
Graphical abstract


Figure 1. NFκB rapidly and transiently translocates to the nucleus following tail amputation, and this requires active NADPH oxidase
(A–C) Western blots showing RelA in cytoplasmic (c) and nuclear (n) protein fractions from stage 55 tail sections as drawn, at indicated hours post amputation (hpa). Arrows indicate approximate site of cuts. Nonspecific (NS) bands for each fraction are included as loading controls. (A) Untreated tadpoles (B) Tadpoles treated with 2 μM DPI from 1 h before amputation, to disrupt ROS production. (C) Graph showing percentage of tadpoles regenerating tails after 7 days, following chemical inhibition of NADPH oxidase (Nox) with 2 μM DPI treatment starting 1 h before amputation of the distal third of stage 43 tadpole tail, or 0.1% DMSO vehicle. Single cohort of tadpoles with N = 14–15 tadpoles per replicate Petri dish; points represent three replicates for each time point. Data were analyzed by twoway ANOVA followed by Sidak's multiple comparisons of all means; adjusted p values ∗p < 0.05, ∗∗p < 0.01 are shown for treatments that differ significantly from the equivalent vehicle control. (C') Bar graphs indicate the distribution of phenotypes observed in (C). Raw data can be found in Data S1.


Figure 2. Activation of NFκB is sufficient for tadpole tail regeneration, and the direct target cox2 is upregulated in the wound epithelium of regenerationcompetent appendages
(A) Tadpoles in refractory stage 46 are more likely to regenerate a tail if exposed to 10 μM prostratin, an indirect activator of NFκB, for 30 min after amputation, compared with vehicletreated controls (0.1% EtOH). Points on scatterplots represent the percentage of regenerating tails with 28–34 individuals in a replicate Petri dish. Analysis by unpaired t test showed the effect of prostratin did not reach significance, although the trend is clear.
(A′) Bar graphs indicate the distribution of phenotypes observed in (A). Sibships are separate cohorts of tadpoles (different parents). Raw data can be found in Data S1.
(B) Representative examples of cox2 expression (dark purple) in tails of stage 52 tadpoles at the indicated time postamputation. Dotted lines indicate planes of amputation, and arrows indicate the limits of cox2 expression in the distal epithelial cells, or tail tip in 6 dpa. Scale bar, 500 μM. Red box shows a zoom of 6 hpa to show specific cox2 expression localized to the wound epithelium (red arrowheads) as distinct from melanophores (scattered black dots).
(C) Representative examples of cox2 expression in limb buds amputated at future ankle level at stages 51 (good regenerators, all 5 digits normally regenerate), 53 (3–5 digits regenerate), and 55 (hypomorphic regenerators, 0–3 digits regenerate). Examples are shown at various times after partial amputation, and dotted lines on controls indicate the plane of amputation. Black arrows indicate expression of cox2 in the distal cells at 6 hpa and 1 dpa. Stronger and broader expression is seen at the most regenerationcompetent stage, 51. Scale bars, 500 μM.


Figure 3. Resident microbes may activate regeneration of tadpole tails
(A–D) Scatterplots showing percentage of tadpoles regenerating after amputation of the distal third of the tail at stage 46, with points representing percentage of tadpoles regenerating in each replicate dish, and stacked bar graphs of the same data by regeneration phenotype (A′’–D′). Raw data can be found in Data S1. (A) Local brief application of either 0.3% or 3% H2O2 to tadpole tail stumps immediately after amputation prevents regeneration completely. This effect can be reversed by addition of heatkilled (HK) E. coli. Data were analyzed using oneway ANOVA and Tukey’s multiple comparisons test, ∗∗∗p < 0.001. Each point represents an experimental dish with N = 14–20 tadpoles. (B) Raising tadpoles in the broadspectrum antibiotic gentamicin (100 μg/mL) to prevent skin bacterial colonization and growth prevents some tadpoles from regenerating. The later the treatment is started, the more tadpoles regenerate. In each case, more tadpoles regenerate if LPScontaining HK E. coli is added just after amputation. Analysis by oneway ANOVA and Tukey multiple comparisons test did not reveal any significant p values, but the data trends are clear. Each point represents an experimental dish with N = 21–48 tadpoles. (C) Significantly more tadpoles regenerate when commercial purified LPS (50 μg/mL, from E. coli 0111:B4) are added to the medium just after tail amputation, for 1 h (unpaired t test, p = 0.002). Each point represents an experimental dish with N = 19–24 tadpoles. (D) Regeneration decreased when tadpoles were either soaked in 0.3% H2O2 for 2 min before amputating the tail or raised with antibiotics (N = 12 to 22, ∗∗p < 0.01, oneway ANOVA, Dunnett's multiple comparisons with controls/MMR). Adding exogenous LPS immediately after amputation rescues the effect of raising in antibiotics.


