TGFBI Expressed by Bone Marrow Niche Cells and Hematopoietic Stem and Progenitor Cells Regulates Hematopoiesis

The interactions of hematopoietic stem and progenitor cells (HSPCs) with extracellular matrix (ECM) components and cells from the bone marrow (BM) microenvironment control their homeostasis. Regenerative BM conditions can induce expression of the ECM protein transforming growth factor beta-induced gene H3 (TGFBI or BIGH3) in murine HSPCs. In this study, we examined how increased or reduced TGFBI expression in human HSPCs and BM mesenchymal stromal cells (MSCs) affects HSPC maintenance, differentiation, and migration. HSPCs that overexpressed TGFBI showed accelerated megakaryopoiesis, whereas granulocyte differentiation and proliferation of granulocyte, erythrocyte, and monocyte cultures were reduced. In addition, both upregulation and downregulation of TGFBI expression impaired HSPC colony-forming capacity of HSPCs. Interestingly, the colony-forming capacity of HSPCs with reduced TGFBI levels was increased after long-term co-culture with MSCs, as measured by long-term culture-colony forming cell (LTC-CFC) formation. Moreover, TGFBI downregulation in HSPCs resulted in increased cobblestone area-forming cell (CAFC) frequency, a measure for hematopoietic stem cell (HSC) capacity. Concordantly, TGFBI upregulation in HSPCs resulted in a decrease of CAFC and LTC-CFC frequency. These results indicate that reduced TGFBI levels in HSPCs enhanced HSC maintenance, but only in the presence of MSCs. In addition, reduced levels of TGFBI in MSCs affected MSC/HSPC interaction, as observed by an increased migration of HSPCs under the stromal layer. In conclusion, tight regulation of TGFBI expression in the BM niche is essential for balanced HSPC proliferation and differentiation.


Introduction
H ematopoietic stem and progenitor cells (HSPCs) interact with specialized bone marrow (BM) niches, of which mesenchymal stromal cells (MSCs) comprise an essential part. Besides cellular factors, secreted and membranebound proteins in these niches are important regulators of HSPCs [1][2][3]. Comparative gene expression profiling of murine HSPCs in homeostatic and regenerative hematopoietic conditions identified transforming growth factor betainduced gene H3 (TGFBI or BIGH3; NCBI Gene ID 7045) as being upregulated in regenerative conditions, provoked by fluorouracil (5FU) treatment [4].
While the role of TGFBI in cell-ECM interactions in different tissues is well described [8], little is known about its function in the hematopoietic BM microenvironment. TGFb, the major activator of TGFBI expression, inhibits the proliferation of primitive HSPCs and skews HSPC fate toward myelocytic progenitors [17][18][19][20][21]. This raises the question whether TGFBI has similar effects on hematopoiesis. Interestingly, HSPC adherence to BM-MSCs increased TGFBI expression in HSPCs, while also increasing their quiescence [22]. Moreover, TGFBI expression is high in murine BM HSPCs compared to fetal liver HSPCs, indicating that TGFBI might become important for HSPCs during migration to and residence in the BM [23]. Furthermore, murine stromal cell lines supportive for HSPCs display elevated TGFBI expression levels, and TGFBI knockdown zebrafish display severely decreased HSPC numbers, indicating that TGFBI is important for HSC specification [24].
These data suggest that TGFBI plays a key role in shaping the BM microenvironment by regulating HSPC development and localization. The aim of this study is to investigate whether TGFBI expression in human stromal and hematopoietic cells affects human HSPC maintenance and differentiation. Our results indicate that tight regulation of TGFBI expression in both HSPCs and MSCs is essential for a balanced proliferation, differentiation, and homeostasis of human HSPCs.

Human cells
Human material was obtained after informed consent, with approval of the local medical ethics committee (MEC). BM was aspirated from patients undergoing cardiac surgery (permit MEC 04/042, No. 04.17.370; AMC, Amsterdam, The Netherlands), mobilized peripheral blood (MPB) was obtained from leukapheresis material, and cord blood (CB) was collected according to the guidelines of NetCord FACT (by the Sanquin Cord Blood bank, The Netherlands). CD34 + cells were selected as described previously [25]. Unless specified otherwise, HSPCs in experiments were CB derived. BM-derived MSCs were isolated and cultured as described previously [26]. L88.5 stromal cells [27]

Gene and protein detection
Quantitative reverse transcriptase PCR (qRT-PCR), western blot assays, and immunofluorescence imaging were performed as described in Supplementary Data.
For long-term culture-colony forming cell (LTC-CFC) and cobblestone area-forming cell (CAFC) assays, CD34 + cells were plated on confluent monolayers of BM-MSCs, in MyeloCult medium (H5100; StemCell Technologies), supplemented with fresh hydrocortisone (1 mM). Colonies were analyzed according to manufacturer's guidelines. Detailed methods are described in Supplementary Data.

