Strategies to reduce variation in Xenopus regeneration studies


  • Holly L.D. Nye,

    1. Department of Cell and Developmental Biology and College of Medicine, University of Illinois at Urbana-Champaign, Urbana, Illinois
    Search for more papers by this author
  • Jo Ann Cameron

    Corresponding author
    1. Department of Cell and Developmental Biology and College of Medicine, University of Illinois at Urbana-Champaign, Urbana, Illinois
    • Department of Cell and Developmental Biology, B106 Chemical and Life Sciences Laboratory, 601 S. Goodwin Avenue, University of Illinois at Urbana-Champaign, Urbana, IL 61801
    Search for more papers by this author


In this study, we present strategies for experimental design that minimize variation in Xenopus hindlimb regeneration results. We have standardized our laboratory culture conditions for older stage Xenopus tadpoles. We have established a normal tadpole growth curve for our laboratory and characterized normal tadpole behaviors in an effort to eliminate abnormal tadpoles from our experiments. We have used large sample sizes and statistical analysis to establish normal regeneration performances for seven amputation planes in stages 55–57 tadpole hindlimbs. We have demonstrated that regeneration performance of abnormal tadpoles is significantly different than that of normal tadpoles. We have examined the kinetics of ossification in developing Xenopus hindlimbs, and have found that increasing ossification rates and rates of regeneration decline are inversely correlated. Developmental Dynamics 234:151–158, 2005. © 2005 Wiley-Liss, Inc.


One of the challenges facing investigators of Xenopus laevis limb regeneration is the amount of variation in the results of similar experiments performed in different laboratories or even between experiments performed at different times within the same laboratory. One type of variation is in the numbers of skeletal elements that regenerate after hindlimb amputation. For example, calculating from the data of Dent (1962), Anton et al. (1988), and Shimizu-Nishikawa et al. (2003) after amputations through the femur in Nieuwkoop and Faber (1967) stage 53 tadpoles, the average numbers of digits regenerated were 2.7, 1.7, and 3.6, respectively. Such variations can also hinder efforts to repeat studies designed to increase the numbers of skeletal elements regenerated by older tadpoles. Yokoyama et al. (2001) applied beads soaked in fibroblast growth factor 10 (FGF10) to stage 56 hindlimbs after knee level amputation. Sixty-four percent of these limbs regenerated, with an average of 1.5 digits per tadpole. Their control tadpoles with saline-soaked beads all failed to regenerate. Doing the same experiments, Slack et al. (2004) found that 45% of the limbs with FGF10 beads regenerated, with an average of 0.5 digits per tadpole; however, 32% of their control tadpoles regenerated, also with an average of 0.5 digits per tadpole. Slack et al. (2004) attributed these dissimilar results to variation in regeneration capability in different batches of tadpoles. They described instances of regeneration failure in younger (stage 55) tadpoles and regeneration success in older (stage 57) tadpoles, and they made the point that these variations could make detection of subtle regeneration changes very difficult to detect. Another type of variability involves which digits regenerate. The identity of Xenopus hindlimb digits is determined by numbers of phalanges and possession of claws. The preaxial (anterior) digits I-III have claws. Digits I and II have 2 phalanges, digit III has 3 phalanges. The postaxial (posterior) digits IV and V have 4 and 3 phalanges, respectively, and do not have claws. When Overton (1963) and Muneoka et al. (1986) performed hindlimb amputations in younger (stage 51–53) and older (stage 54–59) tadpoles, they found that fewer anterior digits regenerated in the older tadpoles. In contrast, similar studies by Anton et al. (1988) and Shimizu-Nishikawa et al. (2003) indicated that fewer posterior digits regenerated with increasing tadpole age.

