC-pupil coupling was strongly correlated with Hit rate, False Alarm rate and decision bias but weakly correlated with overall performance and detection sensitivity. Significantly more LC units (probably 25-30) should be recorded to test if the conclusions are generalizable across LC neurons and animals.

We agree with the reviewer that in order to make such comparisons (i.e., stronger vs. weaker correlation), more recordings will be needed. The main message we wanted to convey here is that the variations in pupil-LC coupling is correlated with behavioral variables. As a result, in the revised manuscript, we do not make such comparisons and only draw the conclusion that decision-biased related variables can explain the variations of pupil-LC coupling (Figure 5). In the meantime, we want to emphasize that making recordings from opto-tagged LC neurons during behavior presents a major challenge to the field and has rarely been done.

Reviewer #3:

The activity of the locus coeruleus (LC) noradrenergic system is thought to play an important role in the regulation of brain states and has been linked to changes in pupil size under constant illumination. However, recent studies, including Yang et al., 2021, have shown that changes in pupil size do not always reliably correlate with LC activity. Here, building on their previous work (Yang et al., 2021), the authors propose to better characterize the relationship between LC activity and pupil dilation, and attempt to explain the source of variability observed in the correlations between these two variables.

Pupillometry, which is an easy-to-use and non-invasive technique, has been used in humans and other animals as a readout of brain states, especially those linked to LC activity. Yet, the neural mechanisms mediating changes in pupil size are complex, and a thorough characterization of the relationship between LC activity and changes in pupil size is still missing in the field. Thus, this paper addresses an important question and should provide relevant information to better understand in which context pupil size can be used to index LC activity.

The data in the paper (pupillometry during a behavioral task paired with extracellular recording of LC neurons and optogenetic stimulation of LC population) are novel and rich and should be adequate to investigate the relationship between pupil size and LC activity. However, some of the analyses performed in this paper are not well designed or properly controlled, and therefore, do not always conclusively support the claims stated in the manuscript.

Some aspects of the paper can be clarified and/or improved:

1) One of the main claims in the paper is that the coupling between LC activity and pupil dilation is variable and that this variability is dependent on specific states (i.e., the decision-bias of mice performing a go/no-go task). Yet, pupil size in awake animals, even under constant luminance and in the absence of locomotion, remains intrinsically variable. In fact, facial movements such as whisking or licking (that are also involved in this go/no-go task) have been associated with pupil dilation. Therefore, different types of movements, as well as their rate and vigor, could potentially account for some of the variability observed in the correlation between pupil and NC activity. Yet, there is no mention of such confounds in the paper.

The reviewer raised an excellent point. In general, we think these are emerging and important questions in the field: what is the relationship between brain states and movements? How could different types of movements affect physiological processes, including pupil-LC coupling? Unfortunately, we do not have whisker tracking in the current dataset. Given that licking is a major type of movement during the task, we performed new analysis to show that licking cannot account for the variability (Figure 5b). However, the reviewer is correct that other types of movements may come into play. We include new sections to discuss this potential confound, line 232-236:

“Although mice licked more during high motivation or high engagement trials (Berditchevskaia et al., 2016; Allen et al., 2019; McBurney-Lin et al., 2020), our data show that licking alone cannot account for the tight correlation between the variations of the behavioral variables and the variations in the pupil-LC relationship (Figure 5), suggesting that pupil-LC coupling is brain state dependent.”

Line 249-254:

“Another possibility is that higher engagement states may be intimately associated with more ‘uninstructed’ movements as revealed by recent work (Musall et al., 2019), which can drive robust neuronal activity throughout the brain (Musall et al., 2019; Steinmetz et al., 2019; Stringer et al., 2019; Salk_off et al., 2020). Future studies with comprehensive movement monitoring will determine whether more frequent movements, both task-related and task-unrelated, during periods of high motivation/engagement underlie the stronger pupil-LC coupling.”

2) Most of the analyses and conclusions of the paper are based on the graded relationship between LC activity and change in pupil size quantified in Figure 1. This relationship relies on the size of spike clusters identified in the activity of single LC neurons and on the maximum change in pupil size within a 6s window after the first spike of a given cluster. However, it is not stated in the paper whether it was ensured that no other clusters occurred within the 6s window. If this condition is not respected, the number of spikes occurring in the time window of interest may themselves be variable, and therefore, not necessarily consistent with the cluster sizes used in the analysis.

