of 45
1
A conserved strategy for inducing appendage regeneration
1
in moon jellyfish,
Drosophila
, and mice
2
Michael J. Abrams
1
, Fayth Tan
1
,
Yutian Li
1
,
3
Ty Basinger
1
, Martin
L.
Heithe
1
,
Anish
A.
Sarma
1
,
Iris
T.
Lee
1
,
Zevin
J.
Condiotte
1
,
4
Misha Raffiee
1
‡, John
O. Dabiri
2
, David A. Gold
1
‡, Lea Goentoro
1
*
5
6
1
Division of Biology and Biological Engineering, California Institute of Technology, Pasadena, CA
7
91125
8
2
Graduate Aerospace Laboratories and Mechanical Engineering, California Institute of Technology,
9
Pasadena
, CA 91125
10
These authors contributed equally to the work.
11
Current addresses: Department of Molecular and Cell Biology, University of California at Berkeley
12
(M.J.A.), Department of Bioengineering, Stanford University (M.R.), Department of Biology and Allied
13
Health Sciences, Bloomsburg University (T.B.), Departmen
t of Earth and Planetary Sciences, University
14
of California at Davis (D.A.G)
15
*
Correspondence: goentoro@caltech.edu
16
A
lternative title:
Amino acid and sugar supplement induces appendage regeneration in
17
cnidarian, insect, and mammal
18
Key words:
Regeneration,
appendage, limb, amino acid, leucine, sugar, insulin, jellyfish,
19
Drosophila
, mouse
20
Abstract
21
Can limb regeneration be induced? Few have pursued this question, and an evolutionarily
22
conserved strategy has yet to emerge. This study reports a strategy for in
ducing regenerative
23
response in appendages, which works across three species that span the animal phylogeny. In
24
Cnidaria, the frequency of appendage regeneration in the moon jellyfish
Aurelia
was increased
25
by feeding with the amino acid L
-
leucine and the g
rowth hormone insulin. In insects, the same
26
strategy induced tibia regeneration in adult
Drosophila
. Finally, in mammals, L
-
leucine and
27
sucrose administration induced digit regeneration in adult mice, including dramatically from
28
mid
-
phalangeal amputation.
The conserved effect of L
-
leucine and insulin/sugar suggests a key
29
role for energetic parameters in
regeneration induction. The simplicity by which nutrient
30
supplementation can induce appendage regeneration provides a testable hypothesis across
31
animals.
32
In
troduction
33
In contrast to humans’ poor ability to regenerate, the animal world is filled with seemingly
34
Homeric tales: a creature that regrows when halved or a whole animal growing from a small
35
body piece. Two views have historically prevailed as to why so
me animals regenerate better than
36
others (Goss, 1992). Some biologists, including Charles Darwin and August Weismann, hold that
37
regeneration is an adaptive property of a specific organ
(Polezh
a
ev, 1972)
. For instance, some
38
lobsters may evolve the ability t
o regenerate claws because they often lose them in fights and
39
food foraging. Other biologists, including Thomas Morgan, hold that regeneration is not an
40
evolved trait of a particular organ, but inherent in all organisms
(Morgan, 1901)
. Regeneration
41
evolvin
g for a particular organ versus regeneration being organismally inherent is an important
42
2
distinction, as the latter suggests that the lack of regeneration is not due to the trait never having
43
evolved, but rather due to inactivation
and
may therefore be ind
uced. In support of Morgan’s
44
view, studies in past decades have converged on one striking insight: many animal phyla have at
45
least one or more species that regenerate body parts (Sanchez
-
Alvarado, 2000; Bely and Nyberg,
46
2010). Further, even in poorly regen
erative lineages, many embryonic and larval stages can
47
regenerate.
In regenerating animals, conserved molecular events (
e.g.,
Cary et al., 2019
,
48
Kawakami et al., 2006
) and
regeneration
-
responsive enhancers
(Wang et al., 2020)
were
49
identified.
Although the
hypothesis
of convergen
t evolution
cannot be fully excluded (
e.g., Lai
50
and Abo
o
bak
e
r, 2018
),
these
findings
begin to build the case that
the ability to regenerate
may
51
be
ancestral (Sanchez
-
Alvarado, 2000; Bely and Nyberg, 2010
)
.
