, 105 (2012);338 Science et al.E. M. Granger
The Efficacy of Student-Centered Instruction in Supporting Science
This copy is for your personal, non-commercial use only.
clicking here.colleagues, clients, or customers by , you can order high-quality copies for yourIf you wish to distribute this article to others
here.following the guidelines can be obtained byPermission to republish or repurpose articles or portions of articles
): October 5, 2012 www.sciencemag.org (this information is current as of
The following resources related to this article are available online at
version of this article at: including high-resolution figures, can be found in the onlineUpdated information and services,
can be found at: Supporting Online Material
, 2 of which can be accessed free:cites 16 articlesThis article
registered trademark of AAAS. is aScience2012 by the American Association for the Advancement of Science; all rights reserved. The title CopyrightAmerican Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005.
(print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by theScience
on October 5, 2012www.sciencemag.orgDownloaded from
tally demonstrated that slip acceleration quickens
fault weakening, and, in light of the transient na-
ture of earthquake slip (1–3), we propose that slip
acceleration controls seismic weakening in addi-
tion to slip distance and slip velocity.
References and Notes
1. T. H. Heaton, Phys. Earth Planet. Inter. 64, 1 (1990).
2. E. Tinti, P. Spudich, M. Cocco, J. Geophys. Res. 110,
3. M. Ohnaka, T. Yamashita, J. Geophys. Res. 94, 4089
4. H. Kanamori, L. Rivera, in Earthquakes: Radiated
Energy and the Physics of Faulting, R. E. Abercrombie,
A. McGarr, G. Di Toro, H. Kanamori, Eds. (AGU,
Washington, DC, 2006), pp. 3–13.
5. D. J. Andrews, J. Geophys. Res. 110, B01307 (2005).
6. D. A. Lockner, P. G. Okubo, J. Geophys. Res. 88, 4313
7. P. G. Okubo, J. H. Dieterich, Geophys. Res. Lett. 8, 887
8. Z. Reches, D. A. Lockner, Nature 467, 452 (2010).
9. Materials and methods are available on Science Online.
10. H. Sone, T. Shimamoto, Nat. Geosci. 2, 705 (2009).
11. D. L. Wells, K. J. Coppersmith, Bull. Seismol. Soc. Am. 84,
12. D. L. Olgaard, W. F. Brace, Int. J. Rock Mech. Min. Sci.
20, 11 (1983).
13. B. Wilson, T. A. Dewers, Z. Reches, J. Brune, Nature 434,
14. R. Han, T. Hirose, T. Shimamoto, J. Geophys. Res. 115,
15. A. Niemeijer, G. DiToro, S. Nielsen, F. Di Felice,
J. Geophys. Res. 116, B07404 (2011).
16. D. L. Goldsby, T. E. Tullis, Science 334, 216 (2011).
17. E. Fukuyama, K. Mizoguchi, Int. J. Fract. 163,15
18. M. Ohnaka, J. Geophys. Res. 108, 2080 (2003).
19. D. E. Grady, M. E. Kipp, J. Appl. Phys. 58, 1210 (1985).
20. Z. Reches, T. A. Dewers, Earth Planet. Sci. Lett. 235, 361
21. E. Y. A. Wornyoh, V. K. Jasti, C. F. Higgs III, J. Tribol. 129,
Acknowledgments: We thank J. Young, J. Fineberg, and
E. Aharonov, and T. Shimamoto and two anonymous reviewers
for their thoughtful reviews. This work was supported by
NSF Geosciences awards 0732715 and 1045414 and
NEHRP2011 award G11AP20008.
