Autocrine production of TGF-β1 promotes myofibroblastic differentiation of neonatal lung mesenchymal stem cells

Abstract

We have isolated mesenchymal stem cells (MSCs) from tracheal aspirates of premature infants with respiratory distress. We examined the capacity of MSCs to differentiate into myofibroblasts, cells that participate in lung development, injury, and repair. Gene expression was measured by array, qPCR, immunoblot, and immunocytochemistry. Unstimulated MSCs expressed mRNAs encoding contractile (e.g., ACTA2, TAGLN), extracellular matrix (COL1A1 and ELN), and actin-binding (DBN1, PXN) proteins, consistent with a myofibroblast phenotype, although there was little translation into immunoreactive protein. Incubation in serum-free medium increased contractile protein (ACTA2, MYH11) gene expression. MSC-conditioned medium showed substantial levels of TGF-β1, and treatment of serum-deprived cells with a type I activin receptor-like kinase inhibitor, SB-431542, attenuated the expression of genes encoding contractile and extracellular matrix proteins. Treatment of MSCs with TGF-β1 further induced the expression of mRNAs encoding contractile (ACTA2, MYH11, TAGLN, DES) and extracellular matrix proteins (FN1, ELN, COL1A1, COL1A2), and increased the protein expression of α-smooth muscle actin, myosin heavy chain, and SM22. In contrast, human bone marrow-derived MSCs failed to undergo TGF-β1-induced myofibroblastic differentiation. Finally, primary cells from tracheal aspirates behaved in an identical manner as later passage cells. We conclude that human neonatal lung MSCs demonstrate an mRNA expression pattern characteristic of myofibroblast progenitor cells. Autocrine production of TGF-β1 further drives myofibroblastic differentiation, suggesting that, in the absence of other signals, fibrosis represents the “default program” for neonatal lung MSC gene expression. These data are consistent with the notion that MSCs play a key role in neonatal lung injury and repair.

mesenchymal stem cells (MSCs) are multipotent cells capable of differentiation into various mesenchymal lineages, including osteoblasts, chrondroblasts, and adipocytes (29). Other cell types of mesenchymal origin include smooth muscle cells and myofibroblasts. Myofibroblsts are essential for distal lung development and the formation of saccules and alveoli. Recently, it has been shown that MSCs undergo myofibroblastic differentiation: bone marrow MSCs contribute to carbon tetrachloride or thioacetamide-induced hepatic fibrosis (35). In addition, lysophosphatidic acid stimulates differentiation of human adipose tissue MSCs to myofibroblasts via an autocrine transforming growth factor (TGF)-β-dependent pathway (16).

Although mesenchymal stem cells were first isolated from bone marrow, it is now generally accepted that most organs carry their own population of MSCs (8). With regard to the lung, MSCs have been isolated from the bronchial tissue of patients undergoing lobectomy for primary lung tumors (36). Our laboratory has isolated MSCs from the tracheal aspirates of premature infants undergoing mechanical ventilation for respiratory distress (14). MSCs of donor sex identity have been found in lung allografts years after transplantation, confirming that MSCs may originate from the lung tissue itself. The physiological role of these cells remains unclear. MSC-derived lung myofibroblasts may participate in lung injury and repair, or in normal lung development, for example, in the alveolarization process. Alveolar myofibroblasts, which express α-smooth muscle actin, elastin, PDGF-Rα, and the actin-binding protein drebrin, are located peripherally at the tips of the developing alveolar septa (23, 24, 44) and are required for the formation of pulmonary alveolar secondary septa (6, 20).

To examine the ability of neonatal lung MSCs to differentiate into myofibroblasts, we monitored the expression of selected genes in MSC isolates from premature infants. We found that neonatal lung MSCs express mRNAs encoding contractile and extracellular matrix proteins, consistent with a myofibroblast progenitor phenotype. Autocrine production of TGF-β1 further drives myofibroblastic differentiation, suggesting that, in the absence of other signals, fibrosis represents the “default program” for neonatal lung MSC gene expression.

METHODS

Cell culture.

MSCs from tracheal aspirates of premature infants were isolated as described previously (14). The study was approved by the University of Michigan Institutional Review Board. Informed consent was obtained from one parent or legal guardian. Cells were cultured in DMEM supplemented with 10% FBS, 1% penicillin-streptomycin, 1% l-glutamine, and 0.5% Fungizone and incubated at 37°C in 5% CO₂. For selected experiments, individual colonies of MSCs were identified and subjected to study. For remaining experiments, colonies were allowed to coalesce, and cells were grown to 50–60% confluence before passage. Passage number 3 or 4 cells, plated in 35- or 100-mm dishes, were used for the remaining studies.

