In vivo Effects of Temperature on the Heart and Pyloric Rhythms in the Crab, Cancer borealis

Summary Statement Temperature elevation increases the frequency of the heart and pyloric rhythms of the crab, Cancer borealis, but the heart rhythm has a higher critical temperature than the pyloric rhythm. Abstract The heart and pyloric rhythms of crustaceans have been studied separately and extensively over many years. Local and hormonal neuromodulation and sensory inputs onto these central pattern generating circuits play a significant role in the animals’ responses to perturbations, but are usually lost or removed during in vitro studies. To examine simultaneously the in vivo motor output of the heart and pyloric rhythms, we used photoplethysmography (PPG). In the population measured (n = 49), the heart rhythm frequencies ranged from 0.3–2.3 Hz. The pyloric rhythms varied from 0.2–1.6 Hz. During multiple hour-long recordings, many animals held at control temperature showed strong inhibitory bouts in which the heart decreased in frequency or become quiescent and the pyloric rhythm also decreased in frequency. Many animals show significant coherence in frequency between the rhythms at the frequency of the heart rhythm. We measured the simultaneous responses of the rhythms to temperature ramps by heating or cooling the saline bath while recording both the heart and pyloric muscle movements. Q10s, critical temperatures (temperatures at which function is compromised), and changes in frequency were calculated for each of the rhythms tested. The heart rhythm was more robust to high temperature than the pyloric rhythm.

Throughout its life, an organism must respond to perturbations that affect the 103 nervous system and the biomechanical structures it controls. Crustaceans are 104 poikilotherms and therefore may experience wide variations in body temperature due to 105 temperature changes in the environment. Because of this, it is relevant to study the 106 effect of temperature on the activity of the STNS and CG, as temperature changes may 107 simultaneously affect cellular processes of both systems and disrupt neuronal function. 108 Throughout their lives, crustaceans experience both short term temperature 109 fluctuations, due to changing tidal patterns during a single day, and long term 110 temperature fluctuations, due to seasonal temperature variations (Soofi et al., 2014;111 Tang et al., Tang et al., 2012). Despite these varied temperature changes, animals 112 must maintain rhythmicity and performance of the foregut and heart. Therefore, 113 temperature is a useful manipulation to study network stability in the face of 114 perturbation. 115 Previous work on crustaceans indicate that both the heart and pyloric rhythms 116 are robust to temperature changes. In the heart, studies have shown that the strength of 117 a heartbeat decreases and heart rate increases with increases in temperature (Camacho 118 et al., 2006;Worden et al., 2006). Therefore, the increase in heart rate partially 119 compensates for the decrease in stroke volume as the CG is further pushed from its 120 normal state at 11°C. The pyloric triphasic rhythm is maintained across a wide range of 121 temperatures. In both the heart and the pyloric rhythm, the maximum frequency  The Q10 is a measure of the sensitivity of a biological process to a 10°C change in 126 temperature. Many biological processes have Q10s between two and three, while some 127 temperature sensitive ion channels have Q10s as high as 50 or 100 . 128 If all Q10s of the components involved in a biological process are similar, this process is

142
In this study we systematically explore the potential relationships between 143 stomach and heart rhythms, and ask whether they are coordinately sensitive to 144 perturbation by temperature. incubator at 10C to 12C. Experiments were done between 9/1/16 and 7/12/17.

155
Prior to each experiment, crabs were weighed and anesthetized on ice for 10 156 minutes. Photoplesmogram (PPG) sensors (Vishay CNY70331) (Fig. 1), as described in 157 Depledge (1983), were placed on the carapace above the heart and pyloric muscles to 158 record the heart and pyloric rhythms, respectively. Sensors were secured to the carapace 159 using dental wax and cyanoacrylate glue (Starbond, EM-2000) and covered in Marine After a period of baseline (10C to 12C) recording, water temperature was 166 manipulated by flowing either cold or warm saline into the tank through a tube inserted 167 through the door of the incubator. A vacuum line was used to pump water out of the 168 tank to maintain a constant volume. Temperature was slowly ramped from 11C to 32C 169 over 1.5 to 2 hours. Heart rate was closely monitored to ensure health during the 170 temperature changes and ramps were halted once the heart rate developed an 171 arrhythmia or decreased to baseline frequencies, indicating that a 'critical temperature' 172 had been reached. Increasing temperatures past this critical temperature lead to death 173 of the animal, as the heart no longer recovered functionality.

