Mesenchymal stromal cells expressing a dominant‐negative high mobility group A1 transgene exhibit improved function during sepsis

Abstract High mobility group (HMG)A proteins are nonhistone chromatin proteins that bind to the minor groove of DNA, interact with transcriptional machinery, and facilitate DNA‐directed nuclear processes. HMGA1 has been shown to regulate genes involved with systemic inflammatory processes. We hypothesized that HMGA1 is important in the function of mesenchymal stromal cells (MSCs), which are known to modulate inflammatory responses due to sepsis. To study this process, we harvested MSCs from transgenic (Tg) mice expressing a dominant‐negative (dn) form of HMGA1 in mesenchymal cells. MSCs harvested from Tg mice contained the dnHMGA1 transgene, and transgene expression did not change endogenous HMGA1 levels. Immunophenotyping of the cells, along with trilineage differentiation revealed no striking differences between Tg and wild‐type (WT) MSCs. However, Tg MSCs growth was decreased compared with WT MSCs, although Tg MSCs were more resistant to oxidative stress‐induced death and expressed less IL‐6. Tg MSCs administered after the onset of Escherichia coli‐induced sepsis maintained their ability to improve survival when given in a single dose, in contrast with WT MSCs. This survival benefit of Tg MSCs was associated with less tissue cell death, and also a reduction in tissue neutrophil infiltration and expression of neutrophil chemokines. Finally, Tg MSCs promoted bacterial clearance and enhanced neutrophil phagocytosis, in part through their increased expression of stromal cell‐derived factor‐1 compared with WT MSCs. Taken together, these data demonstrate that expression of dnHMGA1 in MSCs provides a functional advantage of the cells when administered during bacterial sepsis.


INTRODUCTION
High mobility group (HMG)A proteins are architectural factors that modify the structure of chromatin in a dynamic fashion. 1 These nonhistone chromatin proteins facilitate DNA-directed nuclear processes, Abbreviations: CLP, cecal ligation and puncture; Cox, cyclooxygenase; SDF-1, stromal cell-derived factor-1; ESCs, embryonic stem cells; dn, dominant-negative; HMG, high mobility group; MSCs, mesenchymal stromal cells; PIS, proinflammatory stimuli; qRT-PCR, quantitative real time polymerase chain reaction; RT, reverse transcriptase; Tg, transgenic; WT, wild-type. enhanceosome complexes, resulting in changes in gene transcription. 6 Through these interactions with DNA and other proteins, HMG proteins influence biologic processes such as cell growth, proliferation, differentiation, and death.
The expression of HMG proteins, such as HMGA1, is higher in undifferentiated and embryonic cells, including stem cells and cancer cells, and is linked to transcriptional networks that drive pluripotency and cellular proliferation. 1,[7][8][9][10] During cellular differentiation, and in adult terminally differentiated cells, the expression of HMGA1 is decreased. 10,11 Although HMGA1 is expressed at a lower level in adult, differentiated tissues, the expression of HMGA1 can increase under disease conditions. We have previously shown that vascular smooth muscle cells have low baseline levels of HMGA1. However, expression can be induced by exposure to bacterial endotoxin, the proinflammatory cytokine IL-1 , and to the proliferative stimulus of serum, both in vitro and in vivo. 12,13 To better understand the importance of HMGA1 in mesenchymal cells, we generated a transgenic (Tg) mouse overexpressing a dominant-negative (dn) form of HMGA1 targeted to mesenchymal cells of vascular smooth muscle origin. 14 This dn transgene is not capable of binding to DNA; however, it interacts with transcription factors. The dnHMGA1 Tg mice were less hypotensive to endotoxemia and had lower mortality in bacterial sepsis. In addition, the Tg mice had reduced tissue infiltration of inflammatory cells during experimental sepsis. 14 As mesenchymal stromal cells (MSCs) are another mesenchymal cell with a significant influence on the inflammatory response during sepsis, [15][16][17] we expanded our appraisal of HMGA1 from vascular smooth muscle cells to MSCs. We decided to study adipose-derived MSCs of the Tg mice, a cell population located in the stromal vascular fraction of adipocyte tissue. After determining that the dnHMGA1 transgene was expressed in these MSCs, the focus of the present study was to investigate whether disruption of the HMGA1 pathway would change the MSC phenotype in vitro, and have a functional impact on the cells when they were administered during bacterial sepsis in vivo.
Our hypothesis was that overexpression of the dnHMGA1 transgene in adipose-derived MSCs would improve their ability to mediate the inflammatory response, reduce tissue injury, and improve outcome when administered after the onset of bacterial sepsis.