Human bone marrow MSCs were provided by the Institute for Regenerative Medicine, Texas A&M Health Science Center. Passage 3 cells were used for study. Normal human lung fibroblasts were purchased from Lonza (Walkersville, MD). Passage 4 cells were used for study.

After serum starvation for 24 h, cells were treated with TGF-β1 (PeproTech, Rocky Hill, NJ), the type I activin receptor-like kinase inhibitor SB-431542 (Sigma-Aldrich, St. Louis, MO), and the histone deacetylase inhibitor trichostatin A (Sigma-Aldrich) alone or in combination with TGF-β1. Experiments were performed in the absence of serum.

Gene array.

For gene arrays, individual colonies of MSCs were identified and isolated using cloning disks (PGC Scientifics, Frederick, MD). We examined the gene expression profile of MSCs using the Illumina HumanRefSeq-8 v3 expression BeadChip platform (San Diego, CA). This system covers greater than 18,000 unique genes from the NCBI RefSeq database. Total RNA was extracted using the RNeasy Plus Mini kit (Qiagen, Valencia, CA). Further preparation and analysis was carried out by the University of Michigan Sequencing Core, according to the chip manufacturer's recommended protocol. Hybridized biotinylated cRNA was detected with streptavidin-Cy3 and quantitated using an Illumina BeadArray Reader.

Immunocytochemistry.

Cells were grown in 35-mm plastic dishes or on collagen- or fibronectin-coated glass slides (BD Biosciences, San Jose, CA). Cells were fixed in 1% paraformaldehyde. Cells were permeabilized in 0.1% Triton X-100 in PBS. Alexa Fluor (AF) dye antibody conjugates were prepared using N-hydroxysuccinimide-esterified dye conjugation reactions with 20 μg of antibody and 100 μg of activated dye in 10 mM sodium bicarbonate buffer (pH 8.5) for 1 h. The reaction was quenched by adding 50 μl of 1 M Tris·HCl, pH 7.2, per 50 μl of reaction mix for 1 h. The conjugated antibody was purified by separating it from quenched dye over a G-50 spin column (GE Healthcare, Piscataway, NJ). Slides were probed with AF 488- or AF 594-conjugated mouse anti-α-smooth muscle actin antibody (Sigma-Aldrich), AF 633-conjugated myosin heavy chain antibody (Sigma-Aldrich), AF 594-conjugated SM22 antibody (Abcam, Cambridge, MA), or AF 488-conjugated collagen I antibody (both from Abcam). Nuclei were visualized with 10 μg/ml Hoechst 33258 (Sigma-Aldrich). Cells were imaged using a Zeiss LSM 150 confocal microscope (Thornwood, NY) or Olympus IX71 inverted microscope.

Immunoblotting.

Cell lysates were adjusted for protein concentration, resolved by SDS-PAGE, and transferred to a nitrocellulose membrane. Membranes were blocked in 5% milk for 1 h in room temperature and probed with antibodies against α-smooth muscle actin (Calbiochem, San Diego, CA), myosin heavy chain (Sigma-Aldrich), and SM22 (Cell Signaling, Danvers, MA). Antibody binding was detected with a peroxidase-conjugated anti-mouse or anti-goat IgG and chemiluminescence.

Quantitative real-time PCR.

Total RNA was extracted using the RNeasy Plus Mini kit (Qiagen, Valencia, CA) and then transcribed to first-strand cDNA using Taqman Reverse Transcription Reagents (Applied Biosystems, Foster City, CA). First-strand cDNA was then used to quantify the expression of α-smooth muscle actin, myosin heavy chain, SM22, elastin, collagen 1, drebrin, TGFB1, and GAPDH mRNA levels using specific primers. All primers were from IDT (Coralville, IA). The following primers were used: α-smooth muscle actin (ACTA2) forward primer 5′-TCA ATG TCC CAG CCA TGT AT-3′, reverse primer 5′-CAG CAC GAT GCC AGT TGT-3′; myosin heavy chain (MYH11) forward primer 5′-CAG ATG CTG GAC CTT GAA GA-3′, reverse primer 5′-TCC ATG ACC AGG ATC TCA TC-3′; SM22 (TAGLN) forward primer 5′-GAA TGG CGT GAT TCT GAG C-3′, reverse primer 5′-CTC CAT CTG CTT GAA GAC CA-3′; elastin (ELN) forward primer 5′-GCC AAG TAT GGA GCT GCT G-3′, reverse primer 5′-GAC ACC AAC ACC TGG AAC G-3′; collagen 1 (COL1A1) forward primer 5′-CCC TGG AAA GAA TGG AGA TG-3′, reverse primer 5′-CCA CTG AAA CCT CTG TGT CC-3′; TGF-β1 (TGFB1) forward primer 5′-GGA CAT CAA CGG GTT CAC TA-3′, reverse primer 5′-GCC ATG AGA AGC AGG AAA G-3′; GAPDH forward primer 5′-CGA CCA CTT TGT CAA GCT CA-3′, reverse primer 5′-AGG GGT CTA CAT GGC AAC TG-3′. Reactions were performed on an Eppendorf realplex2 (Westbury, NY). The resulting amplification and melt curves were analyzed to ensure specific PCR product. Threshold cycle (C_T) values were used to calculate the fold change in transcript levels using the 2^−ΔΔCT method (11). For the downregulated genes, fold regulation was calculated as −1/fold change.