Materials and Methods
Figs. S1 to S10
Tables S1 to S4
28 February 2012; accepted 14 August 2012
The Efficacy of Student-Centered
Instruction in Supporting
E. M. Granger,
*T. H. Bevis,
S. A. Southerland,
R. L. Tate
Transforming science learning through student-centered instruction that engages students in a
variety of scientific practices is central to national science-teaching reform efforts. Our study
employed a large-scale, randomized-cluster experimental design to compare the effects of
student-centered and teacher-centered approaches on elementary school students’understanding
of space-science concepts. Data included measures of student characteristics and learning and
teacher characteristics and fidelity to the instructional approach. Results reveal that learning
outcomes were higher for students enrolled in classrooms engaging in scientific practices through
a student-centered approach; two moderators were identified. A statistical search for potential
causal mechanisms for the observed outcomes uncovered two potential mediators: students’
understanding of models and evidence and the self-efficacy of teachers.
The need for a different approach to science
teaching and learning has been the focus
of several recent policy and economic re-
ports (1,2). Research as synthesized by the Na-
tional Research Council suggests that the goal of
science instruction should be to help students
develop four aspects of scientific proficiency, the
ability to (i) know, use, and interpret scientific
explanations of the natural world; (ii) generate
and evaluate scientific evidence and explanations;
(iii) understand the nature and development of
scientific knowledge; and (iv) participate produc-
tively in scientific practices and discourse (3–5).
This approach to science teaching will require a
shift from the teacher-centered instruction com-
mon in science classrooms to more student-centered
methods of instruction. The defining feature of
these instructional methods is who is doing the
sense-making. In teacher-centered instruction,
the sense-making is accomplished by the teacher
and transmitted to students through lecture, text-
books, and confirmatory activities in which each
step is specified by the teacher. In these class-
rooms, the instructional goal is to help students
know scientific explanations, which is only part
of the first aspect of scientific proficiency. In
student-centered instruction, the sense-making rests
with students, and the teacher acts as a facilitator
to support the learning as students engage in sci-
entific practices (3).
The effectiveness of student-centered instruc-
tion in helping students develop scientific pro-
ficiency is supported by a number of largely
small-scale, narrowly focused studies (3,5). De-
spite accumulating support for a student-centered
approach, few large-scale studies have evaluated
the effectiveness of such instruction, and their
results, taken as a whole, are contradictory and
randomized-cluster or quasi-randomized studies
examined separately (6,11,14,15). Many factors
may contribute to the varied results, because
tightly controlling potentially influential variables
is difficult in classroom settings. One central fac-
tor is that the comparison condition (i.e., control
group) is often “undefined or assumed to be
‘traditional’” (14). Likewise, possible “contami-
nation of the untreated teachers”and cases where
investigators did not “vigorously guard”against
special resource materials may have influenced
results (13). Indeed, many studies described in
the literature do not discuss how fidelity to the
curriculum or instructional approach was mea-
sured or whether it was assessed.
We therefore compared the effectiveness of
student-centered with teacher-centered instruc-
tion using a randomized-cluster experimental de-
sign, intended to control as many variables as
possible given the inherent differences between
the two instructional approaches. Specifically, the
effectiveness of the student-centered Great Ex-
plorations in Math and Science Space Science
Curriculum Sequence (SSCS) (16) and profes-
sional development of teachers focused on these
materials (treatment group) was compared with
that of a teacher-centered curriculum (district-
adopted textbook) enacted with a teacher-centered
approach (control group). For details of each cur-
riculum, teacher professional development, and
instructional approach, see the supplementary ma-
terials. Mindful of limits on securing meaningful
data imposed by testing the age group for whom
SSCS is appropriate (fourth and fifth grades), we
selected four student outcomes aligned with the
four aspects of scientific proficiency for this re-
search: space science content knowledge, knowl-
edge about models and evidence in science, views
of scientific inquiry, and attitude toward science.
The research was designed to (i) compare the
effectiveness of the two instructional approaches
in supporting elementary students’science learn-
ing; (ii) identify teacher characteristics (teacher
moderating variables) that might influence the
learning; (iii) identify those for whom this instruc-
tional approach might work (student moderating
variables); and (iv) identify how the treatment
might indirectly affect student outcomes (mediat-
Office of Science Teaching Activities, Florida State University,
Tallahassee, FL 32306–4295, USA.