Isolation, characterization, and differentiation assays of murine MSCs
MSCs were harvested from inguinal fat pads of wild-type (WT) and dnHMGA1 Tg (specifically Tg1 14 ) littermate mice. The Tg mice were generated using a dnHMGA1 construct containing 4 proline to alanine substitutions introduced at amino acids 57, 61, 83, and 87 located in the second and third DNA-binding domains (AT-hooks), as described previously, 18 which prevents the dn construct from binding to DNA. 14 The inguinal fat pads were washed and minced, digested enzymatically with 0.1% collagenase I and 0.25% collagenase II, and the cells were fil-tered via a 70 m cell strainer and centrifuged at 300 g for 10 min at room temperature as described. 19 The supernatant was removed, and the cells were plated for expansion using MesenCult Proliferation Kit (StemCell Technologies, Vancouver, BC).

Chemokine/cytokine assay
For IL-6 analysis of cell and tissue extracts, samples were assessed using the multiplex assay technology by Luminex as described previously. 22 as described. 20 The effectiveness of SDF-1 silencing was assessed at the protein level by ELISA (see above).

Reagents
Murine recombinant IL-1 , IFN-, and TNF-were purchased from PeproTech, Inc (Rocky Hill, NJ, USA). The cytokines were administered together (each at a dose of 10 ng/ml) to produce a proinflammatory stimulus to MSCs. Escherichia coli LPS (serotype O26:B6) was purchased from Sigma-Aldrich, and administered to MSCs at a dose of 100 ng/ml. were monitored over a 7-day period to determine survival.

Models of peritoneal sepsis in mice
Polymicrobial sepsis was induced by cecal ligation and puncture (CLP), as described. 17,25,26 Briefly, C57BL/6 male mice 7-9 weeks of age were anesthetized, the peritoneum was opened, and two-thirds of the cecum was ligated and punctured with 2 21-gauge holes. In sham experiments, the same procedure was performed; however, CLP was not performed. The mice received either WT or dnHMGA1 Tg MSCs

Histology and inflammatory cell infiltration
Mice were sacrificed 24 h following fibrin clot placement or CLP, and the spleens and lungs were harvested for histologic evaluation. The tissues were harvested as described 14 and immunostained with a Ly6G (BioLegend, Dedham, MA) antibody for assessment of neutrophil infiltration, and scored by an investigator who was blinded to the group. Positively stained cells were evaluated per 20× objective, and numerous random fields were assessed per tissue section using Image J software. Cells from the peritoneal fluid were stained with an antibody targeting Ly6G-APC (#127613; BioLegend), to identify neutrophils. The cells were then assessed by flow cytometry using a BD FACS Canto II, and analyzed by FlowJo software.

Assessment of cell death in tissue
Terminal deoxynucleotidetransferase-mediated dUTP nick endlabeling (TUNEL assay) Clontech (Mountain View, CA, USA) was used to detect apoptotic cell death in tissue sections of lungs and spleens from mice in each group per the manufacturer's protocol.
Quantification was performed as described for Ly6G staining.

Bacterial CFUs of peritoneal lavage
Mice underwent fibrin clot or sham surgery. The mice were then sacrificed 24 h after surgery, and peritoneal lavages were performed. 17 Serial dilutions of the peritoneal fluid were made, plated on LB agar plates, and then incubated overnight. CFUs were counted and calculated as described. 17

Isolation of murine neutrophils
For the isolation of neutrophils, mice were given an intraperitoneal injection of Bio-Gel P100 polyacrylamide beads (2% solution; Bio-Rad Laboratories) as described. 17 After 24 h, the mice were anesthetized and 10 ml of sterile PBS was used to lavage the peritoneal cavity, and cells were washed and placed in RPMI 1640 medium with 0.3% BSA and 10 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid.

Phagocytosis assay
MSCs were added at a ratio of 1 MSC to 5 neutrophils. Isolated neutrophils were activated with 10 ng/ml of G-CSF (mouse) for 2 h. Green fluorescent protein (GFP)-labeled E. coli (strain MMB1287) were then added at 10 multiplicity of infection per neutrophil. Bacterial phagocytosis was measured by flow cytometry as described. 17,20

Study approval
Studies using mice were carried out in accordance with the Public Health Service policy on the humane care and use of laboratory animals, and approved by the Institutional Animal Care and Use Committee (IACUC) of Brigham and Women's Hospital.