174
Data Acquisition and Analysis 175 PPG data were acquired through the PPG amplifier (Newshift AMP03) and 176 recorded digitally through a digitizer (Axon Digidata 1550B) into computer software 177 (AxoScope 10.6) with a sampling frequency of 500 Hz. Data were analyzed using custom 178 written C and MATLAB codes.

179
Analysis of heart rhythm frequency 180 Heart rhythm frequency was calculated as the frequency at the peak spectral 181 power. We used the Burg (1967) method to estimate the power spectrum density at each  The fast-Fourier transform (FFT) spectrum is estimated using the previously calculated 185 AR coefficients. This method is characterized by higher resolution in the frequency 186 domain than traditional FFT spectral analysis, especially for a relative short time 187 window (Buttkus, 2000). We used the following parameters for the spectral estimation: The pyloric rhythm frequency was calculated in a similar way as the heart 198 frequency in those cases that showed no interference from heart activity. However, there 199 were instances in which the heart activity was influencing the pyloric rhythm. This was 200 obvious in the pyloric rhythm spectrogram as a peak in the power spectrum density at 201 the frequency of heart rhythm. To identify the intrinsic frequency of the pyloric rhythm 202 in cases with heart interference, a linear regression model was fit to the pyloric signal 203 taking into account the phase difference between the heart and pyloric signals. The 204 heart signal was multiplied by the coefficients of the regression model and subtracted 205 from the pyloric signal. Then spectrogram of the subtracted signal was calculated, and 206 the pyloric frequency was identified as the frequency at peak spectral power. In some 207 cases, the pyloric rhythm frequency could not be determined due to irregularities in the 208 signal.

209
Analysis of the inhibitory bouts 210 We used a hidden Markov model (HMM) to infer the active and inhibitory states 211 of heart rhythms. In HMM, a timeseries is modeled as being generated probabilistically 212 from an underlying discrete-valued stochastic process (Rabiner, 1989). The data can be 213 either discrete-or continuous-valued, while the unobservable 'hidden' state is a discrete 214 random variable that can take n possible values (in our case n=2, representing active 215 and inhibitory states). Estimation of the transition probabilities for HMM was done 216 using the Baum-Welch algorithm, which utilizes an expectation maximization (EM) 217 algorithm (Bilmes, 1998) . The initial parameters used for the detection: transition matrix PAI = PIA = 0.9, PAA = PII = 0.1, where PAI is transition probability from active to 219 inhibitory state, PAI is a transition probability from inhibitory to active state, Pii is a 220 transition probability from inhibitory to inhibitory state and PAA is a transition 221 probability from active to active state. Identification by HMM states was used to 222 calculate the durations of active and inhibitory bouts of heart rhythm.

226
Data were binned into two-minute bins, moved in 5 second steps. The time-bandwidth 227 product was set to 10, and 19 tapers were used. Peak coherence and frequency of peak 228 coherence were calculated for each window. The theoretical confidence level of the 229 coherence was calculated as following: significantly coherent heart and pyloric rhythms and were calculated for each data set.

235
The phase difference between the heart and pyloric rhythms was calculated at the 236 frequency of the peak coherence for each window and median value of phase difference 237 was reported for each dataset with significantly coherent signals.

238
Q10 estimation 239 We estimated the Q10 of frequency of the heart and pyloric rhythms in vivo. The goodness of fit of the linear regression model for each dataset was assessed by 245 calculating the coefficient of determination R 2 , calculated as R 2 = (correlation 246 coefficient) 2 . We report R 2 for heart and pyloric data in the tables below. For majority of 247 the fits we obtained high values of R 2 > 0.8. The critical temperature was defined as the temperature at which the heart and 252 pyloric movements became irregular and the frequency of muscle contraction 253 significantly dropped. This was determined from the spectrograms of heart and pyloric 254 rhythms. In some cases, the critical temperature of the pyloric rhythm was impossible 255 to determine due to irregularity in pyloric rhythm signal.