Our previous work shows that the latency between peak pupil dilation and LC spikes is around 23 s (Yang et al., 2021, eLife). We chose the 6-s window to account for the long latency between pupil responses and large LC clusters (For example, Figure 1d, 6-8 spikes panels). We performed new analysis using a shorter window (3-s) and found that it did not affect the pupil-LC relationship (Figure S2c), and the results in Figure 5 were not affected using the 3-s window (not shown). We can include the latter result if deemed critical.

3) It is briefly mentioned in Figure S2 that subgroups of neurons have instantaneous firing rate changing differently with cluster size. Interestingly, these subgroups have different coupling with pupil dilation. Thus, it is possible that changes in instantaneous firing rate, rather than cluster size, more reliably correlate with pupil dilation, and in turn, explain away some of the variability that is observed with cluster size. This alternative has not been explored in the paper.

As described above, results related to old Figure S2 have been removed due to weak statistics and inconclusive evidence. However, we performed the suggested analysis and found that instantaneous firing rate of the clusters did not better correlate with pupil diameter. We can include this result if deemed critical.

4) In Figure 3, the reliability of the effect of optogenetic stimulation on the population of LC activity is inferred from one single unit across two sessions. Even if one single neuron appears stable between two sessions, it is not unlikely that the population activity varies from session to session. Thus, the data reported in this figure do not convincingly support a consistent transduction of optical stimulation to LC population activity.

We have provided new data to support that the putatively same units were tracked (Figure 3c-g), and all units responded similarly across sessions (Figure 3h). In our opinion, being able to track the putatively same units across days strongly suggest that the tetrodes and optical fiber had minimal shift inside the tissue (< tens of micrometers). Given that the estimated excitable volume is on the order of 0.1 mm³, it is the largely same population of neurons being stimulated. The combined evidence that (1) the largely same population of neurons were activated and (2) all putatively same neurons responded similarly across sessions strongly suggest that population responses were similar. In contrast, pupil responses can be dramatically different (e.g., doubled or halved, Figure 4). Furthermore, some of the new pupil data were acquired a few hours apart (magenta arrows in Figure 4c-f) and still had substantial differences. In our opinion, it further alleviates the concern that LC population responses could be different from session to session, presumably due to slow drift over time. However, due to technical limitations, we are not able to obtain multiple opto-tagged LC units simultaneously, and to our knowledge, no published work has accomplished this. This is now discussed, line 216-219:

“However, optogenetic stimulation tends to synchronize neuronal activity, which may not reflect the physiological condition (Totah et al., 2018). Future experiments with the ability to simultaneously record from multiple opto-tagged LC neurons will further investigate the relationship between pupil diameter and population-level LC activity.”

5) Similarly, variability of pupil responses from "session-to-session", is quantified across only two sessions (i.e., two points), in just a couple of animals. From one session to the next, some experimental parameters that cannot be controlled for may change and add variability to the stimulations and measurements. Even in anesthetized animals, the level of anesthesia could differ from one session to the next. Thus, to more convincingly describe variability across sessions, one would need to add more sessions. Alternatively, the results could be compared to a "baseline variability" quantified from matched non-stimulated trials or from stimulated trials in control mice.

We thank the reviewer for the comments. For consistency, we always used 2% isoflurane in anesthetized recordings. Session-to-session pupil responses were quantified in two sessions because these sessions were baseline-matched, i.e., they had very similar baseline pupil diameter, as it is known that evoked pupil sizes are influenced by baseline pupil. We have included new group data of baseline-matched pupil responses to optical stimulation from more sessions and more mice to support our conclusions (Figure 4c-h).

6) It is stated that pupil dilation is a good proxy for LC activity only for infrequent (<10%) and strong activity events. However, the authors do not comment on the nature of these infrequent events, especially in relation to the decision-bias.

