|Home | About | Journals | Submit | Contact Us | Français|
The adult zebrafish has the potential to become an important model for diabetes-related research. To realize this potential, small-scale methods for analyzing pancreas function are required. The measurement of blood glucose level is a commonly used method for assessing β-cell function, but the small size of the zebrafish presents challenges both for collecting blood samples and for measuring glucose. We have developed methods for collecting microsamples of whole blood and plasma for the measurement of hematocrit and blood glucose. We demonstrate that two hand-held glucose meters designed for use by human diabetics return valid results with zebrafish blood. Additionally, we present methods for fasting and for performing postprandial glucose and intraperitoneal glucose tolerance tests. We find that the dynamics of zebrafish blood glucose homeostasis are consistent with patterns reported for other omnivorous teleost fish.
Our understanding of human endocrine pancreas function has been greatly enhanced by comparative studies on teleost fish. Perhaps the most important work on pancreas function utilizing teleosts was performed by J.J.R. Macleod,1 who used the large, isolated principal islets of angler fish and sculpin to definitively establish that the islets secrete insulin, with no contribution from acinar tissue. This pioneering study confirmed the work of Banting and Best,2 who extracted insulin from the pancreas of dogs, and showed that injecting it into pancreatectomized dogs alleviated diabetes. For their discovery of insulin, Banting and Macleod shared the Nobel Prize in physiology or medicine in 1923. Since then, pancreas function has been studied in a wide variety of fish, and the literature contains a wealth of comparative data for fish blood glucose levels. Recently, zebrafish (Danio rerio) blood glucose studies have been added to this list.3–5 The zebrafish is a small, freshwater teleost that has become a popular and informative developmental and genetic model. Zebrafish pancreas specification and morphogenesis have been well studied (reviewed in Refs.6,7), making this organism attractive for further development into a physiological model of pancreas function. The zebrafish islet, like that of other teleosts, is organized similarly to the human islet and consists of α-, β-, δ-, and -cells, with pancreatic polypeptide antibody-reactive cells also reported.8 Additionally, several zebrafish models have been established for studying β-cell or islet regeneration.5,9–12 These studies demonstrate the ability of the zebrafish islet to regenerate following isletectomy or β-cell-specific ablation. Studies such as these have the potential to significantly enhance diabetes-related research. However, we currently lack tools for evaluating zebrafish islet function.
An important parameter for evaluating islet function is blood glucose level, which is measured, for example, to determine whether the islet is secreting a sufficient amount of insulin. Adult zebrafish are small (typically <4cm in length), which makes them an attractive model for many applications, but which presents a challenge for measuring blood glucose. Existing methods used for other animals need to be scaled-down or redesigned. Fish biologists have traditionally used the glucose oxidase assay, which is also used for the clinical assay of human blood. However, these studies have utilized larger fish for which it is relatively easy to sample blood, and to collect adequate sample volumes. For zebrafish, performing a traditional glucose assay on individuals presents problems. To collect the required sample volume, blood from multiple individuals must be combined, which is less than ideal, or the blood from an individual must be diluted, which is only feasible if the particular assay is sensitive enough.
As an alternative to either of these potential sampling methods, we tested two hand-held meters used by diabetic humans for measuring blood glucose. The meters have several advantages over traditional laboratory methods, including speed and portability: glucose can be measured within seconds of blood collection, on site. More important, the hand-held meters require only a microliter or less of blood, giving this method a distinct advantage over traditional laboratory methods. Here we present the details of tests of both meters. We found that both Meter A, which uses the glucose oxidase method, and Meter B, which uses pyrroloquinoline quinone (PQQ) glucose dehydrogenase, were consistent with results from a laboratory glucose oxidase assay. We also present methods for applying anesthesia, for testing fasted and postprandial glucose levels, and for performing an intraperitoneal glucose tolerance test (IP-GTT). These methods will enable zebrafish researchers to study pancreatic function, and will also open the way to studies of metabolic disorders and diseases that perturb glucose homeostasis. As glucose homeostasis involves multiple organs and cell types, including the liver, skeletal muscle, and adipose tissue, these new methodologies will have very broad application.