ELISA.

The TGF-β1 level of serum-free cell supernatants was measured by ELISA (R&D Systems, Minneapolis, MN). Latent TGF-β1 was activated to immunoreactive TGF-β1 by acidification with 1 N HCl for 10 min in room temperature. Samples were neutralized with 1.2 N NaOH/0.5 M HEPES.

Statistical analysis.

For microarray analysis, background-corrected values for each probe on the BeadChip array were extracted using GenomeStudio Data Analysis Software (Illumina). For unstimulated samples, detection P values were computed using a non-parametric method. Probe signals are ranked relative to signals of negative controls. Detection P value = 1 − R/N, where R is the rank of the gene signal relative to negative controls and N is number of negative controls.

Statistical differences in gene expression following treatment with TGF-β were calculated using the Illumina Custom differential expression algorithm. Multiple tests were corrected by the Benjamini and Hochberg false discovery rate (3). For qPCR, data were normalized for GAPDH, and differences in gene expression were analyzed by paired t-test.

RESULTS

Patient characteristics.

MSCs from 16 individual subjects, 10 of them male, were studied. Average gestational age at birth was 27 ± 2 wk (means ± SD). Two patients had a maternal history of chorioamniotitis, three subjects contracted late-onset sepsis, and nine developed moderate-to-severe bronchopulmonary dysplasia, as defined by a supplemental oxygen requirement after 36 wk of gestation.

Unstimulated neonatal lung MSCs show a gene expression profile consistent with a myofibroblast progenitor phenotype.

We examined the gene expression profile of unstimulated neonatal lung MSCs from passage 3–4 cells of four different patients. Cells demonstrated mRNA expression of genes typically associated with MSCs, including 5′-nucleotidase, ecto (CD73), bone marrow stromal cell antigen 1, endoglin (CD105), galectin 1, galectin 3, stromal antigen 1, stromal antigen 2 (transcript variant 4), and Thy-1 cell surface antigen (CD90). To monitor myofibroblastic differentiation, we focused on mRNAs encoding contractile, extracellular matrix, and actin-binding proteins (Tables 1 and 2, Supplemental Table S1. Supplemental data for this article is available online at the AJP-Lung web site.). Extracellular matrix and actin-binding proteins, which link extracellular matrix components to the cell cytoskeleton through specialized adhesion complexes, are typically found in myofibroblasts and are instrumental in the production of force by myofibroblastic cells (10, 17, 21, 28, 37, 46). We also list genes associated with TGF-β signaling and alveolar septal fibroblasts (Supplemental Table S2). The listed genes showed a detection P value of less than 0.05. We found ample mRNA expression of genes encoding contractile (e.g., ACTA2, CNN1, DES, and TAGLN), extracellular matrix (e.g., COL1A1, COL1A2, ELN1, FN1, LAMA1, and VIM), and actin-binding proteins (e.g., ACTN1, CFL1, DBN1, PFN1, PXN, TLN), consistent with a myofibroblast-like phenotype. In addition to ACTA2 and DBN1, PDGFRA, PDGFRB, ADRP, and PPARG were also highly expressed in unstimulated MSCs. The complete gene expression profile of unstimulated MSCs is provided in the online data supplement (Supplemental Table S3).