We have included new discussions on these points. Line 195-200:

“Large pupil or LC responses have been reported to correlate with a variety of task-related processes, including sensory cue, decision formation, positive feedback, choice bias and action (Rajkowski et al., 1994; Usher et al., 1999; Kalwani et al., 2014b; Bouret and Richmond, 2015; de Gee et al., 2017, 2020; Schriver et al., 2020; Yang et al., 2021). In light of this, our work suggests that the infrequent but strong pupil dilation events can be used as an accurate inference of LC activation in response to sensory stimuli and decision-making processes.”

Line 238-247:

“How may brain states modulate pupil-LC coupling? Pupil size changes have been linked to activity in other brain areas and neuromodulatory systems, including the medial prefrontal cortex, the inferior colliculus and cholinergic signaling (Joshi et al., 2016; Reimer et al., 2016; Okun et al., 2019; Kucyi and Parvizi, 2020; Pais-Roldán et al., 2020; Sobczak et al., 2021). A recent study found that pupil responses to dorsal raphe stimulation exhibited task uncertainty dependent variations (Cazettes et al., 2020). Therefore, it is possible that in high motivation/engagement states, multiple circuits including the LC synergistically influence pupil size changes, resulting in the apparently stronger pupil-LC coupling. Future experiments are needed to elucidate how pupil and LC interact with these brain circuits during different behavioral contexts and cognitive processes.”

Figure 1:

1) During the first read, I had assumed that Figure 1 showed recordings from sessions where mice were awake but passive. After reaching Figure 4, I then became unclear about whether mice were performing the go/no-go task already in Figure 1.

We apologize for this confusion. We now state it clearly that all recordings in Figure 1, 2, 5 were acquired during behavior.

2) It is not specified whether the peak pupil size occurs at a consistent latency after the first cluster spike or whether the latency is cluster size dependent.

We believe this is related to comment #2. Our previous work shows that the latency between peak pupil dilation and LC spikes is around 2-3 s (Yang et al., 2021). Using a 6-s or 3-s window to identify peak pupil responses did not affect the results (Figure S2c), and the results in Figure 5 were not affected using the 3-s window (not shown).

3) A different way to assess the correlation between pupil and LC activity, as well as the latency of the pupil response, is to first compute a smooth instantaneous firing rate signal (convolved with some gaussian kernels for instance), and then perform a cross-correlation between the firing rate and the pupil signal. This analysis will directly reveal the strength of the correlation and lag between the two vectors. Using the instantaneous firing rate could be informative especially with regards to the results in Figure S2.

The suggested analysis has been done in our previous work (e.g., Figure 2b, Yang et al., 2021), and we showed that the latency between peak pupil dilation and LC spikes is around 2-3 s. We believe our current analysis has some advantages both conceptually and methodologically, including (1) the ability to identify distinct LC spiking patterns (clusters), which allows us (2) to quantitatively link the number of LC spikes to pupil diameter, and (3) to use pupil size changes to predict LC activity on a moment-by-moment basis. In addition, Figure S2 has been removed due to weak statistics and inconclusive evidence.

4) Related to comment 1 of the public review: here the authors may want to condition their analysis to epochs when mice are licking, whisking, etc. (assuming that these data have been recorded from the behavioral task). If there are also some video recordings of the face (not just the pupil), the authors can even track facial movements and correlate such signals with the pupil response and LC activity.

We believe this is similar to comment #1. Please see our response above. In brief, the new analysis showed that licking cannot account for the variability (Figure 5b). In addition, we performed the suggested analysis conditioned on licking periods only (± 0.5 s from each licking event). Fewer recordings exhibited monotonic relationships, likely due to fewer LC clusters during licking periods. However, the overall trend in Figure 5 still held (data not shown). We can include this result if deemed critical.

The reviewer raised a great point on how different types of movements may affect pupil-LC coupling, or in a more general sense, affect physiological and cognitive processes. We believe these are emerging and important questions and require separate and comprehensive investigations. We include a paragraph to discuss this topic, line 249-254:

“Another possibility is that higher engagement states may be intimately associated with more ‘uninstructed’ movements as revealed by recent work (Musall et al., 2019), which can drive robust neuronal activity throughout the brain (Musall et al., 2019; Steinmetz et al., 2019; Stringer et al., 2019; Salk_off et al., 2020). Future studies with comprehensive movement monitoring will determine whether more frequent movements, both task-related and task-unrelated, during periods of high motivation/engagement underlie the stronger pupil-LC coupling.”