Regeneration
being
possibly
52
ancestral begs the question: is there a conserved mechanism to activate regenerative state?
53
This study explored how, and whether, limbs can be made to regenerate in animals that do
54
not normally show limb regeneration. In
adult
frogs, studies
from the early 20
th
century and few
55
recent ones have induced various degrees of outgrowth in the limb using strategies including
56
repeated trauma, electrical stimulation, local progesterone delivery, progenitor cell implantation,
57
and Wnt activation (Carlso
n, 2007; Lin et al., 2013; Kawakami et al., 2006).
Wnt
activation
58
restored limb development in chick embryos (Kawakami et al., 2006), but there are no reports of
59
postnatal regeneration induction. In salamanders, a wound site that normally just heals can be
60
induced to grow a limb by supplying nerve connection and skin graft from the contralateral limb
61
(Endo et al., 2004), or by delivery of Fgf2, 8, and Bmp2 to the wound site followed by retinoic
62
acid (Viera et al., 2019).
In
neonatal and
adult
mouse digits,
a model for exploring limb
63
regeneration in mammals, bone outgrowth or joint
-
like structure can be induced via local
64
implantation of Bmp2
(bone)
or
Bmp
9 (
joint;
Yu et al., 2019)
.
Thus far, different strategies gain
65
tractions in different species, and a comm
on denominator appears elusive.
66
However, across animal phylogeny, some physiological features show interesting
67
correlation with regenerative ability (Hariharan et al., 2015; Vivien et al., 2016; Sousounis et al.,
68
2014).
First, regeneration
especially in vertebrates
tends to decrease with age, with juveniles and
69
larvae more likely to regenerate than adults. For instance, the mammalian heart rapidly loses the
70
ability to regenerate after birth and anurans cease to regenerate limbs upon metamorp
hosis.
71
Second, animals that continue to grow throughout life tend to also regenerate. For instance, most
72
annelids continue adding body segments and regenerate well, a striking exception of which is
73
leeches that make exactly 32 segments and one of the few a
nnelids that do not regenerate body
74
segments
(Rouse, 1998)
. Consistent with the notion of regeneration as ancestral, indeterminate
75
growth is thought of as the ancestral state (Hariharan et al., 2015). Finally, a broad correlate of
76
regenerative ability acro
ss animal phylogeny is thermal regulation. Poikilotherms, which include
77
most invertebrates, fish, reptiles and amphibians, tend to have greater regenerative abilities than
78
homeotherms
birds and mammals are animal lineages with poorest regeneration. These
79
p
hysiological correlates, taken together, are united by the notion of energy expenditure. The
80
transition from juvenile to adult is a period of intense energy usage, continued growth is
81
generally underlined by sustained anabolic processes, and regulating bod
y temperature is
82
energetically expensive compared to allowing for fluctuation. Regeneration itself entails
83
activation of anabolic processes to rebuild lost tissues (Hirose et al., 2014; Naviaux et al., 2009;
84
Malandraki
-
Miller et al., 2018). These physiolog
ical correlates thus raise the notion of a key role
85
of energetics in the evolution of regeneration in animals. Specifically, we wondered whether
86
energy inputs can promote regenerative state. In this study, we demonstrate that nutrient
87
3
supplementation can i
nduce regenerative response in appendage and limb across three vastly
88
divergent
species
.
89
Results
90
Leucine and insulin promote appendage regeneration in the moon jelly
Aurelia
91
We reasoned that if there was an ancestral mechanism to promote regeneration, it w
ould likely
92
remain
intact in early
-
branching lineages
with prevalent
regeneration across the species
. In
93
Cnidaria, the ability to regenerate is established in polyps,
e.g.
, hydras and sea anemones. Some
94
cnidarians, notably jellyfish, not only exist as
sessile polyps, but also as free
-
swimming ephyrae
95
and medusae (Figure 1a). In contrast to the polyps’ ability to regenerate, regeneration in ephyrae
96
and medusae appears more restricted
in some
species
(Abrams et al., 2015
;
Sinigaglia et al 2020
;
97
Schmid and
Alder, 1984
). We focused on the moon jellyfish
Aurelia coerulea
(
for
merly
A.
98
aurita
sp. 1 strain)
, specifically on the ephyra, whose eight arms facilitate morphological
99
tracking (Figure 1b).
