Expression and secretion of interleukin-1b, tumour
necrosis factor-a and interleukin-10 by hypoxia- and
serum-deprivation-stimulated mesenchymal stem cells
Implications for their paracrine roles
Zongwei Li, Hua Wei, Linzi Deng, Xiangfeng Cong and Xi Chen
Research Center for Cardiac Regenerative Medicine, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing, China
Introduction
Ischaemic heart disease is a life-threatening condition
that may cause sudden cardiac failure and death.
Many researchers have investigated cell transplantation
as an alternative treatment for heart disease. Bone
marrow-derived mesenchymal stem cells (MSCs) are
easily obtainable and expandable, multipotent progeni-
tor cells [1] that have emerged as attractive candidates
for cellular therapies for heart and other organ-system
disorders [2]. Although several mechanisms have been
proposed for the cardioprotective effects of MSCs,
including cardiomyocyte regeneration, spontaneous cell
fusion and paracrine action [3], there is a growing
Keywords
IL-10; IL-1b; mesenchymal stem cell;
paracrine; TNF-a
Correspondence
X. Chen; X. Cong, Research Center for
Cardiac Regenerative Medicine, The
Ministry of Health, Cardiovascular Institute
& Fu Wai Hospital, Chinese Academy of
Medical Sciences & Peking Union Medical
College, 167 Beilishilu, Beijing 100037,
China
Fax ⁄ Tel: +86 10 88398584

LPS, lipopolysaccharide; MSCs, mesenchymal stem cells; NF-jBp65, nuclear factor kappa Bp65; p38, p38 mitogen-activated protein kinase;
TNF-a, tumour necrosis factor-a.
3688 FEBS Journal 277 (2010) 3688–3698 ª 2010 The Authors Journal compilation ª 2010 FEBS
body of evidence supporting the hypothesis that para-
crine mechanisms mediated by MSC-secreted factors
play an essential role in the reparative process [4,5].
It has been reported that MSC-conditioned medium
under normoxic conditions significantly attenuates car-
diac fibroblast proliferation and type I and III collagen
expression, exerting paracrine anti-fibrotic effects.
However, researchers did not analyse the active compo-
nents of the conditioned medium [6]. Other researchers
have suggested that adrenomedullin and hepatocyte
growth factor are paracrine factors secreted by trans-
planted MSCs, decreasing myocardial fibrosis [7–9].
Whether other paracrine factors released by MSCs
mediate these cells’ anti-fibrotic effects remains largely
unknown.
Interleukin-1b (IL-1b) and tumour necrosis factor-a
(TNF-a) are present in the tissues or systemic circula-
tion in many inflammatory conditions. It has also been
reported that the expression of IL-1b and TNF-a in
MSCs can be augmented by exposure to hypoxia [5].
Furthermore, IL-1b can induce cardiomyocyte growth
but inhibits cardiac fibroblast proliferation in culture
[10]. By contrast, MSC transplantation in rat models
of myocardial infarction has anti-inflammatory effects,
decreasing protein production and gene expression for
IL-1b and TNF-a [11]. To address these paradoxes of
both pro- and anti-inflammatory effects, the secretion

by MSCs.
Results
MSCs-CM inhibits cardiac fibroblast proliferation
and collagen synthesis
The effects of MSCs-CM on cardiac fibroblast prolifer-
ation and collagen synthesis were detected by [
3
H]-thy-
midine and [
3
H]-proline incorporation. As shown in
Fig. 1A, MSC-CM treatment significantly inhibited
[
3
H]-thymidine and [
3
H]-proline incorporation under
normoxic or hypoxic culture conditions. To further
clarify the molecular mass range of important active
factors in the MSCs-CM, the medium was divided into
AB
Fig. 1. MSCs-CM inhibits cardiac fibroblast proliferation and collagen synthesis. (A) The effects of MSCs-CM on the incorporation of [
3
H]-thy-
midine and [
3
H]-proline by cardiac fibroblasts under normoxic or hypoxic conditions. Each data point represents the mean ± SEM of at least
three independent experiments. ***P < 0.001 versus normoxic control (Cont) group; ###P < 0.001 and ##P < 0.01 versus hypoxic control
(Cont + h) group. (B) The effects of the > 30 kDa and < 30 kDa components of MSCs-CM on the incorporation of [
3