Figure 3:

1) The authors claim that they have recorded the same unit across different days. Yet, it is not described in the text nor in the methods what criteria are used to define those units as identical.

We have provided new data to support that the putatively same units were tracked, based on opto-tagging, spike clustering and waveform comparison (Figure 3c-g).

2) I appreciate that examples of single mice are shown (Figure 3, Figure S8-10), but it would also help to see a summary plot with all the mice, all the sessions, and all the different conditions.

We have included new data to show group pupil responses under different conditions (Figure 4c-h).

3) Related to the point above: assuming that there are significant session-to-session fluctuations in the pupil response to identical LC stimulation, could you describe it further? Does the stimulation of LC always induce a smaller pupil response on the second day? Is it consistent across mice?

Pupil responses in an earlier session were not consistently higher or lower than in a later session. This is now discussed in line 144-149:

“Group data from multiple mice further demonstrated that significant session-to-session fluctuations of pupil responses were prevalent but not directional (solid lines in Figure 4c, d), i.e., pupil responses in an earlier session (session 1) were not consistently higher or lower than in a later session (session 2). Therefore, such session-to-session fluctuations were not observable from group comparisons (Figure S8, Privitera et al., 2020).”

4) The 95% confidence intervals seem extremely small compared to their respective P-values.

This is now described in more details in Methods.

Figure 4:

1) This manuscript is formatted as a Research Advances and builds on Yang et al., 2021. Yet, it would help to have a sentence or two to briefly describe the task, either here or in the methods.

We apologize for this oversight. Description of the behavioral task is now included in Methods.

2) It is my understanding that the data shown here are from the same units as the data Figure 1e. However, it is unclear whether the results come from the same sessions. If they do, why are the slopes in Figure 4 not the same as the slope in Figure 1e (right)?

Indeed, the data are in Figure 1e and old Figure 4 (new Figure 5). For the subset of 9 recordings included in Figure 5 (with > 100 trials and R² > 0.6), their slopes shown in the y-axis of Figure 5a and in Figure 1e are the same.

3) The more impulsive the animal (negative bias), the stronger the coupling between pupil and LC. Therefore, here again, we cannot rule out the implication of movements, since impulsive mice may tend to respond stronger and faster (larger false alarm, larger hit rate, which also correlates with stronger pupil-LC coupling). Without this basic control for movement, it is difficult to believe that the LC-pupil relationship directly depends on cognitive states such as motivation and engagement as stated in the paper, rather than difference in movements.

We believe this is similar to comment #1 and recommendation #4. Please see our response above. In brief, we think these are emerging and important questions in the field: What is the relationship between brain states and movements? How could different types of movements affect physiological processes including the pupil-LC relationship studied here? We performed new analysis to show that one major type of movement in this task, licking, cannot account for the reported variability (Figure 5b). However, the reviewer is correct that other task-related or unrelated movements may come into play. We included a new section to discuss this topic, line 249-254:

“Another possibility is that higher engagement states may be intimately associated with more ‘uninstructed’ movements as revealed by recent work (Musall et al., 2019), which can drive robust neuronal activity throughout the brain (Musall et al., 2019; Steinmetz et al., 2019; Stringer et al., 2019; Salk_off et al., 2020). Future studies with comprehensive movement monitoring will determine whether more frequent movements, both task-related and task-unrelated, during periods of high motivation/engagement underlie the stronger pupil-LC coupling.”

[Editors’ note: what follows is the authors’ response to the second round of review.]

The new analyses and more tempered conclusions make the study more convincing compared to the previous version. However, there are remaining methodological and analysis concerns that should be addressed. If these concerns are addressed, there is agreement that the revised study would contribute to the field of LC physiology and how LC activity relates to pupil size, and would be suitable for publication in eLife.

Please address all the concerns outlined by the reviewers below.

In addition, it seems like it would be possible to perform an analysis of variability of optostim-to-dilation *within* each session and show that this variability is similar to the variability across sessions. This would help ameliorate concerns about variability in the LC-pupil relationship being due to variability in the long term stability of recordings and the optical fiber across sessions. The authors should provide information about the within-session vs cross-session variability.