About 3 millimeter in diameter,
Aurelia
ephyrae regenerate tips
of arms and
100
the distal sensory organ rhopalium, but upon more dramatic amputations such as removing a
101
whole arm or halving the body, rapidly reorganize existing body parts and regain radial
102
symmetry (Figure 1c). Observed across four scyphozoan species, sym
metrization occurs rapidly
103
within 1
-
3 days and robustly across conditions (Abrams et al., 2015). Ephyrae that symmetrized
104
matured into medusae, whereas ephyrae that failed to symmetrize and simply healed the wound
105
grew abnormally.
106
107
108
Figure 1.
Aurelia
as a system to identify factors that promote appendage regeneration.
109
(a)
The moon jellyfish
Aurelia aurita
have a dimorphic life cycle, existing as sessile polyps or free
-
110
swimming medusae and ephyrae. Ephyra is the juvenile stage of medusa, a robust sta
ge that can withstand
111
months of starvation. In lab conditions, ephyrae mature into medusae, growing bell tissue and
112
reproductive organs, in 1
-
2 months.
113
(b)
Ephyrae have eight arms, which are swimming appendages that contract synchronously to generate
114
axisy
mmetric fluid flow, which facilitates propulsion and filter feeding. The eight arms are symmetrically
115
positioned around the stomach and the feeding organ manubrium. Extending into each arm is radial
116
muscle (shown in Figure 2) and a circulatory canal that t
ransports nutrients. At the end of each arm is the
117
light
-
and gravity
-
sensing organ rhopalium.
118
(c)
In response to injury, the majority of ephyrae rapidly reorganize existing body parts and regain radial
119
symmetry.
However, performing the experiment in the n
atural habitat, a few ephyrae (2 of 18) regenerated
120
a small arm (arrow).
121
Intriguingly,
in the course of our previous study (Abrams et al., 2015), we observed
in a
122
few symmetrizing ephyrae, a small bud at the amputation site. To follow this hunch, we repeat
ed
123
the experiment in the original habitat of our lab’s polyp population, off the coast of Long Beach,
124
CA (Methods). Two weeks after amputation, most ephyrae indeed symmetrized, but in 2 of 18
125
animals a small arm grew (Figure 1
c
). This observation suggests
that, despite symmetrization
126
4
being the more robust response to injury, an inherent ability to regenerate arm is present and can
127
be naturally manifest. The inherent arm regeneration presents an opportunity: Can arm
128
regeneration be reproduced in the lab, as
a way to identify factors that promote regenerative
129
state?
130
To answer this question, we screened various molecular and physical factors (Figure 2a,
131
Figure
2
figure supplement 1). Molecularly, we tested modulators of developmental signaling
132
pathways as well
as physiological pathways such as metabolism, stress response, immune and
133
inflammatory response. Physically, we explored environmental parameters, such as temperature,
134
oxygen level, and water current. Amputation was performed across the central body removi
ng 3
135
arms (Figure 2a). Parameter changes were implemented or molecular modulators (
e.g.
, peptides,
136
small molecules) were introduced into the water immediately after amputation. Regenerative
137
response was assessed for 1
-
2 weeks until the onset of bell growth
, which hindered the scoring of
138
arm regeneration (Figure 2
figure supplement 2).
139
140
Figure 2. Arm regeneration in
Aurelia
ephyra can be induced using exogenous factors.
141
(a)
Ephyrae were amputated (red line) across the body to remove 3 arms, and then let
recover in various
142
conditions. Figure 2
figure supplement 1
tabulates the factors tested in the screen. Regeneration was
143
assessed over 1
-
2 weeks until bell tissues began developing between the arms and obscured scoring.
The
144
ephyrae shown
are
from high
-
nutr
ient condition (see Figure 3).
145
(b)
Arm regeneration (arrows; from high food condition, see Figure 3a).
146
(c)
Radial circulatory canal in
an uncut arm and is reformed in an arm regenerate.
147
(d)
Muscle (red), as indicated by phalloidin staining, and neuronal
networks (green), as indicated by
148
antibody against tyrosinated tubulin. The orange arrows indicate distal enrichment of tyrosinated
-
tubulin
149
staining, which marks the sensory organ rhopalium (rho). Twenty ephyrae were examined and
150
representative images are
shown.
