Open Access
Add to BibSonomy
Abstract
Intracellular oxygenation is an important parameter for numerous biological studies. While there are a variety of methods available for acquiring in vivo measurements of oxygenation in animal models, most are dependent on indirect oxygen measurements, restraints, or anesthetization. A portable microscope system using a Raspberry Pi computer and Pi Camera was developed for attaching to murine dorsal window chambers. Dual-emissive boron nanoparticles were used as an oxygen-sensing probe while mice were imaged in awake and anesthetized states. The portable microscope system avoids altered in vivo measurements due to anesthesia or restraints while enabling increased continual acquisition durations.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
The negative effects of hypoxia on tumor aggressiveness and therapeutic response are well established. In radiotherapy, hypoxic regions (defined as pO2 less than 10mmHg) are up to ∼3x more radioresistant than well-oxygenated areas [1,2]. Hypoxia is partially caused by insufficient/chaotic vasculature, resulting in regions that exhibit poor drug delivery [3–5]. Mitigating hypoxia requires preclinical models that exhibit repeatable, translatable, and quantitative techniques for measuring hypoxia [4,6–8]. A significant challenge in comparing hypoxia studies is that most fail to report physiological measurements despite anesthetizing the animal and causing major disturbances to normal biological functions. Due to changes in perfusion [9,10], vasodilation/constriction [11,12], oxygen consumption rates, and increased/decreased heart/breathing rate, anesthesia significantly alters tissue-oxygen measurements [13,14], though it is unclear to what extent.
In pre-clinical mouse studies, anesthesia is commonly used during imaging, cancer-cell (or drug) injections, and surgery. Anesthesia can be delivered via injection (e.g. ketamine or pentobarbital) or via inhalation (e.g. isoflurane or sevoflurane) [15]. The American Veterinary Medical Association, the Institute for Laboratory Animal Research, and the Cornell University Institutional Animal Care and Use Committee (IACUC) all recommend 1%-5% isoflurane combined with either 21% O2 or 100% O2 [15]. Isoflurane, though arguably the most common and easy-to-use anesthesia method, causes both respiratory and myocardial activity depression while increasing carbon dioxide and bicarbonate in arterial blood [16,17]. These physiological changes affect the measurable outcomes of many animal studies, notably those that involve the vascular system or hypoxia studies.
Several methods have been developed to circumvent the need for anesthesia in preclinical models, including nanoprobes and electron paramagnetic resonance imaging (EPR). A two-photon phosphorescence probe (PtP-C343) was developed and used in conjunction with two-photon phosphorescence lifetime microscopy to measure pO2 values in the mouse brain [18]. Researchers report physiologically accurate and precise values of pO2 in the vasculature and interstitium of mouse cerebral grey matter. The main limitation in this type of optics-based method is the invasiveness required to mitigate photon attenuation in deep tissue and the requirement for specialized phosphorescence lifetime imaging equipment and need for animal restraint during the measurements.
In that murine brain study, mice were outfitted with cranial windows and were habituated over the course of a week to a custom-made head restraint for data acquisition; the stress inherent to such restraint would be a potential source of bias [18]. An earlier study using EPR spectroscopy reports repeatable, noninvasive pO2 measurements in murine brains without the use of anesthesia [19]. A spin probe was injected into the brain and a resonance loop was attached to the mouse’s head to make pO2 measurements [19]. This is a direct measurement technique, and EPR has been expanded in use over recent years [20,21]. However, it is not readily available at this time [19].
We present a novel method for directly measuring tissue pO2 in awake mice using dual-emissive boron nanoparticles (BNPs) and a portable microscope system attached to the dorsal window chamber. Difluoroboron β-diketonates are a class of fluorophores with excellent optical properties, exhibiting large extinction coefficients, high quantum yields, and tunable emission colors [22,23]. While the fluorescent signal is insensitive to the presence of oxygen, phosphorescent decay changes in response to environmental oxygen through collisional quenching. The ratio of fluorescence to phosphorescence can be correlated to the amount of oxygen, enabling BNPs to function as a ratiometric oxygen-sensing probe. These ratiometric probes enable environmental effects like lamp intensity fluctuations and inter-animal variations to be accounted for.
In this study, we prepared the oxygen sensor by fabricating stereocomplexed nanoparticles from BF2nbm(I)PLLA-PEG and PDLA-PEG. The Boron NanoParticles (BNP) show well-separated fluorescence and phosphorescence peaks, good alignment with RGB channels, and 0-100% oxygen sensitivity. The portable microscope is a customized Raspberry Pi camera and computer; customizations include low-cost, lightweight, 3D printed apparatuses and small optical filters. This system allows mice to be awake and unrestrained during imaging. In this pilot study, we investigate if our system is capable of monitoring oxygen in mice both awake and anesthetized via isoflurane inhalation. We hypothesize that this system can detect and quantify changes in tissue pO2 in mice inhaling various oxygen concentrations while awake or anaesthetized.