TGFBI mRNA is expressed in human hematopoietic cells at various levels
We first assessed steady-state TGFBI mRNA expression in the various cell types of BM tissue (Supplementary Table S1; Supplementary Data are available online at www.liebertpub .com/scd). Compared to BM-derived CD34 + HSPCs, TGFBI mRNA was highly expressed in human primary MSCs (almost 100-fold higher) and endothelial cells (almost five-fold higher; Fig. 1A). Concerning the differentiated hematopoietic lineages, the TGFBI expression in human primary monocytes and NK-cells was highly increased (>20-fold) compared to BM HSPCs and moderately increased (twofold to threefold) in human primary B cells and granulocytes, but decreased in HSPC-derived megakaryocytes (Fig. 1A). The enhanced TGFBI mRNA expression in MSCs compared to HSPCs is in line with transcriptome analyses published by others [30].
( Fig. 2A). Lysates containing equal cell numbers were loaded and a Histone-H3 antibody was used to confirm equal loading. In line with the mRNA expression levels, HSPCs from different origins express much lower levels of TGFBI compared to stromal cells such as primary MSCs and L88-5 cells. Physiological secretion of TGFBI decreases its protein levels detected in cell lysates and therefore impedes the comparison between mRNA and protein levels for the different cell types. TGFBI is detected on western blot at 70 kDa; however, double bands from proteolytic variants are often detected around 65 kDa (truncated C-terminus) and 60 kDa (truncated N-terminus). To examine TGFBI secretion by MSCs, we analyzed the localization of TGFBI in MSCs by widefield microscopy. TGFBI protein was abundantly present in the extracellular space, even on noncoated glass surfaces, indicating that TGFBI is indeed secreted and deposited to serve as ECM protein (Fig. 2B). Interestingly, its secretion is most prominently observed at the regions that resemble focal adhe-sions. Intracellular TGFBI in MSCs is prevalent in vesicular structures, which co-localize with golgin-97, a staining for the Golgi apparatus (Fig. 2C).

TGFBI modifies proliferation of HSPCs
To investigate the effect of TGFBI expression on HSPC expansion, TGFBI was upregulated or downregulated in HSPCs that were subsequently cultured for 10 days in maintenance cocktail. The efficiency of knockdown (Supplementary Fig. S2A, B) and overexpression ( Supplementary  Fig. S2C, D) was verified in HSPCs on mRNA level and protein level. Downregulation of TGFBI in HSPCs reduced the expansion in the maintenance culture by 50% (Fig. 3A), while the number of AnnexinV-positive cells was not increased upon TGFBI knockdown in HSPCs (Fig. 3C), suggesting that reduced expansion was not due to decreased survival. Overexpression of TGFBI did not significantly affect the expansion (Fig. 3B) or the number of AnnexinV-positive cells (Fig. 3D)

FIG. 2. TGFBI expression in BM stromal cells. (A)
TGFBI protein expression in MPB-and CB-derived HSPCs, primary MSCs, and L88-5 stromal cells was analyzed by western blot. Lysates containing equal cell numbers were loaded and a Histone-H3 antibody (17 kDa) was used as a loading control. TGFBI is detected at 70 kDa. (B) Endogenous localization of TGFBI (red), co-stained for Factin (blue) in human primary MSCs after 24 h of adhesion to glass. TGFBI is deposited in the extracellular matrix and shows a footprint-like signature when MSCs are migrating. Scalebar indicates 20 mm. (C) Endogenous localization of TGFBI (red), co-stained for Glogin-97 (green) and F-actin (blue) in human primary MSCs. Zoomed regions show colocalization (yellow) of TGFBI with the golgi-associated protein golgin A1 (golgin-97). Scale bar indicates 50 mm.
in the HSPC maintenance culture. TGFBI upregulation or downregulation in HSPCs did not induce changes in the phenotypic composition of these cell cultures ( Supplementary  Fig. S2E, G) or change the percentage of immature HSPCs defined as CD34 + CD38 - (Supplementary Fig. S2F, H) when cultured in a maintenance cocktail.

