All procedures involving mice were reviewed and approved by Case Western Reserve University, Institutional Animal Care and Use Committee. Case Western Reserve University Animal Assurance #: A3145-01, with IACUC # 2011-0145. All experiments used the congenic B6.129S6-Cftr^tm2Uth (R117H/R117H mutation) and C57 BL/6 J controls (WT). Cftr^tm2Uth mice are a type IV Cftr mutant which predominantly affects the pulmonary response to infections. These animals were chosen to specifically investigate the potential proof of concept toward treating the pulmonary consequences of CF with cell based therapies. To study the therapeutic potential of hMSCs we used a sub-lethal model of airway infection and inflammation without the confounding contribution of gastrointestinal fragility. When chronically infected with Pseudomonas aeruginosa, the Cftr^tm2Uth mice demonstrated cachexia, weight loss and bronchoalveolar lavage (BAL) changes in cellular differential and cytokines without significant mortality. We have used 3 different groups of Cftr^tm2Uth or controls mice for each of 3 different hMSC preparations. In each experiment, we utilized 5 - 8 mice. In a smaller set of studies (n = 2, 5 - 7 animals/group) we had the availability to explore the use of hMSCs in Cftr^tm1Uth (DF508, B6.129S6-Cftr^tm1Kth) mice at day 3. The Cftr^tm1Kth mice (a type II mutation of Cftr) have all of the manifestations of the CF knockout mouse with gastrointestinal blockage and hypersensitivity to bacterial infection. The groups included both Cftr deficient and WT mice: infected without cell based therapy, infected with hMSCs or infected with WT BMM.

To generate a transient chronic infection, mice were infected with 5 × 10⁵ colony-forming units (CFU) Pseudomonas aeruginosa, strain PA-M5715 (a mucoidal clinical isolate) embedded in agarose beads and suspended in 20 μL of PBS [13,14]. All preparations of Pseudomonas aeruginosa impregnated beads were evaluated for relative colony forming units (CFUs) prior to inoculation into the mice. Cftr^tm2Uth, Cftr^tm1Kth and WT mice were anesthetized and then inoculated with bacteria into the trachea with a plastic catheter angled toward the right mainstream bronchus [13]. Brochoalveolar lavage (BAL) and whole lung homogenates were evaluated for CFUs at either day 3 or day 10 (post-hMSCs).

Mice were injected with ketamine (80 mg/kg) and xylazine (10 mg/kg) as previously described [15]. The thoracic cavity was opened and the lungs exposed followed by insertion of a cannula through the trachea into the bronchi and infusion of 1 × 1 ml of warm PBS containing 0.2% lidocaine to do the BAL. The BAL fluid sample was recovered by aspirating the liquid with a syringe for total cell count and cellular differential. BAL fluid was aliquoted and analyzed for cytokines involved in CF pathophysiology by Luminex multi-analyte technology and LL-37 by a commercial kit (Hycult Biotech, Plymouth Meeting, PA. Cat # HK321).

Mice were assessed daily for clinical score which was based upon coat quality, posture, ability to right themselves after being placed in lateral recumbence, ambulation and body weight (Table 1). At 3 days, 10 days, post-mortem or at sacrifice lungs were isolated and assessed for gross lung pathology in addition to quantitative bacteriology (Table 1). Both gross lung pathology and clinical scores were done by two different individuals who did not know the identity of the treatment groups.

Table 1. Key to clinical and gross lung pathology scores.

Concurrent studies were evaluated for lung pathology without BAL using hematoxylin and eosin to define inflammation [16].

hMSCs were obtained from bone marrow aspirates of healthy volunteers after written and verbal informed consent. All procedures were approved by Case Western Reserve: (Case # 12Z05)/University Hospitals # 09-90- 195, Institutional Review Board through the Skeletal Research Center in collaboration with Dr. Arnold Caplan). hMSCs from the bone marrow of healthy volunteers were isolated, cultured and immunophenotyped as described previously [17]. hMSCs were used during log-growth [18]. BMMs were generated as previously described [15]. Briefly, hematopoietic progenitors were obtained from the femurs of C57BL/6 J. Cells were grown in culture for 7 - 10 days in the presence of L929 spent medium. hMSCs and BMM were administered at 10⁶/100 μl PBS through the retro-orbital sinus consistent with previous published observations of using this route of administration [19,20].