Table 1.

mRNA expression of contractile proteins in human lung mesenchymal stem cells

	Average	Average Detection
Symbol	Signal	P Value	Definition
ACTA2	14462.025	0	actin, α2, smooth muscle, aorta
ACTC1	70.525	0.2311725	actin, α, cardiac muscle 1
ACTB	20147.475	0	actin, β
ACTG1	26916.775	0	actin, γ1
ACTG2	7513.6	0	actin, γ2, smooth muscle, enteric
CNN1	508.35	0.003765	calponin 1, basic, smooth muscle
DES	142.975	0.07455	desmin
MYL6	19695.4	0	myosin, light chain 6, alkali, smooth muscle and nonmuscle
MYL6B	3092.05	0	myosin, light chain 6B, alkali, smooth muscle and nonmuscle
SMTN	143.725	0.00017	smoothelin, transcript variant 2
TAGLN	6986.3	0.0088375	transgelin, transcript variant 1
TAGLN2	3178.025	0	transgelin 2
TPM1	5035	0	tropomyosin 1 (α), transcript variant 3
TPM2	12174.275	0	tropomyosin 2 (β), transcript variant 2
TPM3	244	0.0009525	tropomyosin 3, transcript variant 1
TPM4	747.75	0	tropomyosin 4

Table 2.

mRNA expression of extracellular matrix proteins in human lung mesenchymal stem cells

	Average	Average Detection
Symbol	Signal	P Value	Definition
BGN	13263.55	0	biglycan
COL1A1	17794.2	0	collagen, type I, α1
COL1A2	13594.225	0	collagen, type I, α2
COL3A1	8155.6	0	collagen, type III, α1
COL4A1	5400.225	0	collagen, type IV, α1
COL4A2	914.05	0	collagen, type IV, α2
COL4A5	343.05	0.0004025	collagen, type IV, α5
COL5A1	5080.2	0	collagen, type V, α1
COL5A2	5969.625	0	collagen, type V, α2
COL6A1	5057.275	0	collagen, type VI, α1
COL6A2	1266.6	0	collagen, type VI, α2
COL6A3	10647.575	0	collagen, type VI, α3, transcript variant 1
COL7A1	1096.7	0	collagen, type VII, α1
COL8A1	1241.825	0	collagen, type VIII, α1, transcript variant 2
COL12A1	710.375	0.0013375	collagen, type XII, α1, transcript' variant long
COL13A1	269.825	0.0000025	collagen, type XIII, α1, transcript variant 19
COL15A1	276.375	0.0041425	collagen, type XV, α1
COL16A1	1425.675	0	collagen, type XVI, α1
COL18A1	242.85	0.0057075	collagen, type XVIII, α1, transcript variant 2
ELN	131.275	0.0318125	elastin, transcript variant 4
FBN2	2318.675	0	fibrillin 2
FN1	138.175	0	fibronectin 1, transcript variant 6
LAMA1	129.475	0.009245	laminin, α1
LAMA2	140.275	0.0001375	laminin, α2, transcript variant 1
LAMA4	747.625	0	laminin, α4
LAMA5	958.925	0	laminin, α5
LAMB1	596.8	0	laminin, β1
LAMB2	440.3	0	laminin, β2
LAMC1	4422.65	0	laminin, γ1
TNC	754.175	0	tenascin C
VCAN	1369.225	0	versican
VIM	17251.775	0	vimentin

Among the genes that were significantly expressed in unstimulated MSCs was ACTA2, which encodes α-smooth muscle actin, and TAGLN, which encodes the contractile protein SM22. However, when we examined the expression of α-smooth muscle actin and SM22 by immunocytochemical staining and immunoblotting, there was minimal immunoreactive protein (Figs. 1 and 2), suggesting that mRNAs encoding contractile proteins may not be translated in unstimulated cells.

Fig. 1.

Immunocytochemical analysis of contractile protein expression. A–E: unstimulated mesenchymal stem cells (MSCs) show little or no expression of α-smooth muscle actin, SM22, or myosin heavy chain (MHC). Individual channels for α-smooth muscle actin (green, A), SM22 (red, B), MHC (blue, C), and DAPI (black, D) are shown. E represents the merged image (line segment = 20 μM). F–J: treatment with TGF-β1 (10 ng/ml for 6 days) increases the cell size and expression of α-actin (F), MHC (G), and SM22 (H). DAPI staining is shown in I. Colocalization of α-actin, MHC, and SM22 appears white (J). α-Actin is mainly incorporated into contractile filaments, whereas MHC and SM22 are both perinuclear and filamentous in location.

Fig. 2.

Expression of contractile proteins by immunoblot. Unstimulated MSCs show negligible expression of the contractile proteins α-smooth muscle actin and MHC and modest expression of SM22. Following long-term (6-day) serum deprivation, protein abundance of α-actin and MHC increases. Treatment with TGF-β1 (10 ng/ml for 6 days) further increases the expression of these contractile proteins. Expression of contractile proteins in cells undergoing long-term serum deprivation is significantly decreased by treatment with a type I activin receptor-like kinase inhibitor, SB-431542 (10 μM for 6 days).

Evidence of autocrine TGF-β1 production by MSCs.