ulating IL-1b and TNF-a transcription [18,19]. To
investigate the role of this pathway in hypoxia ⁄ SD-
induced transcription, MSCs were exposed to BAY
11-7082, an NF-jB pathway inhibitor, followed by
hypoxia ⁄ SD for 6 h. As shown in Fig. 2C, the tran-
scription of IL-1b and TNF-a was significantly attenu-
ated by BAY 11-7082. Interestingly, the proteasomal
inhibitor MG132 also abrogated hypoxia ⁄ SD-induced
IL-1b and TNF-a transcription.
Next, to clarify the mechanism by which the NF-jB
pathway induces IL-1b and TNF-a transcription, the
nuclear translocation of NF-jBp65 was assessed by
immunocytochemical staining. As shown in Fig. 2D,
NF-jBp65 was mainly distributed in the cytoplasm
of control cells. By contrast, hypoxia ⁄ SD treatment
significantly stimulated the nuclear translocation of
AB
CD
Fig. 2. Hypoxia ⁄ SD induces NF-jB-dependent IL-1b and TNF-a transcription. (A) MSCs were incubated under hypoxia ⁄ SD conditions for the
indicated number of hours, and the relative mRNA levels of IL-1b and TNF-a were determined by real-time PCR. The data are the mean ±
SEM of at least three independent experiments. *P < 0.05 and **P < 0.01 versus control group (0 h). (B) The relative mRNA levels for IL-1b
and TNF-a in MSCs after hypoxia, SD or hypoxia ⁄ SD for 6 h by real-time PCR. **P < 0.01 versus Cont group; #P < 0.05 versus SD group.
(C) MSCs were exposed to BAY 11-7082 or MG132, followed by hypoxia ⁄ SD for 6 h and detection of relative mRNA levels of IL-1b and
TNF-a by real-time PCR. *P < 0.05 and **P < 0.01 versus hypoxia ⁄ SD treatment group. (D) A representative pattern of the nuclear translo-
cation of NF-jBp65, as assessed by immunocytochemical staining of MSCs using anti-(NF-jBp65 primary Ig) (red) and nuclear labelling with
4¢,6-diamidino-2-phenylindone (blue).
Paracrine anti-fibrotic effects of MSCs in vitro Z. Li et al.
3690 FEBS Journal 277 (2010) 3688–3698 ª 2010 The Authors Journal compilation ª 2010 FEBS
NF-jBp65, indicated by strong immunostaining in the
nucleus. Pretreatment with BAY 11-7082 inhibited

Hypoxia
⁄
SD increases the translation of pro-IL-1b
but not TNF- a
Having demonstrated significant transcriptional upreg-
ulation, we next examined protein levels of IL-1b and
TNF-a in MSCs-CM. Unexpectedly, neither IL-1b nor
TNF-a was detectable in MSCs-CM using enzyme-
linked immunosorbent assay (ELISA) analysis. To
determine the reason for this lack of IL-1b and TNF-a
secretion by MSCs, changes in these factors’ transla-
tion in hypoxia ⁄ SD-stimulated MSCs were investi-
gated. As shown in Fig. 4A, hypoxia⁄ SD increased
pro-IL-1b translation in a time-dependent manner,
whereas TNF-a protein expression remained
unchanged at each time point. Furthermore, MG132,
BAY 11-7082 and U0126, all of which abrogated
hypoxia ⁄ SD-induced IL-1b and TNF-a transcription,
also abolished pro-IL-1b translational upregulation
(Fig. 4B,C) but failed to affect TNF-a translation
A
C
B
Fig. 3. IL-1b and TNF-a transcriptional
induction depends on the ERK1 ⁄ 2 pathway.
MSCs were exposed to the ERK1 ⁄ 2
inhibitor U0126 or the p38 inhibitor
SB202190, followed by hypoxia ⁄ SD for 6 h.
(A,B) Relative mRNA levels for IL-1b and
TNF-a, as determined by real-time PCR.