C57BL/6 J or Cftr^tm2Uth BMM was generated as described above [15]. Once a monolayer was generated and the cells were differentiated, the BMM were grown in the presence and absence of 0.5 μg/ml lipopolysaccharide to induce inflammatory cytokines TNFα and IL-6 in vitro. Stimulated BMM were cultured in vitro with or without the addition of supernatants derived from cultured hMSCs. After 24 hours, cells were harvested and evaluated for TNFα or IL-6 gene expression based upon the changes in cytokine production in the mouse model post-hMSC administration [21]. The experiments were done with hMSC supernatants from 3-different donors and all measurements were done in triplicate.

CF epithelial cells have been demonstrated to be hyper-responsive to bacterial exposure resulting in elevated production of IL-8 and IL-6 [22]. Many cell lines are available to study the CF airway epithelial cell inflammatory response. The studies outlined in this manuscript utilized immortalized cell lines developed by transforming human airway tracheal epithelial cells from a CF patient with adenovirus [23]. The immortalized CF cell line is called (IB3-1 cells), the control cells are the same CF derived tracheal epithelial cells transfected with adenovirus with full-length functional CFTR. These control cells are designated S9 (corrected) cells [23]. These cells were kindly provided by the laboratory of Dr. Pamela Davis (Case Western Reserve University, Cleveland, Ohio). Cells were maintained in a 5% CO₂ incubator at 37˚C using LHC-8 media (Biosource, Camarillo, CA). All media contained penicillin/streptomycin and 10% fetal bovine serum. The experiments were done at least 3- times with 3-different donors of 48 hour-cultured hMSC supernatants. After 24 hours, cells were harvested and evaluated for IL-8 or IL-6 based upon the changes in cytokine production in the mouse model post-hMSC administration [22].

Pseudomonas aeruginosa (PA M5715, a clinical isolate) was streaked on Tryptic Soy agarose (TSA) plates then inoculated into flasks [14]. PA M5715 was plated at dilutions of 10⁴ to 10⁷ to define the appropriate dilutions. Growth curve analysis and viability was used to define the CFUs. PA M5715 dilutions (10⁴ to 10⁷) were mixed 1:1 with either PBS, un-stimulated hMSC culture medium (US) or LPS-stimulated hMSC culture medium (LPS, 0.5 μg/ml for 24 hours) for 30 minutes at room temperature followed by plating on TSA plates and incubation overnight at 37.5˚C. Colony counts were quantitatively assessed at 24 hours.

Total RNA is extracted by RNAeasy protocol (Qiagen, Valencia, CA). Expression of mRNA is determined by real time RT-PCR using the ABI Prism 7000 Detection System (Applied Biosystems Inc., ABI, Foster City, CA). RNA specimens were analyzed in duplicate and normalized to GAPDH. Primers (TNFα, IL-6, IL-8 for either mouse or human samples) were purchased from ABI and validated prior to studies.

BAL derived alveolar macrophages (AM) and peritoneal neutrophils (P) were evaluated for the ability to respond to MSC supernatants (n = 3 different donor derived supernatants, using transwells (0.4 μM, Corning, NY). Both alveolar macrophages and neutrophils were obtained as previously described [15,24]. MSC supernatants were put into the lower chamber and cells were put into the upper chamber as previously described [25]. After 4 hours the numbers of cells in the upper chamber and the lower chamber were counted to reflect the ability to respond to MSC supernatant.

Data were analyzed using quantitative and group comparisons with respect to measurements at individual time points. Data are described using means, standard deviations, and appropriate percentiles including medians and extreme values. Graphical representations show the data within different groups and at different time points. For group and time point comparisons, we used analysis of variance (ANOVA) or F values for variance between the means. For pair-wise comparisons non-parametric MannWhitney t-test or Wilcoxon signed-rank test were performed for samples with un-equal numbers or variance respectively, as indicated in the figure legend. Data of significance were established based upon an alpha value of 5% and below (« P ≤ 0.05). In some cases we designated alpha values which are not statistically significant as defined by our guidelines, however alpha values approaching 5% are indicative of how close to significance the data is, particularly when dealing with data that does not follow Gaussian distributions and is often found when dealing with in vivo studies. In this case we designate a specific P value with a dot (l P = 0.07) instead of a star that is used for significance.