Our gene arrays did not reveal significant expression of TGFB1 in unstimulated MSCs. However, TGFBI (for TGF, β-induced) was highly expressed in serum-deprived unstimulated MSCs, consistent with the notion that MSCs themselves produce TGF-β1. To examine this further, we used qPCR to detect TGFB1 mRNA expression in passage 3–4 neonatal lung MSCs. Out of 10 isolates tested, the average cycle number value (C_T) was 23.5 ± 0.3 (means ± SE) compared with 20.8 ± 0.3 for GAPDH, indicating a moderate-to-high level of expression. The explanation for this discrepancy between the gene array and qPCR is unclear but could be due to differences in the probe sequences employed (unlike our sequences, the gene array probe sequence for TGFB1 is from the non-coding part of the gene).

We then examined the conditioned media of nine MSC isolates for immunoreactive TGF-β1 by ELISA. Conditioned media from cells deprived of serum for 24 h showed substantial levels of TGF-β1 (389 ± 43 pg/ml, means ± SE). When we subjected MSCs to prolonged serum deprivation, we noted increases in the protein expression of α-smooth muscle actin and myosin heavy chain (Fig. 2). Finally, treatment of unstimulated, serum-deprived cells with a type I activin receptor-like kinase inhibitor that blocks TGF-β1 signaling, SB-431542 (10 μM), significantly attenuated the expression of mRNA encoding contractile and extracellular matrix proteins, as measured by qPCR (Fig. 3), as well as α-actin, SM22, and myosin heavy chain protein abundance (Fig. 2). Together, these data suggest the presence of a TGF-β1 autocrine loop capable of driving myofibroblastic differentiation.

Fig. 3.

Treatment of unstimulated cells with a type I activin receptor-like kinase inhibitor significantly attenuated the expression of mRNAs encoding contractile and extracellular matrix proteins. After serum starvation for 24 h, unstimulated MSCs were treated with 10 μM SB-431542 for 72 h. Control MSCs were cultured in serum-free medium for 72 h. All bars are significantly different from controls at a level of P < 0.05, n = 3.

Effect of TGF-β1 treatment on MSC gene expression.

We examined MSC gene expression following treatment of three isolates with TGF-β1 (10 ng/ml for 72 h). Gene expression was compared with cells incubated in serum-free medium for 72 h. We noted the significant upregulation of 428 genes and downregulation of 236 genes (Table 3, Supplemental Table S4). TGF-β1 significantly increased the mRNA expression of many genes encoding contractile (CNN1, TAGLN), extracellular matrix (COL4A1, COL5A1, ELN1, FN1), and actin-binding proteins (ACTN1, CFL1, DBN1, FLNA). The expression of selected genes was also examined by qPCR (Fig. 4). TGF-β1 induced the mRNA expression of genes encoding contractile (ACTA2, MYH11, TAGLN) and extracellular matrix proteins (ELN, COL1A1). TGF-β1 also increased the expression of TGFB1. TGF-β1 treatment substantially increased contractile protein abundance and the incorporation of contractile proteins into filaments (Figs. 1 and 2), consistent with myofibroblast differentiation.

Table 3.

Genes regulated by TGF-β1 treatment

	Difference	Difference
Symbol	Score	P Value	Definition
ACTN1	15.222	0.03004	actinin, α1
ACVR1	30.25	0.00094	activin A receptor, type I
ADFP	−19.074	0.01238	adipose differentiation-related protein
BMP1	13.849	0.04122	bone morphogenetic protein 1, transcript variant BMP1-5
BMPR2	17.107	0.01946	bone morphogenetic protein receptor, type II
CNN1	34.697	0.00034	calponin 1, basic, smooth muscle
CFL1	22.488	0.00564	cofilin 1 (nonmuscle)
COL4A1	14.68	0.03404	collagen, type IV, α1
COL4A2	38.162	0.00015	collagen, type IV, α2
COL5A1	56.17	0	collagen, type V, α1
COL6A1	−46.324	0.00002	collagen, type VI, α1
COL6A2	−14.625	0.03447	collagen, type VI, α2, transcript variant 2C2
COL7A1	341.909	0	collagen, type VII, α1
COL8A1	40.596	0.00009	collagen, type VIII, α1, transcript variant 1
COL8A1	34.513	0.00035	collagen, type VIII, α1, transcript variant 2
DBN1	21.286	0.00744	drebrin 1, transcript variant 1
ELN	43.581	0.00004	elastin, transcript variant 4
FLNA	42.722	0.01301	filamin A, α
FN1	89.658	0	fibronectin 1, transcript variant 7
FN1	19.904	0.01022	fibronectin 1, transcript variant 3
FN1	18.039	0.01571	fibronectin 1, transcript variant 6
INHBE	19.187	0.01206	inhibin, βE
MRCL3	17.706	0.01696	myosin regulatory light chain MRCL3
MYL9	47.622	0.00002	myosin, light chain 9, regulatory, transcript variant 1
MYO5A	39.34	0.00012	myosin VA (heavy chain 12, myoxin)
SMAD7	34.803	0.00033	SMAD family member
TAGLN	40.617	0.00009	transgelin, transcript variant 2
TGFB1I1	17.065	0.01966	transforming growth factor β1-induced transcript 1, transcript variant 2
TGFBI	44.134	0.00004	transforming growth factor β-induced, 68 kDa
TPM1	41.718	0.00007	tropomyosin 1 (α), transcript variant 7
TPM1	34.922	0.00032	tropomyosin 1 (α), transcript variant 3
TPM4	23.142	0.00485	tropomyosin 4