2. Methods
2.1 Boron nanoparticle preparation
A full-range (0-100%) oxygen-sensor, boron dye (BF2nbm(I)PLLA-PEG) was prepared for this work (Fig. 1). This was achieved by preparing a phenol-functionalized dye for a postpolymerization modification reaction via Mitsunobu coupling. This method is effective at achieving PEGylated polylactide nanoparticles. The small molecule precursors and polymers were characterized by nuclear magnetic resonance (NMR) and size-exclusion chromatography. Boron nanoparticles were prepared by nanoprecipitation [24]. In this study, a block copolymer of BF2nbm(I)PLLA-PEG (Mn = ∼7 kDa) and a mPEG-PDLA (Mn = ∼9 kDa) were used to form stereocomplexed nanoparticles as previously described (RH = 36.9 nm, PD = 23.8%). It has been demonstrated that PEGylated stereocomplexed nanoparticles also exhibit higher aqueous stability [25]. As shown in Fig. 1(C), boron nanoparticle luminescence is highly sensitive to oxygen concentration, and over the full range. Detailed BNP preparation materials can be found in a previous publication [26].
Fig. 1. (A) Chemical structures of boron nanoparticle material. (B) Nanoprecipitation to prepare the nanoparticles. Polymers were dissolved in a DMF (10 mg/mL) then precipitated into rapidly stirring water (DI) to yield a ∼1 mg/mL solution of nanoparticles. (C) Images of the BNPs in air (∼21% oxygen).
2.2 Raspberry Pi mobile imaging system
The portable microscope system is controlled by a Raspberry Pi 3 computer (RASPBERRYPI3-MODB-1GB, Raspberry Pi) with a ribbon cable connecting a Raspberry Pi Camera (RPI-CAM-V2, Raspberry Pi). The camera’s default specifications include a fixed-focus lens, a pixel size of 1.44µm2, a sensor resolution of 3230 × 2464 pixels, a CMOS image area of 3.68 × 2.78mm, an exposure time range of 1µs-10s, a F-stop of 2.0, and a bit depth of 8-bits per color channel. For our purpose, a custom-machined lens spacer was added to increase the distance between the lens and CMOS detector, decreasing the focus distance from ∼1m to 5mm (while also increasing the magnification). The Raspberry Pi’s general-purpose input/output pins (GPIOs) are used to control a white LED (941-C513A-WSN-CY0Z0231, Mouser Electronics) for brightfield imaging and an ultra-violet LED for fluorescence/phosphorescence excitation. Python (v2.7) scripts controlled the LEDs and Pi Camera.
The system contains two filters: one bandpass filter (264805, Chroma) centered around the excitation range of the UV LED (405 ± 20nm) and a longpass, emission filter (450-750nm) (288747, Chroma). A custom, 3D-printed holder positioned the emission filter in front of the camera. Similarly, 3D printed panels support the LEDs and excitation filter around the dorsal window chamber. A schematic of the system and an image of the portable microscope connected to a dorsal window chamber are shown in Fig. 2.
Fig. 2. (A) A diagram of the portable microscope connected to a window chamber with key components labeled. When obtaining brightfield images with a white LED, filters are not used. For setups involving excitation LEDs, the bandpass filter narrows the wavelengths of light that reach the window chamber, while the longpass filter blocks the excitation wavelengths from being detected by the Pi Camera (while allowing signal from the window chamber to pass on to detection). (B) Drawing of how a simplified system fits onto the dorsal window chamber. (C) The portable microscope system connected to the dorsal window chamber of an awake and freely moving mouse, with Pi Camera and LED panel labeled.
2.3 Boron nanoparticle calibration
To calculate tissue %O2, first the RGB images of UV-excited BNP were converted into oxygen concentration maps. Calibration curves require images of BNP at known oxygen concentrations. At the start of calibration, a cuvette was filled with deionized water for flat field imaging of any light leakage or dark noise. Because these background images cannot be easily obtained while the system is attached to a mouse, the Python code automatically acquired a range of background images at set exposure times. After obtaining background images, the cuvette was filled with a 1:10 solution of BNP:deionized water. Parafilm was wrapped on the cuvette to prevent room air from diffusing into the sample while set %O2 levels were delivered. For delivery of known oxygen concentrations, two GFC mass flow controllers (GFCS-010077, Aalborg) and a MaxO2 oxygen sensor (OM25ME, MaxTec) were connected by gas tubes to a supply of O2 and N2. After mixing the gases, the MaxO2 was used to measure the %O2. With the setup complete, 0%O2 (100% N2) was delivered to the cuvette for 10 minutes. This process was repeated up to 100% O2. Figure 3(A) display images of the UV-excited BNP in the cuvette at increasing oxygen concentrations.
Fig. 3. (A) Images of BNP emissions at various oxygen concentrations shows a color change as oxygen levels increase. The images were analyzed with Matlab to translate the RGB color images of BNP emissions into F and P components (B). These components were then used, together with the known oxygen concentrations, to make a quadratic calibration curve (C).
With calibration images obtained, MATLAB (2018b) was used for creating the required calibrations. For each delivered %O2 calibration image, the background image of the corresponding exposure time was subtracted. Next, a 2D 5 × 5 pixel median filter was applied to decrease noise followed by normalizing each image by their respective exposure time. Intensity-based ROIs were created for each image to focus calculations on regions of UV-excited BNP.