Animals genetically modified to have altered expression of Cftr have been used extensively to study the pulmonary response to chronic infection with Pseudomonas aeruginosa. The benefits of the murine model are that it is a consistent, reproducible model of infection/ inflammation in the context of deficient Cftr and normal mucocilliary clearance [26]. Our CF animal CORE Center has done extensive studies on the kinetics of the Cftr deficient lung response to pathogen exposure using a variety of murine Cftr deficient models (14). The animals used in this study comprised a model that does not have severe gastrointestinal phenotype (Cftr^tm2Uth) to explore the potential therapeutic application of hMSCs in resolving the pulmonary manifestations associated with chronic airway infection and inflammation in CF [27]. Cftr^tm2Uth mice and control animals were inoculated with Pseudomonas aeruginosa-laden agar beads with and without retro-orbital administration of 10⁶ hMSCs or BMM and followed for either 3 or 10 days. Figure 1 shows the mean weight loss of Cftr^tm2Uth mice in response to Pseudomonas aeruginosa infection with and without treatment with 10⁶ hMSCs or BMM. The mice were followed for 10 days with daily weights and clinical scores. At day 3, the Cftr^tm2Uth animals given Pseudomonas aeruginosa lost significant weight when compared to C57Bl/6 mice given the same batch and dosing of Pseudomonas aeruginosa (Figure 1(a), n = 3 different experi-

ments). Animals given 10⁶ hMSCs or BMM retro-orbitally initially lost weight but by 3-day began to resolve the cachexia (Figure 1(a), n = 3 different hMSC or BMM preparations, P ≤ 0.05). By day 10, the Cftr^tm2Uth mice treated with hMSCs had weights comparable to WT mice infected at the same time (Figure 1(b), n = 3 different hMSC preparations, P ≤ 0.05). Treatment with BMM caused a statistical increase (P ≤ 0.05) in weight gain over the WT controls and the Cftr^tm2Uth at day 10 consistent with the important role of myeloid cells in the CF lung response to infection (Bonfield, T.L., J. Leuk. Biol., 92: 1111-1122). Treatment of Cftr^tm1Kth mice (delF508) with hMSCs also statistically decreased the amount of weight loss relative to the untreated infected control (Table 2) showing the consistency of the hMSC effect.

Weight loss is often used as a parameter in conjunction with clinical score as defined in Table 1. When the animals were assessed for clinical score, Cftr^tm2Uth mice with infection but without hMSCs had a higher clinical score (P ≤ 0.05) than controls which was attenuated with hMSCs even by 3-days (the higher the value the sicker the animals, Figure 2(a), P = 0.07), suggesting a trend toward improvement. At day 10, there was a statistical difference between the WT and Cftr^tm2Uth mice in terms of clinical score (Figure 2(b), P ≤ 0.05) which was significantly decreased by hMSC or BMM therapy (P ≤ 0.05). The improvement in clinical score also reached significance in our Cftr^tm1Kth mice (Table 2). When animals were sacrificed and evaluated for gross lung pathology (Figure 3), Cftr^tm2Uth animals tended to have a greater focal lung consolidation and pathology scores compared to C57Bl/6 J controls (Figure 3, P = 0.07). Administration of the hMSCs statistically changed the pathology scores in the Cftr^tm2Uth to have levels comparable with C57BL/6 mice infected with Pseudomonas aeruginosa (P ≤ 0.05). Treatment of the Cftr^tm2Uth and WT mice with BMM had no statistical impact on lung pathology. Similar results were also found in the studies using Cftr^tm1Kth mice (Table 2).

In order to investigate how the hMSCs impact the mu-

Table 2. MSCs in Pseudomonas aeruginosa infected Cftr^tm1Kth (delF508) animals and controls.

rine model of CF lung infection and inflammation, animals were euthanized followed by BAL for differentials, and total cell counts. As has been published previously, Cftr^tm2Uth animals had higher numbers of BAL leukocytes than WT mice given the same inoculums (Figure 4, P ≤ 0.05). Cell based therapy, whether it was hMSCs or BMM resulted in a statistical decrease in the overall numbers of BAL leukocytes (P ≤ 0.05, for BMM and hMSCs) in the Cftr^tm2Uth mice which was not observed in the WT controls. The leukocyte recruitment in the Cftr^tm2Uth showed decreased relative numbers of alveolar macrophages (Figure 5(a), P ≤ 0.05) and increased numbers of neutrophils (Figure 5, P ≤ 0.05). Both hMSCs and BMM enhanced recruitment of alveolar macrophages (Figure 5(a), P ≤ 0.05) while attenuating the relative numbers of neutrophils (Figure 5(b), P ≤

0.05). We obtained similar observations in the Cftr^tm1Kth model (Table 2).