We previously performed amputations through three planes in the ankle region of stage 53–59 Xenopus tadpoles: the ankle joint, the tarsus (also known as the tibiale and fibulare), and the tarsal-metatarsal joint. We found that amputations through the ossified center of the tarsus regenerated poorly compared to amputations performed through cartilaginous portions located only a few millimeters either proximally or distally (Wolfe et al., 2000). Some of the variable responses in Xenopus hindlimb regeneration might, therefore, result from slight variations in the amputation planes used by different investigators. In addition, the amount of ossification present in the femur, tibia-fibula, and tarsus in tadpole stages 53–59 in previous studies was unknown, so it was not possible to gauge whether an amputation plane passed through ossified or cartilaginous skeletal elements. Trueb and Hanken's (1992) comprehensive study of Xenopus laevis skeletal development found that ossified tissue is not definitely present in hindlimbs until stage 57, but they did not describe the percentages of the skeletal elements that were ossified. They also noted variation in the timing of skeletal ossification with respect to the external Nieuwkoop and Faber (1967) stage morphologies. This variation in the extent of ossification may help explain the differences between younger and older tadpole regeneration capabilities observed by Slack et al. (2004).

In this study, we extended our ossification investigations in order to reduce variance in the numbers of skeletal elements produced in Xenopus regeneration experiments in our laboratory. We measured the lengths of stained ossified tissue in the femur, tibia-fibula, and tarsus of stage 54–59 Xenopus tadpoles and calculated percentages of ossification for those skeletal elements. We computed mean tadpole regeneration percentages by counting the numbers of regenerated skeletal elements for seven planes of amputation from the femur through the tarsus for stages 55–57. We found that increasing sample sizes reduced variation in hindlimb regeneration. We employed statistical methods to correlate rates of increasing ossification and regeneration decline. We also showed that another major cause for variation is the inclusion of unhealthy or developmentally abnormal (DA) tadpoles in regeneration studies. We characterized these abnormalities and established a normal tadpole laboratory growth curve that allows us to remove abnormal tadpoles from experimental populations.


Generation of Tadpole Growth Curves and Identification of Developmentally Abnormal Tadpoles

While conducting large numbers of limb regeneration experiments over a period of several years, we noticed that some tadpoles developed abnormally. In their reviews, McDiarmid and Altig (1999) and Alford (1999) describe how environmental factors such as overcrowding, bacterial contamination, and changes in temperature can adversely affect tadpole development. The developmental abnormalities we observed could not be explained in this way, since we maintained our tadpoles in the constant environmental conditions described in the Experimental Procedures section throughout the study. A few of our abnormal tadpole phenotypes were consistent with three of the 42 known inherited mutations: bent tail, bloat, and delayed metamorphosis (Droin, 1991). However, the majority of tadpoles we classified as developmentally abnormal (DA) did not fit any genetic phenotype. They were pale in color and were smaller than normal. Xenopus tadpoles are obligate suspension feeders and swim continuously to maintain a central position in their container (reviewed by Hoff et al., 1999). By comparison, DA tadpoles were sluggish. They swam slowly, and were often found resting on the bottom of their container. They responded less well to startling stimuli, and produced less fecal matter than normal tadpoles, indicating that they fed less well. Their mortality rate was higher than that of other tadpoles. Most of them died prior to or during the tail resorption phase of metamorphosis. Growth rate is an important measure of larval fitness (reviewed by Harris, 1999). The defining characteristic of DA tadpoles was that their growth rate was much slower than that of other tadpoles in the same batch.

We noted that DA tadpoles appeared to regenerate very few skeletal elements after hindlimb amputation. This suggested that their regeneration data could skew that of normal tadpole hindlimbs towards lower levels of performance. To explore this idea, we decided to establish growth curves for normal and DA tadpoles so that these populations could be separated. We collected growth rate data from each new batch of tadpoles arriving in the laboratory, which were all from the same commercial source. Tadpole stages at arrival were relatively homogenous (all tadpoles in one or two stages), but as time progressed, they grew at different rates. The final developmental stages they attained were recorded when they were collected for hindlimb regeneration data analysis.

The growth data for DA and normal tadpoles were separated and analyzed by non-linear regression to fit a quadratic model. A similar regression was calculated for the wild-type tadpoles described by Nieuwkoop and Faber (1967). The result was a set of three growth curves (see Fig. 1). The regression P values for the Cameron lab normal and abnormal tadpoles were <0.0001 and 0.0015, respectively. A P value was not calculated for the Nieuwkoop-Faber tadpoles, as sample sizes were unknown. The curves for the animals we identified as normal were very similar to those of wild-type tadpoles, except that our tadpoles grew slightly more slowly (reflected in the slightly steeper slope for the Cameron Lab tadpoles). The curve for the abnormal tadpoles was markedly steeper in slope, describing their much slower growth rate. We could now use these curves to track tadpoles that were growing more slowly, commencing at the time they arrived in the laboratory. In this way, they could either be studied separately or euthanized and removed.