Knowing that the acquired RGB data is the summed result of the underlying factors of fluorescence (F) and phosphorescence (P), the first calibration required is one that disambiguates the contribution of each of these factors to the individual RGB channels. For each delivered %O2, 200 RGB pixels were randomly selected from within the ROI. Parallel factor analysis was performed on all selected pixels by the MATLAB function PARAFAC [27]. PARAFAC was initialized to search for two positive underling factors, generating a relative loading matrix that can convert RGB channels to F/P channels (Fig. 3(B)). A quadratic fit was performed on the delivered %O2 vs the mean F/P from all calibration images (Fig. 3(C)), producing the following quadratic Eq. (1) where we solved for %O2:
With both calibrations created, BNPs can be used as an in vivo oxygen-sensing probe. During in vivo measurements, background images of the window chambers are obtained prior to injection of BNP. The relative loading matrix converts RGB images of UV-excited BNP into F/P images. The F/P images are fed into the quadratic equation to obtain oxygen concentration maps.
2.4 Animal methods
Five female, nude mice were obtained from the Duke Breeding Core for use in this study. All procedures performed with animals were in accordance with the Institutional Animal Care and Use Committee Guidelines at Duke University. All five animals were surgically outfitted with skinfold-dorsal window-chambers. The window chamber surgery is described by Palmer, et al. [28], but a brief description follows.
Mice were anesthetized with 100mg/kg of ketamine and 10mg/kg xylazine intraperitonially for surgery, and 1 mg/kg of buprenorphine SR was injected subcutaneously for pain management. After anesthetization, dorsal skin was spread by suturing the skinfold to a custom machined C-holder. With dorsal skin spread, the front flap of skin was removed for the window and holes were punched through the skin for attaching the titanium window. Saline was used to clean/fill the window chamber, while a glass cover slip and metal retaining ring were used to enclose the window tissue. Sutures connected the top of the dorsal skin to the window chamber. Finally, the mouse was removed from the C-holder and given time to awaken from anesthesia. At the end of the study, mice were humanely sacrificed.
All mice were allowed 24h recovery post-op before being entered into the study.
2.5 Physiological measurements
Internal temperature, arterial pO2, breathing rate, and heart rate were all monitored while the mouse was anesthetized and awake. The rectal probe for measuring temperature was only used during anesthetized measurements. A pulse oximeter collar sensor (MouseOx Plus, Starr Life Sciences Corp.) was placed around the mouse’s neck to obtain signal from the carotid arteries. This sensor is designed to provide data on mice awake or anesthetized in a prone position. The sensor was connected to the MouseOx Plus software (v 1.6) for recording arterial pO2, breathing rate, and heart rate. In the anesthetized state, the signal was strong and continuous; however, while awake, the signal was temporarily lost if the mouse was moving too rapidly.
The rectal temperature sensor (MouseOx Plus, Starr Life Sciences Corp.) was used while the mouse was anesthetized. The mouse was kept on a water-circulating heating pad (Gaymar TP 650 T/Pump/Pad) at a constant 37°C while the mouse was both awake and asleep.
2.6 Autofluorescence photobleaching
The autofluorescence in the window chamber was minimized by photobleaching with UV light prior to the experiment. For at least one hour, the window was continuously and directly exposed to UV LED radiation. During this time, the mouse was lightly anesthetized (1%-3% isoflurane with 100% O2) and kept warm while the UV LED was attached to the cover-slip side of the window. The UV LED is the same used during image acquisition to excite the BNP.
2.7 Raspberry PI image acquisition
After anesthetization under 1.5% isoflurane in 100% O2, the camera and white LED were attached to the window chamber. Using a live image under the white LED, the focus was manually adjusted, lengthening or shortening the camera lens. Next, a brightfield image was acquired (no filters in place) to visualize the anatomy within the window chamber. The white LED panel was removed and replaced by filters with the UV LED panel. Autofluorescent (background) images were obtained at a variety of exposure times for background subtraction; a range of pre-BNP images (200-2000µs, 50-100µs increments) was required as the post-BNP exposure time (camera shutter speeds) was unknown at the time and retroactive background images (a day or so later) would have been a less accurate representation of the actual background noise during the experiment. The UV panel was temporarily removed to inject 50µL of BNP through the back of the window between the cover glass and tissue surface. Imaging commenced after 20 minutes, allowing for the BNP to saturate into the tissue.
Each mouse was imaged while awake in a small, isoflurane induction chamber repurposed as an enclosure that the animal could freely move in while the environmental oxygen concentration was tightly controlled. While the chamber’s lid was slightly opened for the camera’s wires, the input rate of the oxygen (>2L/min) was sufficient to equilibrate the chamber to the desired oxygen concentration. This was confirmed with an oxygen monitor in series with the chamber’s output port.
For anesthetized measurements, mice were anesthetized with 1%-3% isoflurane and kept warm on the heating pad. The isoflurane strength was dynamically adjusted to keep the breathing rate between ∼80bpm and ∼110bpm.