We thank the editors for this suggestion. We performed the analysis where across-session variability was estimated by resampling trials pooled from all sessions and found that within session and across-session variability were comparable (new Figure 4—figure supplement 3). Furthermore, for the paired sessions that exhibited significantly different responses to optical stimulation, we tested whether across-session variability was comparable to within-session variability by resampling trials pooled from both sessions to generate a distribution of across session variability and examined whether the variability of individual sessions fell outside 5% of the distribution. We found that 2 session pairs out of 12 under awake 10-ms stimulation condition had larger across-session variability. 3 session pairs out of 11 under awake 50-ms stimulation condition had larger across-session variability. Altogether, we conclude that across-session and within-session variability were largely comparable. This is now included in the revised manuscript, e.g., line 160-164:

“Further analysis showed that across-session variability of pupil responses was comparable to within-session variability (Figure 4—figure supplement 3, Methods). In addition, for the paired sessions that exhibited significantly different responses to optical stimulation (solid lines in Figure 4c-f), only a small subset exhibited larger across-session variability than within-session variability (2 pairs out of 12 under 10-ms condition, and 3 pairs out of 11 under 50-ms condition, Methods).”

Reviewer #1:

The authors main claim is that pupil diameter is not an accurate real-time readout of locus coeruleus activity. With the revised text, figures, and new analyses, the claim is convincing. I think the title change more accurately conveys the findings of the study.

With the addition of new analyses, the authors convincingly show that there is substantial session-to-session variability in the pupil dilation response to optogenetic stimulation (Figure 4C, D, etc.). In response to the prior review, the authors addressed the differences in relation to Privitera, et al. (2020) regarding pupil response variability across sessions. Moreover, with the addition of a new analysis of waveform correlations for LC single units across days (Figure 3G) the authors convincingly show that the same units can be tracked across days. The authors then point out that, in contrast to the pupil response to opto-stim having variable responses across days, the response of the same unit to opto-stim on different days is not variable (Figure 3H). Based on these data, the authors suggest that LC neurons are not an "accurate real-time readout of locus coeruleus activity". I think these new analyses and the data collectively do support this statement.

We thank the reviewer for the positive evaluation.

There is one caveat to this, which is that different size populations of LC neurons could be driven by optical stimulation on different days. This could occur because slow changes in the tissue over days (gliosis, blood flow, inflammation) could occlude light or alter light transmission. Under this scenario, the neurons tracked across days were likely not affected by changes in inflammation/gliosis (which is why they could be tracked), but this does not mean that LC neurons which could not be monitored by the electrode did not receive different light transmission on different days. If different size populations were stimulated, then the overall pupil dilation might change. This caveat is of course nothing that the authors could address given the current technology limitations and it is mentioned just for their consideration and could be discussed if they agree and find this warranted.

We agree with this potential caveat. We have expanded the relevant Discussion section to incorporate this caveat more explicitly, line 224-227:

“The fact that the putatively same neurons tracked across days exhibited similar responses to optical stimulation cannot fully warrant the long-term stability of population LC response because slow changes in the tissue due to tetrode/optical fiber implant (gliosis, inflammation, etc.) could alter light transmission to the neurons that were not recorded.”

In addition, based on the editors’ comment above, we performed new analyses to show that within-session and across-session variability of pupil responses were comparable (new Figure 4figure supplement 3 and related discussion), to ameliorate concerns about long-term stability of LC responses.

There is a question about the number of units in Figure 3H. The legend states 5 awake units (3H, left panel) and does not report the number of units for anesthetized (3H, right panel). However, it appears that there are only 4 awake units (dots) in the left panel.

We apologize for this confusion. All 5 units were tested under the anesthetized condition (3 subsets tested with 3 different patterns of optical stimulation), and 4 out of the 5 units were tested in the awake condition (1 stimulation pattern). We now make this clear in the legends of Figure 3h.

There remains one methodological concern regarding single unit recordings.