⁄
SD-stimulated MSCs require a second
signal for IL-1b and TNF-a release
Although significant cleavage of pro-IL-1b and pro-
caspase 1 occurred intracellularly in hypoxia ⁄ SD-stim-
ulated MSCs, mature IL-1b was undetectable in
MSCs-CM (Fig. 6A). However, significant release of
IL-1b by hypoxia ⁄ SD-stimulated MSCs in the presence
of ATP was detected. Furthermore, when both LPS
and ATP were present, hypoxia ⁄ SD-stimulated MSCs
released a larger amount of IL-1b (Fig. 6A). We also
examined TNF-a expression in hypoxia ⁄ SD-stimulated
MSCs in the presence of LPS. As shown in Fig. 6B,
LPS relieved the translational inhibition of TNF-a.
Moreover, TNF-a release by MSCs was detectable
after hypoxia⁄ SD treatment for 6 h in the presence of
LPS (Fig. 6C). These findings demonstrate that hypoxia ⁄
SD-stimulated MSCs require a second stimulatory
signal in order to secrete IL-1b and TNF-a.
Hypoxia
⁄
SD induces the transcription and
secretion of IL-10
Because of the lack of secretion of the inflammatory
cytokines IL-1b and TNF-a from hypoxia ⁄ SD-stimu-
lated MSCs, as well as the significant anti-inflamma-
tory effects of MSCs, expression and secretion of the
anti-inflammatory cytokine IL-10 by these cells was
investigated. As shown in Fig. 7A, hypoxia ⁄ SD
A

Next, the secretion of IL-10 from hypoxia⁄ SD-stimu-
lated MSCs was examined by ELISA. As shown in
Fig. 7C, a small amount of IL-10 release from MSCs
was detected at the 6-h time point, and this release
was elevated at the 12-h time point. Furthermore,
IL-10 secretion was augmented by the presence of LPS
at each time point.
IL-10 inhibits cardiac fibroblast proliferation and
collagen expression
The molecular mass of IL-10 is 19 kDa, which is
< 30 kDa and thus part of the MSCs-CM fraction
that inhibited cardiac fibroblast proliferation and colla-
gen synthesis (Fig. 1B). To investigate the potential
contribution of IL-10 to the paracrine effects of MSCs,
the influence of IL-10 on cardiac fibroblast prolifera-
tion was characterized using a 5-bromodeoxyuridine
(BrdU) incorporation assay. As shown in Fig. 8A,B,
different IL-10 concentrations significantly inhibited
A
B
C
Fig. 6. Hypoxia ⁄ SD-stimulated MSCs
require a second signal for IL-1b and TNF-a
release. (A) The results of ELISA analysis of
supernatants from MSCs after hypoxia ⁄ SD
stimulation for 12 h in the presence and
absence of ATP and LPS. (B) Representative
western blots for TNF-a expression in MSCs
stimulated by hypoxia ⁄ SD in the presence
or absence of LPS for the indicated number

fibroblasts (Fig. 8C). Moreover, IL-10 effectively limited
angiotensin II-induced type I and III collagen protein
expression (Fig. 8D). These results indicate that IL-10
can inhibit cardiac fibroblast proliferation and collagen
expression, suggesting a paracrine, anti-fibrotic role for
this factor.
Discussion
In this study, we focused on the paracrine effects of
MSCs on cardiac fibroblast proliferation and collagen
expression, as well as the possible paracrine roles of
IL-1b, TNF-a and IL-10 in cardiac fibrosis. First, our
results demonstrate that MSCs-CM have significant
anti-fibrotic effects, as indicated by decreased [
3
H]-thy-
midine and [
3
H]-proline incorporation by cardiac
fibroblasts. Moreover, we found that < 30 kDa compo-
nents of MSCs-CM play the dominant anti-fibrotic
role, suggesting that these anti-fibrotic factors may be
soluble small molecules. Second, our data show that
hypoxia ⁄ SD induces NF-jB-dependent IL-1b and TNF-
a transcriptional upregulation. However, these two fac-
tors are not secreted from hypoxia ⁄ SD-stimulated
MSCs unless a second signalling stimulus is present.
This finding suggests that the paracrine roles of TNF-a
and IL-1b after MSC transplantation may be negli-
gible. Third, we determined that hypoxia ⁄ SD induces
transcription and secretion of IL-10, which signifi-