For both awake and anesthetized measurements, mice breathed the same concentrations for the same amount of time. The mice were randomly chosen to be imaged awake or anesthetized first. Every ten minutes, beginning with 100% oxygen, the oxygen concentration was decreased via increasing nitrogen components: 100% O2, 50% O2, 20% O2, 17% O2, 15% O2, 13% O2. Images were acquired once per minute, and the physiological measurements (breathing rate, heart rate, temperature and arterial pO2) were acquired continuously throughout the experiment. All images were analyzed in the same manner as the calibration images described in Section 2.3
2.8 Statistical analysis
A 2-way ANOVA determined if anesthesia or environmental %O2 had a significant effect on physiological parameters: arterial %O2, breathing rate and heart rate. Tukey’s Post Hoc test determined the significance between specific data points. A multivariable linear regression analysis was used to determine if any of the physiological measurements were linearly related to tissue %O2. The tissue %O2 is a continuous, dependent variable while the breathing rate, heart rate, environmental oxygen (100% O2, 50% O2, etc.) and arterial %O2 are continuous, independent variables. The animal’s state, anesthetized or awake, was treated as a categorical, independent variable. Subjects (mice) were included as a categorical, independent variable. Finally, 2-way interactions between the physiological parameters, the environmental %O2 and the anesthesia state were accounted for. The multivariable linear regression was only considered valid if the residuals passed the Shapiro-Wilk test for normality (p > 0.05).
All p-values < 0.05 were considered significant. All statistical analyses were performed in GraphPad Prism (v 9.0.2, 161).
3. Results
Figure 4 shows an example of the three image types we acquired. The first (Fig. 4(A)) is the brightfield image under white light to visualize anatomy. The second image (Fig. 4(B)) is the emission pattern of UV-excited BNP. The third image (Fig. 4(C)) is acquired by analyzing the BNP image in Matlab and, using the calibration curve, translating the images into oxygen maps. Note that BNPs do not extravasate readily into blood vessels, so they appear hypoxic (blue) while the surrounding tissue is well-oxygenated (yellow).
Fig. 4. (A) For each mouse, a single anatomical (brightfield) image was recorded under white light. (B) The UV-excited image shows the green emission spectrum of excited BNP. (C) Once the images have been analyzed, they are translated into an oxygen map. Areas of low oxygen concentrations are blue while the well-oxygenated regions surrounding the vasculature are yellow.
Additional oxygen maps are shown in Fig. 5(A). There is a clear, visual difference between the awake and anesthetized images where, for this mouse, the tissue was notably more well-oxygenated while awake and breathing 100% O2 compared to anesthetized. Conversely, there is little difference between awake and anesthetized at 13% O2. Figure 5(B) and (C) show the difference in tissue %O2 as the mouse breathes decreasing oxygen concentrations. The colors of the symbols and lines in the graphs denote specific mice as they move from anesthetized states. In Fig. 5(D)-(F), the physiological measurements are showed, summarizing all mice – both anesthetized and awake – as they experience decreasing environmental %O2. In the supplementary document, Fig. S1 shows more explicitly the inter-mouse variability in arterial %O2, the breathing rate and heart rate for each mouse for each state (awake or anesthetized).
Fig. 5. (A) The tissue increases in oxygenation (yellow) as the mouse breathes higher %O2 concentrations. There is a visual discrepancy between the anesthetized and awake tissue %O2. (B) and (C) Each mouse, of different colors/symbols, has a different tissue %O2 that depends on both the environmental %O2 and the anesthetized state. (D) The arterial %O2 has no significant difference between awake and anesthetized states; however, the environmental %O2 has a significant effect on arterial %O2 (p < 0.0001, 2-way ANOVA). (E) Similarly, the heart rate is not significantly different between states but does significantly change based on environmental %O2 (p = 0.0221, 2-way ANOVA). (F) As expected, the breathing rate changes dramatically when the mouse is awake compared to anesthetized (p = 0.0369, 2-way ANOVA and Tukey’s Post Hoc test). Environmental %O2 also has a significant effect when tested within mice that are all anesthetized or all awake (p = 0.0027, 2-way ANOVA).
A 2-way ANOVA showed that for all physiological parameters, the amount of oxygen in the environment had a significant effect: arterial %O2 – p < 0.0001, heart rate – p = 0.00221, and breathing rate – p = 0.0227. Light anesthesia had no significant effect on these parameters, except for breathing rate.
A multivariable linear regression analysis (Fig. 6) shows that the interaction between breathing rate and anesthesia significantly changes the value of the tissue %O2, p = 0.0107. Mouse 2 also significantly affected the model outcomes; referring to Fig. 5 and Fig. S1, this mouse’s tissue %O2 did not change in response to changing environmental %O2, while its physiological parameters were comparable to the other mice. For this analysis, the tissue %O2 was treated as the dependent variable, while all other physiological parameters were treated as independent, continuous variables. Anesthesia state was treated as a categorical variable while subjects (mice) were also treated as categorical variables. Figure 6 shows the parameter covariance plot and the tissue %O2 prediction plot. Below the parameter covariance plot is a description of the variables, their respective estimate in the linear model and their p-values. All continuous variables passed the Shapiro-Wilk test for normality of residuals (p = 0.0791). Figure 6(B) and (C) show the goodness of the model; the overall R2 is 0.8224.