151
5
(e)
Higher magnification of the phalloidin staining shows the striated morphology of the regrown muscle
152
in the arm regenerate (called radial muscle), which extends seamlessly from circular muscle in the body.
153
3 supplements: Figure 2
figure
supplement 1
-
3
.
Source file: Aurelia screen.xls
154
After 3 years of screening, only three factors emerged that strongly
promoted
arm
155
regeneration (Figure 2b). The ephyrae persistently symmetrized in the majority of conditions
156
tested. In the few conditions
where regeneration occurred, arm regenerates show multiple tissues
157
regrown in the right locations: circulatory canals, muscle, neurons, and rhopalium (Figure 2c
-
e
)
.
158
The arm regenerates contracted synchronously with the original arms (Video 1), demonstratin
g a
159
functional neuromuscular network. Thus, arm regeneration in
Aurelia
that was observed in the
160
natural habitat can be recapitulated in the lab by administering specific exogenous factors.
161
The extent of arm regeneration varied, from small to almost fully
sized arms (Figure 2b).
162
The variation manifested even within individuals: a single ephyra could grow differently sized
163
arms. Of the three arms removed, if regeneration occurred, generally one arm regenerated (67%),
164
occasionally 2 arms (32%), and rarely 3 a
rms (1%, of the 4270 total ephyrae quantified in this
165
study). Finally, the frequency of regeneration varied across clutches,
i.e.
, strobilation cohorts.
166
Some variability may be due to technical factors,
e.g.
, varying feed culture conditions; however,
167
varia
bility persisted even with the same feed batch. We verified that the variability was not
168
entirely due to genetic differences, as it manifested across clonal populations (Figure 2
figure
169
supplement 3). Thus, there appears to be stochasticity in the occurren
ce of arm regeneration in
170
Aurelia
and the extent to which regeneration proceeds.
171
What are the factors that promote arm regeneration? Notably, modulation of developmental
172
pathways often implicated in regeneration literature (
e.g.
, Wnt, Bmp, Tgfß) did not p
roduce
173
effect in the screen (Figure 2
figure supplement 1
)
although
we do not
rule
out their
174
involvement
in other capacity, e.g., in downstream patterning.
We first identified a necessary
175
condition: water current.
The requirement for current for promoting
regeneration is interesting
176
because ephyrae can recover from injury by symmetrizing in stagnant water (Figure 1c). Thus, a
177
specific physiological state is required for enabling regenerative response.
Behaviorally,
the
178
presence of current
promotes
more
swimming, while in stagnant water ephyrae tend to rest at the
179
bottom and pulse stationarily (Figure 3
figure supplement 1 and Video 2 show the aquarium
180
setup used to implement current). In this permissive condition, the first factor that
promotes
181
regenerat
ion is the nutrient level: increasing food amount increases the frequency of arm
182
regeneration. To measure the regeneration frequency, we scored any regenerates with lengths
183
greater than 15% of that of an uncut arm (Figure 3a). This threshold was chosen to
184
predominantly exclude non
-
specific growths or buds that show no morphological structures
185
(Figure 3b) while including small arm regenerates that show clear morphological features,
i.e.
,
186
lappets, radial canal, and radial muscle sometimes showing growing ends
(Figure 3b). Given the
187
clutch
-
to
-
clutch variability, control and treatment were always performed side by side using
188
ephyrae from the same clutch. The effect size of a treatment was assessed by computing the
189
change in regeneration frequency relative to the
internal control. Statistical significance of a
190
treatment was assessed by evaluating the reproducibility of its effect size across independent
191
experiments (
Methods
). With
th
e
s
e
measurement and statistical methodologies, we found that
192
although the baseline
regeneration frequency varied across clutches, higher food amounts
193
reproducibly increased regeneration frequency (Figure 3c). The magnitude of the increase varied
194
(Figure 3g, 95% CI [4.7, 12.1
-
fold]), but the increase was reproducible (95% CI excludes 1)
and
195
statistically significant (p
-
value<10
-
4
).
196
6
The second factor that promotes regeneration is insulin (Figure 3d). We verified that the
197
insulin receptor is conserved in
Aurelia
(Figure 3
figure supplement 2). Administering insulin
198
led to a reproducible (Fi
gure 3g, 95% CI [1.1, 5.0
-
fold]) and statistically significant (p
-
199
value<0.05)
200
201
Figure 3. Nutrient level, insulin, hypoxia, and leucine increased regeneration frequency in
202
Aurelia
.