Adult zebrafish were maintained following standard procedures.13 Fish were fed dry meal (Starter for Cultured Fish; Rangen, Inc.) supplemented with live brine shrimp. Fish were from the AB wild type line, multiple transgenic lines, and from the heterozygotes of multiple mutant lines. For most experiments, fish were utilized from multiple genetic backgrounds. For the fasting experiment, glucose tolerance testing, and glucose oxidase assay, fish were outbred wild type. Procedures were approved by the Institutional Animal Care and Use Committee at The University of Chicago.
Length (mm) was measured from the anterior-most point of the mouth, to the posterior-most region of the caudal peduncle, using digital calipers (Tresna). Weight (g) was measured by putting the fish into a small beaker of facility water on a scale and subtracting the nonfish weight.
MS-222 (tricaine; Sigma) was used at 0.02% in facility water, at 28.5°C.13 Hypothermia used ice-cold facility water.14 The cold water was held in a beaker in an ice bucket to maintain temperature. To anesthetize fish for measuring blood glucose, fish were transferred to a beaker containing either MS-222 or ice-cold water and then monitored for signs that they had reached stage III, plane 2 of anesthesia, namely, loss of equilibrium, loss of operculum movements, and loss of reactivity.14,15 For both anesthetics, stage III was typically reached within 60s. Fish were sacrificed (see below) and blood glucose was measured immediately.
To anesthetize fish for IP injection, fish were placed into 17°C facility water, and the container was not on ice. Ice made from facility water was gradually added to the water, to decrease water temperature to 12°C over several minutes. This procedure gradually brought the fish to stage II of anesthesia, in which muscle tone is decreased enough to allow handling of the fish, and opercular movements are decreased but still present.14,15 Recovery from stage II anesthesia was typically within several seconds of returning the fish to 28.5°C facility water.
To obtain whole blood, fish were anesthetized then decapitated by cutting cleanly through the pectoral girdle with scissors. The cut was immediately anterior to the articulation of the pectoral fin with the girdle, and severed the heart. Unless stated otherwise, whole blood was analyzed immediately by applying a test strip directly to the cardiac blood. For repeated measurement of a sample, whole blood was collected by holding either a heparinized 100μL microcapillary tube (Sarstedt) adjacent to the severed heart, or a heparinized 40mm microhematocrit tube (StatSpin). With this method, the quantity of blood collected depends on the size of the fish. We found that approximately 5μL was typical, but as much as 10μL was not uncommon. The tube was briefly spun to collect the sample, and glucose was measured immediately unless stated otherwise. With improper blood handling, hemolysis may occur, which can potentially affect glucose measurement.15a Potential causes of hemolysis are shearing with a needle and storage at incorrect temperature. Our blood-handling procedures avoided those practices, and we used collection tubes designed specifically for blood collection.
Whole blood was collected using 40mm heparinized microhematocrit tubes (StatSpin). Samples were spun for 120s at 13,700 g in a CritSpin hematocrit centrifuge (StatSpin), and hematocrit value was read with a digital reader (StatSpin).
The following glucose meters and test strip sample volumes were used—Meter A: OneTouch Ultra, 1μL sample (LifeScan); Meter B: FreeStyle Lite, 0.3μL sample (Abbott); Meter C: Accu-Chek Aviva, 0.6μL sample (Roche Diagnostics); Meter D: Accu-Chek Compact Plus, 1.5μL sample (Roche Diagnostics). For each meter, the appropriate control solution was used each time a new vial of test strips was opened, following manufacturer's instructions.