Fig. 6. (A) The parameter covariance plot shows the R2 correlations between the parameters with dark blue being highly positively correlated and dark red being highly negatively correlated. The table below describes how the parameters are labeled, their estimations in the linear model, and their respective significant (or insignificant) effects. (B) The actual versus predicted tissue %O2 with its line of identity shows that, though there is some noise, there is a correlation between the multivariable linear model and the actual data. The overall R2 for this model is 0.8224. (C) While the residual plot is not perfectly randomly distributed around zero, it still shows a lack of a heteroscedastic pattern and no outliers. This, along with passing the residual test for normality, suggests that the linear regression analysis is appropriate.
4. Discussion
4.1 Portable Raspberry Pi microscope
The portable microscope was well-tolerated by the mice and did not notably hinder movement. Currently, the unit weighs 8g, which is ∼40% of an average mouse’s body weight. This weight is counterbalanced by holding wires above the mouse, reducing the weight on the mouse. For future studies where movement and behavior are important endpoints, a ballast system should be implemented to reduce weight. Alternately, a wireless design without the ribbon cable and LED wires would provide more freedom of movement. One potential wireless design is using the electric components in capsule endoscopy [29]. With a wire-free system, data could be acquired continually for days without constant supervision.
The spatial and temporal resolution of the CCD array is high: the pixels are 1.44µm2, which can resolve some microvasculature and larger cells. The exposure time can be decreased to 200µs. Additionally, standard deviation of the %O2 calculated from all sample pixels for each delivered %O2 during calibration suggests that oxygen measurements have an uncertainty of ± 1.5% O2 (± 10mmHg).
Once we bleached the autofluorescence, we considered scattering and absorption effects on the measurement of O2% to be negligible. Because we rely on a ratiometric approach, only absorption and scattering effects which differ in each of these components would have an effect on the measurement of O2%. The dominant chromophores between 450nm and 600nm are hema- and myoglobin [30]. We were careful to exclude obvious vasculature in our ROI analysis, and the BNPs did not perfuse into the vasculature within the duration of the experiment. It is certain that some signal was lost to absorption and scattering, but as the BNPs were applied topically, resting above the tissue chromophores, these effects were less apparent on the extracted O2%. Though we discarded these effects, any light-based microscope system will lose appreciable signal to absorption and scattering, and when these effects significantly reduce the SNR, they must be accounted for.
4.2 Effects of environmental %O2
Tissue pO2 values calculated from BNP emissions cover a large gamut of values. Because of the scarcity of publications that include absolute tissue measurements, there is little to compare our values to. For awake mice inhaling 100% O2, the mean tissue pO2 was ∼29mmHg (range 2.76-61.8mmHg). This range is comparable to oxygen measurements between 0-20µm away from arterial vessels in a study of oxygen diffusion in rat mesentery tissue (∼20mmHg at 10µm distance) [31]. The same study provides the blood-tissue interface and subsequent diffusion for low-oxygen vasculature: ∼80% decrease in pO2 at 20µm from blood/vasculature barrier (beginning pO2 around 25mmHg). In our data, the average change in tissue %O2 from 100% down to 20% environmental oxygen (still in awake mice) is -11.8%. Our %O2 values are subject to variation between mice because the placement of the image ROI in relation to vasculature was inconsistent. While our camera can resolve large vessels with ease and visualize some microvessels, we are unable to determine where small arterioles, venules, and capillaries are in relation to our ROI. Higher tissue %O2 might be due to close relation to an artery (perhaps even including the artery) while lower tissue-oxygen values might be due to ROIs placed at a greater distance from vessels. Without multimodal imaging with high-powered optical microscopes (during which the animal must be anesthetized), it is impossible to analyze the BNP emission images consistently in awake mice.
The effect of environmental %O2 was apparent in the physiological parameters as each parameter exhibited a significant change as the environmental %O2 changed, both within the anesthetized data set as well as within the awake data set (treated as independent datasets). However, there are several mice that do not appear to acclimate consistently to the decreasing environmental %O2 (Fig. 5(E)-(F)). 2/5 mice exhibit increasing heart rate with decreasing oxygen – both awake and anesthetized. The remaining three mice exhibit decreasing heart rates in response to a low-oxygen environment. Cowburn, et al, performed a cardiovascular study in mice where they exposed the mice to hypoxic environments over 48h and recorded temperature, blood pressure, heart rate and breathing rate [32]. They found, as is supported by other literature [33–36], that heart rate increases dramatically and acutely (within 10min) in hypoxic environments: 606 to 717 BPM. Our data (in the 2/5 awake mice that exhibit an increased heart rate) shows a median heart rate increase of 210 BPM. The other three mice exhibit a -122 BMP decrease. Interestingly, the same mice in the 48h study by Cowburn, et al, show that following the acute increase in heart rate, it dramatically decreases (-53% from baseline) [32]. Our low-heart rate mice experienced ∼60min of decreasing oxygen conditions (∼20min of hypoxia exposure) and a -13% decrease in heart rate compared to baseline. Cowburn, et al, found that their decreased heart rate correlated with increasing peripheral vascular resistance and decreasing physical activity [32]. Therefore, the deviation between our mice might be due to the natural inter-mouse variation in physiological responses to hypoxia.