Bülent Ecevit Üniversitesi
Ereğli Eğitim Fakültesi, Turkey.
FSU-Teach/School of Teacher
Education, Florida State University, Tallahassee, FL 32306–4459,
Educational Evaluation and Research, Florida State Uni-
versity (retired), 415 Castleton Circle, Tallahassee, FL 32312, USA.
*To whom correspondence should be addressed: E-mail:
www.sciencemag.org SCIENCE VOL 338 5 OCTOBER 2012 105
on October 5, 2012www.sciencemag.orgDownloaded from
Data were collected from 125 fourth- and
fifth-grade classrooms. Randomization occurred
at the level of assignment of teachers to treatment
or control group; control and treatment groups
economic status (SES), school statewide assess-
ment performance, and student ethnic diversity.
Student demographics were collected (table S1).
Contexts included urban, suburban, and rural
and high- and low-SES schools; 2594 students
participated—1418 in the classrooms of 66 treat-
ment teachers and 1176 in the classrooms of 59
control teachers (for details, see the supplemen-
tary materials). Student conceptual development
and affective dimensions were assessed by means
of four instruments: (i) Space Science Content
Tes t (17); (ii) Homerton Science Attitudes Sur-
vey (18); (iii) Models and Evidence Question-
naire (19); (iv) Views of Scientific Inquiry (VOSI)
Elementary Version Questionnaire (20). Both
groups were assessed immediately before the
unit, immediately after the unit, and 5 months T
2 weeks after the unit (see the supplementary ma-
terials for test and scoring details). Each teacher’s
fidelity to the assigned teaching approach was
assessed with the Reformed Teacher Observa-
tion Protocol (RTOP) (21) two to three times dur-
ing the unit. RTOP is a measure of the degree to
which lesson enactment is aligned with student-
centered science instruction. To help identify
potential teacher moderators, we also assessed
teachers’space-science content knowledge, sci-
ence attitudes, views of scientific inquiry, self-
efficacy, and beliefs about science teaching before
and after their participation in the project.
Multilevel (hierarchical linear) modeling was
used to estimate the SSCS’seffectswithmoder-
ation and to account for interdependencies of
student outcomes within teachers (22,23). Each
student outcome was reflected in two measures,
the postmeasure and delayed postmeasure. Po-
tential explanatory variables for each outcome
included (i) pretest measure; (ii) treatment vari-
able (SSCS or control); (iii) teacher cohort (year
1 or 2); (iv) grade level (4 or 5); (v) interactions
among the treatment, cohort, and grade variables;
(vi) teacher years of experience; (vii) preassess-
ments of teacher outcomes; (viii) student ethnic-
ity; (ix) student SES, based on participation in the
free or reduced lunch program (FRL); (x) student
gender; and (xi) primary language of the student.
Inspection of bivariate correlations and backward
elimination processes produced the final models.
Special attention was given to the possibility of
interactions involving student and teacher mod-
erators. Multilevel mediation models addressed
the question of how the treatment might indi-
rectly affect student outcomes (mediating varia-
bles) by two analytic approaches: an application
of multilevel modeling based on separate models
for each of the variables explained by the model
and a simultaneous solution by multilevel path
analysis (24). Positive mediation results should
be considered as evidence that the data are con-
sistent with the hypothesized mediation. To com-
pare effect sizes for outcomes with different scales,
we also present effects in standardized form [0.2 is
small, 0.5 medium, and 0.8 large (25)] but note
that a small effect should not be interpreted as
trivial (15,25). The VOSI outcome could not be
similarly standardized, because it is an ordered
binary variable with two levels. For VOSI, a non-
linear version of multilevel modeling describes
effects as an odds ratio (i.e., the odds of SSCS
students giving a response coded as “transitional/
informed”rather than “naïve”divided by the odds
of control students doing so); a ratio of 1.1 is at the
threshold of practical importance. (See the supple-
mentary materials for details of statistical analysis.)