Before blood collection, fish were fasted for 24h. To generate a range of values, a subset of fish were additionally fed their normal meal 30min before testing, or injected with 1mg/g glucose solution (see below). Whole blood was collected from each individual, and measured 2–6 times with either Meter A, which uses glucose oxidase, or Meter B, which uses PQQ glucose dehydrogenase. A new test strip was used for each measurement. The resulting values were sorted into range categories (Table 1), and coefficient of variation (CV) was calculated. For some species of vertebrates, including several species of teleosts (reviewed in Ref.16), there are sex differences with respect to blood glucose levels. For zebrafish, we were unable to detect an effect of sex on blood glucose level (unpublished data), and therefore combined results from both sexes. A study of several other freshwater teleost species also found no sex differences with respect to glucose level.17
To perform a glucose oxidase assay, we collected approximately 80–90μL of whole blood and then separated plasma as follows. Whole blood was collected using a tube assembly consisting of a 40mm heparinized microhematocrit tube (StatSpin) inserted through the split cap of a StatSpin #SS1E blood collection tube, and held by a 100μL microcentrifuge tube (Sarstedt). Both the Sarstedt tube and the StatSpin #SS1E are designed for whole blood collection and consist of a capillary tube inserted into a larger microcentrifuge-style tube. However, we found that substituting the narrower microhematocrit tube for the Sarstedt microcapillary tube was more efficient for collecting microsamples than using either the Sarstedt or StatSpin collection tubes. Further, the short height of the StatSpin #SS1E microcentrifuge tube prevented efficient collection of the sample from the microhematocrit tube during centrifugation. We found that transferring the split cap with microhematocrit tube to the taller Sarstedt microcentrifuge tube resulted in an assembly that allowed complete collection of the blood sample into the centrifuge tube by spinning down briefly using a mini centrifuge.
Whole blood was collected from 16 large fish, using one tube assembly per fish; then, the blood was pooled for a sample size of 80–90μL, and gently mixed. Blood was collected within 10min. Glucose was measured twice with Meter A and twice with Meter B; then, plasma (approximately 30μL) was immediately separated from the remaining sample and used for glucose measurement with a YSI 2300 STAT Plus Analyzer (Yellow Springs Instruments, Inc.). To maximize the volume of plasma that could be separated, we made a scaled-down version of the StatSampler Micro Blood Collector (StatSpin, #SS2U), an assembly that includes a microtube containing a barrier gel that separates blood cells from plasma. We made a smaller version by transferring approximately one-third of the gel to a 0.5mL PCR tube, and spinning for 20s at 13,000 g. The whole-blood sample was centrifuged for 14s at 10,000 g to separate plasma. The plasma sample was measured twice with the YSI Analyzer.
For this procedure, there was an unavoidable time delay of approximately 20min between blood collection and measurement. Therefore, tubes were kept on a cold rack (4°C) throughout the prodecure to minimize glycolytic activity. To test for an effect of the time delay on the glucose value, we collected whole blood as described, and measured the sample at 10min postcollection using Meter A and Meter B. After an additional 20min, the sample was measured again and the values were compared. The sample was measured twice with both meters at both time points and the entire experiment was performed four times. We found that the time delay had a negligable affect on the average value for Meter B (glucose increased by 2.5mg/dL, on average, after 20min). However, for Meter A there was a significant decrease (11.63mg/dL, on average, after 20min), thus indicating a technical constraint in our ability to directly compare the performance of Meter A with the YSI Analyzer. We concluded that, given the delay between blood collection and measurement, a correction factor of 11.63mg/dL should be applied for Meter A.
Tanks were taken offline during the fasting period and food was withheld. We observed (unpublished data) that stress can raise glucose levels in zebrafish, as reported for other teleosts (reviewed in Refs.17,18). To minimize stress, we avoided overcrowding by housing fish at 12–15 fish per 9L tank, and maintaining the normal mixed-sex environment. These practices are in line with general recommendations for maintaining zebrafish.13,19,20 Because mating behavior appeared to be normal, and adults will eat their eggs if they are accessible, the tank bottoms were covered with 2–3 layers of glass marbles to allow the eggs to settle out of reach. Tanks were maintained daily by siphoning eggs and waste off the bottom, and replacing 10%–15% of the tank volume with fresh facility water. Fish were fasted for a maximum of 4 days, and blood glucose was measured in a subset of fish every 24h, using Meter B.