The increase in breathing rate as %O2 decreases (average +32 breaths per min) in awake mice is supported by literature; it is common for hypoxic conditions to induce rapid, shallow breathing [32,37,38]. In anesthetized mice, the breathing rate is a direct correlation to the depth of anesthesia and not necessarily hypoxic conditions; thus, the breathing rate (and the subsequent significant difference between anesthetized and awake breathing rates) are due to anesthesia.
4.3 Relationship between anesthesia, physiological parameters and tissue %O2
There is a consistent, significant difference between awake versus anesthetize mice that may not solely be due to anesthesia (Fig. 5(B)-(C)). The two experiments (awake versus anesthetized) occurred consecutively, and the different tissue %O2 values are likely due to a combination of all the physiological parameters as they change in response to isoflurane. A multivariable linear regression model showed that the interaction between breathing rate and anesthesia significantly affected tissue oxygen levels (p = 0.0107). The arterial %O2 variable is approaching significance at p = 0.0787; a higher powered experiment with more subject may clarify whether this has a significant predictive value in tissue %O2.
The physiological impact of isoflurane suggests that there are multiple factors within our “anesthetized” variable that could contribute to the change in tissue %O2. Lyons, et al, measured the effect of isoflurane (compared to awake mice) in the brain, and found that awake cortex, capillary and interstitial pO2 values were nearly half that of when the mice were anaesthetized under isoflurane [18]. In a PET study measuring the difference on mouse physiology between isoflurane when breathing 100% O2 or room air, Mahling, et al, report that, when breathing 100% oxygen, tumor and muscle pO2 is increased (6.25mmHg vs 0.45mmHg in tumor and 95.64mmHg vs 56.85mmHg in muscle) [39]. Our results require a much higher number of subjects and physiological measurements to fully describe the effect of isoflurane on tissue pO2; however, it is important to be aware that anesthesia significantly affects tissue oxygen levels.
The overall system, including both the MouseOx for measuring respiratory rate, etc. and the camera, worked perfectly in tandem. Should the system be expanded to non-window chamber applications where respiration and heart rates cause significant artifacts, the MouseOx would be insufficient for gating purposes. We have noted the importance of physiological variables on functional imaging, but they are also important to minimize in in vivo anatomical imaging. Ventilation, though invasive, is reliable for respiratory gating. Heart rate gating, on the other hand, is much more complex and requires very fast electronic responses and imaging acquisition times [40].
5. Conclusion
As hypoxia remains a major hinderance in good therapeutic outcomes for most cancers, preclinical techniques for quantifying hypoxia are paramount to tackling clinical problems. Our development of the portable microscope system shows promise as a method for acquiring accurate pO2 measurements without the measurement-altering effects of anesthesia. By avoiding anesthesia, continual measurements of pO2 can be extended from hours to days, allowing for studies of long-term trends.
However, it is unreasonable to suppose that preclinical hypoxia studies can continue without anesthetics. At this point, it is important to simply be aware that isoflurane does significantly affect tissue oxygen (along with breathing rate and likely a whole host of other parameters) and to account for it in hypoxia studies.
Funding
National Institutes of Health (R01 CA167250); UVA Cancer Center (P30 CA44579).
Acknowledgments
We utilized the Duke Optical Molecular Imaging and Analysis shared resource at Duke, which is supported by the Duke Cancer Institute and Duke School of Medicine.
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request. This limitation is due to the large size of the dataset (on the order of gigabytes) where secure online hosting for large datasets is dependent on subscription fees that Duke University does not support at this time.
References
1. S. Rockwell, I. T. Dobrucki, E. Y. Kim, S. T. Marrison, and V. T. Vu, “Hypoxia and radiation therapy: past history, ongoing research, and future promise,” Curr. Mol. Med. 9(4), 442–458 (2009). [CrossRef]
2. D. Hennessey, L. M. Martin, A. Atzberger, T. H. Lynch, D. Hollywood, and L. Marignol, “Exposure to hypoxia following irradiation increases radioresistance in prostate cancer cells,” Urol. Oncol. 31(7), 1106–1116 (2013). [CrossRef]
3. M. W. Dewhirst, Y. Cao, and B. Moeller, “Cycling hypoxia and free radicals regulate angiogenesis and radiotherapy response,” Nat. Rev. Cancer 8(6), 425–437 (2008). [CrossRef]
4. C. T. Lee, M. K. Boss, and M. W. Dewhirst, “Imaging tumor hypoxia to advance radiation oncology,” Antioxid. Redox Signaling 21(2), 313–337 (2014). [CrossRef]
5. M. C. Skala, A. Fontanella, L. Lan, J. A. Izatt, and M. W. Dewhirst, “Longitudinal optical imaging of tumor metabolism and hemodynamics,” J. Biomed. Opt. 15(1), 011112 (2010). [CrossRef]
6. S. Matsumoto, H. Yasui, J. B. Mitchell, and M. C. Krishna, “Imaging cycling tumor hypoxia,” Cancer Res. 70(24), 10019–10023 (2010). [CrossRef]
7. A. A. Manzoor, H. Yuan, G. M. Palmer, B. L. Vigilanti, and M. W. Dewhirst, “Imaging hypoxia,” in Molecular imaging: Principles and practice (People’s Medical Publishing House, 2010), pp. 756–780.