TGFBI modifies differentiation of HSPCs and proliferation of lineage-committed cells
To investigate the effect of TGFBI expression on HSPC lineage commitment and differentiation, CD34 + -selected HSPCs with upregulated TGFBI were cultured in the presence of different cytokine cocktails (See Supplementary Methods for detailed culture conditions) for 14 days and subsequently analyzed for proliferation and specific cell surface markers.
Monocytic differentiation is characterized by increased surface CD14 expression, which was not affected by TGFBI overexpression (data not shown). At day 14 of the monocytic differentiation culture, HSPCs overexpressing TGFBI were decreased to 60% -12% compared to control cells (n = 3, P < 0.05, Fig. 4G).
Thus, the effects of increased TGFBI expression depended on the hematopoietic lineage. High levels induced megakaryocytic differentiation and inhibited granulocytic proliferation and differentiation. In the erythroid and monocytic lineage, differentiation was unaffected by TGFBI, but proliferation of the culture was reduced.

Tight regulation of TGFBI expression in HSPCs is important for colony-forming capacity
To examine how TGFBI expression affects the colonyforming capacity of HSPCs, TGFBI was upregulated or downregulated in CD34 + HSPCs. Sorted GFP + cells were seeded in MethoCult to assess CFU-GM and BFU-E/CFU-GEMM colonies or in Megacult to assess megakaryocytic colonies. Compared to empty vector controls, TGFBI overexpression in CB-derived HSPCs reduced CFU-GM colony formation by 22% (n = 4, P < 0.01, Fig. 5A) and megakaryocytic colonies by 51% (n = 3, P < 0.05). No statistically significant decrease in BFU-E/CFU-GEMM colonies was detected, although this result is not clear due to a large variation in colony numbers in the different experiments (n = 4, P = 0.08).
The small reduction in CFU upon TGFBI overexpression and the more pronounced CFU reduction upon TGFBI knockdown are in line with the data concerning total proliferation as shown in Fig. 3A and B. Although, the cellular  processes perturbed by increased or decreased TGFBI expression may be different, these data imply that tight regulation of TGFBI expression is essential during hematopoiesis.

TGFBI expression in HSPCs reduces the number of immature hematopoietic cells in co-cultures
To investigate whether TGFBI affects maintenance and differentiation of HSPCs in a BM niche environment, we cocultured HSPCs and primary MSCs. The CAFC assay allows for the quantification of immature HSPCs in a limiting dilution setup. The differentiation and proliferation potential of input HSPCs are measured retrospectively and in the constant presence of MSCs. Results are based on visual endpoints eliminating the effects of medium changes or trypsinization. At week 2 and 4, cobblestone areas represent progenitors and mature hematopoietic cells, whereas cobblestone areas at week 6 represent the more immature HSPCs [2]. TGFBI was either upregulated or downregulated in HSPCs and GFPexpressing cells were seeded in CAFC assays. TGFBI overexpression in HSPCs did not affect CAFC formation in the first 4 weeks, but at week 6, CAFC frequencies were reduced compared to mock transduced HSPCs (Fig. 6A).
Reduced TGFBI expression did not induce a statistically significant change in cobblestone areas at week 2. However, at week 4 and 6, the CAFC frequencies were increased in all individual experiments with reduced TGFBI expression. A significant increase was observed with sh4 or sh5 (n ‡ 3, P < 0.05 at week 4, P < 0.01 at week 6), and a similar trend with sh2 (n = 3, P = 0.19 at week 4 and P = 0.12 at week 6, Fig. 6B). Because the efficiency of TGFBI knockdown was comparable for the different shRNA constructs and similar responses were observed, data were pooled as indicated. Thus, the number of week-6 CAFCs, which is a measure of the more immature hematopoietic cells, was inversely correlated with TGFBI expression. The data suggest that HSPC maintenance was suppressed by increased expression of TGFBI and enhanced by reduced TGFBI expression.

TGFBI expression in HSPCs decreases hematopoietic progenitor frequencies in co-cultures
Next we investigated how TGFBI expression affected progenitor production in LTC-CFC assays. Results of these assays are determined by both HSPC maintenance during culture and colony-forming capacity of progenitors, described in the previous paragraphs. Transduced CD34 + cells were co-cultured with primary MSCs and nonadherent (NA) cells derived from the supernatant were seeded in methylcellulose medium after 2 and 4 weeks of co-culture to asses CFCs. After 6 weeks, the complete culture was harvested and both NA-and stroma-adherent (SA) cell fractions were seeded in methylcellulose to determine CFC.
The results of the LTC-CFC assay corroborated the results of the CAFC assay. An increased TGFBI expression in HSPCs reduced expansion of early progenitors, whereas a decreased TGFBI expression in HSPCs increased late CAFC numbers and maintained the production of SA progenitors.