Fish were fasted for 4 days, then fed their normal dry meal, and allowed to feed until no food remained, which took approximately 3min. Blood glucose was measured using Meter B.
D-glucose (Sigma) was dissolved in Cortland salt solution, pH 7.45, at 0.5mg/μL (formulation shown in Table 2). Glucose solution or vehicle was injected IP at 1mg/g fish weight.
For IP injection, a surgical table was constructed consisting of a 60mm Petri dish holding a soft sponge (#L800-D; Jaece Industries), and set into a pipet tip box lid. The surgical table was set into a larger, deeper container (2.4L Rubbermaid container). All containers were flooded with cold facility water. The sponge was cut in half, and a shallow trough was cut into the flat face. The trough was used for holding the fish securely during injection. The sponge was saturated with cold facility water. Saturation of the sponge was important for maintaining anesthesia during injection as well as for providing water to the gills for respiration.
Each fish was weighed, anesthetized in the largest container of the surgical table setup, and transferred to the saturated sponge as soon as stage II anesthesia was reached (see above), and body movements had slowed enough for handling. The surgical table (sponge, Petri dish, and box lid) with fish was immediately transferred to an adjacent dissecting microscope stage (Leica M165 FC) for injection. Injection was performed under 7.3× magnification using a 35-gauge beveled steel needle and a 10μL NanoFil syringe (World Precision Instruments) attached to an UltraMicroPump III driven by a Micro4 Controller (World Precision Instruments). The needle was inserted into the midline of the ventral posterior abdomen, between the pelvic fins. The injection site was closer to the insertion of the fins on the pelvic girdle than to the anus. The needle was directed cranially, to reduce the possibility of damage to internal organs, as recommended by Reavill.21 Immediately following injection, the fish was returned to 28.5°C facility water for recovery. Blood glucose was measured using Meter A.
Analysis was performed using GraphPad Prizm, v5.02. All t-tests were two-tailed, and were unpaired unless stated otherwise. Averages are reported as mean and standard deviation.
We compared two methods of anesthesia, MS-222 (tricaine) and ice-cold facility water (hypothermia). MS-222 is a known ion channel blocker,22 and thus may interfere with β-cell function. Further, MS-222 has been reported to increase blood glucose in teleost fish.15,16 To test its effect on zebrafish, fish were fed their normal meal, fasted for 2h, anesthetized, and measured for glucose level using Meter B. As shown in Figure 1, hypothermia yielded more consistent glucose readings than MS-222 treatment. Although the mean glucose concentration was similar for both anesthesia methods (p=0.3376, t-test), the CV was much higher for MS-222-treated fish (MS-222, CV=54%; hypothermia, CV=29%). We concluded that MS-222 may be interfering with normal ion channel function in the β-cells, and therefore chose hypothermia for subsequent analyses.
To measure zebrafish blood glucose, we tested two hand-held devices designed for measuring glucose in whole blood of human diabetics. First, we tested whether zebrafish blood is within the recommended hematocrit range for both meters. We measured the hematocrit of individual fish and found a mean of 31.50%±5.71% (n=12). This value was within the recommended range of both meters (Meter A, 30%–55%; Meter B, 15%–65%; ranges according to manufacturers's instructions). Additionally, the zebrafish hematocrit was consistent with values reported for a variety of freshwater teleosts.17,23,24 For some teleost species, hematocrit has been found to vary by sex and/or size.18 For zebrafish, we found no difference in hematocrit for males versus females (males, 30.90%±5.52%, n=5; females, 29.67%±4.37%, n=6; p=0.6880, t-test). To test for a correlation between hematocrit and size, we measured length (millimeters) and weight (grams) of each fish. We performed linear regression to test the strength of the relationship between hematocrit and length, and calculated r2=0.05212 (n=9). For the relationship between hematocrit and weight, r2=0.002928 (n=11). We concluded that hematocrit cannot be predicted by either length or weight, and therefore fish were not separated by size for glucose testing.