8. S. K. Chitneni, G. M. Palmer, M. R. Zalutsky, and M. W. Dewhirst, “Molecular imaging of hypoxia,” J. Nucl. Med. 52(2), 165–168 (2011). [CrossRef]
9. H. Toyama, M. Ichise, J. S. Liow, D. C. Vines, N. M. Seneca, K. J. Modell, J. Seidel, M. V. Green, and R. B. Innis, “Evaluation of anesthesia effects on [18f]fdg uptake in mouse brain and heart using small animal pet,” Nucl. Med. Biol. 31(2), 251–256 (2004). [CrossRef]
10. S. K. Woo, T. S. Lee, K. M. Kim, J. Y. Kim, J. H. Jung, J. H. Kang, G. J. Cheon, C. W. Choi, and S. M. Lim, “Anesthesia condition for (18)f-fdg imaging of lung metastasis tumors using small animal pet,” Nucl. Med. Biol. 35(1), 143–150 (2008). [CrossRef]
11. C. Constantinides, R. Mean, and B. J. Janssen, “Effects of isoflurane anesthesia on the cardiovascular function of the c57bl/6 mouse,” ILAR J 52, e21–31 (2011).
12. C. Constantinides, S. Angeli, and R. Mean, “Murine cardiac hemodynamics following manganese administration under isoflurane anesthesia,” Ann. Biomed. Eng. 39(11), 2706–2720 (2011). [CrossRef]
13. C. Constantinides and K. Murphy, “Molecular and integrative physiological effects of isoflurane anesthesia: the paradigm of cardiovascular studies in rodents using magnetic resonance imaging,” Front. Cardiovasc. Med. 3, 23 (2016). [CrossRef]
14. M. K. Huss, H. H. Chum, A. G. Chang, K. Jampachairsi, and C. Pacharinsak, “The physiologic effects of isoflurane, sevoflurane, and hypothermia used for anesthesia in neonatal rats (rattus norvegicus),” J. Am. Assoc. Lab. Anim. Sci. 55, 83–88 (2016).
15. S. Gargiulo, A. Greco, M. Gramanzini, S. Esposito, A. Affuso, A. Brunetti, and G. Vesce, “Mice anesthesia, analgesia, and care, part i: Anesthetic considerations in preclinical research,” ILAR J 53(1), E55–69 (2012). [CrossRef]
16. N. Cesarovic, F. Nicholls, A. Rettich, P. Kronen, M. Hassig, P. Jirkof, and M. Arras, “Isoflurane and sevoflurane provide equally effective anaesthesia in laboratory mice,” Lab. Anim. 44(4), 329–336 (2010). [CrossRef]
17. S. C. Thal and N. Plesnila, “Non-invasive intraoperative monitoring of blood pressure and arterial pco2 during surgical anesthesia in mice,” J. Neurosci. Methods 159(2), 261–267 (2007). [CrossRef]
18. D. G. Lyons, A. Parpaleix, M. Roche, and S. Charpak, “Mapping oxygen concentration in the awake mouse brain,” eLife 5, e12024 (2016). [CrossRef]
19. K. J. Liu, G. Bacic, P. J. Hoopes, J. Jiang, H. Du, L. C. Ou, J. F. Dunn, and H. M. Swartz, “Assessment of cerebral po2 by epr oximetry in rodents: Effects of anesthesia, ischemia, and breathing gas,” Brain Res. 685(1-2), 91–98 (1995). [CrossRef]
20. M. Afeworki, G. M. van Dam, N. Devasahayam, R. Murugesan, J. Cook, D. Coffin, J. H. Larsen, J. B. Mitchell, S. Subramanian, and M. C. Krishna, “Three-dimensional whole body imaging of spin probes in mice by time-domain radiofrequency electron paramagnetic resonance,” Magn. Reson. Med. 43(3), 375–382 (2000). [CrossRef]
21. K. Saito, S. Matsumoto, H. Yasui, N. Devasahayam, S. Subramanian, J. P. Munasinghe, V. Patel, J. S. Gutkind, J. B. Mitchell, and M. C. Krishna, “Longitudinal imaging studies of tumor microenvironment in mice treated with the mtor inhibitor rapamycin,” PLoS One 7(11), e49456 (2012). [CrossRef]
22. P. Z. Chen, L. Y. Niu, Y. Z. Chen, and Q. Z. Yang, “Difluoroboron beta-diketonate dyes: spectroscopic properties and applications,” Coord. Chem. Rev. 350, 196–216 (2017). [CrossRef]
23. C. L. F. Christopher and A. DeRosa, “Tetracoordinate boron materials for biological imaging,” in Main Group Strategies towards Functional Hybrid Materials, F. J. Thomas Baumgartner, ed. (John Wiley & Sons Ltd., 2017).