265
Heart and pyloric muscle movements of Cancer borealis were recorded in vivo 266 using photoplesmography (PPG) (Fig. 1). In most of the experiments described, both 267 rhythms were recorded from the same animals, although in some only the heart 268 rhythms were recorded. It is straightforward to place a PPG sensor over the heart 269 because the heart is dorsal, and situated just under the carapace, and its movements are  a stable high frequency pyloric rhythm. Animal 5 has a more variable pyloric rhythm. The 315 pyloric rhythm frequency decreases after the saline injection (indicated by vertical white line). 316 Note that the waveforms of the pyloric rhythms are more complex than the waveforms of the 317 heart rhythm due to the complex movement patterns of the pyloric muscles. 318 Figure 3 summarizes pooled frequency data for the heart from 49 animals and 319 the pylorus from 29 animals. In all cases, the data came from stretches of recordings in 320 excess of 30 minutes. All of the pyloric rhythm data came from animals that were also 321 used for heart measurements. The histogram in Figure 3A shows an apparent   Figure 3B shows a more normal distribution of the pyloric rhythm frequencies.

329
The spread in pyloric rhythm frequencies was much smaller than in heart frequencies, 330 ranging from 0.2 Hz to 1.6 Hz. In Figure 3C, we plotted the frequency of the pyloric 331 rhythm as a function of the heart rhythm for the 29 animals for which we had 332 measurements of both. Note that more than half of the points are not found close to the 333 identity line, suggesting that the pyloric and heart recordings are picking up rhythms in 334 the same general frequency range, but are not identical. Heart rhythms often displayed periods of bradycardia, during which the heart 345 considerably slowed or halted for a significant period (Fig. 4). We defined these periods 346 as inhibitory bouts using a hidden Markov model (procedure described in the methods 347 section). Periods of bradycardia were marked by a decrease in both amplitude and 348 frequency (Fig. 4) of heart rhythm PPG recordings by at least 33% that lasted at least 10 349 seconds. An example of a 24 hr recording with multiple inhibitory bouts can be seen in 350 Figure 4A. Termination of inhibitory bouts was associated with a return of amplitude 351 and frequency. A temporary increase in amplitude of the heart signal could sometimes 352 be observed immediately following the inhibitory bout (Fig A inset). Inhibitory bouts 353 were seen in 20/49 animals of the population tested (41%). In most cases, when 354 inhibitory bouts were seen, they occurred repeatedly over extended periods of time, 355 such as seen in the 24 hr recordings shown in Figure 4. Bout durations and frequency 356 were variable both across and between animals (Fig. 4B). The occurrence of inhibitory  show that the amplitude of the pyloric signal decreases during the inhibitory bout of the 381 heart ( Fig 4A). Spectral analysis revealed that the frequency of the pyloric rhythm 382 modestly decreases during the heart inhibitory bouts (Fig. 4 C). However, there is 383 considerable variability in the changes in frequency of pyloric rhythm, as can be seen 384 from the spectrograms calculated during single inhibitory heart bout in 4 animals (Fig 4   385   D). For example, the experiments shown in the first panel illustrates an example when 386 the pyloric rhythm was reliably moving with a frequency of 1.5 Hz while the heart was 387 beating at 1.6 Hz. In this animal when the heart temporarily stopped, the pyloric rhythm 388 slowed to 0.9 Hz. The second example also shows a strong decrease in pyloric frequency 389 during the heart inhibitory bout. The third example showed a transient increase 390 followed by a decrease, while the fourth example showed a slight increase in frequency.

391
Relationships between the heart and pyloric rhythms. Although it is clear that 392 the pyloric rhythm and heart rhythms are often at different frequencies, and the pyloric 393 rhythm continues during the inhibitory bouts, spectrograms of the pyloric rhythm 394 frequently reveal a band at the heart frequency. This is very clearly illustrated in the 395 third example in Fig. 4D, where the heart rhythm is seen as a tight band at about 1.2 Hz.

396
That same band is seen below in the pyloric rhythm traces. When the heart stops, the 397 heart band disappears from both recordings.