F I G U R E 1 Characterization of adipocyte-derived MSCs harvested from dominant-negative (DN) HMGA1 transgenic (Tg) mice. (A)
Total RNA was extracted from wild-type (WT) and dnHMGA1 Tg MSCs, and reverse transcriptase PCR was performed using 5 ′ and 3 ′ primers designed according to the sequences of the dnHMGA1 cDNA construct and the HA-tag designed to generate the mice. 14

Statistics
For comparisons between 2 groups, we used Student's unpaired ttest. For analysis of more than 2 groups, one-way ANOVA was used.
The assessment of cell growth between groups, over a period of 5 days, was analyzed by two-way ANOVA. When data were not normally distributed, nonparametric analyses were performed using Mann- then log-rank test to assess for differences in survival. Statistical significance was accepted at P < 0.05.

Characterization of dnHMGA1 Tg MSCs
MSCs were harvested from adipose tissue of WT and dnHMGA1 Tg mice. The dnHMGA1 transgene was present in the Tg cells, but not WT cells (Fig. 1A). Expression of the transgene did not alter endogenous mRNA levels of HMGA1, which was not different between WT MSCs and dnHMGA1 Tg MSCs (Fig. 1B). However, Cox-2 mRNA (known to be regulated by HMGA1 28

MSC cell growth and response to oxidative stress/inflammatory stimuli
The number of MSCs in the WT and Tg groups was counted at baseline in low serum medium, and then over the next 5 days in full growth medium. Cell growth, as depicted by cell number in each group of MSCs, demonstrated that Tg MSCs have significantly fewer cells on days 3, 4, and 5 compared with WT MSCs (Fig. 2A). We next exposed the cells to oxidative stress, an important component of the pathophysiologic injury response during sepsis, using hydrogen peroxide (H 2 O 2 ). 19,29 WT MSCs exposed to 125 M H 2 O 2 resulted in a significant decrease in viability, with 50.0 ± 2.6% of cells alive at 24 h (Fig. 2B). In contrast, Tg MSCs had more cell survival when exposed to H 2 O 2 , with 74.3 ± 8.2% viable cells. Finally, we assessed the production of IL-6, an inflammatory mediator important in the pathobiology of sepsis with higher levels associated with worse clinical outcome. 30,31 Tg MSCs produced less IL-6 protein after exposure to the proinflammatory stimuli (PIS: mIL-1 + mIFN-+ mTNF-) than WT MSCs (Fig. 2C). Taken together, these data demonstrated that even though the growth response of dnHMGA1 Tg MSCs is reduced compared with WT MSCs, the dnHMGA1 TG cells are more resistant to oxidative stress-induced death and they expressed less IL-6 than WT cells. In these studies, we focused on the lung, an organ with inflammation and injury that contributes significantly to the pathobiology of sepsis.

Administration of dnHMGA1 Tg MSCs after the onset of E. coli-induced sepsis exhibits improved cell efficacy with increased survival and decreased tissue injury
Compared with the marked increase in IL-6 mRNA in the lungs of septic mice receiving PBS, the level of IL-6 was significantly decreased in mice receiving Tg MSCs, and this level was also decreased compared with mice receiving WT MSCs (Fig. 6A). Similar to the decreased infiltration of Ly6G + cells, the expression of MIP-2 and KC was decreased in the lungs of mice with E. coli-induced sepsis receiving Tg MSCs compared with mice receiving PBS ( Fig. 6B and C).