Figure 1.

Laboratory growth curves for Xenopus laevis tadpoles. The Nieuwkoop and Faber (1967) curve is for wild-caught tadpoles in natural pond water. The Cameron Laboratory curves are for tadpoles maintained in the culture conditions described in the Experimental Procedures section. Curves were calculated using non-linear regression to fit a quadratic model.

From these data, we conclude that developmentally abnormal tadpoles have significantly reduced growth rates. We find that eliminating these tadpoles from our experimental analyses removes one very important source of regenerative variation in our data (abnormal hindlimb regeneration).

Xenopus laevis Hindlimb Regeneration for Normal and DA Tadpoles in Stages 55–57

Xenopus hindlimb regenerative ability declines in a proximal to distal direction within each tadpole stage as the hindlimb progressively differentiates (Dent, 1962; Overton, 1963; Anton et al., 1988). In the present study, we investigated the regenerative ability of joint, cartilaginous, and ossified regions along the entire length of the normal hindlimb for stages 55–57. We also used much larger sample sizes than in Wolfe et al. (2000). Our expectations were that variation in regeneration results would decrease for each amputation plane, and that we would obtain a regeneration performance baseline that would be useful for comparison of results after experimental manipulations. We performed amputations through the femur, knee joint, tibia-fibula, ankle joint, tarsus, tarsal-metatarsal joint, and metatarsals. The seven amputation planes and regeneration scoring system are shown in Figure 2. The mean regeneration scores and standard errors calculated for the seven different amputation planes in normal tadpole stages 55–57 are listed in Table 1. The mean regeneration scores are illustrated in Figure 3. Multiple pair-wise comparisons measured the amounts of statistically significant differences between sets of scores. These comparisons indicated that the femur score significantly differed from that of the knee; the knee score differed significantly from that of the tibia-fibula, and so on at the 0.05 significance level and with P values ranging from <0.0001 to 0.0068 in stages 55 and 56. The pair-wise comparison did not find a significant difference in the scores of the tarsal-metatarsal joint and the metatarsal bones. In stages 55 and 56, these two sites consist predominantly of cartilaginous tissue, and it is not surprising that their regeneration scores were comparable. The differences between all of the adjacent amputation planes became less significant in stage 57 tadpole hindlimbs. In that stage, the P value for the femur versus knee scores comparison was 0.5319; for the knee versus the tibia-fibula scores, it was 0.8536, and so on. This disappearance of regeneration score differences indicates that stage 57 hindlimbs had lost most of their regeneration capability. Similarly, pair-wise comparisons were made for the ankle and knee joints, and for the tarsus and tibia-fibula bones for all three stages. As anticipated, those P values ranged from 0.0604 to 1.000, indicating that joint (cartilaginous) site regeneration scores were similar to each other, as were mid-bone site scores to each other. We found that cartilaginous and soft tissue amputation planes consistently produced larger numbers of regenerated skeletal elements than ossified tissue planes all along the length of the hindlimb. For example, the more proximal knee joint amputation plane regenerates significantly better than the ossified tibia-fibula distal to it in stages 55 and 56 (P values 0.0068 or less). This difference may be related to the state of terminal differentiation present in the tissues at the amputation plane. Cartilaginous tissue matrix has been shown to dedifferentiate more quickly than bone matrix in amputated urodele limbs (Stocum, 1979), and it is possible that this is also true in Xenopus.

Figure 2.

Xenopus hindlimb amputation planes and regeneration scoring system. A: Amputations were performed in seven different amputation planes for tadpole hindlimbs in stages 54–59: femur (F), knee (K), tibia-fibula (TF), ankle joint (AJ), tarsus (T), tarsal-metatarsal joint (TMTJ), and metatarsals (MT). B: Using joints as boundaries, the skeletal element scoring system assigns 10 points for regeneration of the skeletal elements of each limb segment. Raw scores are computed by counting the regenerated skeletal elements for each segment. An index (percent) of regeneration is computed by dividing the raw score by the total possible score. The scores can then be compared within a stage, or between stages, and can all be graphed on a scale of 0–100% regeneration.