24. A. Pfister, G. Zhang, J. Zareno, A. F. Horwitz, and C. L. Fraser, “Boron polylactide nanoparticles exhibiting fluorescence and phosphorescence in aqueous medium,” ACS Nano 2(6), 1252–1258 (2008). [CrossRef]
25. C. Kerr, C. A. DeRosa, M. L. Daly, H. Zhang, G. M. Palmer, and C. L. Fraser, “Luminescent difluoroboron beta-diketonate pla-peg nanoparticle,” Biomacromolecules 18(2), 551–561 (2017). [CrossRef]
26. A. G. Rickard, M. Zhuang, C. A. DeRosa, X. Zhang, M. W. Dewhirst, C. L. Fraser, and G. M. Palmer, “Dual-emissive, oxygen-sensing boron nanoparticles quantify oxygen consumption rate in breast cancer cells,” J Biomed. Opt. 25(11), 116504 (2020). [CrossRef]
27. R. Bro, “Parafac. Tutorial and applications,” Chemom. Intell. Lab. Syst. 38(2), 149–171 (1997). [CrossRef]
28. G. M. Palmer, A. N. Fontanella, S. Shan, and M. W. Dewhirst, “High-resolution in vivo imaging of fluorescent proteins using window chamber models,” Methods Mol Biol. 872, 31–50 (2012). [CrossRef]
29. “Capsule endoscopy,” https://www.mayoclinic.org/tests-procedures/capsule-endoscopy/about/pac-20393366.
30. C. P. Sabino, A. M. Deana, T. M. Yoshimura, D. F. da Silva, C. M. Franca, M. R. Hamblin, and M. S. Ribeiro, “The optical properties of mouse skin in the visible and near infrared spectral regions,” J. Photochem. Photobiol., B 160, 72–78 (2016). [CrossRef]
31. A. G. Tsai, B. Friesenecker, M. C. Mazzoni, H. Kerger, D. G. Buerk, P. C. Johnson, and M. Intaglietta, “Microvascular and tissue oxygen gradients in the rat mesentery,” Proc. Natl. Acad. Sci. U. S. A. 95(12), 6590–6595 (1998). [CrossRef]
32. A. S. Cowburn, D. Macias, C. Summers, E. R. Chilvers, and R. S. Johnson, “Cardiovascular adaptation to hypoxia and the role of peripheral resistance,” eLife 6, e28775 (2017). [CrossRef]
33. M. J. Campen, Y. Tagaito, J. Li, A. Balbir, C. G. Tankersley, P. Smith, A. Schwartz, and C. P. O’Donnell, “Phenotypic variation in cardiovascular responses to acute hypoxic and hypercapnic exposure in mice,” Physiol Genomics 20(1), 15–20 (2004). [CrossRef]
34. A. Y. Yu, L. A. Shimoda, N. V. Iyer, D. L. Huso, X. Sun, R. McWilliams, T. Beaty, J. S. Sham, C. M. Wiener, J. T. Sylvester, and G. L. Semenza, “Impaired physiological responses to chronic hypoxia in mice partially deficient for hypoxia-inducible factor 1alpha,” J. Clin. Invest. 103(5), 691–696 (1999). [CrossRef]
35. M. Dematteis, C. Julien, C. Guillermet, N. Sturm, S. Lantuejoul, M. Mallaret, P. Levy, and E. Gozal, “Intermittent hypoxia induces early functional cardiovascular remodeling in mice,” Am. J. Respir. Crit. Care Med. 177(2), 227–235 (2008). [CrossRef]
36. J. T. Pearson, M. Shirai, C. Yokoyama, H. Tsuchimochi, D. O. Schwenke, A. Shimouchi, K. Kangawa, and T. Tanabe, “Alpha(2)-adrenoreceptor mediated sympathoinhibition of heart rate during acute hypoxia is diminished in conscious prostacyclin synthase deficient mice,” Pfluegers Arch. 454(1), 29–39 (2007). [CrossRef]
37. P. Kiatchoosakun, I. A. Dreshaj, J. M. Abu-Shaweesh, M. A. Haxhiu, and R. J. Martin, “Effects of hypoxia on respiratory neural output and lower esophageal sphincter pressure in piglets,” Pediatr. Res. 52(1), 50–55 (2002). [CrossRef]
38. C. Arias-Reyes, J. Soliz, and V. Joseph, “Mice and rats display different ventilatory, hematological, and metabolic features of acclimatization to hypoxia,” Front. Physiol. 12, 647822 (2021). [CrossRef]
39. M. Mahling, K. Fuchs, W. M. Thaiss, F. C. Maier, M. Feger, D. Bukala, M. Harant, M. Eichner, J. Reutershan, F. Lang, G. Reischl, B. J. Pichler, and M. Kneilling, “A comparative po2 probe and [18f]-fluoro-azomycinarabino-furanoside ([18f]faza) pet study reveals anesthesia-induced impairment of oxygenation and perfusion in tumor and muscle,” PLoS One 10(4), e0124665 (2015). [CrossRef]
40. C. Badea, L. W. Hedlund, and G. A. Johnson, “Micro-ct with respiratory and cardiac gating,” Med. Phys. 31(12), 3324–3329 (2004). [CrossRef]