Average RTOP scores for SSCS teachers were
27.3 points (out of 100) higher than those of con-
trol teachers, a statistically significant difference
(P= 0.001); groups overlapped only modestly,
indicating that the two instructional approaches
were implemented with fidelity. To examine within-
group RTOP effects, we transformed total RTOP
scores to within-group deviation scores. For each
student outcome for which the SSCS effect was
statistically significant (content knowledge, mod-
els and evidence, and VOSI) (see Table 1), the
effect of the RTOP deviation score was positive
and statistically significant (P= 0.002 to 0.017).
That is, after the SSCS treatment effect was con-
trolled for, the RTOP deviation score could be
considered as a “dosage”variable within each
group resulting in an increase in student outcomes.
This finding suggests that student engagement in
Table 1. SSCS total effects for student outcomes. OR, odds ratio.
Content knowledge‡0.488* 0.002 0.171*
VOSI§ 0.464* 0.002 OR = 1.59
Models and evidence
No FRL 0.354* <0.001 0.682*
FRL 0.261* <0.001 0.503*
Attitude toward science 0.009 0.753 0.014
Delayed posttest outcome
Content knowledge‡0.187 0.193 0.067
VOSI║0.333* 0.015 OR = 1.40
Models and evidence 0.285* <0.001 0.573*
Attitude toward science 0.020 0.530 0.029
*Statistically significant, with a family-wise error rate of 0.10 (i.e., for a test family of four post-outcomes or four delayed post-
outcomes, P< 0.10/4 = 0.025). †For all outcomesexcept VOSI, standardized effects were obtained by division of raw-score SSCS
coefficients by the outcome standard deviations. ‡Results are from a sample from which one extreme outlier was
removed. ║The student variable of VOSI is dichotomous. The associated unstandardized SSCS effect is the SSCS coefficient
in a model for the log odds (logit) of the outcome, and the standardized SSCS effect is the OR for the transitional/informed
Fig. 1. Mediation model
for student posttest con-
tent knowledge, with path
coefficients indicated. *,
significant at the 0.05 lev-
el; NS, not significant; TSE
pre, teacher pretest self-
efficacy; TSE post, teacher
posttest self-efficacy; ME
pre, student pretest mod-
els and evidence; ME post,
student posttest models
and evidence; CK pre, pre-
test student content knowl-
edge; CK post, posttest
student content knowledge.
0.767* - 0.799*SSCS
3.27* - 0.799*TSE pre
- 0.067 (NS)
TSE p re TSE p os t
ME p re ME p os t CK post
5 OCTOBER 2012 VOL 338 SCIENCE www.sciencemag.org106
on October 5, 2012www.sciencemag.orgDownloaded from
learning (student-centeredness), as indicated by
RTOP scores, is a feature of more effective teaching.
Table 1 summarizes estimated total effects of
the SSCS curriculum on student post- and delayed
posttest outcomes, including student-level mod-
erators. When a family-wise error rate of 0.10
was controlled for, students in the SSCS group
scored significantly higher than control studen ts
on content-knowledge, models-and-evidence, and
VOSI posttest outcomes. For delayed posttests,
the SSCS group remained significantly higher for
VOSI and models and evidence. For content-
knowledge and models-and-evidence outcomes,
the standardized effect magnitudes were ~0.2 and
0.7, respectively. The odds ratio representing the
SSCS effect for the VOSI outcome was 1.59; that
is, the odds of SSCS students giving transitional/
informed responses were 1.59 times greater than
those for control students. Interpreting the results
of the content-knowledge delayed posttest requires
accounting for factors potentially affecting it; fore-
most is the timing, which placed these assessments
within about 2 weeks of statewide assessments
and all their concomitant drill and practice in
both treatment and control classrooms. The large
size (25) and persistence of the models-and-
evidence and VOSI outcomes are notable.