Table 1. Percentages of Normal and Abnormal Tadpole Hindlimb Regeneration Performance With P Values for Pair-Wise Comparisons in Stages 55–57a
Stage, planeMean normal regeneration (%)Std. errornMean abnormal regeneration (%)Std. errorn
  • a

    MT, metatarsal; TMTJ, tarsal-metatarsal joint; T, tarsus; AJ, ankle joint; TF, tibia-fibula; K, knee; F, femur; n, sample size.

55 MT68.03.2631   
55 TMTJ59.41.65124   
55 T25.91.7810317.36.0023
55 AJ36.11.27204   
55 TF17.51.6212713.53.8016
55 K29.22.3561   
55 F5.673.2631   
56 MT40.33.0740   
56 TMTJ38.91.55138   
56 T17.01.671186.484.3223
56 AJ26.91.39170   
56 TF13.01.541403.572.4012
56 K25.02.4157   
56 F4.152.6547   
57 MT19.53.5626   
57 TMTJ22.52.5055   
57 T9.601.072872.681.5213
57 AJ11.52.1374   
57 TF5.741.651231.771.4012
57 K11.12.5253   
57 F0.493.2132   
Figure 3.

Regeneration mean scores and standard errors for 7 amputation planes for stages 55–57. Bars represent standard errors. Where error bars are not visible, they were small enough to be contained within the plotting symbols. Commencing with the tarsus bone and proceeding proximally, regeneration score maxima occur at the joints, and regeneration score minima occur in ossified regions of the bones. In stages 55 and 56, the differences between maxima and minima are significant at the 0.05 level. In stage 57, the maxima and minima are still present, but with the exception of the maximum for the tarsal-metatarsal joint, they are no longer significantly different at the 0.05 level.

One possible source of regenerative variation in results within and between laboratories is the difficulty of performing amputations with precision. The hindlimbs are very small, and at stages used in many investigations, the cartilaginous portions are very close to the desired ossified amputation planes. The differences between the cartilaginous and ossified amputation plane scores are so large that only a few incorrectly placed amputations will change regeneration results, especially with the 20–30 limb sample sizes often seen in hindlimb experiments.

Next, we tested the hypothesis that abnormal tadpole hindlimb regeneration scores would be lower than those of normal tadpoles. DA hindlimb regeneration mean scores and standard errors for stages 55–57 are given in Table 1. We conducted pair-wise comparisons as described previously. There were 19 normal-abnormal regeneration score sets. Fifteen of the abnormal mean scores differed significantly from those of the normal mean scores at the 0.05 significance level with P values ranging from <0.0001 to 0.0130. A graphic presentation of the comparisons for the tarsus and tibia-fibula amputation planes is shown in Figure 4. Even though the abnormal data set sample sizes are small, it is possible to see that if their data were combined with those of the normal tadpoles, overall regeneration scores would be reduced. For example, addition of the DA data to the normal data for stage 55 and 56 tibia-fibulas triples the size of the standard errors and decreases the scores by 4.5 and 3%, respectively.

Figure 4.

Pair-wise comparison of normal and developmentally abnormal (DA) tadpole tarsus and tibia-fibula scores for stages 55–57. All pair-wise combinations were calculated with the General Linear Model Least Squares Means Multiple Comparison protocol (with Tukey-Cramer adjustments). A visual representation for 6 of these pairs is shown above. The bars represent the standard errors. The differences are all significant at the 0.05 level.

We conclude that experiments designed to change regeneration capabilities may require large sample sizes. We also conclude that the presence of even relatively small numbers of abnormal tadpoles can adversely affect Xenopus hindlimb regeneration scores and introduce regenerative variation.