Figure 4. Raising tadpoles in antibiotics influences tail regeneration
(A–C) Scatterplots showing percentage of tadpoles regenerating after amputation of the distal third of the tail at stage 46, percentage of tadpoles regenerating in each replicate dish, and stacked bar graphs of the same data by regeneration phenotype (A′–C′). Raw data can be found in Data S1. (A) Raising tadpoles in either gentamicin 50 μg/mL or penicillin/streptomycin from 2 to 4cell stage significantly reduces the number of tadpoles that regenerate tails in the refractory period in two sibling cohort groups (twoway ANOVA, Tukey's multiple comparisons test all means). Each point represents an experimental dish with N = 20–33 tadpoles. (B and C) Antibiotics do not directly alter the regeneration process of Xenopus tadpole tails. Culturing embryos from 2 to 4 cells in penicillin/streptomycin (B) or 100 μg/mL gentamicin (C) significantly reduces the number of tadpoles regenerating, but adding the same antibiotics after cutting does not hinder regeneration (oneway ANOVA, Tukey's multiple comparisons test of all means). Each point represents an experimental dish with N = 28–34 tadpoles. ∗p < 0.05, ∗∗p < 0.01.


Figure 5. Shortterm chemical genetic activation of NFκB enhances hindlimb regeneration at stage 56, but regeneration is limited to 3 digits
(A) Stage 56 hindlimbs were amputated at knee level (red arrowheads). The right limb was treated with 1 μL topical 100 μM prostratin in 1% ethanol and the left limb treated with vehicle (1% ethanol), for 17 min. Both limbs were then treated with 40 μM celecoxib for 90 min. Tadpoles were allowed to regenerate until stage 58 and are viewed from the ventral side (so the left control hindlimb appears on the right of the panels). Numbers in top right of each image indicate the score for each limb (right, prostratin and celecoxib; left, vehicle and celecoxib).
(B) Violin plot to show number of digits regenerated after immediate treatment of stage 56 limb stumps with 1% ethanol for 17 min (control) or 100 μM prostratin for 17 min (n = 20), or the same two treatments followed by 40 μM celecoxib for 90 min (n = 17, tadpoles shown in A). Prostratin followed by celecoxib after 17 min resulted in significantly more digits regenerating than any other treatments, oneway ANOVA and Tukey's multiple comparisons to all means, ∗∗∗∗p < 0.0001, median shown by solid line and quartiles by dotted lines. Number above violin indicates size of sample (N).


Figure 6. Brief activation of NFκB using chemical genetics (prostratin for 20 min followed by celecoxib) does not enhance forelimb regeneration, but does increase blastema size
(A) Numbers on limbs indicate days after amputation. A full stage series is shown for froglets 6 and 9, with just the 17 dpa limbs, which have clear blastemas, shown for froglets 10, 11, 12, 13, 14, and 15. In each case the left forelimb has been treated with vehicle (1% EtOH) and the right one with 100 μM prostratin in 1% EtOH for 20 min following amputation at the mid forearm level, and then with 40 μM celecoxib for 90 min. Scale bar, 5 mm.
(B) Blastema size was estimated from photographs at 17 dpa in n = 15 froglets using ImageJ. Left control limbs formed significantly smaller blastemas, p = 0.0017, paired t test. Left and right limbs from the same froglet are connected by a dotted line.
(C) Bone and cartilage were stained in 5 froglets, 2 months after amputation; 4 sets of forelimbs are shown here. The alizarin red bone (red) ceases at the point the forelimbs were amputated, and a cartilage spike formed, with no joints or branches, in each case (pale blue). Numbers on the top right indicate froglet number; L and R indicate left and right forelimb, respectively; scale bar, 2 mm, ∗∗p < 0.01. Raw data can be found in Data S1.


Figure 7. Model for how skin bacteria could result in tail regeneration in refractorystage tadpoles
ROS are produced rapidly when the tail is cut. We propose a model where gramnegative bacteria on tadpole skin can activate TLR4 receptors on tissueresident macrophages when the tail is cut. This leads to rapid activation of the NFκB transcription factor, which in turn upregulates expression of NADPH oxidases (Nox2, Nox4) allowing sustained production of ROS, as well as upregulation of the proinflammatory enzyme Cox2. In this model, raising tadpoles with antibiotics to prevent Gramnegative bacteria colonization of skin or denaturing LPS on skin would result in reduced activation of TLR4, no activation of NFκB, and loss of sustained ROS.