Only one student characteristic, SES, mod-
erated the SSCS effect on one posttest outcome,
models and evidence (Table 1). Although the two
SES groups differed in achievement, both high-
and low-SES students in the treatment group
performed better than did students in the control
group. This difference between the groups in the
SES achievement gap disappeared by delayed
posttesting. This result is consistent with a wide
body of research that indicates that students from
low-SES groups initially need more support to
participate in the practices of science (here, using
models and evidence) (6,11,26).
Only one teacher characteristic, pretest self-
efficacy, moderated one student posttest outcome,
content knowledge. Self-efficacy, a well-researched
construct (see the supplementary materials for more
details), is defined as a teacher’s“judgement of
his or her capabilities to bring about desired
outcomes of student engagement and learning”
(27). The SSCS effect was positive and large
for low values of teacher pretest self-efficacy
but decreased as teacher pretest self-efficacy in-
creased (table S2). That is, students in classrooms
of teachers who had low teaching self-efficacy
at the outset of the study showed a statistically
significant increase in their posttest content-
knowledge scores, whereas students in classrooms
with teachers who had high initial self-efficacy
did not. Much research has examined how teach-
ers’knowledge, attitudes, and beliefs about their
abilities shape and are shaped by their classroom
experiences [e.g., (28)]; our results suggest that
teachers’underlying beliefs, such as self-efficacy,
might influence the overall effectiveness of a
student-centered curriculum. No teacher modera-
tors were apparent in the delayed posttest results.
A search for possible indirect mechanisms by
which the treatment produced its effects on post-
test student outcomes resulted in two mediation
models. In the first (Fig. 1), both teacher self-
efficacy and student models-and-evidence varia-
bles mediate the SSCS effect on the student
posttest content-knowledge outcome. Indices in-
dicated a good fit of this model to the data. The
indirect effect mediated by the teacher posttest
self-efficacy measure varied with the level of its
pretest measure, being positive and large for low
values of teacher pretest self-efficacy but decreas-
ing with increasing levels of the teacher pretest
self-efficacy. The indirect effect mediated by student
posttest models and evidence was positive, con-
stant, and relatively strong. In this model, the es-
timated direct effect of the SSCS treatment on
student achievement was not statistically signif-
icant, so the total effect consisted entirely of the
indirect effects of teacher self-efficacy and student
understanding of models and evidence. The re-
sults of this analysis therefore suggest that the
two mediators influence the student content-
knowledge outcome somewhat equally in class-
rooms taught by teachers with lower self-efficacy
at the beginning of the project, but the posttest
models-and-evidence outcome was the only sig-
nificant mediator in classrooms taught by teach-
ers who began the project with relatively high
levels of self-efficacy (Fig. 2). (See the supple-
mentary materials for more details.)
These results suggest that an emphasis on
models and evidence supports students’learning
about space-science content. The models-and-
evidence instrument was not designed to assess
knowledge about models separately from knowl-
edge about evidence, so their individual influ-
ences cannot be separated. SSCS students explicitly
learned about the nature of both models and
evidence in science, as did control-group students,
but SSCS students further engaged in activities in
2.5 3.0 3.5 4.0 4.5 5.0
Total effect = Total indirect effect (both paths)
Indirect through TSE post
Indirect through ME post
Fig. 2. Standardized effects of SSCS on the student posttest content knowledge outcome. Abbreviations
as in Fig. 1.
Fig. 3. Mediation model
for posttest VOSI, with path
coefficients indicated. Mod-
el is for the log odds (logit)
of student posttest VOSI. *,
significant at the 0.05 level;
VOSI pre, pretest student
VOSI; VOSI post, posttest
student VOSI; other abbre-
viations as in Fig. 1.
VOSI postME post
www.sciencemag.org SCIENCE VOL 338 5 OCTOBER 2012 107
on October 5, 2012www.sciencemag.orgDownloaded from
which they used and evaluated models. They also
were required to provide explicit evidence that
supported new science concepts throughout the
curriculum, including activities in which they used
evidence to support their arguments about sci-
entific explanations. Experience with models and
evidence is therefore supported as underpinning
In the second mediation model, the models-
and-evidence outcome mediates the SSCS effect
on views of scientific inquiry (Fig. 3). This indi-
rect effect was somewhat smaller than the direct
effect of SSCS, but the odds ratio of 1.21 for the
indirect path was nevertheless large enough to be
of practical importance. The results of this anal-
ysis suggest that the emphasis on models and
evidence supports students’learning about the
endeavor of science.