398
To look at the potential influence of the heart rhythm on the pyloric rhythm, we 399 calculated the time-frequency coherence between simultaneously recorded heart and 400 pyloric rhythms in 2-minute bins moved in 5 second steps. In 70% of the animals, the 401 heart and pyloric rhythms were significantly coherent at the frequency of the heart more 402 than 50% of the time during the baseline period. Because the pyloric frequency often 403 changes when the heart stops, this suggests that some kind of biomechanical coupling or 404 common drive is influencing the two structures. Figure 5 A illustrates an example of 405 such coupling in an individual animal. The coherence peaks at 1.1 Hz frequency (Fig. 5   406 A), which is the frequency of heart oscillations shown in the spectrogram of the heart 407 signal (Fig 5A). The spectrogram of the pyloric rhythm has two frequency bands: one at 408 the frequency of approximately 0.5 Hz, which is intrinsic frequency of the pyloric 409 oscillations, and another on the heart rhythm frequency. By calculating the phase at the 410 frequency of peak coherence we determined that the pyloric rhythm is shifted by 190 411 degrees relative to the heart rhythm in this animal. This can also be seen from the cross-412 correlation function, which has a minimum at 0.075 s lag. We also calculated auto-413 correlations for the heart and pyloric rhythms. The pyloric rhythm has a more complex 414 auto-correlation function than the heart rhythm featuring two peaks, one peak on the  Effects of temperature on heart and pyloric rhythms. We tested the effects of 450 increasing temperature on the heart movements of twelve animals and on pyloric 451 movements in nine animals (Fig. 6). Figure 6A shows raw data from a typical 452 experiment showing heart and pyloric movements during a temperature ramp from 11 o C 453 to 28 o C and then back to 11 o C. Figure 6B shows plots of the heart and pyloric rhythm 454 frequency for three animals as a function of the temperature. Frequency was calculated 455 for both heart and pyloric rhythms as the frequency at the peak of the power spectrum  ramp. The pyloric rhythm is less robust to the increases in temperature than the heart rhythm 477 and crashes at a much lower temperature. "Crash" is evident by significant decrease in 478 frequency and amplitude of the pyloric rhythm. C) Frequencies of the heart and pyloric rhythms 479 during increasing portion of temperature ramps plotted as a function of a temperature in a 480 logarithmic scale, each color corresponds to an individual animal. A line was fit to data points 481 for each animal's heart frequencies to estimate Q10s. D) Critical temperature of the heart 482 rhythm is significantly higher than of the pyloric rhythm (mean heart critical temperature Frequencies of the heart and pyloric rhythms are plotted as a function of 488 temperature in a logarithmic scale in Figure 6C. Linear models were fitted to the data 489 points for each animal to estimate heart and pyloric frequency Q10s. The pyloric 490 rhythms consistently crashed at lower temperatures than the heart rhythm in the same 491 animal. The critical temperature was defined as the temperature at which cardiac or 492 pyloric stability and regularity was lost, with a subsequent drop in contraction frequency 493 to near baseline values. Critical temperatures of the heart muscle movements were 494 collected for all twelve animals and for the pyloric muscle movements for nine animals 495 (Fig. 6D). The mean heart critical temperature was 25.0°C (s.d. = 1.62) and the mean 496 pyloric critical temperature was 19.1°C (s.d. = 2.76) (Fig. 6D). The critical temperatures 497 of the heart and pyloric muscle movements were significantly different (one-way 498 ANOVA, p=0.0005, F(1,19)=21.07). 499 Q10, a measurement of the rate of change of a biological process in response to a 500 change in temperature, was calculated for both the heart and pyloric rhythm for each 501 animal tested (Fig. 6E). The frequency of the heart and pyloric rhythms was plotted as a 502 function of environmental temperature in a logarithmic scale and linear regression 503 model was fitted into data points (Fig. 6E). Coefficients of determinations (R 2 ) for each 504 fit are shown in table in methods section. Q10s were calculated as slopes of linear 505 models. Q10s for heart rhythms, ranged from 1.3 to 4.2 with a mean of 2.007 (s.d. = 506 0.854). Pyloric rhythm Q10s ranged from 1.33 to 2.7 with a mean of 2.04 (s.d. = 0.47).