The authors added additional quantification of spike train and spike clustering characteristics. The new Figure S1A shows that the inter-spike interval (ISI) histogram has most of its density around a mean of 20-25 msec. However, the figure legend reports a median of 200 msec. That median does not appear to match the distribution plotted, given that nearly the entire distribution plotted in Figure S1A is <0.2 sec. In prior work that recorded awake monkey LC single units, the ISI histogram has been shown to peak between ~150 msec and 350 msec with ISI's <100 msec rarely being observed (Aston-Jones, G., Rajkowski, J., Kubiak, P. and Alexinsky, T. Locus coeruleus neurons in monkey are selectively activated by attended cues in a vigilance task. J Neurosci 14, 4467-4480 (1994)). Other studies of multi-unit activity have found a median ISI of 53 msec with an inter-quartile range of 20 to 157 msec (Kalwani, R. M., Joshi, S. and Gold, J. I. Phasic activation of individual neurons in the locus ceruleus/subceruleus complex of monkeys reflects rewarded decisions to go but not stop. J Neurosci 34, 13656 13669 (2014)). This range appears to be in line with the authors' presentation of the ISI distribution in Figure S1A. On the other hand, the example waveforms look convincing. Further quantification is necessary, such as showing some ISI histograms for individual single units as well as some waveform clustering examples and average waveforms. Given the small number of units in the study, I think it would be helpful for the field to see for each unit the average waveform and the ISI histogram for each unit in a supplementary figure. This can help the field to better interpret their work in the context of others. Additionally, a presentation of the distribution of waveform durations against firing rate would also provide helpful support for interpreting the results.

In Figure 1—figure supplement 1a (original Figure S1a), ISI was linearly binned and logarithmically plotted. We apologize that this plot did not clearly reveal the right side of the distribution. It is now changed to logarithmic binning for better visualization. This unit indeed has a median ISI of 200 ms (first example in the new Figure 1—figure supplement 1a). We now include additional examples of ISI distribution, spike waveform and spike clustering in Figure 1—figure supplement 1a, and list ISI and spike waveform from all recordings in the new Figure 4—figure supplement 6.

While the authors' claim that pupil diameter is not an accurate real-time readout of locus coeruleus activity is supported by their data, I am still not convinced that this is not already a known and accepted idea. The field clearly knows that many different neuronal circuits affect pupil size. For example, it is known that LC output (axonal activity) tracks specific aspects of pupil dynamics but cholinergic neurons another aspect (Reimer, J. et al. Pupil fluctuations track rapid changes in adrenergic and cholinergic activity in cortex. Nat Commun 7, 13289 (2016)). With that being said, I think these data support their claim and add to the literature on this topic.

We thank the reviewer for this comment. We hope our current manuscript will reinforce this understanding and contribute to the field.

Reviewer #3:

The authors have performed new analyses (Figure 3, 4, S1, S2c, S8, S9) in an attempt to address most of the reviewers' comments. Missing information in the text and methods has been added, especially regarding the optogenetics experiments, and the discussion has been extended. This helps clarify some aspects of the paper. Some of the results are now more convincing (especially Figure 3 and 4) but other analyses remain oversimplified and therefore do not strongly support several conclusions.

1) In Figure 1 and Figure S2, depending on the method used to quantify the relationship between pupil dilation and cluster size, a different number of neurons meet the R2>0.6 criterion. This seems problematic since following analyses only focus on this subset of neurons. Are there always the same neurons that consistently fall below the 0.6 criterion? Do the results hold when all the recorded neurons are included?

When the pupil-LC relationship was quantified using % changes from baseline (original Figure S2a, new Figure 1—figure supplement 3a), the exact same 13 recordings in Figure 1e were identified with R² > 0.6. When the relationship was quantified using a 3-s window (new Figure 1—figure supplement 3c) instead of 6-s, 12 out of the 13 recordings in Figure 1e were identified with R² > 0.6. Using the time derivative of pupil yielded a slightly lower number of linear pupil-LC relationship (new Figure 1figure supplement 3b, 9 recordings with R² > 0.6), and 8 out of the 9 were the same as in Figure 1e. We expanded the legends of new Figure 1—figure supplement 3 to include this information. Furthermore, the results in Figure 1g held when all recordings were included. This is now shown in the new Figure 1—figure supplement 4.