mRNAs encoding contractile, extracellular matrix, actin-binding, and TGF-β-related proteins are shown.

Fig. 4.

Treatment of MSCs with TGF-β1 induced the mRNA expression of genes encoding contractile and extracellular matrix proteins. After serum starvation for 24 h, unstimulated MSCs were treated with TGF-β1 (10 ng/ml for 72 h). All bars are significantly different from controls at a level of P < 0.05, n = 3.

Experiments on primary cells.

To determine whether primary colonies of neonatal lung MSCs demonstrate similar characteristics as later passage cells, supernatants from 35-mm plates were collected for measurement of TGF-β1 by ELISA. Conditioned media from cells deprived of serum for 24 h showed an average level of 590 pg/ml (range 320–904). qPCR showed expression of mRNAs encoding contractile and extracellular matrix proteins which increased significantly following TGF-β treatment (Fig. 5). Immunocytochemical stains showed minimal protein expression of α-actin and collagen I in unstimulated cells, and substantial expression after TGF-β1 stimulation. Together, these data demonstrate that primary cells produce TGF-β1 and undergo myofibroblastic differentiation in an identical manner as passage 3 cells.

Fig. 5.

Experiments on primary cells. A: qPCR shows a significant increase in the expression of mRNAs encoding contractile and extracellular matrix proteins following TGF-β1 treatment. B–G: immunocytochemical stains showed minimal protein expression of α-actin (red) and collagen I (green) in unstimulated cells (B–D) and substantial expression after TGF-β1 stimulation (E–G). B and E show merged images; individual channels for α-smooth muscle actin (C, E) and collagen I (D, G) are also shown.

Effect of TGF-β1 treatment on human bone marrow MSCs and normal human lung fibroblasts.

We examined whether TGF-β1 induces myofibroblastic differentiation in human bone marrow MSCs and normal human lung fibroblasts. The baseline patterns of mRNA expression for genes encoding contractile and extracellular matrix proteins, as examined by qPCR, was similar between bone marrow MSCs, normal human lung fibroblasts, and neonatal lung MSCs. However, following TGF-β1 treatment, only normal human lung fibroblasts demonstrated increased expression of contractile and extracellular matrix proteins, similar to neonatal lung MSCs (Fig. 6). Results were confirmed by immunocytochemical staining for α-smooth muscle actin and collagen I. Thus, lung- and bone marrow-derived MSCs demonstrate different capacities for myofibroblastic differentiation.

Fig. 6.

Effect of TGF-β1 treatment on human bone marrow MSCs and normal human lung fibroblasts. A: following 10 ng/ml TGF-β1 treatment, normal human lung fibroblasts but not human bone marrow-derived MSCs demonstrate increased expression of contractile and extracellular matrix proteins, similar to neonatal lung MSCs. B–M: Immunocytochemical stains for α-smooth muscle actin (red) and collagen I (green) in unstimulated normal human lung fibroblasts (B–D), TGF-β1-stimulated normal human lung fibroblasts (E–G), unstimulated human bone marrow MSCs (H–J), and TGF-β1-stimulated human bone marrow MSCs (K–M). B, E, H and K show merged images; individual channels for α-smooth muscle actin (C,F,I,L) and collagen I (D,G,J,M) are also shown.

Trichostatin A inhibits TGF-β1-induced mRNA expression.

Trichostatin A, a histone deacetylase (HDAC) inhibitor, suppresses myofibroblastic differentiation of rat hepatic stellate cells in primary culture (41). We therefore tested the effects of trichostatin A on TGF-β1-induced myofibroblastic differentiation by qPCR. Trichostatin A significantly inhibited the expression of mRNAs encoding contractile and extracellular matrix proteins (Fig. 7).

Fig. 7.