507
The Q10s of the heart and pyloric rhythms were not significantly different as determined 508 by one- way ANOVA (p=0.127, F(1,19)=2.54). 509 Finally, we calculated the coherence between the heart and pyloric rhythms 510 during the temperature ramps to determine whether temperature perturbation affects 511 the relationship between signals. An example time-frequency coherence from an 512 individual animal is shown in Figure 7A. The coherence peaks at the frequency of heart 513 oscillations at baseline as well as during the temperature ramps. Examples of coherence 514 and phase differences calculated at different stages of the experiment (baseline, rising 515 phase of temperature ramp, and decreasing phase of temperature ramp) show that the 516 amplitude of the coherence remains high throughout the whole experiment and the 517 temperature perturbation does not affect the phase relationship between rhythms. In 518 this example the rhythms oscillate in phase (phase difference at the frequency of 519 maximal coherence is shown by arrows in Figure 7C). Peak coherence at baseline and 520 during the rising phase of the temperature ramp as well as the percent of time the The crustacean heart is neurogenic, and therefore the activity of the cardiac 568 ganglion directly regulates the heart beat frequency. The CG must be able to produce 569 activity regularly and across a wide range of perturbations. It must be modifiable to Robertson and Money, 2012) and reliable enough to ensure that the heart continues to 572 pump hemolymph at all times. Both frequency and contraction amplitude, two factors 573 which must be regulated to ensure the animals' success, are expressions of the activity of 574 the CG, specifically its interburst interval, rate, and firing patterns. 575 We captured heart activity using PPG recording techniques while limiting the 576 invasiveness and stress placed on the animal. PPG recordings of the heart musculature 577 were reliable over time and showed differences in frequency of heart beats within the 578 population (Fig. 2). In the majority of experiments, an animal's baseline heart rate was 579 relatively stable (Fig. 2), but in some animal's frequency switches were spontaneously 580 seen ( Fig. 2A, animal 3). During baseline recordings, animals were not subjected to 581 changes in temperature, light, salinity, stress or environmental factors that could 582 interact with the metabolic needs of the animal or the functions of the CG. During this 583 time, the firing of the CG, and therefore the activity of the heart, would not be expected 584 to appreciably change. Nonetheless, the heart rate of resting animals falls into a 585 multimodal distribution. These data suggest that heart activity, and therefore activity of 586 the CG, falls into states of higher or lower activity. These states may be a consequence of 587 variable metabolic needs of the animal during the time of recording, such as digestion, 588 movement, or excretion. The CG may therefore have mechanisms for switching between 589 high and low activity states through neuromodulation or extrinsic neural input.  (McMahon, 1999). This is likely due to overflow of ocean 600 water, and therefore oxygen, in the gills, causing the organism's heart to stop for a 601 significant period. It is interesting, however, that this occurs in only about 41% of 602 animals tested. This implies a mechanism affecting both the heart and the gills that  In the experiments presented here, PPG sensors recording the muscle movement 615 of the pylorus were placed on the carapace above the dorsal dilator muscle. While heart 616 rhythm waveforms were relatively simple, with one maximum and minimum, pyloric 617 rhythm waveforms were more complex (Fig. 2B). The complexity of the PPG waveform 618 is likely due to the positioning of the muscles being recorded and their movements in 619 relation to the PPG sensor. In vivo, the pyloric rhythm frequency drifts more than that 620 of the heart rhythm frequency over a baseline period. Across the population of tested 621 and analyzed animals, the pyloric rhythm had a more normal distribution, unlike that of 622 the heart rhythm (Fig. 3).

623
Comparison of the Heart and Pyloric Rhythms 624 While the generation and movement of the heart and pylorus have been 625 extensively studied separately in the past, here we examined the potential interactions 626 between these two central pattern generated movements. To determine how these two 627 essential rhythms may interact in a single animal, we calculated time-frequency 628 coherence between the rhythms of animals in controlled environments. In a majority of 629 animals, the heart and pyloric rhythms were coherent at the frequency of the heart 630 rhythm (Fig. 5A). We know that this coherence is not simple cross-talk between the 631 sensors themselves, as the pyloric rhythm continues when the heart stops (Fig. 4C,D), 632 and because there are numerous instances when the two rhythms are quite different. 633 We suspect that the coherence between the two rhythms is a bio-mechanical coupling In the data presented here, it is clear that temperature affects both rhythms and 650 increases their overall frequency, with similar Q10s (Fig. 6E). This indicates that 651 increases in temperature cause similar changes in the two frequencies, until a crash 652 point occurs. Interestingly, the heart is more robust to extreme temperature changes 653 than the pyloric rhythm.

654
Crabs can live for days and weeks without eating, but presumably cannot survive 655 for extended periods of time without hemolymph oxygenation and circulation. From 656 this perspective, it is easy to justify the fact that the critical temperature for the heart 657 rhythm is higher than that for the pyloric rhythm. The additional 4-5 o C might make a 658 big difference for an animal caught in shallow water during the summer, and give it time 659 to find its way to more hospitable environments. Interestingly, the mean critical