Next, we tested the precision and accuracy of both meters with zebrafish blood, as the meters use different chemistries to measure glucose. Meter A uses glucose oxidase and Meter B uses PQQ glucose dehydrogenase. The principles of immobilized enzyme technology in glucose testing have been reviewed.25 To measure precision, whole blood was collected from individuals using heparinized capillary tubes, and measured repeatedly. According to the manufacturers, both test strips are compatible with heparinized blood. Meter precision was tested over three ranges for both meters: Meter A low, 28–59mg/dL; mid, 60–129; high, 130–550; Meter B low, 21–44mg/dL; mid, 45–105; high, 106–280. For both test strip chemistries, the CVs were similar to those reported in test strip package inserts for human blood (Table 1).
In a further comparison of the two meters, blood from 31 individuals was measured once with each meter, and the values were compared. Meter B returned values that were 28mg/dL lower, on average, than Meter A (p<0.0001, paired t-test, Fig. 2A). To confirm that order of meter use did not affect blood glucose values in this comparison, five of the samples were measured first with Meter B followed by Meter A, and five of the samples were measured first with Meter A followed by Meter B. Order of meter use did not significantly affect the value (Meter B first vs. second, p=0.4873; Meter A first vs. second, p=0.2831, paired t-tests). These tests suggested that at least one of the meters may not be accurate.
To further test the accuracy of the meters, we asked whether the lower values given by Meter B were a consequence of the different test strip chemistries, or a consequence of the difference in sample volume (0.3μL for Meter B vs. 1.0μL for Meter A). We measured blood glucose using two additional meters that employ the same enzyme as Meter B, PQQ glucose dehydrogenase, but that use larger sample volumes. Meter C requires a 0.6μL blood sample, and Meter D requires 1.5μL. The zebrafish hematocrit falls into the range for both Meter C (20%–70%) and Meter D (25%–65%). We collected a blood sample from individual fish, and measured each sample once with each meter. We found that for test strips employing PQQ glucose dehydrogenase, sample volume may have an effect on the glucose value, as smaller sample volume is associated with lower glucose measurement (Fig. 2B, repeated measures ANOVA, p<0.0001). However, the average blood glucose value from Meter D, which uses the largest sample volume of the four meters, was slightly lower than the average blood glucose value from Meter A, which uses glucose oxidase chemistry (Meter D, 42.06±12.15mg/dL vs. Meter A, 45.15±16.27mg/dL; p<0.05 Tukey's multiple comparison test). This suggested that differences in test strip technology, beyond possible effects of the sample volume, may be contributing to the discrepancy between the meters. For example, although the test strips for Meter C and D both use the same enzyme, coenzyme, and mediator, they employ different indicators.25
Next, we directly tested the accuracy of Meter A and B by comparing meter results with those from a clinical laboratory glucose oxidase assay. We measured a pooled sample using both meters and a YSI 2300 Analyzer and found that Meter A measured 106.63±8.49mg/dL; Meter B measured 114±9.90mg/dL; and the YSI Analyzer measured 121±0mg/dL. The performance standard set by the International Organization for Standardization (ISO 15197) requires that meters measure within±20% of the laboratory measurement for 95% of the samples (cited in Ref.26). Measurements with Meter A were within 11.9% and measurements with Meter B were within 5.8%. For Meter A, we applied a correction factor as detailed in the Materials and Methods section.
To demonstrate the utility of blood glucose measurement in zebrafish, we performed three sets of experiments in which we monitored changes in blood glucose levels over time. We fasted the fish over the course of 4 days to determine the changes in glucose level in the absence of food, and followed the fasting with a postprandial glucose test. Next, we performed an IP-GTT to determine the time required for the zebrafish to return glucose levels to homeostasis.