To determine if the hMSCs had the capacity to contribute to the shift in inflammatory cell recruitment away from neutrophils toward alveolar macrophages, chemotaxis studies were performed using alveolar macrophages and peritoneal neutrophils from WT and Cftr^tm2Uth mice (Figure 6). There was no difference between WT and Cftr^tm2Uth (data not shown) neutrophils or alveolar macrophages when cultured with medium alone. Using three different hMSC derived supernatants, there was a significant chemotactic effect of the supernatants on recruiting Cftr^tm2Uth alveolar macrophages but not on any other cell type (P ≤ 0.05), with a suggestive suppressive effect on neutrophil recruitment.

Cytokines are essential in the process of leukocyte recruitment and define the cell type and inflammatory response. We used Luminex multi-analyte technology to measure cytokines known to be involved in CF. KC, IL-6 and IL-1β were measured for chronic inflammation and neutrophil recruitment (Figure 7). MIP-2, resistin and adiponectin were measured due to their implications in regulating macrophage responses and inflammation [28- 30]. Not surprisingly, Cftr^tm2Uth mice had elevated levels of KC (Figure 7(A)), IL-6 (Figure 7(B)), IL-1β (Figure 7(C)). Both KC and IL-6 levels were attenuated by both hMSC and BMM cell based therapy (P ≤ 0.05). Only BMM attenuated the IL-1β levels (Figure 7(C), P ≤ 0.05). Cftr^tm2Uth mice had elevated BAL adiponectin levels (Figure (7D)) but not MIP-2 (Figure 7(E)) or resistin (Figure 7(F)) compared to infected WT controls. Only BMM attenuated MIP-2 and adiponectin concentrations (P ≤ 0.05). Resistin concentrations are shown for comparison of a non-response. We did not detect IL-10 in any of the BAL samples (data not shown). Comparative studies in the Cftr^tm1Kth mice are on-going and the focus of another manuscript.

To investigate the mechanisms behind the anti-inflammatory impact of the hMSCs and whether it is related to hMSC soluble products, we used two different in vitro assays of cytokine production: LPS stimulation of bone marrow derived macrophages and epithelial cells. In the first set of studies we obtained bone marrow cells from WT and Cftr^tm2Uth mice and differentiated them into

bone marrow derived macrophages (BMM) in vitro. The WT and Cftr^tm2Uth BMM preparations were stimulated with 0.5 μg/ml LPS for 24 hours to induce the production of pro-inflammatory cytokines TNFα and IL-6 mRNA expression. The LPS treated cultures were evaluated with or without the addition of hMSC supernatants to determine if hMSC soluble products could decrease the proinflammatory response to LPS in these in vitro cultures of BMM. We measured mouse IL-6, TNF-α mRNA expression and as predicted, LPS stimulated both IL-6 (Figure 8(a), P ≤ 0.05) and TNFα (Figure 8, P ≤ 0.05) with more being expressed by Cftr^tm2Uth BMM. hMSC supernatants decreased both IL-6 (Figure 8(a), P ≤ 0.05) and TNF-α (Figure 8(b), P = 0.07) mRNA expression.

Since primary epithelial cells are difficult to culture from CF and WT mice, we took advantage of established human airway epithelial cell models of CF and controls (corrected; see description in the methods). LPS stimulation of the CF and corrtcted cells showed elevated IL-8 (Figure 8(c)) and IL-6 (Figure 8(d)) gene expression, again with the CF cells expressing significantly more of both cytokines (P ≤ 0.05). hMSC supernatant suppressed both IL-8 and IL-6 mRNA expression (P ≤ 0.05). These data suggest that soluble products produced by hMSC contribute to decreased cytokine production in both macrophages and epithelial cells. The mechanisms of this effect are the focus of our on-going studies.

Recent literature has suggested that not only can hMSCs be anti-inflammatory, but they may also have

anti-bacterial potential [5,31]. We evaluated the colony forming units of Pseudomonas aeruginosa remaining in the lungs of the Cftr^tm2Uth at day 3 to test this function in our in vivo model. Whole lung homogenates were made of the animals and cultured overnight on TSA plates. Both WT and CF animals had significant and comparable bacterial loads at day 3 (Figure 9(a)). CF animals treated with hMSCs or BMM, had significantly less bacterial counts (Figure 9(a), P ≤ 0.05), than the animals infected at the same time but without cell based therapy. These observations were consistent in our Cftr^tm1Kth animal studies (Table 2).