203
(a)
An ephyra is regenerating if it has at least one growth from the cut site with a length greater than 0.15
204
of the uncut arm length. The uncut arm length was determined in each ephyra by measuring 3 uncut arms
205
and taking the average. Lappets, the distal pai
red flaps, were excluded in the length measurement because
206
their shapes tend to vary across ephyrae. The measurements were performed in ImageJ.
207
7
(b)
The threshold 0.15 was chosen to balance excluding non
-
specific growths that show no morphological
208
structure
s (
e.g.
, as shown, lack of phalloidin
-
stained structures) and retaining rudimentary arms that show
209
morphological structures, including radial muscle sometime with growing ends (shown, phalloidin
210
stained).
211
(c
-
f)
In each experiment, treated (blue) and contro
l (grey) ephyrae came from the same strobilation.
212
(c)
Regeneration frequency in lower amount of food (LF) and higher amount of food (HF). The
213
designation “high” and “low” is for simplicity,
and does not presume the nutrient level in the wild. If we
214
were t
o speculate, the LF amount is likely closer to typical nutrient level in the wild, based on two lines of
215
evidence. First, regeneration frequency in LF is comparable to that observed in the natural habitat
216
experiment. Second, in many of the wild populations
studied, ephyrae mature to medusae over 1
-
3
217
months (Lucas, 2001), comparable to the growth rate in LF (by contrast, ephyrae in HF mature to
218
medusae over 3
-
4 weeks).
219
(d)
Regeneration frequency in 500 nM insulin.
220
(e)
Regeneration frequency in ASW with redu
ced oxygen.
221
(f)
Ephyrae recovering in low food, with or without 100 mM L
-
leucine.
222
(g)
The effect size of a treatment was computed from the ratio between regeneration frequency in treated
223
and control group within an experiment,
i.e.
, the metric Risk Ratio
(RR; RR =1 means the treatment has
224
no effect [Borenstein et al., 2009]).
The statistical significance and reproducibility of a treatment was
225
assessed by analyzing the effect size across experiments using the meta
-
analysis package, metafor
226
(Viechtbauer 2010
), in R with statistical coefficients based on normal distribution. See
Methods
for
227
details. A treatment was deemed reproducible if the 95% confidence intervals (95% CI) of RR exclude 1.
228
The p
-
value evaluates the null hypothesis that the estimate RR is 1.
Reproducibility and statistical
229
significance of each treatment were verified using another common size effect metric, Odds Ratio (Figure
230
3
figure supplement 3).
231
6 supplements: Figure 3
figure supplement 1
-
7
.
Source files:
Aurelia data.xls
.
Source codes:
R
-
codes.rtf
232
233
increase in regeneration frequency. The insulin effect was unlikely to be due to non
-
specific
234
addition of proteins, since bovine serum albumin at the same molarity showed no
statistically
235
significant
effect
(Figure 3
figure supplement 4)
. Finally, the third promoter of regeneration is
236
hypoxia (Figure 3e). We verified that the ancient oxygen sensor HIFα is present in
Aurelia
237
(Figure
3
figure supplement 2). Hypoxia led to a reproducible (Figure 3g, 95% CI [1.4, 12.0
-
238
fold]) and statistically
significant (p
-
value<0.01) increase in regeneration frequency. To reduce
239
oxygen, nitrogen was flown into the seawater, achieving ~50% reduction in dissolved oxygen
240
level (Methods). We verified that the effect was due to reduced oxygen rather than increase
d
241
nitrogen, since reducing oxygen using argon flow similarly increased regeneration frequency
242
(
Figure 3
figure supplement 4
). The factors can act synergistically (
e.g.
, insulin and high
243
nutrient level), but the effect appears to eventually saturate (
e.g.
,
hypoxia and high nutrient
244
level).
245
In addition to quantifying the number of ephyrae that regenerate, we further quantified the
246
regeneration phenotypes in each ephyra,
i.e.