dnHMGA1 Tg MSCs increase the phagocytic function of neutrophils, and demonstrate increased production of SDF-1
Eradication of the invading microorganism(s), along with subsequent resolution of the inflammatory response, leads to improved outcomes in sepsis. 20 Interestingly, the level of SDF-1 mRNA in the cells, and SDF-1 protein in the conditioned medium, was 3-fold higher in Tg MSCs compared with WT MSCs (Fig. 7B). Thus, we next silenced SDF-1 in Tg MSCs to determine whether the increased expression of SDF-1 contributes to improved neutrophil phagocytosis, and thus bacterial clearance. In both WT and Tg MSCs, silencing of SDF-1 (shSDF-1) led to a marked decrease in protein expression compared with the scrambled control (shSCR; Fig. 7C). WT shSCR MSCs significantly increased the phagocytosis of E. coli by neutrophils, compared with neutrophils not exposed to cells (Fig. 7D). However, when neutrophils were exposed to It has previously been shown that elevated levels of HMGA1 were associated with increased production of cellular reactive oxygen species, and decreased efficiency of mitochondrial DNA repair due to oxidative injury. 41 Interestingly, Tg MSCs were more resistant to death due to oxidative stress than WT MSCs and also produced less IL-6 in the presence of PIS. Thus, we administered WT and Tg MSCs to WT mice after the onset of E. coli-induced peritoneal sep- Quantitative real time PCR (qRT-PCR) was used to assess SDF-1 mRNA levels, whereas -actin was employed for normalization of gene expression. Gene expression is presented as fold-increase compared with WT MSCs. Data are presented as mean ± SEM, n = 3 per group. Analysis by unpaired t-test (2 tailed), with significant comparisons † versus WT MSCs (P < 0.0001). SDF-1 protein levels (pg/ml) were assessed by ELISA in conditioned medium of WT and Tg MSCs. Data are presented as mean ± SEM, n = 4 per group. Analysis by unpaired t-test (2 tailed), with significant comparisons † versus WT MSCs (P = 0.0278). (C) ELISAs for SDF-1 were performed on cell culture supernatants of WT MSCs (shSCR and shSDF-1) and Tg MSCs (shSCR and shSDF-1). Data are presented as mean ± SEM, n = 3 in shSCR and n = 4 in shSDF-1 groups. Analysis was performed by oneway ANOVA (P < 0.0001), with significant comparisons † versus WT MSC (shSCR), and § shSDF-1 versus shSCR groups. (D) Isolated neutrophils were incubated with GFP-labeled E. coli in the presence of WT or dnHMGA1 Tg MSCs, silenced for SDF-1 (shSDF-1) or scrambled control (shSCR). Data are presented as mean ± SEM, n = 3 per group. Analysis by one-way ANOVA (P < 0.0001), with significant comparisons * versus no MSCs, and † versus WT shSCR MSCs, and § shSDF-1 versus shSCR MSCs neutrophils into the peritoneal fluid. Thus, the MSCs limited nonspecific infiltration of neutrophils into tissues, but maintained neutrophil migration to the site of infection. Moreover, tissue production of IL-6, and also the neutrophil chemokines MIP-2 and KC were significantly decreased in mice receiving Tg MSCs. Although Tg MSCs expressed less IL-6 when exposed to PIS, it is likely that a decrease in the inflammatory response in tissues (such as the lung) explains the decreased local production of IL-6, which is associated with less tissue injury. To determine whether this effect on tissue injury and inflammation was also seen in another clinically relevant model of sepsis, we assessed lung tissue from mice after CLP surgery plus antibiotic therapy. MSCs were able to decrease cell death and neutrophil infiltration into the lungs, and Tg MSCs were able to promote a further decrease compared with mice receiving WT MSCs.
An important aspect of resolving the inflammatory response during sepsis is to eliminate the invading microorganism(s). In the mice receiving Tg MSCs, we found a decrease in the peritoneal bacterial counts compared with mice receiving WT MSCs or PBS at 24 h after the onset of E. coli-induced sepsis. We recently demonstrated that expression of SDF-1 in MSCs played a critical role in promoting neutrophil phagocytosis and bacterial clearance during sepsis. 24 Moreover, HMGA1 is known to bind to the SDF-1 promoter. 42 Interestingly, we found that the expression of SDF-1 was 3-fold higher in Tg MSCs (mRNA), and equally increased in the conditioned medium (protein) of Tg MSCs, compared with WT MSCs. We previously demonstrated that WT MSCs increase neutrophil phagocytosis, in part via paracrine actions. 17,20,24 This improved neutrophil response in the presence of WT MSCs (shSCR) was also apparent in the present study; however, Tg MSCs (shSCR) significantly enhanced the increase in bacterial phagocytosis by neutrophils. Moreover, silencing of SDF-1 (shSDF-1) in Tg MSCs eliminated the enhanced phagocytosis by neutrophils. These data confirm that increased levels of SDF-1 in dnHMGA1 Tg MSCs were critical for the improved phagocytic response by neutrophils, and thus bacterial clearance.
Previously, we reported the impact of expressing a dnHMGA1 transgene to improve outcomes in endotoxin exposure and bacterial sepsis, with a focus on vascular smooth muscle cells. 14