To determine if this was a direct effect of the hMSCs, supernatants from hMSCs were cultured with Pseudo-

monas aeruginosa in vitro. The supernatants were derived from hMSCs with or without stimulation with LPS (0.5 μg/ml for 24 hours) to determine if the hMSCs would generate products with enhanced bactericidal activity in response to endotoxin. The Pseudomonas aeruginosa was cultured with the different hMSC supernatants and then plated out on TSA plates and allowed to grow overnight. Bacterial counts were evaluated and compared to controls of bacteria not treated or treated with PBS. The hMSC supernatant obtained from nonstimulated hMSCs (US) and LPS-stimulated hMSC (LPS) culture supernatant significantly decreased the ability of the bacteria to grow in vitro (Figure 9(b), P ≤ 0.05) over the PBS control. BMM supernatants also decreased bacterial load in vitro (data not shown), but is the focus of a separate manuscript.

To identify the potential agent associated with the anti-inflammatory and anti-bacterial properties of hMSC supernatants, we evaluated the BAL supernatants from both Cftr^tm2Uth and Cftr^tm1Kth (data not shown) and controls for the presence of LL-37, because it has been reported to be both anti-inflammatory and anti-microbial [5,31,32]. BAL fluid was obtained from mice chronically infected with Pseudomonas aeruginosa with and without hMSC or BMM therapy (Figure 10, n = 3 experiments with 4-6 BAL samples/group). All of the BAL fluid had detectable levels of LL-37, with the highest levels found in the Cftr deficient animals treated with hMSCs (P ≤ 0.05) using the F-test for analysis of variance, P = 0.07 for the Mann-Whitney t-test). The trend of increased levels of LL-37 in both Cftr deficient models supports the potential role of LL-37 in the effectiveness of hMSCs at de-

creasing bacterial load and inflammation. Although the BMM had anti-inflammatory and anti-microbial effects in vivo, they did not appear to associated with LL-37 levels, suggesting the involvement of other anti-microbial proteins besides LL-37 maybe important in the antimicrobial effects of BMM. This is the focus of on-going studies in our laboratory.

With differences in LL-37 concentrations in the in vivo model, we investigated the potential source of the LL-37. hMSCs cultured in vitro for 24 hours secreted LL-37 (Figure 11). LPS stimulation for 24 hours did not appear to significantly change the amount of secreted LL-37 relative to the un-stimulated control. Incubation of the hMSCs with the CFTR inhibitor I-172 (10 μg/ml for 48 hours), significantly reduced the ability of hMSCs to secrete LL-37 relative to the un-stimulated control (P ≤ 0.05). Further, when cells were stimulated with LPS after inhibiting CFTR activity (for 48 hours), the amount of LL-37 was even further suppressed relative to the LPS control (P ≤ 0.05). These data suggest that the hMSCs express functional CFTR and that blocking CFTR impacts the ability to produce LL-37 and the ability to respond to LPS. To determine if CF MSCs express CFTR to validate the I-172 studies, we took advantage of an immortal-mouse derived MSC clone BMC9 [33], especially since we have a highly reproducible mouse Cftr gene expression assay in our CF animal CORE center [26,34]. These cells are abundant and have a MSC

phenotype when grown at 37˚C. Our data showed that the BMC9 cells have 0.36% ± 0.14% Cftr expression (Ct value of 31 ± 1) compared to intestinal epithelium (Ct value around 20 ± 3), which expresses extremely high levels of Cftr. The sensitivity and specificity of our Cftr expression assay is 40 ± 2 Ct. If we had used a lower expressing tissue, the % of Cftr levels would be higher. BMC9 cells cultured with I-172 (10 μg/ml, for 48 hours) secreted 37% ± 13% (Mean ± SEM, n = 3) less LL-37 than BMC9 cells not cultured with I-172 (P ≤ 0.05). These data suggest that MSCs express CFTR and that CFTR function impacts the ability of MSCs to produce products such as LL-37. Our future studies are aimed at studying the differences between and Cftr deficient MSCs and control MSCs.