cardiac fibroblasts grown in DMEM with 10% fetal bovine serum at 24 h after IL-10 treatment at different concentrations. **P < 0.01 and
***P < 0.001 versus 10% fetal bovine serum treatment group. (C) The relative mRNA levels of collagen I, collagen III and a-smooth muscle
actin (a-SMA) in cardiac fibroblasts in the presence and absence of IL-10. *P < 0.05 and **P < 0.01 versus Cont group. (D) Representative
western blots for collagen I and III in the presence and absence of 0.1 l
M angiotensin II and IL-10.
Paracrine anti-fibrotic effects of MSCs in vitro Z. Li et al.
3694 FEBS Journal 277 (2010) 3688–3698 ª 2010 The Authors Journal compilation ª 2010 FEBS
requires caspase 1 [29], which is itself activated by a
molecular scaffold termed the inflammasome [23]. It is
generally accepted that such IL-1b generation and
secretion by monocytes occurs in two steps. First, an
inflammatory signal, such as the endotoxin LPS, pro-
motes the synthesis and cytoplasmic accumulation of
pro-IL-1b. A second signal, in the form of exogenous
ATP, triggers caspase 1-mediated processing of pro-IL-
1b and secretion of the mature cytokine [30,31]. In our
study, hypoxia ⁄ SD enhanced the transcription and
translation of pro-IL-1b as well as the cleavage of pro-
IL-1b into mature IL-1b. However, IL-1b was not
released from hypoxia ⁄ SD-stimulated MSCs unless
ATP or LPS was present.
Interestingly, although hypoxia ⁄ SD induced signifi-
cant TNF-a transcription, the translation of TNF-a
remained unchanged even when TNF-a transcription
was inhibited by MG132 or BAY 11-7082. The exact
reason for the translational repression of TNF-a is
unclear, but there are at least two possibilities: micro-
RNA-mediated TNF-a mRNA translational silencing
or TNF-a mRNA AU-rich element-mediated post-
transcriptional regulation involving AU-rich element-

mediator of the cells’ paracrine anti-fibrotic effects.
These findings help to improve our understanding of
the cellular and molecular basis of MSCs’ anti-inflam-
matory and paracrine effects.
Materials and methods
Materials
Iscove’s modified Dulbecco’s medium (IMDM), Dulbecco’s
modified Eagle’s medium (DMEM) and Trizol reagent were
purchased from Invitrogen (Carlsbad, CA, USA). M-MLV
reverse transcriptase was obtained from Promega (Madison,
WI, USA) and Power SYBR Green PCR Master Mix was
purchased from Applied Biosystems (Foster City, CA,
USA). SB202190, U0126, MG132, BAY 11-7082, LPS and
angiotensin II were obtained from Sigma (St. Louis, MO,
USA). The BrdU cell proliferation assay kit was acquired
from Calbiochem (Gibbstown, NJ, USA). ELISA detection
kits for IL-1b, TNF-a and IL-10 as well as antibodies
against IL-1b and TNF-a were obtained from R&D Sys-
tems (Minneapolis, MN, USA), whereas antibodies against
ERK, phospho-ERK1 ⁄ 2, p38 and phospho-p38 were pur-
chased from Cell Signalling Technology (Danvers, MA,
USA). Antibodies against NF-jBp65, caspase 1, collagen I,
collagen III and b-actin and horseradish peroxidise-conju-
gated secondary antibodies were manufactured by Santa
Cruz Biotechnology (Santa Cruz, CA, USA).
Cell culture, inhibitor treatment and conditioned
medium collection
Isolation and expansion of MSCs were conducted as previ-
ously reported [20]. Briefly, bone marrow was harvested
from the tibias and femurs of 80 g rats, plated in IMDM

and < 30 kDa components using 30 kDa molecular mass
cut-off ultrafiltration membranes (Millipore, Billerica, MA,
USA) if necessary. As a control, plates containing medium
alone were also subjected to the same conditions.
Neonatal cardiac fibroblasts were isolated from Sprague–
Dawley rats (1–3 days old) and characterized as previously
described [38]. All experiments were performed on the sec-
ond or third passage of cardiac fibroblasts after starvation
in serum-free DMEM for 24 h. The cells were then treated
with control medium or MSCs-CM.
[
3
H]-Thymidine and [
3
H]-proline uptake assays
Cardiac fibroblasts were transferred to 24-well plates,
starved of serum for 24 h and then stimulated with stan-
dard medium or MSCs-CM for 24 h. [
3
H]-Thymidine or
[
3
H]-proline (Institute of High Energy Physics, Chinese
Academy of Sciences, Beijing, China) was added to each
well to a final concentration of 1 lCiÆmL
)1
during the last
6 h of incubation. Stimulation was terminated by rinsing
the cardiac fibroblasts three times with NaCl⁄ P
i

