expected for a mineralization process in super-
saturated solution ( 22 ) (fig. S6).
Second, tendon samples were immersed
into different solutions with the same total
concentration of ions, to test whether the in-
teraction with ions or the mineralization pro-
cess was the origin of the contraction (fig. S7).
No stress generation occurred if only either
Sr2+or CO 32 −ions were present in the solu-
tion. Stress increased solely if both mineralizing
ions (Sr2+and CO 32 −) were present and caused
intrafibrillar mineralization. Therefore, the
deposition of minerals inside collagen fibrils
(intrafibrillar mineralization) plays a dominant
role for stress generation. Third, by modifying
the pH value of the mineralizing solution, the
degree of mineralization could be controlled.
A higher pH value of the solution led to a fast
increase of stress in the initial stage of tendon
mineralization, ranging from 0.03 MPa/hour
atapHof8.5to0.095MPa/houratapHof9
(inset of Fig. 2A). That pH influences the rate
of stress generation could also indicate a role
of collagen charges in the process.
The generation of contractile stress during
mineralization corresponds to a contraction
of tendons along their longitudinal direction.
To reveal structural changes of tendon tissue
during the mineralization, we performed in-
operando synchrotron SAXS measurements
(Fig. 3A and fig. S8). Throughout these mea-
surements, a constant force of 0.06 N was ap-
plied to the tendon during mineralization (fig.
S9A), and the motor position was recorded to
evaluate the tissue strain (fig. S9B). After water
was replaced by a SrCO 3 solution, the free ions
quickly diffused into the interfibrillar matrix,
causing a slight expansion of the tendon. Sub-
sequently, a fast drop of tissue strain to−0.44%
could be observed at 1.1 hours, and a slow con-
traction to−1.7% after 9 hours of mineralization
(fig. S9B).
The axial staggering of tropomolecules in
collagen fibrils results in an alternation of
stripes with high and low molecular density,
which not only can be visualized by transmis-
sion electron microscopy (TEM) but also can
be measured by SAXS ( 23 , 24 ). The SAXS pat-
terns of original tendons in water exhibited a
series of Bragg peaks (fig. S10A). Theq-positions
of the nth order (n= 1, 3, 5...) corresponds to
qn=2pn/D, whereDis the periodic spacing
(~67 nm) within collagen fibrils according to
the gap and overlap zones. The nanoscopic var-
iation of theDspacing (inverse relation toq) is
an ideal measure to evaluate the microscopic
stress generation in tendons during minerali-
zation. Because in comparison to the first-order
scattering peak, the relatively strong third-
order peak is more sensitive to changes ofD
spacing, it was employed to determine the peak
positions by fitting with a Gaussian (fig. S10).
In-operando SAXS measurements were per-
formed to evaluate both (i) the degree of min-
eralization at different regions across the tendon
sample and (ii) changes inDspacing during
the process. The latter measurement is related
to the internal,“microscopic”strain of the col-
lagen fibrils (fibril strain), whereas the me-
chanical setup (fig. S8) yields the“macroscopic”
strain of the tissue (tissue strain), when the
clamps are moved so as to keep the total stress
in the tendon close to zero. If clamps are kept
fixed instead, the setup can measure the mac-
roscopic contractile stress that develops dur-
ing mineralization.
Figure 3B shows SAXS patterns where the
intensity of the reflection in the trapezoidal
box was monitored, starting from 1 min to
8 hours (see also fig. S11A). The integrated SAXS
intensity increased with the mineral content in
some areas but does not distinguish between
crystalline or amorphous. In other regions no
mineralization occurred, as evidenced by the
SAXS intensity that remained low (Fig. 3C).
Clear differences between mineralized and
unmineralizedregionscouldbeobserved(fig.
S11B). The time evolution of theDspacing incollagen fibrils and the calculated fibril strain
during mineralization is summarized in fig. S11,
C to D. Both mineralized and unmineralized
regions exhibited a decrease of theDspacing.
However, mineralized regions showed a faster
decrease of fibril strain than the unmineral-
ized ones (see also fig. S12).
This fibril strain can be correlated with the
macroscopic tissue strain measured through
the movement of the clamps (Fig. 3D), which
is a weighted average of all local strains. The
black dashed line indicates a 1:1 correlation of
fibril and tissue strain. Deviations from this
value are due to the local inhomogeneities in
mineralization and in fibril strain. In the re-
gion that mineralized (red symbols in Fig. 3,
C and D), the SAXS intensity (which monitors
the amount of mineral) increased sharply in
the first 2 hours, with only a slight increase in
local contractile fibril strain. After this, mine-
ral content continued to increase more slow-
ly, with an associated strong contraction of
the fibril (up to nearly 2% contraction). In the
region that did not mineralize (blue symbols),SCIENCEscience.org 8 APRIL 2022•VOL 376 ISSUE 6589 191
Fig. 4. Lattice strains of nanocrystals in collagen tissues.(A) Two-dimensional WAXS pattern of SrCO 3 -
mineralized tendon. The direction of the mineralized tendon is vertical. To detect peak shifts along and
perpendicular to the collagen fibril direction, the integrated areas of (110), (002), and (200) rings are marked
by red boxes. (B) Lattice strains as calculated from scattering patterns shown in (A) for the three indicated
lattice directions. (C) TEM image of an isolated SrCO 3 -mineralized collagen fibril. Inset shows the
corresponding SAED pattern. The orientation of nanocrystals along the (200) direction is marked by a red
arrow. (D) Generality of stress generation in collagen tissue. The final stress and mineral volume per collagen
mass are shown for the following minerals: SrCO 3 , SrWO 4 , SrSO 4 , CaF 2 , CaCO 3 , and Ca 10 (PO 4 ) 6 (OH) 2.RESEARCH | REPORTS