Xenopus Tadpole Hindlimb Regeneration Capability Decline Is Correlated With Rates of Increasing Ossification in the Skeletal Elements

Since the presence of ossification in a skeletal element amputation plane is related to its regeneration potential, we quantified ossification. An average of 50 each femurs, tibia-fibulas, and tarsus bones of tadpoles in stages 54–59 were stained for ossified tissue with Alizarin Red. The skeletal elements were photographed, and the resulting slide film images were projected onto a screen for large-scale measurement with a ruler (see Experimental Procedures section). The lengths of ossified tissue were divided by the total lengths of the skeletal elements. Mean ossification percentages and standard errors were calculated. The results are given in Table 2.

Table 2. Percentages of Ossification and Regeneration Performance in Femurs, Tibia-Fibulas, and Tarsus Bones in Tadpole Stages 54–59a
Stage boneMean ossification (%)Std. errornMean regeneration (%)Std. errorn
  • a

    F, femur; TF, tibia-fibula; T, tarsus; n = sample size.

54 F25.21.012025.45.7820
54 TF23.80.613740.85.6123
54 TNoneNoneNone55.84.0446
55 F30.32.13155.673.2661
55 TF29.10.772517.51.62127
55 T20.81.002025.91.76103
56 F33.01.00424.152.6531
56 TF33.60.548613.01.53140
56 T23.40.518517.01.67118
57 F43.70.86490.493.2132
57 TF49.30.661435.741.65123
57 T39.80.711309.601.07287
58 F51.60.73220.553.1510
58 TF57.10.80461.572.1025
58 T52.40.82474.401.9831
59 F57.82.1990.513.1914
59 TF68.21.26442.400.2924
59 T63.61.64363.804.4021

The tadpole growth curve (see Fig. 1) for our laboratory revealed that an average of 6 days separates each of the tadpole stages beginning at stages 54 through 57, then 4 days separate stages 57–58 and 58–59. Linear regressions were performed to calculate the daily rates of increase of percentages of ossified tissue for the femur, tibia-fibula, and tarsus bones (see Fig. 5A). These rates were 1.27, 1.72, and 2.25% for the femur, tibia-fibula, and tarsus, respectively. The average daily ossification rate for the three skeletal elements together was 1.75%.

Figure 5.

Change in percent ossification and regeneration with tadpole growth in Xenopus stage 54–59 hindlimbs. Percent ossified tissue in femurs, tibia-fibulas, and tarsus bones was calculated by dividing the length of the ossified portion of the bone stained by Alizarin Red by the total length of the bone. Regeneration percent values are from Table 2. Percentages were related to daily tadpole growth by linear regression. A: The average daily increase of ossification for all three bones is 1.75%. B: The average daily rate of regeneration capability decline for the three bones is −1.35%. Spearman Rank Order Correlation calculations determined that increasing ossification and decreasing regeneration rates are inversely correlated at the 0.05 significance level with P < 0.0001.

Linear regression calculations were also performed for femur, tibia-fibula, and tarsus amputation planes for stages 54–59, plotting mean regeneration scores (Y axis) against daily tadpole growth (X axis, see Fig. 5B). The P values for both regressions together ranged from <0.0001 to 0.0416, so these regression values were significant at the 0.05 level. If the daily rates of decline in the percentage of regeneration performance for the three skeletal elements are compared, the most proximal skeletal element (the femur) is found to have the least steep decline in regeneration capability (-1.27% per day), followed by the tibia-fibula (-1.72% per day). The most distal element (the tarsus) has the steepest decline (-2.25% per day). The average decline of the three skeletal elements together is -1.35% per day. To assess the degree of correlation between ossification and regeneration in the three skeletal elements, the average ossification and regeneration rates were compared. The Pearson Correlation Coefficient did not meet the rigor we set for the statistical significance in this study (the P value was 0.093, larger than 0.05). This is not surprising, since the average rate of increase in ossification (1.75%) and the average rate of regeneration decline (-1.35%) are not exactly equal and opposite. In addition, the ossification data were parametric, while the regeneration data were non-parametric (Pagano and Gauvreau, 2000). When the non-parametric Spearman Rank Order Correlation was calculated for the data sets, the two are found to be inversely related with P < 0.0001, indicating that this relationship is highly significant.