With fasting, glucose levels rose slightly after 2 days, and dropped significantly by 3 days (Fig. 3). We concluded that fasting for 3 days is sufficient to bring blood glucose to a baseline level. After 4 days of fasting, the mean was not reduced further compared to 3 days; however, the CV was reduced after 4 days of fasting, compared to shorter fasts (Fig. 3A). Refeeding the fish following a 4-day fast revealed that postprandial glucose peaks by 30min (Fig. 3B).
Next, we developed a GTT that utilized IP injection. For performing a GTT, glucose could simply be added to the tank water, to be taken up orally and through the gills. Indeed, such a test has been performed on zebrafish with success.4 However, there may be experimental questions in which it is desirable to give each fish the same amount of glucose, relative to weight. Therefore, we developed a method for IP injection into zebrafish. We followed the general recommendations found in Perry and Reid27 with respect to vehicle, injection volume, and controls. For vehicle we used Cortland salt solution, a physiological saline formulated for freshwater teleosts.28,29 We used an injection volume of 2μL/g body weight, and controls included vehicle-only injections and no-injection controls. Before testing, fish were fasted for 3 days and then challenged with high glucose, and clearance time was determined. We found that blood glucose peaks at 30min postinjection and recovers to normal levels by 6h (Fig. 4). Additionally, glucose levels following injection with vehicle alone were not significantly elevated relative to noninjected controls. This demonstrates that any stress caused by the injection itself has a minimal effect on glucose level.
How important is glucose or carbohydrate for zebrafish metabolism? Studies on carbohydrate utilization in fish show that warm water, omnivorous teleosts utilize carbohydrates to a much greater extent than do coldwater, carnivorous teleosts (for reviews, see Refs.30,31). As zebrafish are warm water omnivores,32 carbohydrates are likely an important component of their diet. Indeed, recent work on adult zebrafish has demonstrated that the amount of carbohydrate in the diet is positively correlated with growth rate.33 In light of these studies, it is not surprising that zebrafish express 18 glucose transporters,34 as well as a number of genes, including hexokinases, that are important for glucose metabolism.33,35 Additionally, deficiency in the glucose transporter s/c2a1a (solute carrier family 2 [facilitated glucose transporter], member 1a, formerly glut1) has been shown to cause a suite of neural defects in embryonic zebrafish.36 We anticipate that studies of glucose metabolism utilizing the zebrafish model will become increasingly important. Here, we have presented methods for the analysis of blood glucose homeostasis in adult zebrafish and have found that important considerations for measuring blood glucose include choice of anesthetic, blood collection method, and glucose assay method.
We found that anesthetics should be tested for obvious effects on glucose level. In our tests on zebrafish, treatment with the commonly used MS-222 rapidly affected blood glucose levels, as reported for other freshwater teleosts.16 MS-222 treatment produced highly variable glucose levels compared to anesthetizing with cold water. Although it is known that MS-222 is a nerve ion channel blocker,22 it is not known whether MS-222 also directly affects β-cell ion channels and therefore insulin secretion. Anesthetics in general alter glucose levels,37 and some have been shown to alter insulin secretion by acting directly on β-cell channels. To cite two examples, tetracaine alters Ca2+ uptake or efflux from β-cells, depending on dosage,38,39 and isoflurane decreases insulin release from β-cells by opening ATP-sensitive potassium channels.40 Thus, for glucose metabolism studies, appropriate anesthetic is critical.