Figure S1: In situ hybridisation using nox1, 2 and 4 probes. Stage 51 Xenopus
hindlimbs amputated at future knee level (a) or stage 41 tails following amputation of the
distal third (b) and fixed at the indicated hours or days after amputation (hpa/dpa).
Representative examples are shown, expression staining is dark purple and indicated by
black arrows. (a) Nox1 was not detectable, nox2 (cybb) is expressed in punctate cells near
the wound and nox4 in cells near the wound and maintained in the regeneration bud cells of
stage 51 limbs. None of the probes showed detectable staining in control uncut limbs (stage
52 shown). (b) In situ hybridisation using nox2 probe in tadpole tails during regeneration
again shows staining in punctate cells, present in the tail and fin before amputation (control),
that gather near the wound site from 6 hpa. Dotted lines indicate approximate level of
amputation in control limbs/tails and scale bar is 500µM.


Figure S2: Adding purified LPS from P. aeruginosa (PALPS) at 50µg/ml after tail amputation
increases the number of tadpoles regenerating tails. Top: left and right scatter plots indicate two
sibships of tadpoles (different mothers), with each point for a treatment being a replicate petri dish.
Gentamicin treatment was 100µg/ml from 4 cell stage. Statistical analysis with one way ANOVA and
Tukey’s multiple comparisons of all means for each cohort. Bottom: same data presented as
regeneration phenotypes. Sample sizes: N for left graph control (30,26,33); PALPS (26,25,25);
Gentamicin (18,18,20) Gentamicin + PALPS (17,17) and for right graph: control (27,23,26); PALPS
(34,35,37); Gentamicin (27,27,28) Gentamicin + PALPS (26,25,25). * p<0.05, ** p<0.01, *** p<0.001.
Note that the right cohort did not generate significant data, but the trend is the same for both cohorts,
and that data from control and gentamicin sibship 1 are the same data as Figure S2 sibship 1, Raw
data can be found in Supplemental Data S1.


Figure S3: Different regeneration competence is observed when tadpoles are raised in
different antibiotics. Tadpoles were raised from the 4 cell stage with gentamicin sulphate (50µg/ml),
streptamycin (100µg/ml, penicillin G (100U/ml), or no antibiotics (control). Top: left and right scatter
plots graphs indicate two sibships of tadpoles (different parents), with each point for a treatment being
a replicate petri dish. Statistical analysis with one way ANOVA and Dunnet’s multiple comparisons to
the control for each cohort. Bottom: same data presented as regeneration phenotypes. Sample sizes:
N for sibship 1 control (30,30, 32, 27); gentamicin (28, 32, 31, 32); streptamycin (27, 22, 26) penicillin
G (25, 24, 28) and for sibship 2 control (30, 26, 33); gentamicin (18, 18, 20); streptamycin (31, 34, 37,
36) penicillin G (33, 31, 32, 31). * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001. Note that the base
regeneration differs between sibships of tadpoles, and that the naturally poor regenerating sibship 1
has no response to penicillin G. Raw data can be found in Supplemental Data S1.


Figure S4: Brief activation of NFkB using chemical genetics (DPP for 20 minutes followed by
celecoxib) increases blastema size but does not enhance forelimb regeneration. White numbers
in top right indicate froglet number, only 4 froglets were used. (a) numbers on limbs indicate days
after amputation. A full stage series is shown for froglets 3 and 4. In each case the left forelimb has
been treated with vehicle (10% DMSO) and the right one with 750µM DPP in 10% DMSO for 20
minutes following amputation at the mid forearm level, then with 40µM celecoxib for 90 minutes (b)
blastema size was estimated from photographs at 17 dpa in n=4 froglets using Image J. Left control
limbs formed smaller blastemas in all 4 froglets, but the mean size was not significant (Paired ttest).
Left and right limb data from the same froglet are linked via dotted line. (c) Bone and cartilage were
stained in two froglets, two months after amputation. The alizarin red bone (red) ceases at the point
the forelimbs were amputated, and a cartilage spike formed, with no joints or branches, in each case
(pale blue), except for two tiny bone fragments in the right limb of froglet 3 (3R, red arrow). Box
indicates area of zoom in side view of limb 3R. Numbers in top right indicate froglet number, L and R
indicate left and right forelimb. Raw data can be found in Supplemental Data S1. Scale bars =
500µM.