CTGTACATCAAGGA; alpha smooth muscle actin
(a-SMA): AGCCAGTCGCCATCAGGAAC and CCGG
AGCCATTGTCACACAC; and glyceraldehyde-3-phosphate
dehydrogenase: 5¢-GGCACAGTCAAGGCTGAGAATG-3¢
and 5¢-ATGGTGGTGAAGACGCCAGTA-3¢.
Immunocytochemical staining for NF-jBp65
MSCs in IMDM supplemented with 10% fetal bovine
serum were plated on six-well glass slides. When the cells
reached 70–80% confluence, they were preincubated with
U0126 or BAY 11-7082 as described above and exposed to
hypoxia ⁄ SD for 6 h. The cells were then fixed in 2% para-
formaldehyde in NaCl ⁄ P
i
for 30 min, washed twice with
NaCl ⁄ P
i
and permeabilized with 0.3% Triton X-100 in
NaCl ⁄ P
i
for 1 0 min. N ext, the MSCs were blocked in 2% goat
serum for 1 h and incubated with rabbit anti-(NF-jBp65
primary IgG) for 1–2 h. The cells were then washed and
incubated with rhodamine-labelled goat anti-(rabbit second-
ary IgG). After three NaCl ⁄ P
i
washes and incubation with
the nuclear stain 4¢,6-diamidino-2-phenylindone for 20 min,
the MSCs were washed in NaCl ⁄ P
i
for 10 min and

tions. Production of IL-1b, TNF-a and IL-10 by MSCs
Paracrine anti-fibrotic effects of MSCs in vitro Z. Li et al.
3696 FEBS Journal 277 (2010) 3688–3698 ª 2010 The Authors Journal compilation ª 2010 FEBS
was then determined by ELISA using the commercially
available kits mentioned earlier according to the manufac-
turer’s instructions. Absorbance was measured at 450 nm
using a microplate reader. Results were compared with a
standard curve constructed by titrating rat IL-1b, TNF-a
and IL-10.
BrdU incorporation assay
Cardiac fibroblasts were transferred to 96-well plates,
starved of serum for 24 h and stimulated with IL-10 for
24 h. DNA synthesis at 24 h was measured using a BrdU
ELISA kit. Briefly, the cells were incubated for 4 h at
37 °C with 20 lLÆwell
)1
of BrdU. The supernatant was then
removed and the cells were fixed in 200 lL Æwell
)1
of FixDe-
nat for 30 min at room temperature. Subsequently, anti-
BrdU Ig, horseradish peroxidase-conjugated goat anti-
(mouse IgG) and substrate solution were applied to the
wells. The absorbance of the samples was measured at
450 nm using a microplate reader.
Statistical analysis
Data are expressed as the mean ± SEM. Differences
among groups were tested by one-way analysis of variance
(ANOVA). Comparisons between two groups were
evaluated using Student’s t-test. A value of P < 0.05 was

678–685.
6 Ohnishi S, Yasuda T, Kitamura S & Nagaya N (2007)
Effect of hypoxia on gene expression of bone marrow-
derived mesenchymal stem cells and mononuclear cells.
Stem Cells 25, 1166–1177.
7 Li L, Zhang S, Zhang Y, Yu B, Xu Y & Guan Z
(2009) Paracrine action mediates the antifibrotic effect
of transplanted mesenchymal stem cells in a rat model
of global heart failure. Mol Biol Rep 36, 725–731.
8 Li L, Zhang Y, Li Y, Yu B, Xu Y, Zhao S & Guan Z
(2008) Mesenchymal stem cell transplantation attenu-
ates cardiac fibrosis associated with isoproterenol-
induced global heart failure. Transpl Int 21, 1181–1189.
9 Tang J, Wang J, Guo L, Kong X, Yang J, Zheng F,
Zhang L & Huang Y (2010) Mesenchymal stem cells
modified with stromal cell-derived factor 1 alpha
improve cardiac remodeling via paracrine activation of
hepatocyte growth factor in a rat model of myocardial
infarction. Mol Cell 29 , 9–19.
10 Palmer JN, Hartogensis WE, Patten M, Fortuin FD &
Long CS (1995) Interleukin-1 beta induces cardiac myo-
cyte growth but inhibits cardiac fibroblast proliferation
in culture. J Clin Invest 95, 2555–2564.
11 Guo J, Lin GS, Bao CY, Hu ZM & Hu MY (2007)
Anti-inflammation role for mesenchymal stem cells
transplantation in myocardial infarction. Inflammation
30, 97–104.
12 Liu N, Chen R, Du H, Wang J, Zhang Y & Wen J
(2009) Expression of IL-10 and TNF-alpha in rats with
cerebral infarction after transplantation with mesenchy-