Effect of trichostatin A, a histone deacetylase inhibitor, on TGF-β-induced mRNA expression. After serum starvation for 24 h, unstimulated MSCs were treated with trichostatin A (125 nM) with or without TGF-β1 for 72 h. Control MSCs were cultured in serum-free medium for 72 h. All bars are significantly different from controls at a level of P < 0.05, n = 3.

DISCUSSION

In this manuscript, we examined the capacity of neonatal lung MSCs to differentiate into myofibroblasts. Unstimulated MSCs expressed mRNAs encoding contractile, extracellular matrix, and actin-binding proteins, consistent with a myofibroblast phenotype, although there was little translation into immunoreactive protein. However, prolonged incubation in serum-free media was accompanied by autocrine production of TGF-β1 and type I activin receptor-dependent expression of contractile and extracellular matrix proteins. Together, these data suggest that, in the absence of other signals, fibrosis represents the default program for neonatal lung MSC gene expression.

Under mechanical stress, myofibroblast progenitors (“protomyofibroblasts”) develop cytoplasmic actin-containing stress fibers that terminate in fibronexus adhesion complexes. Under the influence of TGF-β, protomyofibroblasts become differentiated myofibroblasts, which are characterized by the de novo expression of α-smooth muscle actin, more extensively developed stress fibers, large fibronexus adhesion complexes, and greater contractile force. Differentiated myofibroblasts also express high levels of extracellular matrix proteins and fibrogenic cytokines (10, 17, 21, 26, 28, 37, 46).

It is generally accepted that myofibroblasts represent key cells in the physiological reconstruction of connective tissue after injury, and in generating the pathological tissue deformations that characterize fibrosis. In addition, based on recent evidence that fibroblastic elements have considerable influence on the gene expression of overlying epithelial cells (45), myofibroblasts in injured lungs could also have a profound effect on the neighboring airway and alveolar epithelial cells, in a manner conducive to promotion of fibrosis (26). Myofibroblast precursor cells may be recruited from different sources. Myofibroblasts have been assumed to originate from local recruitment of fibroblasts in the surrounding wound tissue (34). Other sources include pericytes and smooth muscle cells of the vascular intima (12, 31) and bone marrow-derived CD34-, CD45-positive circulating cells known as fibrocytes (1, 25, 27, 38). Myofibroblasts may also originate by means of epithelial-mesenchymal transition (15). More recently, it has been recognized that pericytes exhibit the surface markers and differentiation potential of MSCs (7). Thus, MSCs may also undergo myofibroblastic differentiation (16, 35). It is also conceivable that MSCs play a role in normal lung development and function. The presence of stable protomyofibroblasts in normal alveolar septa is well established (19, 40). During development, PDGF-Rα-expressing alveolar myofibroblasts, which normally contain α-actin and produce elastin, are required for the formation of pulmonary alveolar secondary septa (6, 20).

In the present study, unstimulated neonatal lung MSCs expressed mRNAs encoding contractile, extracellular matrix, and actin-binding proteins, consistent with a myofibroblast phenotype. However, immunoblotting and immunocytochemistry showed little translation of mRNA into contractile protein, and no incorporation of proteins into contractile filaments. Primary colonies of MSCs showed similar patterns of mRNA and protein synthesis as later passage MSCs. Thus, these unstimulated MSCs may be considered myofibroblast progenitors. On the other hand, TGF-β1 stimulation of both primary and later passage MSCs induced the mRNA expression of multiple genes encoding contractile and extracellular proteins and increased protein abundance of α-actin, SM22, and myosin heavy chain. Furthermore, immunocytochemical stains of TGF-β1-stimulated cells showed that contractile proteins were incorporated into contractile filaments. Together, these results demonstrate myofibroblastic differentiation of neonatal lung MSCs in culture and imply a role for MSCs in lung injury and repair. Indeed, while myofibroblasts are normally present along terminal airways in the developing lung, increased myofibroblast number and persistence has been noted in babies with bronchopulmonary dysplasia (42), a chronic lung disease of prematurely born infants.

Studies in rats and mice show the presence of two subpopulations of interstitial cells in the secondary alveolar septa, which subdivide primitive alveolar cavities into primary and secondary alveolar saccules (23, 24, 44). One type possesses well-developed rough endoplasmic reticulum and Golgi apparatus in the cytoplasm and expresses α-actin, elastin, PDGF-Rα, and the actin-binding protein drebrin, and is located peripherally at the tips of the alveolar septa, indicative of an alveolar myofibroblast. As noted above, alveolar myofibroblasts are required for the formation of pulmonary alveolar secondary septa (6, 20). Similar to these cells, MSCs from tracheal aspirates of premature infants born from 24 to 32 wk postconceptual age, corresponding to the saccular stage of lung development, express ACTA2, ELN, PDGFRA, and DBN1.