These findings should also be considered in
light of the contradictory results of the previous
large-scale studies on the effects of student-
centered instruction. In contrast to the common
research practice of comparing the treatment group
to a control group in which the instructional ap-
proach is not specified (i.e., to whatever else is
present in schools), our research design tightly
controlled the curriculum and instructional approach
employed by the treatment and control groups.
Further, fidelity was monitored through classroom
observations and assessed with RTOP. We argue
that this attention to the control group and our ef-
study apart from others, and the failure to identify
central components of the control group may ac-
count for the contradictory nature of previous results.
The increased outcomes of the treatment group
in comparison with the control group in content
knowledge, models and evidence, and VOSI and
the size and persistence of the latter two indicate
that student-centered instruction supports the de-
velopment of students who are more proficient in
the four strands of scientific proficiency. More
specifically, the mediation models suggest that
student-centered instruction that engages students
in scientific practices such as using models and
evidence is important for developing more scientif-
ically proficient students. Taken together, the results
of our study lend empirical support to the view
put forth by the National Research Council that
“teaching content alone is not likely to lead to
proficiency in science, nor is engaging in inquiry
experiences devoid of meaningful science content.
In current practice, content and an oversimplified
view of scientific processes are often the primary
or even sole foci of instruction…[and] leads to a
very impoverished understanding of science and
masks the complex process involved in developing
scientific evidence and explanations”[(3), p. 335)].
References and Notes
1. Rising Above the Gathering Storm, Revisited, National
Academy of Sciences, National Academy of Engineering,
Institute of Medicine (National Academies Press,
Washington, DC, 2010).
2. U.S. Congress Joint Economic Committee, STEM
Education: Preparing for the Jobs of the Future
(U.S. Congress Joint Economic Committee, Washington,
DC, April 2012).
3. R. Duschl, H. Schweingruber, A. Shouse, Eds., Taking
Science to School: Learning and Teaching Science in
Grades K-8 (National Academies Press, Washington, DC,
4. S. Michaels, A. Shouse, H. Schweinberger, Ready, Set,
Science: Putting Research to Work in K-8 Science Classrooms
(National Academies Press, Washington, DC, 2008).
5. M. S. Donovan, J. Bransford, Eds., How Students Learn:
Science in the Classroom (National Academies Press,
Washington, DC, 2005).
6. M. Blanchard et al., Sci. Ed. 94, 577 (2010).
7. P. A. Kirschner, J. Sweller, R. Clark, Educ. Psychol. 41,75
8. D. Klahr, M. Nigam, Psychol. Sci. 15, 661 (2004).
9. W. H. Leonard, G. R. Cavana, L. F. Lowery, J. Res. Sci.
Teach. 18, 497 (1981).
10. W. H. Leonard, J. Res. Sci. Teach. 20, 807 (1983).
11. S. Lynch, J. Kuipers, C. Pyke, M. Szesze, J. Res. Sci. Teach.
42, 912 (2005).
12. R. W. Marx et al., J. Res. Sci. Teach. 41, 1063 (2004).
13. J. A. Shymansky, L. Yore, J. Anderson, J. Res. Sci. Teach.
41, 771 (2004).
14. S. Lynch et al., What’s up with the comparison group?
How a large quasi-experimental study of high rated
science curriculum units came to grips with unexpected
results, in Annual Meeting of the American Educational
Research Association (American Educational Research
Association, San Francisco, 2006).
15. D. Viadero, Educ. Week 28, 14 (2009).
16. Lawrence Hall of Science, Great Explorations in Math and
Science: Space Science Curriculum Sequence (Univ. of
California Press, Berkeley, CA, 2007).