Reduced TGFBI in MSCs alters HSPC differentiation and increases the migration of HSPCs under the stromal layer
TGFBI expressed and secreted by MSCs might affect hematopoiesis in co-culture experiments. To investigate this, we downregulated TGFBI in MSCs with two different shRNA constructs (Fig. 7A) and co-cultured these cells together with HSPCs. Efficient TGFBI knockdown in primary MSCs did not affect proliferation (Fig. 7B) or cell morphology (Fig. 7C). Multilineage differentiation capacity of MSCs toward adipocytes or osteoblasts was also not affected by TGFBI knockdown (Supplementary Fig. S3A-D).
In co-culture experiments, a reduction of TGFBI in MSCs by sh2 resulted in a significantly decreased output of NA hematopoietic cells, while a similar trend was observed with sh3 (Fig. 7D). To examine whether reduced TGFBI levels in co-cultures had an effect on the colony-forming capacity of hematopoietic progenitors, the NA cells were seeded in methylcellulose medium after 1 week of co-culture. When CFU-GM colonies were counted after 12-14 days of culture, no differences in colony number was detected (Fig. 7E).
To investigate whether TGFBI expression by MSCs can direct lineage commitment of HSPCs, different surface markers for the myeloid lineage were assessed by flow cytometry (Fig. 7F). Both CD14 and CD36 were significantly upregulated after 1 week of co-culture with MSCs expressing reduced levels of TGFBI, indicating that monocyte and macrophage maturation are enhanced. In addition to cocultures of TGFBI knockdown MSCs with HSPCs, we also performed co-cultures with TGFBI knockdown L88-5 stromal cells to assess the number of mature and more immature hematopoietic cells. However, these co-cultures did not affect the CAFC capacity of HSPCs ( Supplementary Fig. S4A-C).
Since TGFBI is part of the ECM, and is known to decrease adhesion in HSPCs [16], we investigated whether HSPC migration patterns are also altered in our co-culture assays. To evaluate HSPC migration, co-cultures were live imaged through differential interference contrast microscopy and the number of HSPCs migrating under the mesenchymal stromal layer was quantified after 1 week of co-culture ( Fig. 7G and Supplementary Movie M1). In these experiments, reduced levels of TGFBI in MSCs resulted in an increased adhesion of HSPCs under the MSC stromal layer. These results corroborate with the HSPC count of the NA fraction (Fig. 7D) and emphasize the importance of TGFBI regulation for the MSC-HSPC interaction [16].

Discussion
Environmental factors that control maintenance and differentiation of HSC include the ECM protein TGFBI. In this study, we investigated the role of TGFBI in stromal and hematopoietic cells, in particular, its role in HSPC maintenance and differentiation. We demonstrate that TGFBI expression maintains HSPC immaturity, as shown by a decrease in late-cobblestone area formation (late-CAFC) in TGFBI-overexpressing HSPCs and an increase in CAFC in TGFBI knockdown HSPCs. Enhanced TGFBI expression impaired the colony-forming capacity of HSPCs in the absence of stromal support, and in the presence of MSCs, late-CAFC, a measure for HSPC immatu-rity, was reduced. TGFBI knockdown in HSPCs enhanced late-CAFC and adherent LTC-CFC, indicating HSC maintenance. Furthermore, TGFBI enhanced megakaryocytic differentiation, while it reduced expansion during monocytic, granulocytic, and erythrocytic differentiation.
Knockdown of TGFBI in MSCs has no notable effect on culture-expanded MSCs themselves; however, their hematopoietic support alters, resulting in an increased monocyte/macrophage outgrowth and an increased migration of HSPCs under the stromal cell layer. We conclude that, tight regulation of TGFBI expression in both HSPCs and stromal cells is essential for normal proliferation, differentiation, and homeostasis of HSPCs.