For sampling blood, we found that decapitation was a reliable and easy method. A drawback is that it precludes repeatedly measuring individuals for time course studies. A potential alternative is cardiac puncture on live fish, which would allow repeated measurements from the same individual. In theory, it should be possible to withdraw 30%–50% of total blood volume without adverse affects, based on work in larger teleost fish.18 In practice, the small size of the zebrafish heart makes this procedure quite difficult. Although the location of the beating heart can easily be determined externally, inserting the needle into the heart chamber accurately requires precise anatomical knowledge, and significant surgical skill. One group has performed zebrafish cardiac puncture for blood sampling, and has utilized pulled glass capillary pipettes to remove a blood volume of approximately 50nL.5 With currently available meters, the smallest blood sample that can be measured for glucose is 300nL. Therefore, the small amount of blood that can be retrieved requires that the sample be diluted, or combined with other samples, to bring it to assay volume. The relative advantages and disadvantages of cardiac puncture should be carefully evaluated when designing experiments.
To collect blood for downstream assays requiring whole blood or plasma, we used two different collection tubes, a 100μL heparinized microcapillary tube, and a 40mm heparinized microhematocrit tube. For both tubes, blood is drawn up via capillary action. We ultimately preferred the microhematocrit tube over the microcapillary tube, as its smaller diameter and thinner walls permitted collection of a larger sample volume. To minimize the possibility of collecting fluids other than blood, we found it was important to hold the collection tube in close proximity to, but not touching, the exposed heart chamber.
A large number of studies have compared various glucose meters using human blood and have shown that there can be significant differences in performance, for example.41,42 Such studies have shown that, although each meter may meet the accuracy requirements of the International Organization for Standardization, they do not necessarily agree with each other precisely. Therefore, human patients are advised to use one meter for managing blood glucose rather than comparing across meters. Additionally, studies have compared meters for use on nonhumans, including dogs, cats, and birds, and have reached similar conclusions with respect to meter use.43–45 Here, we tested the suitability of two human glucose meters for use with zebrafish blood. We conclude that although both meters are accurate with respect to a laboratory glucose assay, for consistency within and between experiments, only one meter should be used.
The two meters tested, Meter A and Meter B, use different chemistries for measuring glucose in whole blood. In an initial test of both meters, we determined that the average zebrafish hematocrit fell into the range of both meters, and that neither body weight, length, nor sex influenced hematocrit value. For reading the hematocrit value, we tried three commonly used commercially available readers: a card-style reader, in which the microhematocrit tube sits over a sliding scale and the packed cell volume is read manually; a disc-style reader, in which the scale is placed directly over the tube rotor for reading manually; and a digital reader, in which the tube is read electronically. We found that the manual readers were not practical, because they require a minimum 9μL blood volume, which we could not typically draw. By contrast, the digital reader could measure hematocrit using as little as approximately 5μL blood volume. Hematocrit measurement is useful for the study of a variety of diseases, including anemias, infections by parasites, and some metabolic disorders.46,47
After determining that both glucose meters were suitable with respect to zebrafish hematocrit, we went on to test precision and accuracy. Both meters were slightly less precise than reported for human blood. However, as the CV was <8% across a range of glucose values for either meter, we concluded that both meters give reliable measures.
In our comparison of all four meters, we found that we could discriminate statistically between them, with the exception that Meter B could not be discriminated from Meter C. Although we tested for differences attributable to sample volume, this parameter is simply a proxy for general differences in test strip technology. The strips differ in a number of ways that could potentially affect glucose measurement, including the specific mediator and indicators employed, and number and orientation of electrodes.25 In further testing, we directly compared the performance of Meter A and B with a clinical laboratory test. We found that both meters fall within the 20% error cutoff deemed acceptable for human patients. Ultimately, we concluded that although all four brands of test strips and meters give somewhat different values, they perform similarly from a clinical perspective. The practical significance of these data is that only one meter should be used for experimental tests so that there is consistency for comparing across measurements.