17. P. Sadler et al., Misconception-Oriented, Standards-
Based Assessment Resources for Teachers (MOSART)
(Harvard College, Cambridge, MA, 2007).
18. M. Warrington, M. Younger, J. Williams, Br. Educ. Res. J.
26, 393 (2000).
19. E. Granger, Y. Saka, T. H. Bevis, Models and Evidence
Questionnaire (2007); available from http://ret.fsu.edu.
20. R. S. Schwartz, N. Lederman, J. S. Lederman, An instrument
to assess views of scientific inquiry: The VOSI questionnaire,
International Conference of the National Association for
Research in Science Teaching (National Association for
Research in Science Teaching, Baltimore, 2008).
21. D. Sawada et al., Sch. Sci. Math. 102, 245 (2002).
22. S. Raudenbush, A. Bryk, Hierarchical Linear Models:
Applications and Data Analysis Methods (Sage, Newbury
Park, CA, ed. 2, 2002).
23. S. Raudenbush et al., HLM 6: Hierarchical Linear and
Nonlinear Modeling (SSI Scientific Software International,
Lincolnwood, IL, 2004).
24. L. K. Muthen, B. O. Muthen, Mplus User’s Guide (Muthen
& Muthen, Los Angeles, CA, ed. 5, 2007).
25. J. Cohen, Statistical Power Analysis for the Behavioral
Sciences, revised ed. (Academic Press, New York,
26. O. Lee, A. Luykx, Science Education and Student Diversity
(Cambridge Univ. Press, New York, 2006).
27. M. Tschannen-Moran, A. W. Hoy, Teach. Teach. Educ. 17,
28. L. K. Smith, S. A. Southerland, J. Res. Sci. Teach. 44, 396
Acknowledgments: This work was supported by a grant
from the Florida Center for Research in Science,
Technology, Engineering, and Mathematics Education
(FL DOE 371-96700-7SF01, 371-96700-8SS01, and
371-96700-9SF01). Data are archived in the supplementary
Materials and Methods
Tables S1 and S2
Student Data Archive
23 April 2012; accepted 16 August 2012
Wnt5a Potentiates TGF-bSignaling
to Promote Colonic Crypt Regeneration
After Tissue Injury
*Christine T. Luo,
Terry P. Yamaguchi,
†Thaddeus S. Stappenbeck
Reestablishing homeostasis after tissue damage depends on the proper organization of stem
cells and their progeny, though the repair mechanisms are unclear. The mammalian intestinal
epithelium is well suited to approach this problem, as it is composed of well-delineated units called
crypts of Lieberkühn. We found that Wnt5a, a noncanonical Wnt ligand, was required for crypt
regeneration after injury in mice. Unlike controls, Wnt5a-deficient mice maintained an expanded
population of proliferative epithelial cells in the wound. We used an in vitro system to enrich
for intestinal epithelial stem cells to discover that Wnt5a inhibited proliferation of these cells.
Surprisingly, the effects of Wnt5a were mediated by activation of transforming growth factor–b
(TGF-b) signaling. These findings suggest a Wnt5a-dependent mechanism for forming new crypt
units to reestablish homeostasis.
Tissue regeneration requires proper spa-
tial allocation and organization of stem
cells for efficient return to homeostasis
(1,2). Crypts of Lieberkühn are subunits that
house intestinal stem cells and are lost in re-
sponse to a variety of insults, including ischemia,
infection, irradiation, and inflammatory bowel
disease (3). Although individual crypts undergo
fission to replicate during homeostasis (fig. S1A)
(4,5), the mechanism of their regeneration is un-
known. Thus, crypt regeneration is a proxy for
proper stem cell organization and provides an ex-
cellent system to uncover the principles underlying
stem cell replacement and/or organization in vivo.
To model crypt/epithelial stem cell loss, we
previously developed an injury system to focally
5 OCTOBER 2012 VOL 338 SCIENCE www.sciencemag.org
on October 5, 2012www.sciencemag.orgDownloaded from