In addition to alveolar myofibroblasts, the alveolar septum includes a second fibroblast subpopulation containing lipid droplets in the cytoplasm. These lipofibroblasts, located at the base of the alveolar septum, produce triglycerides and other lipids that can be transferred to alveolar type II cells and used in the synthesis of pulmonary surfactant (4). While the present study does not address the capacity of MSCs to produce lipids, we have previously shown that MSCs from tracheal aspirates undergo adipogenic differentiation in the appropriate medium (14). MSCs also demonstrate significant mRNA expression of ADRP and PPARG (detection P < 0.05), each of which are expressed in human lung lipofibroblasts (33). Thus, tracheal aspirate MSCs may represent a common progenitor for both alveolar fibroblast subtypes. Consistent with this, TGF-β1 treatment was associated with a significant reduction in ADRP mRNA expression, implying a shift towards the myofibroblast phenotype and away from the lipofibroblast phenotype.

We found that serum-deprived MSCs produce significant levels of TGF-β1. While the level of TGF-β1 did not approach the ng/ml range used in experiments examining the effects of exogenous TGF-β treatment, autocrine production appeared to have physiological effects. First, we noted substantial increases in the protein expression of α-smooth muscle actin, myosin heavy chain, and SM22 after prolonged incubation in serum-free medium. Second, treatment with the type I activin receptor-like kinase inhibitor, SB-431542, decreased the expression of mRNA encoding contractile and extracellular matrix proteins, as well as α-actin, myosin heavy chain, and SM22 protein abundance. In addition, TGF-β1 treatment of MSCs increased TGF-β1 mRNA. Together, these data suggest the presence of a TGF-β1 autocrine loop capable of driving myofibroblastic differentiation. Furthermore, in the absence of other signals, these data indicate that myofibroblastic differentiation may be the default program for MSC gene expression. Autocrine production of TGF-β1 leading to myofibroblastic differentiation has been noted in hepatic stellate cells (5) and coronary artery adventitial fibroblasts after endoluminal injury (39), as well as TGF-β-stimulated rabbit keratocytes (18), hypertrophic scar fibroblasts (9), and human lung fibroblasts (43).

While neonatal lung MSCs may be primed for myofibroblastic differentiation, we found that the HDAC inhibitor trichostatin A attenuated the expression of mRNAs encoding contractile and extracellular matrix proteins. Specific regulatory regions of contractile protein genes are hyperacetylated during differentiation to smooth muscle (22, 30). Trichostatin A has previously been shown to suppress myofibroblastic differentiation of rat hepatic stellate cells in primary culture (41) and TGF-β-induced myofibroblastic differentiation of human skin fibroblasts (11). In the latter study, α-actin expression was abrogated by siRNA against HDAC4. It is therefore conceivable that chemical modifiers such as HDAC inhibitors might divert the cells from repair towards regeneration. It is also conceivable that TGF-β is opposed by endogenous antifibrotic factors. For example, basic fibroblast growth factor and interferon-β have each been shown to have antifibrotic effects (2, 32).

While we harvest MSCs from the tracheal aspirates of premature infants, we have not yet determined the precise origin of these cells. Bleomycin-induced injury has been shown to promote the migration of bone marrow-derived, type I collagen-positive fibroblasts to the lung. However, unlike lung fibroblasts, these bone marrow-derived cells do not undergo myofibroblastic differentiation following TGF-β stimulation (13). Similarly, human bone marrow-derived MSCs failed undergo myofibroblastic differentiation after TGF-β treatment.

These data are consistent with the notion that neonatal lung MSCs originate in the lung itself, rather than the bone marrow. Our preliminary studies examining the pattern of Hox gene expression in these cells also suggest a lung origin (P. Bozyk, A. Popova, M. Hershenson, unpublished data). An alternative approach to definitively determine the origin of neonatal lung MSCs would be lineage tracing studies.

In conclusion, unlike bone marrow MSCs, neonatal lung MSCs from premature infants with respiratory distress undergo myofibroblastic differentiation in culture, suggesting that these cells play a role in neonatal lung injury and repair. In addition, based on their pattern of gene expression, these cells may function as progenitors of lung alveolar myofibroblasts in vivo.

GRANTS

This work was supported by National Heart, Lung, and Blood Institute Grant HL-90134 (M. B. Hershenson). Human bone marrow MSCs were obtained from the Texas A&M Health Science Center College of Medicine Institute for Regenerative Medicine at Scott & White Hospital through National Institutes of Health Grant RR-017447.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).