, the number of arms regenerating, the length of arm
247
regenerates, and the formation of
rhopalia (Figure 3
figure supplements
5
and
6
). Nutrient level
248
strikingly improved all phenotypic metrics: not only more ephyrae regenerated in higher
249
nutrients, more ephyrae regenerated multiple arms, longer arms, and arms with rhopalia. Insulin
250
and hypo
xia, interestingly, show differential phenotypes. Most strikingly, while insulin induced
251
more ephyrae to regenerate multiple arms, hypoxia induced largely single
-
arm regenerates,
e.g.
,
252
hypoxia experiments 3 and 5 in Figure 3
figure supplement
5
c. Thus, whi
le all factors
253
increased the probability to regenerate, they had differential effects on the regeneration
254
8
phenotypes, suggesting a decoupling to a certain extent between the regulation of the decision to
255
regenerate and the regulation of the subsequent morp
hogenesis.
256
Of the three factors identified in the screen, nutrient input is the broadest, and prompted us
257
to search if a more specific nutritional component could capture the effects of full nutrients in
258
promoting regeneration. Jellyfish are carnivorous an
d eat protein
-
rich diets of zooplanktons and
259
other smaller jellyfish (Graham and Kroutil, 2001). Notably, all three factors induced growth:
260
treated ephyrae are larger than control ephyrae (Figure 3
figure supplement
7
). The growth
261
effect is interesting bec
ause of essential amino acids that must be obtained from food, branched
262
amino acids supplementation correlates positively with protein synthesis and growth, and in
263
particular, L
-
leucine appears to recapitulate most of the anabolic effects of high amino aci
d diet
264
(Lynch and Adams, 2001; Stipanuk, 2007).
Motivated by the correlation between growth and
265
increased regeneration frequency, we wondered if leucine administration could
promote
266
regeneration. Animals typically have a poor ability to metabolize leucine, such that the
267
extracellular concentrations of leucine fluctuate with dietary consumption (Wolfson et al., 2016).
268
As a consequence, dietary leucine directly influences cellular metab
olism.
Feeding amputated
269
ephyrae with leucine indeed led to increased growth (Figure 3
figure supplement 6). Assessing
270
arm regeneration in the leucine
-
supplemented ephyrae, we observed a significant increase in the
271
regeneration frequency (Figure 3f
-
g, 95%
CI [2.5, 6.6
-
fold], p
-
value<10
-
4
). Furthermore, leucine
272
treatment phenocopies the effect of high nutrients, improving all measured phenotypic metrics:
273
increasing multi
-
arm regeneration, the length of arm regenerate, and the frequency of rhopalia
274
formation
(Figure 3
figure supplement
5
-
6
).
275
These experiments demonstrate that abundant nutrients, the growth factor insulin, reduced
276
oxygen level, and the amino acid L
-
leucine promote appendage regeneration in
Aurelia
ephyra.
277
The identified factors are fundamental
physiological factors across animals. Might the same
278
factors promote appendage regeneration in other animal species?
279
Leucine and insulin induce regeneration in
Drosophila
limb
280
To pursue this question, we searched for other poorly regenerating systems, wh
ich fortunately
281
include most laboratory models.
Drosophila
, along with beetles and butterflies, belong to the
282
holometabolans
a vast group of insects that undergo complete metamorphosis, and that as
283
whole, do not regenerate limbs or other appendages as adul
ts (Hopkins and Das, 2015).
Larval
284
stages have imaginal disks, undifferentiated precursors of adult appendages such as the legs and
285
antennae, and portions of imaginal disks have been shown to regenerate (Worley et al., 2012).
286
Motivated by findings in
Aure
lia
, we asked if leucine and insulin administration can induce
287
regenerative response in the limb of adult
Drosophila
. We focused on testing leucine and insulin
288
in this study because of considerations of specificity (
i.e.
, nutrients are broad and compositio
n of
289
nutritional needs vary across species), pragmatism (
i.e.
, administering hypoxia requires more
290
complex setups), and in the case of
Drosophila
specifically,
Drosophila
being resistant to
291
hypoxia (Haddad et al., 1997).
292
We amputated
Drosophila
on the hind
limb, across the fourth segment of the leg, the tibia
293
(Figure 4a).
After amputation, flies were housed in vials with standard food (control) or standard
294
food supplemented with leucine and insulin, with glutamine to promote leucine uptake (Nicklin
295
et al., 2009) (treated) (Figure 4c). Each
fly
was examined multiple times,
t
wice in the first week,
296
and then once weekly over the course of
2
-
4 weeks.
297