TGFBI controls HSPCs in the context of supporting MSCs
In the hematopoietic BM niche, TGFb and TGFBI contribute to microniches that regulate hematopoiesis [35]. Compared to HSPCs, MSCs express and secrete more TGFBI [8,24,36,37], but modulation of TGFBI expression in HSPCs themselves controls their maintenance, even in cocultures with MSCs.
Several observations support the conclusion that TGFBI mediates its effect by both cell intrinsic and paracrine mechanisms. Our data clearly show an intrinsic effect of TGFBI through its knockdown and overexpression in HSPCs. Colony-forming capacity of HSPCs in semisolid colony assays is decreased by both, knockdown and overexpression of TGFBI, indicating that the intrinsic pathways depend on a tight balance of TGFBI expression. Moreover, even in the presence of MSCs that secrete high levels of TGFBI, modulation of the intrinsic TGFBI expression in HSPCs controls their functional outcome as shown by the CAFC and LTC-CFC assays, supporting the idea that TGFBI acts cell autonomously in HSPCs. The paracrine effect of TGFBI is observed in co-cultures; decreased TGFBI expression in MSCs results in decreased total hematopoietic proliferation and a relative increase in hematopoietic cells skewed toward monocyte macrophage differentiation.
The TGFBI paracrine effect might also be attributed to changes in HSPC adhesion in these co-cultures. We previously showed that HSPCs bind TGFBI through integrins on their cell membrane. TGFBI overexpression in HSPCs reduced their binding to fibronectin by competitive binding of integrins [9,16]. In this study, we showed that TGFBI can be detected both intracellular and on the cell surface of both HSPCs and MSCs. The interaction of HSPCs with MSCs is dynamic and crucial for HSC maintenance [2,38]. HSPCs that express low TGFBI levels may readily bind to the ECM produced by MSCs, whereas increased TGFBI production by HSPCs may block this interaction and reduce HSPC-MSC interactions. The same mechanism may explain the observed increase of HSPC-MSC interaction in co-cultures of MSCs expressing lower levels of TGFBI; adhesion molecules secreted by the MSCs at the substrate surface are more accessible for HSPC integrins in the absence of competitively secreted TGBFI from the MSC (Fig. 7G). Together, this implies that TGFBI modifies intracellular signals of HSPCs and alters the interaction with MSCs. Both mechanisms may contribute to the functional outcome of the HSPCs.

A B
C D G F E TGFBI CONTROLS HEMATOPOIESIS TGFBI during homeostatic and regenerative hematopoiesis in the BM niche TGFBI is one of many TGFb target genes in HSPCs. In vitro, TGFb is known to inhibit HSPC proliferation and promote HSPC quiescence [17,20,39]. In vivo studies show similar results in regenerative conditions [18,40,41], although in homeostatic conditions, HSPCs can remain quiescent, even at low levels of TGFb, indicating that the context of TGFb signaling in HSPCs determines its effect [35,40].
In our experiments, TGFBI downregulation in HSPCs increased the maintenance of immature HSPCs compared to control HSPCs in the context of stromal cells.
This opposite effect compared to TGFb signaling might be because upregulation of TGFBI in HSPCs negatively regulates the other signaling pathways of TGFb that initiate HSPC quiescence, explaining the differences in CFU outgrowth in the presence or absence of stroma. TGFb is not the only factor inducing TGFBI expression. TGFBI is also upregulated in murine HSPCs following HSPC mobilization by G-CSF [42], or after 5FU treatment [4]. In these regenerative conditions, active TGFb is initially reduced [40,43], possibly related to hematopoietic recovery, while it is upregulated at later time points, possibly to induce quiescence in HSPCs again [40]. Lineage-committed progenitors are less responsive to TGFb and expand during hematopoietic recovery, while HSCs should be maintained. We showed that high TGFBI levels decrease colony formation in vitro, which may represent increased HSPC differentiation in regenerative conditions.
TGFBI expression in HSPCs depends on the source of the cells. We detected the highest TGFBI expression in BMderived HSPCs, compared to other sources of HSPCs. TGFBI is upregulated in HSPCs upon binding to MSCs [25], and only BM-derived HSPCs are in close contact with MSCs. MPBderived HSPCs, on the other hand, are mobilized by G-CSF that increases TGFBI expression in BM-derived HSPCs [42]. Nevertheless, the G-CSF-mobilized HSPCs isolated from peripheral blood showed lower TGFBI mRNA expression compared to BM-derived HSPCs. The higher TGFBI expression in BM-derived HSPCs may also be correlated with the increased percentages of active cycling HSPCs in BM compared to MPB [25], suggesting that TGFBI expression is more pronounced in HSPC populations with a higher mitotic activity.

TGFBI regulates hematopoietic lineages differentially
We found that megakaryocytic differentiation was enhanced upon TGFBI overexpression, whereas granulocytic differentiation was inhibited. Thus, the effect of enhanced TGFBI expression depends on the hematopoietic lineage. Increased maturation of megakaryocytic cells upon enhanced TGFBI expression is also in line with the recently identified critical role of TGFBI in platelet activation and function [44]. Altogether our data suggest that tight regulation of human TGFBI levels in HSPCs and MSCs is essential for normal proliferation, differentiation, and homeostasis of HSPCs.