2) I am still not fully convinced about the validity of the pupil dilation vs. cluster size analysis. Yet, this result (and its variability) is at the essence of the main conclusion of the paper. In particular, this analysis assumes that discrete events (spike clusters) linearly increase pupil size after a fixed latency (2-3s from Yang et al., 2021). Not accounting for the number of spikes occurring after a given cluster in the 3 s or 6 s time window disregard the fact that the latency may be variable or that the relationship between LC spiking and changes in pupil size could be nonlinear. This could reflect true variability in the LC-pupil relationship.

We apologize for not being clear in our previous response to a similar question. In addition to our recent work showing that the peak correlation coefficient between pupil and LC spiking has a lag of 2-3 s (Yang et al., 2021), other studies in rodents have demonstrated that activating the LC induced peak pupil dilation with a comparable latency (e.g., Breton-Provencher and Sur, 2018; Liu et al., 2017 and our own data in Figure 4). In our opinion, these lines of evidence strongly suggest that the pupil responds to LC activity with a temporal delay on the order of 2-3 s. Furthermore, based on the reviewer’s comment, we performed new analyses to show that the latency of peak pupil diameter following LC clusters did not significantly vary with cluster size (correlation coefficient = 0.37, P = 0.37), and ranged between 2.5 and 4 s (new Figure 1—figure supplement 2). We also found that only in 1 out of the 13 recordings (with R² > 0.6) did the number of spikes occurring after a given cluster (in-between spikes) significantly correlate with LC cluster size or peak pupil diameter (P = 0.049 and 0.047, respectively), and overall the in-between spikes did not significantly correlate with LC cluster size (correlation coefficient = 0.35, P = 0.39; R² = 0.10) or peak pupil diameter (correlation coefficient = 0.29, P = 0.48; R² = 0.15). In our opinion, altogether these results strongly suggest that the in-between spikes did not significantly contribute to the variability of the pupil-LC relationship. This information is now included in the legends of Figure 1—figure supplement 3.

3) In Figure 5b, the non-liking period is defined as the activity outside a one second window around each licking events (clusters occurring within {plus minus}0.5 s from a lick were excluded). This choice is somewhat arbitrary, or at least not well motivated in the text. Besides, the pupil response is slow and the authors mentioned that the latency between peak pupil dilation and LC spikes is around 2 to 3 s. Therefore, the 1 s window may not be the most appropriate. A more convincing approach would be to condition the analysis on the number of licks performed in a trial for instance.

Our choice of the ± 0.5 s window was motivated by previous results that peak LC response occurred within a few hundred milliseconds of licking onset (Yang et al., 2021). This is now included in Methods. In addition, we did further analyses to exclude pupil responses in a 4-s window following each licking event. Fewer recordings passed the inclusion criterion (6 instead of 9), but the overall relationship held (Left panel in Author response image 1). We also conditioned the analysis on no-licking trials only, as suggested by the reviewer, and similar results held (Right panel in Author response image 1).

Author response image 1

Download asset Open asset

Author details

1. Marine Megemont

   Department of Molecular, Cell and Systems Biology, University of California, Riverside, Riverside, United States

   Contribution

   Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Software, Validation, Visualization, Writing - original draft, Writing - review and editing

   Contributed equally with

   Jim McBurney-Lin

   Competing interests

   No competing interests declared

2. Jim McBurney-Lin

   1. Department of Molecular, Cell and Systems Biology, University of California, Riverside, Riverside, United States
   2. Neuroscience Graduate Program, University of California, Riverside, Riverside, United States

   Contribution

   Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Software, Validation, Visualization, Writing - original draft, Writing - review and editing

   Contributed equally with

   Marine Megemont

   Competing interests

   No competing interests declared

3. Hongdian Yang

   1. Department of Molecular, Cell and Systems Biology, University of California, Riverside, Riverside, United States
   2. Neuroscience Graduate Program, University of California, Riverside, Riverside, United States

   Contribution

   Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing - original draft, Writing - review and editing

   For correspondence

   hongdian@ucr.edu

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-5203-9519

• 4,528

  Page views

• 474

  Downloads

• 15

  Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.