their interleukin-10 production. Nat Med 15, 42–49.
18 Lu WQ, Qiu Y, Li TJ, Tao X, Sun LN & Chen WS
(2009) Timosaponin B-II inhibits pro-inflammatory
cytokine induction by lipopolysaccharide in BV2 cells.
Arch Pharm Res 32, 1301–1308.
19 Shao DZ & Lin M (2008) Platonin inhibits LPS-induced
NF-kappaB by preventing activation of Akt and
IKKbeta in human PBMC. Inflamm Res 57, 601–606.
20 Chen J, Baydoun AR, Xu R, Deng L, Liu X, Zhu W,
Shi L, Cong X, Hu S & Chen X (2008) Lysophospha-
tidic acid protects mesenchymal stem cells against
hypoxia and serum deprivation-induced apoptosis. Stem
Cells 26, 135–145.
21 Li Z, Wei H, Liu X, Hu S, Cong X & Chen X (2010)
LPA rescues ER stress-associated apoptosis in hypoxia
and serum deprivation-stimulated mesenchymal stem
cells. J Cell Biochem, doi:10.1002/jcb.22731.
22 Rutault K, Hazzalin CA & Mahadevan LC (2001)
Combinations of ERK and p38 MAPK inhibitors
ablate tumour necrosis factor-alpha (TNF-alpha)
mRNA induction. Evidence for selective destabilization
of TNF-alpha transcripts. J Biol Chem 276, 6666–6674.
23 Martinon F, Burns K & Tschopp J (2002) The inflam-
masome: a molecular platform triggering activation of
inflammatory caspases and processing of proIL-beta.
Mol Cell 10, 417–426.
24 Li Q & Verma IM (2002) NF-kappaB regulation in the
immune system. Nat Rev Immunol 2, 725–734.
25 Richmond A (2002) Nf-kappa B, chemokine gene tran-
scription and tumour growth. Nat Rev Immunol 2,

Kollias G (2001) Interleukin-10 targets p38 MAPK to
modulate ARE-dependent TNF mRNA translation and
limit intestinal pathology. EMBO J 20
, 3760–3770.
34 Yao Y, Zhang F, Wang L, Zhang G, Wang Z, Chen J
& Gao X (2009) Lipopolysaccharide preconditioning
enhances the efficacy of mesenchymal stem cells
transplantation in a rat model of acute myocardial
infarction. J Biomed Sci 16, 74–84.
35 Wilson MS, Elnekave E, Mentink-Kane MM, Hodges
MG, Pesce JT, Ramalingam TR, Thompson RW,
Kamanaka M, Flavell RA, Keane-Myers A et al. (2007)
IL-13Ralpha2 and IL-10 coordinately suppress airway
inflammation, airway-hyperreactivity, and fibrosis in
mice. J Clin Invest 117, 2941–2951.
36 Liu CH & Hwang SM (2005) Cytokine interactions in
mesenchymal stem cells from cord blood. Cytokine 32,
270–279.
37 Kitamura H & Okazaki M (2005) AG-041R, a novel
indoline-2-one derivative, stimulates chondrogenesis in a
bipotent chondroprogenitor cell line CL-1. Osteoarthri-
tis Cartilage 13, 287–296.
38 Chen J, Han Y, Zhu W, Ma R, Han B, Cong X, Hu S
& Chen X (2006) Specific receptor subtype mediation of
LPA-induced dual effects in cardiac fibroblasts. FEBS
Lett 580, 4737–4745.
Supporting information
The following supplementary material is available:
Fig. S1. Dose–response data for U0126 and SB202190.
This supplementary material can be found in the