We conclude that the percent of hindlimb skeletal element ossification is an excellent predictor of the regeneration potential present. We also observe that the use of statistical methods in the analysis of morphogenetic and regeneration phenomena yields more detailed information, aids in experimental design, and decreases variability in observed results. There are many sources of regenerative variation in Xenopus hindlimb research, but it is possible to identify and minimize them by eliminating developmentally abnormal animals, using adequate sample sizes, and placing amputation planes as precisely as possible.


Source and Care of Tadpoles

Xenopus laevis tadpoles were obtained from Xenopus I, Inc. (Dexter, MI). Tadpoles were kept in dechlorinated water and 7% Holtfretter's solution at 22°C. Tadpole population density was maintained at one tadpole per 500 ml. Containers were cleaned and tadpoles were fed daily, minimizing bacterial contamination as described in McDiarmid and Altig (1999). The food source was tadpole powder (Xenopus Express, Homossassa, FL; Xenopus I, Inc., Dexter, MI).

Anesthetization, Amputation, Regeneration, and Euthanization

Prior to surgery, tadpoles were anesthetized in a 0.04% MS222 (amino benzoic acid ethyl ester), then removed to 7% Holtfreter's solution to prevent over-anesthetization. Anesthetized Nieuwkoop and Faber (1967) stages 54–59 tadpoles were amputated at one of seven locations along the proximodistal length of the left hindlimb as shown in Figure 5. The tadpoles were returned to normal culture conditions and allowed to regenerate for 4 weeks. Prior to collection of samples, tadpoles were placed in a 10× (0.4%) MS222 solution, which causes euthanization within a few minutes.

Whole Skeletal Element Cartilage Staining and Ossified Tissue Staining

The cartilage staining protocol was based on Bryant and Iten (1974). Collected hindlimbs were fixed in Gregg's fixative, dehydrated, and stained with 1% Victoria Blue solution for 2 h. After de-staining, the hindlimbs were cleared in methyl salicylate. The limbs were viewed under a binocular microscope, and skeletal elements distal to the plane of amputation were identified and counted. The ossified tissue staining protocol was modified from Kimmel and Trammel (1981). Collected hindlimbs were placed without fixation in 0.0025% Alizarin Red in 2% KOH for 1–3 h, then cleared in 50% glycerol in distilled water for 6 h. Skeletal elements were photographed with a Nikon 35-mm camera mounted on a Nikon Optiphot microscope. Images of femurs, tibia-fibulas, and tarsus bones were enlarged with a slide projector. The lengths of Alizarin Red-stained tissue and the total lengths of the three skeletal elements were measured in inches with a ruler. The tibia-fibula and tarsus bones have anterior and posterior elements that ossify at the same rate (data not shown). The anterior elements were used for measurements in most cases.

Whole Skeletal Element Regeneration Scoring System

The whole limb scoring system is modified slightly from Wolfe et al. (2000) and it is based on the Stocum (1968) designation of joint sites as boundaries for the segments of the limb. The scoring system is described in Figure 3. Regeneration scores constituted non-parametric discrete data as described in Pagano and Gauvreau (2000), with sample sizes large enough to be able to apply the Central Limit Theorem. This theorem states that if sample sizes are large, and large degrees of freedom are considered for comparisons of data sets, then assumptions and calculations can be made that essentially amount to referring to a standard normal distribution for P values.

Statistical Analysis

Statistical Analysis System (SAS Institute Inc., Cary, NC) software was used for all analyses. Mean regeneration scores, mean ossification percentages, and most linear regressions were calculated with the General Linear Model procedure. Tadpole growth curve piecewise regressions were calculated with the Non-Linear Model procedure. Multiple pair wise comparisons and P values were performed with the General Linear Model Least Square Means procedure calculated with Tukey-Cramer adjustments.


We acknowledge David Zimmerman for conducting studies that characterized developmentally abnormal tadpole hindlimb regeneration results. Special thanks are due to Patrick Redwood for assisting in the ossification study and assembling our data for statistical analysis. We are grateful to the University of Illinois Statistics Department Consulting Division for analysis of the regeneration data and especially Professor Jeffrey Douglas for helpful discussions. Funds for this study were provided by the Cell and Developmental Biology Department at the University of Illinois.