Glucose metabolism studies commonly employ measurement of blood glucose following fasting, feeding, or a glucose challenge. Here, we have presented methods for performing these tests on adult zebrafish, and have used the tests to demonstrate that zebrafish blood glucose is dynamically regulated. We monitored blood glucose level over the course of 4 days of fasting, and found that blood glucose rose after 2 days of fasting, and then decreased to a baseline level by 3 days. A similar pattern for fasting glucose, in which glucose was elevated on day 2, was reported for the related goldfish species Carassius auratus.16,18 Like zebrafish, C. auratus is a warm water omnivorous teleost, and is in the family Cyprinidae. It is possible that short-term fasting causes stress, which may in turn increase plasma cortisol. This could explain the increased glucose level observed on day 2 of the fast. Cortisol has been shown to increase liver gluconeogenesis in teleosts (reviewed in Ref.48), but the effect of fasting on cortisol level has not been studied in zebrafish. During long-term fasting, maintenance of blood glucose at a baseline level has been observed in a variety of teleosts, and this maintenance has been attributed to gluconeogenesis.49
We found that injection of glucose produced peak blood glucose levels within 30min following injection, and that blood glucose levels returned to the control level within 6h. This 30-min peak was consistent with our postprandial experiment, which showed a peak in glucose 30min after feeding. Additionally, this time frame for clearance is consistent with studies in another freshwater omnivorous teleost, Cyprinus carpio (carp), which is in the same family as zebrafish. Oral administration of glucose raised blood glucose in C. carpio to a maximum within 1h, and glucose returned to baseline by 5h.50 Similarly, intravenous injection of glucose into the marine omnivore Opsanus tau (oyster toadfish) resulted in peak blood glucose by 30min, followed by clearance by around 6h.51 A recent article has reported IP-GTT in zebrafish and found peak glucose at 45min rather than 30min, with clearance time not determined.35 The difference in peak time between their study and the current one might be attributed to a difference in the two fish populations, and/or the different test strip chemistry employed (phenanthroline quinone glucose dehydrogenase used in the previous study, whereas we used glucose oxidase). There may also have been a difference in the level of stress induced by the injection procedure, as the control values were higher than we found in the current study.
We have shown that zebrafish blood glucose is dynamically regulated, and that glucose metabolism is consistent with reports on other omnivorous fish. Our studies show that zebrafish, like other omnivores, metabolize glucose faster than carnivorous teleosts.30 The relatively fast glucose metabolism of zebrafish should facilitate laboratory studies of pancreas and liver function. Other tissues involved in glucose homeostasis include skeletal muscle and adipose, and numerous factors affect circulating glucose level, including insulin sensitivity of peripheral tissues and plasma glucocorticoid levels. Thus, the methods presented here will be useful for studying the function of multiple tissues with respect to glucose metabolism in zebrafish. We anticipate that additional methods and techniques for studying glucose metabolism will be developed in the near future. An important tool that is currently lacking for pancreatic islet studies is an assay for determining circulating insulin level. The ability to assay insulin would be a powerful complement to glucose measurement, and would enhance our ability to study carbohydrate metabolism in the zebrafish model. As pointed out by Moon and Foster, “Carbohydrates are key to the metabolism of all vertebrates, including fish species.”52 Here we have demonstrated that the zebrafish is an important new tool for carbohydrate utilization studies in a tractable, genetic vertebrate model.
We thank Oni Mapp, Gökhan Dalgin, and Robert Risley for donating fish; Isaac Skromne for technical advice on setting up a surgical table; Nandini Kaluvakolanu for technical assistance with the YSI Analyzer; and Sean Burns, Robert Risley, Matthew Rowe, Elizabeth Sefton, and Molly Steele for fish care. This study was supported by Juvenile Diabetes Research Foundation Grant 5-2007-97 (to V.E.P.), by National Institute of Child Health and Human Development (NICHD) Ruth L. Kirschstein NRSA F32HD050031 (to M.D.K.), National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) Grants R01DK48494 (to L.H.P.), R01DK064973 (to V.E.P.), Training Grant T32DK07074 (supporting S.C.E.), K01DK083552 (to M.D.K), and by P60DK20595 to The University of Chicago Diabetes Research and Training Center. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIDDK, NICHD, or the National Institutes of Health.
No competing financial interests exist.